docs: point README at GitBook and restore the legal-notice heading [skip ci]

Product docs already live on crackproof-research; keep a minimal README (usage, legal notice, license) so CI can still extract LEGAL-NOTICE.md.
chore: bump version to 1.3.1
2026-09-19 03:57:59 -04:00 · 2026-09-15 23:27:24 +08:00 · 2026-09-10 20:58:44 +08:00 · 2026-09-10 20:42:39 +08:00 · 2026-09-10 09:32:58 +08:00 · 2026-09-10 09:29:23 +08:00
97 changed files with 16378 additions and 6204 deletions
@@ -19,7 +19,7 @@ jobs:
    runs-on: windows-latest
    steps:
      - uses: actions/checkout@v4
-      - run: cargo clippy --all-targets -- -D warnings
+      - run: cargo clippy --workspace --all-targets -- -D warnings

  test:
    # The test suite exercises Windows path semantics, so it runs on Windows.
@@ -28,7 +28,7 @@ jobs:
    runs-on: windows-latest
    steps:
      - uses: actions/checkout@v4
-      - run: cargo test --release
+      - run: cargo test --release --workspace

  check-portable:
    # Build-only portability gate: non-Windows host and the wasm target the
@@ -39,8 +39,8 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
-      - run: cargo clippy --all-targets -- -D warnings
-      - run: cargo check --target wasm32-unknown-unknown
+      - run: cargo clippy --workspace --all-targets -- -D warnings
+      - run: cargo check --workspace --target wasm32-unknown-unknown

  cli:
    strategy:
@@ -82,11 +82,12 @@ jobs:
    steps:
      - uses: actions/checkout@v4
      - run: cargo install wasm-pack --locked
-      # `-- --locked` forwards to cargo: web/Cargo.lock is committed on
+      # `-- --locked` forwards to cargo: senbei-wasm/Cargo.lock is committed on
      # purpose, so the wasm build must be pinned by it rather than silently
      # re-resolving (which is how it drifted out of sync with the manifest).
-      - run: wasm-pack build --target web --release -- --locked
-        working-directory: web
+      # --out-dir emits the JS/wasm package into the static frontend's web/pkg/.
+      - run: wasm-pack build --target web --release --out-dir ../web/pkg -- --locked
+        working-directory: senbei-wasm
      - uses: actions/upload-artifact@v4
        with:
          name: senbei-web
@@ -62,11 +62,12 @@ jobs:
    steps:
      - uses: actions/checkout@v4
      - run: cargo install wasm-pack --locked
-      # `-- --locked` forwards to cargo: web/Cargo.lock is committed on
+      # `-- --locked` forwards to cargo: senbei-wasm/Cargo.lock is committed on
      # purpose, so the wasm build must be pinned by it rather than silently
      # re-resolving (which is how it drifted out of sync with the manifest).
-      - run: wasm-pack build --target web --release -- --locked
-        working-directory: web
+      # --out-dir emits the JS/wasm package into the static frontend's web/pkg/.
+      - run: wasm-pack build --target web --release --out-dir ../web/pkg -- --locked
+        working-directory: senbei-wasm
      - name: Stage static site
        run: |
          mkdir dist
@@ -153,6 +153,8 @@ target/
 !/samples/README.md

 ### senbei web build ###
+# The wasm crate builds out-of-workspace in senbei-wasm/ and wasm-pack emits
+# the JS/wasm package into web/pkg/ for the static frontend.
+/senbei-wasm/target/
 /web/pkg/
-/web/target/
 /web/.playwright-cli
@@ -1,79 +1,40 @@
 # AGENTS.md

-Guidance for AI coding agents (and human contributors) working in this repo.
+Guidance for contributors working in this repository.

 ## Project

-Senbei is a static unpacker for Crackproof-protected PE files: a pure,
-panic-free, no-I/O unpacker core (`src/unpacker/`) plus a thin CLI shell
-(`src/`), an il2cpp metadata de-obfuscator (`src/metadata.rs`), and a
-WebAssembly browser frontend (`web/`). Read `docs/design.md` first.
+Senbei is a static unpacker for protected PE files and Android AArch64 shared libraries. The workspace contains `senbei-cli`, `senbei-crypto`, `senbei-elf`, `senbei-engine`, `senbei-io`, `senbei-metadata`, and `senbei-pe`; `senbei-wasm` is a separate crate for the browser frontend.
+
+Read the [Senbei design notes](https://xn--ri8h.gitbook.io/crackproof-research/senbei) before changing architecture or pipeline boundaries.

 ## Commands

 ```cmd
-cargo build --release            :: CLI
-cargo test --release             :: full suite (golden corpus: samples/, git-ignored)
-cargo clippy --all-targets -- -D warnings
+cargo build --release
+cargo test --release --workspace
+cargo clippy --workspace --all-targets -- -D warnings
 cargo fmt --all
-cd web && wasm-pack build --target web --release   :: browser build
+cd senbei-wasm && wasm-pack build --target web --release --out-dir ../web/pkg
 ```

-The `samples/` corpus is user-managed and absent on CI; without it the
-samples test is a no-op pass. `SENBEI_REQUIRE_SAMPLES=1` makes an absent
-corpus fail (use this on a private CI that *does* have the corpus). Do not
-delete `samples/` with `rm -rf` — it may be a junction; use git
-worktree-aware cleanup.
+The optional `samples/` corpus is user-managed and ignored by Git. The Android corpus is under `samples/android/` when present. Do not delete sample directories as part of routine cleanup.

-## Hard rules
+## Crate Boundaries

- **The unpacker core stays pure**: no file I/O, no `unsafe`, no panics across
-  the public boundary, no platform-specific code. It must keep compiling to
-  `wasm32-unknown-unknown` (`cargo check --target wasm32-unknown-unknown`).
- **`catch_unwind` does not work on wasm** (the prebuilt std can't unwind; a
-  caught panic becomes a fatal `unreachable` trap). Native code may rely on
-  `catch_unpack`, but any routing decision must also work without a catchable
-  panic: spliced companion inputs route straight to the EXE pipeline, and the
-  web app isolates every unpack in a disposable Web Worker, retrying trapped
-  DLLs with `job::unpack_bytes_force_exe`. Never make correctness on wasm
-  depend on catching a panic.
- **Byte-identical output is the contract.** Any pipeline change must re-run
-  the full golden corpus; a byte mismatch on any golden is a regression.
- **Trial-and-validate, never trust a heuristic.** A silently wrong offset
-  produces a silently broken binary — worse than an error. Every layout
-  candidate must be validated (checksum / structural oracle) with fall-through
-  to the next candidate.
- **Determinism under parallelism.** Block fan-out must stay byte-identical
-  regardless of thread count (`SENBEI_THREADS=1` is the sequential reference).
- **Folder scanning: deny-list, never allow-list.** Targets are recognised by
-  content, not extension, and can carry arbitrary names — there is no closed
-  set of target extensions an allow-list could enumerate. Only known
-  bulk-asset formats are excluded.
- **No binaries in the repo** — not as fixtures, not in commits. The only
-  corpus is the local git-ignored `samples/`. (Issue attachments of
-  protected inputs are fine when the user is authorized to share them, but
-  never commit them.)
+`senbei-pe` and `senbei-elf` contain format parsing, address mapping, and ELF dynamic-table helpers only. `senbei-engine/src/windows/` contains the PE unpacking pipeline; `senbei-engine/src/android/` contains Android extraction and ELF restoration. `senbei-crypto/src/windows/` and `senbei-crypto/src/android/` contain platform-specific primitives; seeded Android metadata code is under `senbei-metadata/src/android/`, while the structural metadata transform is shared at the metadata crate root. Shared source stays directly under `src/`.

-## Public-repo hygiene (important)
+The format crates and PE engine remain free of filesystem I/O. Native Android extraction and restoration may memory-map inputs and write temporary workspaces. The browser binding must continue to compile for `wasm32-unknown-unknown`.

-This is a public research repository. In code comments, docs, tests, and
-commit messages:
+## Hard Rules

- **Never name specific games, publishers, or product codenames.** Refer to
-  build families generically ("older EXE-64 builds", "the marker-less
-  layout", "external-companion builds"). Keep offsets/numbers — drop names.
- **Never name specific protected filenames** from real distributions. Test
-  fixtures use generic names (`app.exe`, `managed.dll`, `daemon.exe`).
-  Exceptions (platform-standard technology names, allowed): `il2cpp`,
-  `Unity`, `global-metadata.dat`, the Crackproof magic `KONN`.
- **Never reference other tools, projects, implementations, or paths outside
-  this repo.** Describe behavior and layout directly; do not mention prior
-  art, porting, or where any algorithm came from.
+- Outputs must be byte-identical to the available golden corpus.
+- Layout heuristics must trial and validate every candidate before accepting it.
+- Deterministic parallel and sequential paths must produce identical bytes.
+- Folder scanning must not open bulk assets. Windows candidates are `.exe`, `.dll`, and `global-metadata.dat`; Android candidates are `.so` and `global-metadata.dat`. Matching `.exe._` and `.dll._` files are auxiliary payloads and are not counted as skipped targets.
+- APK, APKS, and XAPK processing must inspect manifests first and extract only `.so` and `global-metadata.dat` entries.
+- Do not commit protected or restored binaries. Use generic fixture names and do not add product-specific names or external tool references to public code, docs, tests, or commit messages.

-## Conventions
+## Documentation

- Comments explain *why* (layout rationale, observed variants, failure modes),
-  not *what*.
- Rust 2024 edition; clippy-clean at `-D warnings`; rustfmt default style.
- CLI behavior (flags, exit codes, output naming) is documented in
-  `docs/usage.md` — update the doc when changing behavior.
+Use one line for each normal Markdown paragraph. Keep code blocks, table rows, and list items structurally separate. Product documentation lives at <https://xn--ri8h.gitbook.io/crackproof-research/senbei>; update that site (not this repository) when CLI behavior changes.
@@ -2,6 +2,17 @@
 # It is not intended for manual editing.
 version = 4

+[[package]]
+name = "aes"
+version = "0.9.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "35f0f96ce78e38c3dc6d8948aa8163d06385be74000f3c7a95bf1eef35d3ea32"
+dependencies = [
+ "cipher",
+ "cpubits",
+ "cpufeatures",
+]
+
 [[package]]
 name = "anyhow"
 version = "1.0.104"
@@ -14,6 +25,15 @@ version = "2.13.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b588b76d00fde79687d7646a9b5bdf3cc0f655e0bbd080335a95d7e96f3587da"

+[[package]]
+name = "block-buffer"
+version = "0.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d2f6c7dbe95a6ed67ad9f18e57daf93a2f034c524b99fd2b76d18fdfeb6660aa"
+dependencies = [
+ "hybrid-array",
+]
+
 [[package]]
 name = "bumpalo"
 version = "3.20.3"
@@ -26,6 +46,16 @@ version = "1.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"

+[[package]]
+name = "cipher"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e8cf2a2c93cd704877c0858356ed03480ff301ee950b43f1cbe4573b088bfa6c"
+dependencies = [
+ "crypto-common",
+ "inout",
+]
+
 [[package]]
 name = "console"
 version = "0.16.4"
@@ -38,12 +68,68 @@ dependencies = [
 "windows-sys",
 ]

+[[package]]
+name = "const-oid"
+version = "0.10.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a6ef517f0926dd24a1582492c791b6a4818a4d94e789a334894aa15b0d12f55c"
+
+[[package]]
+name = "cpubits"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "15b85f9c39137c3a891689859392b1bd49812121d0d61c9caf00d46ed5ce06ae"
+
+[[package]]
+name = "cpufeatures"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5ca28b0ae3115b884660db4118d803791fd6756b6e88f39c0f3f7859060d7566"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "crc32fast"
+version = "1.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8498c871161e1742aaa9d52551b2d6ebdd4c3d45a3be423e3728f33b955be550"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "crypto-common"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ce6e4c961d6cd6c9a86db418387425e8bdeaf05b3c8bc1411e6dca4c252f1453"
+dependencies = [
+ "hybrid-array",
+]
+
+[[package]]
+name = "digest"
+version = "0.11.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f1dd6dbb5841937940781866fa1281a1ff7bd3bf827091440879f9994983d5c2"
+dependencies = [
+ "block-buffer",
+ "const-oid",
+ "crypto-common",
+]
+
 [[package]]
 name = "encode_unicode"
 version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "34aa73646ffb006b8f5147f3dc182bd4bcb190227ce861fc4a4844bf8e3cb2c0"

+[[package]]
+name = "equivalent"
+version = "1.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f"
+
 [[package]]
 name = "errno"
 version = "0.3.14"
@@ -61,22 +147,31 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "da7c62ceae207dd37ea5b845da6a0696c799f85e97da1ab5b7910be3c1c80223"

 [[package]]
-name = "futures-core"
-version = "0.3.33"
+name = "flate2"
+version = "1.1.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2cd50c473c80f6d7c3670a752354b8e569b1a7cbfdc0419ec88e5edad85e0dc7"
+checksum = "6e634e2e0ebac1ee034020da1ca582e17ffe4e0f5e985823721e168928136dcb"
+dependencies = [
+ "zlib-rs",
+]
+
+[[package]]
+name = "futures-core"
+version = "0.3.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "92d699e522242e69e3003b94ecc1f960f3a5e015aa7c5d7486e65ad01dd94f5e"

 [[package]]
 name = "futures-task"
-version = "0.3.33"
+version = "0.3.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b231ed28831efb4a61a08580c4bc233ec56bc009f4cd8f52da2c3cb97df0c109"
+checksum = "cd417de3d1d015fc3bfd2b1ea46dfc7bab72ef86f1cc7cc9c78e728b34a6d1fd"

 [[package]]
 name = "futures-util"
-version = "0.3.33"
+version = "0.3.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a77a90a256fce34da66415271e30f94ee91c57b04b8a2c042d9cf3220179deaa"
+checksum = "0d50a92467f8ba5dd6e3ee5d4bd04d73ab2e4e1c44474a0674821dfce14b79bc"
 dependencies = [
 "futures-core",
 "futures-task",
@@ -95,6 +190,42 @@ dependencies = [
 "r-efi",
 ]

+[[package]]
+name = "goblin"
+version = "0.10.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "17582616a7718cca54cec18e534a76c7c4aec11a8b9a85695712f262fd15a4c8"
+dependencies = [
+ "log",
+ "plain",
+ "scroll",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.17.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ed5909b6e89a2db4456e54cd5f673791d7eca6732202bbf2a9cc504fe2f9b84a"
+
+[[package]]
+name = "hybrid-array"
+version = "0.4.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "707114b52a152fa7bdb290cd7cd5912d9467273b6d74e21b8d81aca1f8533f6b"
+dependencies = [
+ "typenum",
+]
+
+[[package]]
+name = "indexmap"
+version = "2.14.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "07aa2048142242915a31d35844fb311e0e53fcca590c3a0a40dcf1b841fa09eb"
+dependencies = [
+ "equivalent",
+ "hashbrown",
+]
+
 [[package]]
 name = "indicatif"
 version = "0.18.6"
@@ -109,10 +240,25 @@ dependencies = [
 ]

 [[package]]
-name = "js-sys"
-version = "0.3.103"
+name = "inout"
+version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53b44bfcdb3f8d5837a46dae1ca9660a837176eee74a28b229bc626816589102"
+checksum = "4250ce6452e92010fdf7268ccc5d14faa80bb12fc741938534c58f16804e03c7"
+dependencies = [
+ "hybrid-array",
+]
+
+[[package]]
+name = "itoa"
+version = "1.0.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8f42a60cbdf9a97f5d2305f08a87dc4e09308d1276d28c869c684d7777685682"
+
+[[package]]
+name = "js-sys"
+version = "0.3.104"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0e0c1080212aad755ea003d18543e8768dd432c48819efd73a7bf1e39b7a5a3a"
 dependencies = [
 "cfg-if",
 "futures-util",
@@ -131,6 +277,27 @@ version = "0.12.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "32a66949e030da00e8c7d4434b251670a91556f4144941d37452769c25d58a53"

+[[package]]
+name = "log"
+version = "0.4.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f9f8bd3e56ce4dfc153cf470fffbfa98c7620958b312ca5c3a4b8d5181fd13c6"
+
+[[package]]
+name = "memchr"
+version = "2.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf8baf1c55e62ffcace7a9f06f4bd9cd3f0c4beb022d3b367256b91b87513d98"
+
+[[package]]
+name = "memmap2"
+version = "0.9.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d1219ed1b7f229ee7104d281dd01d6802fe28bb6e95d292942c4daacdeb798c0"
+dependencies = [
+ "libc",
+]
+
 [[package]]
 name = "once_cell"
 version = "1.21.4"
@@ -139,9 +306,9 @@ checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"

 [[package]]
 name = "owo-colors"
-version = "4.3.0"
+version = "4.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d211803b9b6b570f68772237e415a029d5a50c65d382910b879fb19d3271f94d"
+checksum = "13c45bb4a6ae1280ec0803b1ef9d3455eb50f01efbbe1447ab020f1d54fba9d8"

 [[package]]
 name = "pin-project-lite"
@@ -150,10 +317,16 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"

 [[package]]
-name = "portable-atomic"
-version = "1.14.0"
+name = "plain"
+version = "0.2.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d20d5497ef88037a52ff98267d066e7f11fcc5e99bbfbd58a42336193aacec3"
+checksum = "b4596b6d070b27117e987119b4dac604f3c58cfb0b191112e24771b2faeac1a6"
+
+[[package]]
+name = "portable-atomic"
+version = "1.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "05c8b63e8d9609db387f0324918f81d68fe27748f084ef092fb35954d0539a85"

 [[package]]
 name = "proc-macro2"
@@ -208,19 +381,163 @@ dependencies = [
 ]

 [[package]]
-name = "senbei"
-version = "1.0.1"
+name = "scroll"
+version = "0.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c1257cd4248b4132760d6524d6dda4e053bc648c9070b960929bf50cfb1e7add"
+dependencies = [
+ "scroll_derive",
+]
+
+[[package]]
+name = "scroll_derive"
+version = "0.13.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e1a36a382ed65dbcc0ab47fd5e9a94112417ccd34560a392ef3b7b0f0ec39148"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 3.0.4",
+]
+
+[[package]]
+name = "senbei-cli"
+version = "1.3.1"
+dependencies = [
+ "senbei-engine",
+ "senbei-io",
+ "senbei-metadata",
+ "sha2",
+ "tempfile",
+]
+
+[[package]]
+name = "senbei-crypto"
+version = "1.3.1"
+dependencies = [
+ "aes",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-elf"
+version = "1.3.1"
+dependencies = [
+ "goblin",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-engine"
+version = "1.3.1"
+dependencies = [
+ "memmap2",
+ "senbei-crypto",
+ "senbei-elf",
+ "senbei-pe",
+ "serde",
+ "serde_json",
+ "sha2",
+ "tempfile",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-io"
+version = "1.3.1"
 dependencies = [
 "anyhow",
 "indicatif",
 "libc",
+ "memmap2",
 "owo-colors",
+ "senbei-crypto",
+ "senbei-elf",
+ "senbei-engine",
+ "senbei-metadata",
+ "senbei-pe",
+ "sha2",
 "tempfile",
- "thiserror",
 "walkdir",
 "windows",
+ "zip",
 ]

+[[package]]
+name = "senbei-metadata"
+version = "1.3.1"
+dependencies = [
+ "serde",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-pe"
+version = "1.3.1"
+dependencies = [
+ "thiserror",
+]
+
+[[package]]
+name = "serde"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4148590afebada386688f18773da617792bf2ef03ffc1e4cbd2b1d45b023e0ba"
+dependencies = [
+ "serde_core",
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_core"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "67dca2c9c51e58a4791a4b1ed58308b39c64224d349a935ab5039aa360942a48"
+dependencies = [
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_derive"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e7a5d71263a5a7d47b41f6b3f06ba276f10cc18b0931f1799f710578e2309348"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 3.0.4",
+]
+
+[[package]]
+name = "serde_json"
+version = "1.0.151"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c841b55ecdae098c80dcae9cf767f6f8a0c2cdb3416bbef72181df4d0fe73f14"
+dependencies = [
+ "itoa",
+ "memchr",
+ "serde",
+ "serde_core",
+ "zmij",
+]
+
+[[package]]
+name = "sha2"
+version = "0.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "446ba717509524cb3f22f17ecc096f10f4822d76ab5c0b9822c5f9c284e825f4"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "digest",
+]
+
+[[package]]
+name = "simd-adler32"
+version = "0.3.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3a219298ac11a56ea9a6d2120044824d6f01aeb034955e7af7bc16858527deea"
+
 [[package]]
 name = "slab"
 version = "0.4.12"
@@ -240,9 +557,9 @@ dependencies = [

 [[package]]
 name = "syn"
-version = "3.0.3"
+version = "3.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53e9bae58849f64dfa4f5d5ae372c8341f7305f82a3868709269343628b659a3"
+checksum = "e6275cddf4610d1775e6d1fe9469b2e77d0f39fd98fb7450901b821e0c53649f"
 dependencies = [
 "proc-macro2",
 "quote",
@@ -264,24 +581,36 @@ dependencies = [

 [[package]]
 name = "thiserror"
-version = "2.0.19"
+version = "2.0.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09a43598840e33d5b0331f38c5e30d13bb11c11210a4b58f0d9b18a5a5eefcd9"
+checksum = "ec86235f5fcc2a73650310756d2ac5b138a5780bbbdfae3eeccec992c435ba4f"
 dependencies = [
 "thiserror-impl",
 ]

 [[package]]
 name = "thiserror-impl"
-version = "2.0.19"
+version = "2.0.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "43cbfe0cf76104d42a574802844187e84a305e531ed54455f11fbde0f10541cd"
+checksum = "bc04cd3e1236dd4a98afca4569f2deb3f120e5422a4023be2cb683f8486292af"
 dependencies = [
 "proc-macro2",
 "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]

+[[package]]
+name = "typed-path"
+version = "0.12.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e28f89b80c87b8fb0cf04ab448d5dd0dd0ade2f8891bae878de66a75a28600e"
+
+[[package]]
+name = "typenum"
+version = "1.20.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6f5e870be6c3b371b77fe0ee0bafb859fa4964b4404c27de1d380043c4dda20"
+
 [[package]]
 name = "unicode-ident"
 version = "1.0.24"
@@ -312,9 +641,9 @@ dependencies = [

 [[package]]
 name = "wasm-bindgen"
-version = "0.2.126"
+version = "0.2.127"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4b067c0c11094aef6b7a801c1e34a26affafdf3d051dba08456b868789aaf9a4"
+checksum = "1b70935747edd64d89de3efa29d73789b806c15798f8e7dca4d8ac356b50ce70"
 dependencies = [
 "cfg-if",
 "once_cell",
@@ -325,9 +654,9 @@ dependencies = [

 [[package]]
 name = "wasm-bindgen-macro"
-version = "0.2.126"
+version = "0.2.127"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "167ce5e579f6bcf889c4f7175a8a5a585de84e8ff93976ce393efa5f2837aab1"
+checksum = "77775f8f3f7217702089053b94958f8f54061a3f663417df76e19cbdcca29bc1"
 dependencies = [
 "quote",
 "wasm-bindgen-macro-support",
@@ -335,9 +664,9 @@ dependencies = [

 [[package]]
 name = "wasm-bindgen-macro-support"
-version = "0.2.126"
+version = "0.2.127"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f3997c7839262f4ef12cf90b818d6340c18e80f263f1a94bf157d0ec4420380e"
+checksum = "e11d33f857dc2fb11b8bc75aee111aa9cbeb12cd9f25efd3d4c2a3dd4e235284"
 dependencies = [
 "bumpalo",
 "proc-macro2",
@@ -348,9 +677,9 @@ dependencies = [

 [[package]]
 name = "wasm-bindgen-shared"
-version = "0.2.126"
+version = "0.2.127"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dc1b4cb0cc549fcf58d7dfc081778139b3d283a081644e833e84682ad71cea24"
+checksum = "7ef64dbcc55df09c7e5a46182d181c2cfa3e925f3da937ea764728b4bbb9dcbf"
 dependencies = [
 "unicode-ident",
 ]
@@ -492,3 +821,41 @@ checksum = "3949bd5b99cafdf1c7ca86b43ca564028dfe27d66958f2470940f73d86d75b37"
 dependencies = [
 "windows-link",
 ]
+
+[[package]]
+name = "zip"
+version = "8.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2d04a6b5381502aa6087c94c669499eb1602eb9c5e8198e534de571f7154809b"
+dependencies = [
+ "crc32fast",
+ "flate2",
+ "indexmap",
+ "memchr",
+ "typed-path",
+ "zopfli",
+]
+
+[[package]]
+name = "zlib-rs"
+version = "0.6.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34b31d188d9d685a4f9c7b46d6e36631b07058d2cfe190267adce54dc230bf12"
+
+[[package]]
+name = "zmij"
+version = "1.0.23"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "29666d0abbfad1e3dc4dcf6144730dd3a3ab225bbbdac83319345b1b44ccfc1b"
+
+[[package]]
+name = "zopfli"
+version = "0.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f05cd8797d63865425ff89b5c4a48804f35ba0ce8d125800027ad6017d2b5249"
+dependencies = [
+ "bumpalo",
+ "crc32fast",
+ "log",
+ "simd-adler32",
+]
@@ -1,39 +1,60 @@
-[package]
-name = "senbei"
-version = "1.0.1"
+[workspace]
+members = [
+    "senbei-cli",
+    "senbei-crypto",
+    "senbei-elf",
+    "senbei-engine",
+    "senbei-io",
+    "senbei-metadata",
+    "senbei-pe",
+]
+default-members = ["senbei-cli"]
+# The wasm bindings crate is its own crate (own Cargo.lock, cdylib) and stays
+# outside the workspace.
+exclude = ["senbei-wasm"]
+resolver = "2"
+
+[workspace.package]
+version = "1.3.1"
 edition = "2024"
-description = "Static unpacker for Crackproof-protected PE files"
+rust-version = "1.98.1"
 license = "AGPL-3.0-only"
-keywords = ["unpacker", "reverse-engineering", "pe", "security-research"]
-categories = ["command-line-utilities"]

-[lib]
-name = "senbei"
-path = "src/lib.rs"
-
-[[bin]]
-name = "senbei"
-path = "src/main.rs"
-
-[dependencies]
+[workspace.dependencies]
+aes = "0.9"
 anyhow = "1"
+goblin = "0.10"
+indicatif = "0.18"
+libc = "0.2"
+memmap2 = "0.9"
+owo-colors = "4"
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+sha2 = "0.11"
+tempfile = "3"
 thiserror = "2"
 walkdir = "2"
-indicatif = "0.18"
-owo-colors = "4"
-
-[target.'cfg(windows)'.dependencies]
 windows = { version = "0.62", features = [
    "Win32_Foundation",
    "Win32_System_Console",
    "Win32_System_SystemInformation",
 ] }
+zip = { version = "8", default-features = false, features = ["deflate"] }
+senbei-crypto = { path = "senbei-crypto" }
+senbei-elf = { path = "senbei-elf" }
+senbei-engine = { path = "senbei-engine" }
+senbei-io = { path = "senbei-io" }
+senbei-metadata = { path = "senbei-metadata" }
+senbei-pe = { path = "senbei-pe" }

-[target.'cfg(all(not(windows), not(target_arch = "wasm32")))'.dependencies]
-libc = "0.2"
+[workspace.lints.rust]
+unsafe_op_in_unsafe_fn = "deny"

-[dev-dependencies]
-tempfile = "3"
+[workspace.lints.clippy]
+correctness = { level = "deny", priority = -1 }
+suspicious = { level = "warn", priority = -1 }
+complexity = { level = "warn", priority = -1 }
+perf = { level = "warn", priority = -1 }

 [profile.release]
 opt-level = 3
@@ -1,77 +1,32 @@
 # Senbei

-A static unpacker for Crackproof-protected 64-bit and 32-bit PE files. Point it
-at a file or a folder and it writes decrypted copies — no launch of the
-protected program, no kernel driver, no code runs out of the protected binary.
+A static unpacker for CrackProof-protected Windows PE files and Android AArch64 shared libraries.

 > _"Crackproof"? It's senbei (煎餅 — rice cracker). Cracks itself._

-Senbei reads a protected `.exe` or `.dll`, replays the unpacking algorithm
-entirely in memory, and writes the recovered image to a new file. The core is a
-pure, panic-free library with no file I/O; the CLI wraps it with scanning, a
-progress bar, and a run log. A browser version (WebAssembly, fully client-side)
-lives in [`web/`](web/).
+## Usage
+
+```cmd
+cargo build --release
+senbei protected.exe
+senbei game.apk
+senbei "C:\Games\MyGame"
+```
+
+Outputs are written below an `unpack` directory unless `--out` is supplied.
+
+Full documentation: <https://xn--ri8h.gitbook.io/crackproof-research/senbei>

 ## Legal notice and intended use

 **Read this before using Senbei.**

- Senbei is a research and interoperability tool. It exists to enable lawful
-  reverse engineering, security research, preservation, and interoperability
-  with software you already legitimately possess.
- **Only process binaries you own or are explicitly authorized to analyze.**
-  Depending on your jurisdiction and license agreements, circumventing
-  technological protection measures may be restricted (for example under
-  DMCA §1201 in the United States, which contains exemptions for security
-  research and interoperability). It is your responsibility to ensure your use
-  is lawful.
- Senbei does not bypass any access control for you: it performs a purely
-  static transformation of a file already on your disk. It derives everything
-  it needs from the input file itself, contains no vendor code or secrets, and
-  distributes no keys, cracks, or copyrighted content.
- Senbei does not enable online play, license fraud, or cheating, and must not
-  be used to redistribute decrypted binaries. Do not upload outputs anywhere.
- The authors provide this software "as is", without warranty of any kind, and
-  accept no liability for misuse. See [LICENSE](LICENSE) (AGPL-3.0).
- "Crackproof" is a trademark of its respective owner; this project is not
-  affiliated with or endorsed by the protection vendor or any software
-  publisher. Names are used for identification only.
-
-## What it handles
-
-| Kind | Description |
-| --- | --- |
-| `NativeExe` | Crackproof-protected native executable (PE32+ and PE32). |
-| `ManagedExe` | Protected .NET executable (has a CLR data directory). |
-| `NativeDll` | Protected native (unmanaged) DLL. |
-| `ManagedDll` | Protected .NET assembly (has a CLR data directory). |
-| `._` companion | Stub + external encrypted payload layout, spliced automatically. |
-| `global-metadata.dat` | il2cpp metadata with obfuscated method tokens, de-obfuscated in place. |
-
-Detection is content-based (header key-table at offset 4096, magic `KONN`),
-not extension-based. Anything unrecognized is left untouched.
-
-## Quick start
-
-```cmd
-cargo build --release
-
-senbei protected.exe
-:: -> unpack\protected.unpack.exe
-
-senbei "C:\Games\MyGame"
-:: -> C:\Games\MyGame\unpack\...  (recursive, skips non-targets)
-```
-
-Every output is sanity-checked statically; structurally broken results are
-flagged as suspect rather than silently trusted.
-
-## Documentation
-
- [Usage reference](docs/usage.md) — CLI flags, exit codes, integrity check
- [Design](docs/design.md) — architecture, routing, and error model
- [Development](docs/development.md) — building, testing, environment variables
- [Web version](web/README.md) — run Senbei in a browser
+- Senbei is a research and interoperability tool. It exists to enable lawful reverse engineering, security research, preservation, and interoperability with software you already legitimately possess.
+- **Only process binaries you own or are explicitly authorized to analyze.** Depending on your jurisdiction and license agreements, circumventing technological protection measures may be restricted (for example under DMCA §1201 in the United States, which contains exemptions for security research and interoperability). It is your responsibility to ensure your use is lawful.
+- Senbei does not bypass any access control for you: it performs a purely static transformation of a file already on your disk. It derives everything it needs from the input file itself, contains no vendor code, and distributes no cracks or copyrighted content. (One Android packaging variant's embedded metadata layer is unwrapped with an XOR keystream recovered from a ciphertext/plaintext pair during analysis of a single build; that keystream is research output shipped with the unpacker, not a vendor-distributed key, and builds it doesn't match are left alone.)
+- Senbei does not enable online play, license fraud, or cheating, and must not be used to redistribute decrypted binaries. Do not upload outputs anywhere.
+- The authors provide this software "as is", without warranty of any kind, and accept no liability for misuse. See [LICENSE](LICENSE) (AGPL-3.0).
+- "Crackproof" is a trademark of its respective owner; this project is not affiliated with or endorsed by the protection vendor or any software publisher. Names are used for identification only.

 ## License

@@ -1,143 +0,0 @@
-# Design
-
-Senbei is a fully static unpacker: it replays the unpacking algorithm on the
-file bytes in memory and writes the recovered PE image. No code from the
-protected binary is ever executed, no process is launched or attached to, and
-no driver or proxy DLL is involved.
-
-## Crate layout
-
-The crate is split into a pure core and a thin CLI shell:
-
- **`src/unpacker/`** — the core. Pure functions over byte slices: no file
-  I/O, no environment access (beyond a few debugging overrides, see
-  [development.md](development.md)), panic-free at the public boundary (all
-  internal panics are trapped and converted to `UnpackError::Corrupt`). This
-  is what the WebAssembly build embeds.
- **`src/` (top level)** — the CLI shell: argument parsing, recursive folder
-  scanning, per-run log file, progress bar, Explorer-friendly exit pause, and
-  the single-file/folder orchestration in `job.rs`.
- **`src/metadata.rs`** — il2cpp `global-metadata.dat` method-token
-  de-obfuscation (format version 31; other versions are left untouched).
-
-```
-src/
-├── main.rs                argument parsing + dispatch
-├── lib.rs                 module roots
-├── job.rs                 single-file + folder orchestration, out-naming,
-│                          companion splice, stub overlay/TLS restore,
-│                          pipeline routing (incl. the wasm-safe byte API)
-├── scan.rs                recursive Crackproof + metadata discovery
-├── metadata.rs            il2cpp global-metadata.dat de-obfuscation
-├── logfile.rs             per-run timestamped log
-├── ui.rs                  progress bar + status lines
-├── pause.rs               Explorer-friendly exit pause
-└── unpacker/              pure, panic-free, no-I/O core
-    ├── mod.rs             detection + unpack_auto dispatch
-    ├── exe.rs             EXE pipeline (PE32+ and PE32)
-    ├── dll.rs             native + managed DLL pipeline
-    ├── integrity.rs       static post-unpack sanity check
-    ├── primitives.rs      decrypt_data* steps, key/shift selection
-    ├── bytecode.rs        bytecode VM
-    ├── parallel.rs        deterministic block-parallel fan-out
-    ├── tables.rs          constant tables
-    └── crc32.rs           checksum
-```
-
-## Detection and routing
-
-Detection is content-based (`unpacker::detect`), never extension-based: the
-key table is derived from the file header and checked against the format
-magic, then the PE characteristics classify the input as EXE or DLL and the
-CLR data directory splits each into native vs managed (`NativeExe` /
-`ManagedExe` / `NativeDll` / `ManagedDll`).
-
-`unpack_auto` then dispatches:
-
- `NativeExe` / `ManagedExe` → the EXE pipeline (handles both PE32+ and
-  PE32). Managed EXEs take the same path: their import-string table is null
-  (imports are the CLR bootstrap stub), the entry point comes from the
-  protected header (the config block stores 0 for managed images), and the
-  COR20 header, BSJB metadata stream, and CLR resources are restored verbatim
-  from the protected file, mirroring the managed-DLL restore.
- `NativeDll` / `ManagedDll` → the DLL pipeline first; on failure, the EXE
-  pipeline as a fallback. Two DLL layouts exist in the wild: an older layout
-  the DLL pipeline parses, and a newer one that protects DLLs with the
-  EXE-style shell layout instead. The DLL-first order keeps old-layout outputs
-  byte-identical (the EXE pipeline also "succeeds" on old-layout DLLs but
-  produces different bytes); the fallback handles the new layout (including
-  the managed-DLL .NET metadata restore).
-
-One routing shortcut bypasses `unpack_auto`: inputs spliced from an external
-companion (`job.rs`, both the CLI and the wasm byte API) go **straight to the
-EXE pipeline**. The companion layout is definitionally the EXE-style shell,
-so the DLL probe can never be right for it — and the probe's rejection of
-EXE-shell DLLs relies on a caught panic, which is a fatal trap on targets
-without unwinding (WebAssembly). Output bytes are identical to the
-probe-then-fallback route.
-
-## External-companion inputs
-
-Some builds split a protected module into an on-disk loader stub plus an
-encrypted `._` companion. When a `<name>._` sibling matches the stub's header
-region, `job.rs` splices the two before unpacking and afterwards overlays the
-export table and TLS directory from the stub — pieces the encrypted companion
-does not carry. All overlay steps are best-effort no-ops when their inputs
-can't be mapped, so a malformed stub can never corrupt an otherwise-good
-unpack.
-
-## Pipelines
-
-Both pipelines are **heuristic with trial-and-validate**: where a layout
-leaves ambiguity (e.g. which block is the real file decryptor, or a page-XOR
-shift), the pipeline tries candidates and validates the result structurally
-(an entry-stub oracle, checksum stamps, cluster stamps) instead of trusting
-the first match. A validation failure falls through to the next candidate
-rather than producing silently wrong output.
-
-Several protected stages are themselves little bytecode programs. The core
-includes a small VM (`bytecode.rs`) that generates and interprets those
-programs rather than hardcoding each variant's constants.
-
-## Integrity check
-
-Every produced image passes through `integrity::check` — a static, execution-
-free sanity check that only flags defects impossible in a correctly unpacked
-image (malformed headers, unmapped/non-executable/all-zero/all-int3 entry
-point, a native DLL with no base-relocation directory, any import descriptor
-whose DLL name is still ciphertext, a managed image whose COR20 header or BSJB
-metadata did not survive). See [usage.md](usage.md#integrity-check).
-A clean report is not a proof of correctness; a non-clean report is a reliable
-"broken" signal.
-
-## Parallelism
-
-Section decrypt/decompress blocks write disjoint output spans and read only
-immutable input plus snapshotted key tables, so `parallel.rs` fans them out
-across worker threads with **byte-identical** output regardless of thread
-count. There is no `unsafe`: the buffer is carved with safe `split_at_mut`
-chains so the borrow checker proves spans never alias. Overlapping spans (only
-possible on corrupt input) degrade to the sequential whole-buffer pass,
-preserving the deterministic last-writer-wins behavior of the serial
-pipeline. `SENBEI_THREADS=1` forces the sequential path; on targets without
-threads (WebAssembly) the sequential path is used automatically.
-
-## Error model
-
-The public API never panics: every pipeline runs under a `catch_unwind`
-wrapper (`catch_unpack`) that converts a trapped panic to
-`UnpackError::Corrupt`, with the default panic hook transiently suppressed.
-Size requests are bounds-checked against a 1 GiB `MAX_IMAGE_SIZE` before
-allocation so a crafted header cannot abort the process with a huge
-allocation. In folder mode each file is isolated: one file's failure is logged
-and counted, never fatal to the run.
-
-**WebAssembly caveat:** the prebuilt wasm std cannot unwind, so a caught
-panic becomes a fatal `unreachable` trap there. The DLL-routing probe relies
-on this mechanism to reject EXE-shell-layout DLLs, so the web build routes
-around it instead of through it: spliced companion inputs skip the probe
-entirely (see "Detection and routing"), and the web app isolates every unpack
-in a disposable Web Worker — a trapped DLL is retried once in a fresh worker
-with the forced-EXE pipeline (`job::unpack_bytes_force_exe`), reproducing the
-probe-then-fallback outcome without a catchable panic. A trap on any other
-input is reported as a clean error rather than freezing the page.
@@ -1,116 +0,0 @@
-# Development
-
-## Building
-
-Requires a Rust toolchain (MSVC backend is the default on Windows;
-`rustup-init.exe` from <https://rustup.rs> installs it). The pinned toolchain
-and targets are in `rust-toolchain.toml`.
-
-```cmd
-cargo build --release
-```
-
-Output: `target\release\senbei.exe`. The binary is self-contained — no driver,
-no proxy DLL, no external assets.
-
-The library and CLI also build for Linux/macOS (`cfg`-gated platform code
-only) and for `wasm32-unknown-unknown` (see the [web version](../web/README.md)).
-
-## Testing
-
-```cmd
-cargo test --release
-```
-
-The suite covers CLI behavior, detection, the folder driver, the run log, and
-byte-exact golden tests over `samples/` — a user-managed corpus (git-ignored,
-see `samples/README.md`) of real Crackproof inputs plus `<base>.golden.<ext>`
-reference outputs. Every input goes through `job::unpack_bytes` — the same
-routing the CLI uses, so an `<input>._` companion in the corpus is spliced and
-the stub export/TLS overlays run — and is gated on **two** checks: the static
-integrity check (catches runtime-broken outputs even when a stale golden would
-still byte-match) and, when a golden exists, a bit-for-bit comparison. il2cpp
-`*.dat` inputs are routed through `metadata::deobfuscate` instead. An empty or
-absent corpus is a no-op pass; set `SENBEI_REQUIRE_SAMPLES` to make it fail
-instead (useful on a private CI that has the corpus — public CI never does,
-since binaries are not committed).
-
-> **Note:** goldens encode expected *bytes*, not runtime behavior. A golden
-> produced before a pipeline fix may byte-match while still being wrong — the
-> integrity check is the second gate for exactly this reason. Re-verify
-> goldens against real runs when touching the affected pipeline stages.
->
-> **The corpus only protects what it contains.** Wire the test to the routing
-> the CLI actually takes (it is), and keep a sample for every layout family —
-> marker-based, marker-less, external-companion, PE32, PE32+, native, managed,
-> metadata. An unrepresented family has no regression gate at all, which is
-> how a "re-run the golden corpus" rule can pass while silently covering
-> nothing.
-
-## Debugging levers (environment variables)
-
- `DD8_SHIFT` — override the `decrypt_data8` page-XOR shift (`99` skips dd8
-  entirely).
- `SEL_DIAG` — print the dd8 selector's scores: the per-shift `0xCC` counts and
-  the plaintext baseline they are compared against (PE32+), and the per-formula
-  counts, baseline and net gain (PE32).
- `SENBEI_THREADS` — cap the block-parallel fan-out (`1` forces the fully
-  sequential path).
- `SENBEI_SCAN_ALL` — same as `--scan-all` (probe every file in a folder).
-
-## Conventions
-
- The `src/unpacker/` core is pure: no file I/O, no panics across the public
-  boundary, no `unsafe`. Keep it that way — it is what the WebAssembly build
-  embeds.
- Layout heuristics must **trial-and-validate**: never pick a candidate offset
-  on shape alone and trust it; validate by decryption/checksum and fall
-  through to the next candidate on failure. A silent wrong offset produces a
-  silently broken output, which is worse than an error.
- Output must remain byte-identical against the golden corpus for every
-  supported layout. When fixing one build family, re-run the full golden
-  corpus to prove no other family regressed.
- Folder scanning uses a size floor plus an extension **deny**-list, never an
-  allow-list: targets are recognised by content, not extension, and can carry
-  arbitrary names, so only known bulk-asset extensions are excluded. The
-  pre-filter exists because folder-scan cost is per-file I/O latency, not the
-  walk — probe fewer files, don't parallelize the probe loop.
- `cargo fmt` and `cargo clippy` must stay clean (CI enforces both).
-
-## Repository layout
-
-```
-senbei/
-├── Cargo.toml                 senbei lib + bin package
-├── rust-toolchain.toml        pinned toolchain + targets
-├── src/                       CLI shell + pure unpacker core (see docs/design.md)
-├── tests/                     CLI, detection, golden, and folder tests
-├── samples/                   local-only test corpus (git-ignored)
-├── web/                       WebAssembly browser build
-├── docs/                      usage, design, and development documentation
-└── .github/                   CI workflows and issue templates
-```
-
-## Web build
-
-See [web/README.md](../web/README.md). In short:
-
-```cmd
-cd web
-wasm-pack build --target web --release
-```
-
-then serve `web/` statically and open `index.html`. Everything runs
-client-side; no file leaves the browser.
-
-## Contributing
-
-Issues and pull requests are welcome. A few ground rules:
-
- **Never commit binaries** (protected or decrypted) to the repository —
-  the only corpus is the local git-ignored `samples/`. Attaching a protected
-  input file to an issue is welcome if it helps diagnose the problem; only
-  attach files you are authorized to share.
- Run `cargo test --release`, `cargo clippy`, and `cargo fmt` before
-  submitting.
- Keep the unpacker core free of I/O, `unsafe`, and platform-specific code.
@@ -1,126 +0,0 @@
-# Usage
-
-```
-senbei <file|folder> [--out DIR] [-v|--verbose] [-q|--quiet]... [--scan-all]
-                     [--no-log] [--no-pause] [-V|--version] [-h|--help]
-```
-
-Real runs print `Senbei <version>` once at start. Use `-V` / `--version` to
-print the version and exit.
-
-## Single file
-
-The decrypted image is written under `<parent>/unpack/` with `.unpack` inserted
-before the extension. A `senbei-<timestamp>.log` is written in the same
-directory. With `--out DIR`, both the output and the log go into `DIR` instead:
-
-```cmd
-senbei app.exe
-:: -> unpack\app.unpack.exe
-:: -> unpack\senbei-YYYYMMDD-HHMMSS.log
-
-senbei app.exe --out C:\out
-:: -> C:\out\app.unpack.exe
-:: -> C:\out\senbei-YYYYMMDD-HHMMSS.log
-```
-
-Pointing senbei directly at an il2cpp `global-metadata.dat` rewrites its
-obfuscated method tokens back to the contiguous per-module range il2cpp
-expects; the output is `global-metadata.unpack.dat`, written only when tokens
-actually changed. Only metadata format version 31 is rewritten; other versions
-are reported and left untouched.
-
-## Folder mode
-
-Senbei walks the directory recursively, skips any subdirectory literally named
-`unpack`, and unpacks every file it recognises as Crackproof-protected (by
-content, not extension — renamed files and `.bak` backups are still found).
-Results land under `<root>/unpack/` (or `--out DIR`), mirroring the input
-tree's relative paths. The run log is written **in that same out directory**:
-
-```cmd
-senbei "C:\Games\MyGame"
-:: -> C:\Games\MyGame\unpack\...
-:: -> C:\Games\MyGame\unpack\senbei-YYYYMMDD-HHMMSS.log
-```
-
-Folder mode also picks up `global-metadata.dat` files and external-companion
-`._` payloads: a module whose `<name>._` sibling matches its header region is
-spliced with the companion automatically (no flag needed) and unpacked as one
-image, with the output named for the stub.
-
-Each file is processed in isolation: an error or panic on one file is caught,
-counted, and logged, and the run continues. Folder mode finishes with a summary
-line, then duration:
-
-```
-12 unpacked · 3 skipped · 0 errors · 1 suspect · 2 metadata
-done in 1234 ms
-```
-
-## Integrity check
-
-A successful unpack is not always a runnable one: a layout heuristic can pick
-the wrong offset and leave the entry-point stub or import strings encrypted, so
-the pipeline reports success but the OS loader faults at runtime (typically
-`0xC0000005`, STATUS_ACCESS_VIOLATION). To catch this, senbei runs a static
-sanity check over every output it produces — inspecting the bytes alone, with
-no reference image and no execution.
-
-It flags only defects that cannot occur in a correctly unpacked image:
-
- malformed DOS/PE headers, bad optional-header magic, implausible section
-  count, zero `SizeOfImage`, or section raw-data ranges that run past EOF;
- an entry point that doesn't map into a section, isn't in an executable
-  section, or whose stub is all zeros or all `0xCC` int3 padding (the classic
-  left-encrypted symptom);
- a native (unmanaged) DLL with no base-relocation directory — it cannot
-  survive being mapped at a non-preferred base;
- **any** import descriptor whose DLL name doesn't resolve or isn't readable
-  ASCII (imports left encrypted) — the whole table is walked, not just the
-  first entry;
- for a managed assembly, a COR20 header whose `cb` isn't `0x48` or a
-  MetaData stream missing its `BSJB` signature (the CLR would reject the
-  image outright).
-
-The entry-point and import checks are skipped for managed assemblies, whose
-native EP and import stub are legitimately not what the native loader expects.
-
-The check is deliberately conservative: a clean report is **not** a proof of
-correctness, but a non-clean report is a reliable "this is broken" signal. A
-suspect file is still written (the bytes are the best available) and flagged —
-single-file mode prints a warning to stderr, folder mode prints a yellow `!`
-line, adds a `SUSPECT` entry to the run log, and counts it in the summary's
-`suspect` total (which is additive to `unpacked`).
-
-## Flags
-
-| Flag | Behavior |
-| --- | --- |
-| `--out DIR` | Write outputs (and the log, unless `--no-log`) under `DIR`. |
-| `-v`, `--verbose` | Print detailed `[N/9]` per-stage unpack progress (and the destination path) for each file. In folder mode this replaces the progress bar. |
-| `-q`, `--quiet` | Once: hide progress bar and per-file lines; keep banner, summary, and duration. Twice (`-q -q`): suppress all stdio (exit code only). |
-| `--no-log` | Do not write `senbei-*.log`. Console output is unchanged by this flag alone. |
-| `--scan-all` | Probe every file in a folder, including ones the scan pre-filter skips (under 4128 bytes, or a bulk-asset extension like `.ab`/`.xml`/`.acb`). Much slower on large game trees; finds the same targets in practice. |
-| `--no-pause` | Skip the "Press Enter to exit" prompt (for scripted runs). |
-| `-V`, `--version` | Print `Senbei <version>` and exit. |
-| `-h`, `--help` | Show usage. |
-
-On Windows, when launched from Explorer (the process owns its console) senbei
-pauses for Enter before exiting so the window doesn't vanish. `--no-pause`
-disables this; it has no effect when stdout is piped or run from another
-process.
-
-## Exit codes
-
-| Code | Meaning |
-| --- | --- |
-| `0` | Success (single file unpacked, or folder run with no errors). |
-| `1` | At least one file failed, a scan probe was unreadable, or a single-file unpack errored. |
-| `2` | Usage error: no path given, unknown option, missing `--out` value, or multiple input paths (help printed). |
-
-A folder run also fails with `1` when parts of the tree could not be scanned
-(unreadable directory entries or files that failed the content probe) — those
-are potential missed targets, not clean skips. An il2cpp metadata blob whose
-format version senbei does not handle is *not* an error: it is reported, left
-untouched, and counted as skipped.
@@ -1,3 +1,7 @@
 [toolchain]
-channel = "stable"
+channel = "1.98.1"
+# CI invokes rustfmt/clippy through the rustup shim with no setup action, so
+# the pinned toolchain must declare its components here — the runner images
+# only preinstall them for their default toolchain.
+components = ["rustfmt", "clippy"]
 targets = ["x86_64-pc-windows-msvc", "wasm32-unknown-unknown"]
@@ -1,84 +1,23 @@
-# senbei/samples
+# Samples

-Drop-in corpus for the `samples` integration test (`tests/samples.rs`).
+This ignored directory is the optional local corpus used by the samples integration test. Protected binaries and restored outputs must never be committed.

-This folder is **git-ignored** (only this `README.md` is tracked), so it holds
-whatever Crackproof binaries happen to be on your machine. Nothing here is
-committed.
+Place protected `.exe` and `.dll` files, exact `global-metadata.dat` files, and matching `.exe._` or `.dll._` companion payloads here. A golden output may sit beside an input as `<base>.golden.<ext>`.

-## What to put here
-
-Place protected inputs directly in this folder:
-
- `*.exe` — Crackproof-protected executables (PE32 or PE32+)
- `*.dll` — Crackproof-protected DLLs (native or managed)
- `*.dat` — il2cpp `global-metadata.dat` blobs (method-token de-obfuscation)
-
-For an **external-companion** module, copy the `<name>._` payload in as well,
-keeping the exact `._` suffix on the full file name. The test splices it the
-same way the CLI does; without it the loader stub alone is meaningless and the
-splice / export-overlay / TLS-restore code is never exercised.
-
-Optionally, place a **golden** next to each input — the known-good unpacked
-output, named `<base>.golden.<ext>`:
-
-```
+```text
 samples/
-  app.exe                   <- input
-  app.golden.exe            <- golden (optional)
-  managed.dll               <- input
-  managed.golden.dll        <- golden (optional)
-  stub.dll                  <- input  (external-companion layout)
-  stub.dll._                <- its encrypted payload  (NOT an input itself)
-  stub.golden.dll           <- golden
-  global-metadata.dat       <- input
-  global-metadata.golden.dat<- golden
-  mystery.exe               <- input, no golden
+  app.exe
+  app.golden.exe
+  managed.dll
+  stub.dll
+  stub.dll._
+  stub.golden.dll
+  global-metadata.dat
+  global-metadata.golden.dat
 ```

-The type (EXE vs native/managed DLL vs metadata) is auto-detected from the file
-contents, not the extension, so you don't need to classify anything by hand.
+Run `cargo test --release --test samples -- --nocapture`. A missing golden prints a warning; a mismatched golden or failed restore fails the test. An empty corpus is a no-op pass.

-Since the corpus is the only regression gate on byte-identical output, keep it
-broad: each build family, each layout (marker-based and marker-less), and at
-least one external-companion pair. A family with no sample here is a family no
-test protects.
+## Android Corpus

-## How the test treats each input
-
-Run with:
-
-```
-cargo test --release --test samples
-```
-
-For every input file, the test runs the same routing the CLI uses
-(`job::unpack_bytes`, so companions splice and the stub overlays run) — or
-`metadata::deobfuscate` for an il2cpp blob — and then:
-
-| Situation                                   | Result                          |
-| ------------------------------------------- | ------------------------------- |
-| Golden present, bytes **identical**         | **pass**                        |
-| Golden present, bytes **differ**            | **fail** (test fails)           |
-| **No golden** found                         | **warning** (needs manual check)|
-| Unpack errored / file unreadable            | **fail**                        |
-
-Warnings are printed but do not fail the test — they flag outputs you should
-eyeball or promote to a golden once verified. Failures fail the test. An empty
-or absent folder is a no-op pass.
-
-To see the per-file warning/pass/fail summary, run with output shown:
-
-```
-cargo test --release --test samples -- --nocapture
-```
-
-## Naming rules
-
- An **input** is any `*.exe` / `*.dll` / `*.dat` whose name does **not**
-  contain the `.golden.` segment.
- A **golden** is `<base>.golden.<ext>` sitting next to its input. Files with
-  `.golden.` in the name are never treated as inputs.
- A **companion** is `<input file name>._` (e.g. `stub.dll._` for `stub.dll`).
-  Its extension is `_`, so it is never picked up as an input of its own; it is
-  read only when its base module is processed.
+`samples/android/` may contain one extracted app tree per subdirectory. Protected libraries are restored through the real pipeline and can carry SHA-256 sidecars named `<base>.golden.so.sha256` and `<base>.golden.metadata.sha256`. An empty `<base>.restore-fails` marker documents a known restore gap.
@@ -0,0 +1,22 @@
+[package]
+name = "senbei-cli"
+version.workspace = true
+edition.workspace = true
+description = "Command-line entry point for Senbei"
+license.workspace = true
+keywords = ["unpacker", "reverse-engineering", "pe", "security-research"]
+categories = ["command-line-utilities"]
+
+[[bin]]
+name = "senbei"
+path = "src/main.rs"
+
+[dependencies]
+senbei-io.workspace = true
+senbei-engine.workspace = true
+
+[dev-dependencies]
+senbei-io.workspace = true
+senbei-metadata.workspace = true
+sha2.workspace = true
+tempfile.workspace = true
@@ -1,4 +1,4 @@
-use senbei::{job, pause};
+use senbei_io::{job, pause, scan};
 use std::path::Path;

 fn main() -> std::process::ExitCode {
@@ -27,9 +27,6 @@ fn main() -> std::process::ExitCode {
            "--no-log" => no_log = true,
            "--scan-all" => scan_all = true,
            "--out" => match args.next() {
-                // Reject a missing value (and a following flag swallowed as the
-                // value): previously `--out` at end of argv silently fell back
-                // to the default output directory.
                Some(v) if !v.starts_with('-') => out = Some(v),
                _ => {
                    eprintln!("error: --out requires a directory argument");
@@ -42,7 +39,6 @@ fn main() -> std::process::ExitCode {
                return std::process::ExitCode::from(2);
            }
            other => {
-                // Previously the last positional silently won.
                if let Some(prev) = &path {
                    eprintln!("error: multiple input paths given ('{prev}' and '{other}')");
                    return std::process::ExitCode::from(2);
@@ -63,33 +59,29 @@ fn main() -> std::process::ExitCode {
            }
            let p = Path::new(&p);
            let out_path = out.as_deref().map(Path::new);
-            let r = if p.is_dir() {
+            let result = if p.is_dir() {
                job::run_folder_opts(
                    p,
                    out_path,
                    quiet,
                    verbose,
                    no_log,
-                    scan_all || senbei::scan::scan_all_env(),
+                    scan_all || scan::scan_all_env(),
                )
            } else {
                job::run_file_v(p, out_path, quiet, verbose, no_log)
            };
-            match r {
-                Ok(s) => {
+            match result {
+                Ok(summary) => {
                    if quiet < 2 {
-                        println!(
-                            "{} unpacked · {} skipped · {} errors · {} suspect · {} metadata",
-                            s.unpacked, s.skipped, s.errors, s.suspect, s.metadata
-                        );
-                        println!("done in {} ms", s.duration_ms);
+                        println!("{}", summary.line());
+                        println!("done in {} ms", summary.duration_ms);
                    }
-                    if s.errors > 0 { 1 } else { 0 }
+                    if summary.errors > 0 { 1 } else { 0 }
                }
-                Err(e) => {
-                    // Fatal: out-dir/log create, etc.
+                Err(error) => {
                    if quiet < 2 {
-                        eprintln!("error: {e:#}");
+                        eprintln!("error: {error:#}");
                    }
                    1
                }
@@ -106,8 +98,13 @@ fn print_help() {
        "senbei <file|folder> [--out DIR] [-v|--verbose] [-q|--quiet]... [--scan-all] [--no-log] [--no-pause] [-V|--version] [-h|--help]"
    );
    println!(
-        "  --scan-all   probe every file in a folder, including ones the scan\n\
-         \x20              pre-filter skips (under 4128 bytes, or a bulk-asset\n\
-         \x20              extension like .ab/.xml/.acb). Much slower on game trees."
+        "  input        a Crackproof PE (.exe/.dll), an il2cpp global-metadata.dat,\n\
+         \x20              a protected Android AArch64 library (.so), an Android app\n\
+         \x20              package (.apk/.apks/.xapk), or a folder containing any of\n\
+         \x20              these"
+    );
+    println!(
+        "  --scan-all   probe selected .exe/.dll/.so/metadata names below the\n\
+         \x20              size floor; other filenames remain excluded."
    );
 }
@@ -0,0 +1,231 @@
+//! Corpus test over the user-managed Android samples.
+//!
+//! Each immediate subdirectory of `samples/android/` that contains a `lib/`
+//! tree is one app-package sample (an extracted APK layout). For every
+//! protected AArch64 `.so` found by content probe, the test runs the real
+//! restore pipeline and checks the result:
+//!
+//! - `<name>.golden.so.sha256` next to the input pins the restored bytes
+//!   (byte-identity through the digest; absent sidecar -> WARNING).
+//! - `<name>.restore-fails` (empty marker) documents an input whose restore
+//!   is known to fail; the test then *requires* failure, so a future fix
+//!   surfaces as a test failure too. Without the marker a failed restore is
+//!   a test failure.
+//! - A restored library carrying an unwrappable embedded metadata blob must
+//!   produce one, pinned by `<name>.golden.metadata.sha256`.
+//!
+//! The folder-mode driver is then run over each app dir to exercise the
+//! scan/restore/write path end to end; its error count must equal the number
+//! of marked known-failures.
+//!
+//! The corpus is git-ignored and absent on CI (no binaries in the repo);
+//! `SENBEI_REQUIRE_SAMPLES=1` turns an absent corpus into a failure, and
+//! `SENBEI_ANDROID_SAMPLES` overrides the corpus location.
+
+mod common;
+
+use std::path::{Path, PathBuf};
+
+use senbei_io::{android, job};
+
+fn corpus_dir() -> PathBuf {
+    if let Some(dir) = std::env::var_os("SENBEI_ANDROID_SAMPLES") {
+        return PathBuf::from(dir);
+    }
+    common::samples_dir().join("android")
+}
+
+/// Immediate subdirectories of `root` that hold an app tree (a `lib/`
+/// folder) — research notes, dumps, and other non-app material in the corpus
+/// never match.
+fn app_dirs(root: &Path) -> Vec<PathBuf> {
+    let mut dirs: Vec<PathBuf> = std::fs::read_dir(root)
+        .unwrap_or_else(|e| panic!("read {}: {e}", root.display()))
+        .filter_map(|entry| entry.ok().map(|entry| entry.path()))
+        .filter(|path| path.is_dir() && path.join("lib").is_dir())
+        .collect();
+    dirs.sort();
+    dirs
+}
+
+/// Every regular `.so` below `dir`, skipping previous output trees.
+fn collect_so_files(dir: &Path, out: &mut Vec<PathBuf>) {
+    let mut entries: Vec<_> = std::fs::read_dir(dir)
+        .unwrap_or_else(|e| panic!("read {}: {e}", dir.display()))
+        .filter_map(|entry| entry.ok().map(|entry| entry.path()))
+        .collect();
+    entries.sort();
+    for path in entries {
+        if path.is_dir() {
+            if path
+                .file_name()
+                .is_some_and(|name| !name.eq_ignore_ascii_case("unpack"))
+            {
+                collect_so_files(&path, out);
+            }
+        } else if path
+            .extension()
+            .and_then(|ext| ext.to_str())
+            .is_some_and(|ext| ext.eq_ignore_ascii_case("so"))
+        {
+            out.push(path);
+        }
+    }
+}
+
+fn sha256_hex(data: &[u8]) -> String {
+    use sha2::Digest;
+    let mut digest = sha2::Sha256::new();
+    digest.update(data);
+    let mut out = String::with_capacity(64);
+    for byte in digest.finalize() {
+        out.push_str(&format!("{byte:02x}"));
+    }
+    out
+}
+
+/// `<stem>.golden.so.sha256` next to `input`.
+fn golden_sidecar(input: &Path, artifact: &str) -> PathBuf {
+    let file = input.file_name().unwrap().to_string_lossy();
+    let stem = file.strip_suffix(".so").unwrap_or(&file);
+    input.with_file_name(format!("{stem}.golden.{artifact}.sha256"))
+}
+
+fn read_sidecar(path: &Path) -> Option<String> {
+    std::fs::read_to_string(path)
+        .ok()
+        .map(|text| text.trim().to_ascii_lowercase())
+}
+
+#[test]
+fn android_samples_restore_against_goldens() {
+    let root = corpus_dir();
+    // Same opt-in gate as the PE corpus test: an absent corpus is a no-op
+    // pass unless CI explicitly requires it.
+    let require = std::env::var_os("SENBEI_REQUIRE_SAMPLES").is_some();
+    if !root.is_dir() {
+        assert!(
+            !require,
+            "android samples: {} does not exist — corpus required (CI)",
+            root.display()
+        );
+        eprintln!(
+            "android samples: {} does not exist, nothing to test",
+            root.display()
+        );
+        return;
+    }
+    let apps = app_dirs(&root);
+    if apps.is_empty() {
+        assert!(
+            !require,
+            "android samples: no app trees under {} — corpus required (CI)",
+            root.display()
+        );
+        eprintln!("android samples: no app trees under {}", root.display());
+        return;
+    }
+
+    let mut passed = 0usize;
+    let mut warnings: Vec<String> = Vec::new();
+    let mut failures: Vec<String> = Vec::new();
+
+    for app in &apps {
+        let mut so_files = Vec::new();
+        collect_so_files(app, &mut so_files);
+        let protected: Vec<PathBuf> = so_files
+            .into_iter()
+            .filter(|path| android::is_protected_so_file(path))
+            .collect();
+        let mut known_failures = 0usize;
+
+        for input in &protected {
+            let name = input.file_name().unwrap().to_string_lossy().to_string();
+            let known_fails = input.with_file_name(format!(
+                "{}.restore-fails",
+                name.strip_suffix(".so").unwrap_or(&name)
+            ));
+            let temp = tempfile::tempdir().expect("tempdir");
+            let dest = temp.path().join("restored.so");
+            match android::restore_so_file(input, &dest, false) {
+                Ok(embedded) => {
+                    if known_fails.is_file() {
+                        failures.push(format!(
+                            "{name}: restore succeeded but a restore-fails marker exists \
+                             (delete the marker — the gap is fixed)"
+                        ));
+                        continue;
+                    }
+                    let bytes = std::fs::read(&dest).expect("read restored output");
+                    match read_sidecar(&golden_sidecar(input, "so")) {
+                        Some(expected) if expected == sha256_hex(&bytes) => passed += 1,
+                        Some(expected) => failures.push(format!(
+                            "{name}: restored bytes differ from golden\n  expected sha256 {expected}\n  actual   sha256 {}",
+                            sha256_hex(&bytes)
+                        )),
+                        None => warnings.push(format!(
+                            "{name}: no golden sidecar — restored sha256 {}",
+                            sha256_hex(&bytes)
+                        )),
+                    }
+                    if let Some(blob) = embedded {
+                        match read_sidecar(&golden_sidecar(input, "metadata")) {
+                            Some(expected) if expected == sha256_hex(&blob) => {}
+                            Some(expected) => failures.push(format!(
+                                "{name}: embedded metadata differs from golden\n  expected sha256 {expected}\n  actual   sha256 {}",
+                                sha256_hex(&blob)
+                            )),
+                            None => warnings.push(format!(
+                                "{name}: no embedded-metadata sidecar — sha256 {}",
+                                sha256_hex(&blob)
+                            )),
+                        }
+                    }
+                }
+                Err(error) => {
+                    if known_fails.is_file() {
+                        known_failures += 1;
+                    } else {
+                        failures.push(format!("{name}: restore failed: {error:#}"));
+                    }
+                }
+            }
+        }
+
+        // Folder-mode smoke run: the scan must route every protected library,
+        // and only the marked known-failures may error.
+        let out_temp = tempfile::tempdir().expect("tempdir");
+        match job::run_folder_opts(app, Some(out_temp.path()), 2, false, true, false) {
+            Ok(summary) => {
+                if summary.errors != known_failures {
+                    failures.push(format!(
+                        "{}: folder run errors {} != known-failure markers {known_failures}",
+                        app.display(),
+                        summary.errors
+                    ));
+                }
+                if summary.unpacked < protected.len().saturating_sub(known_failures) {
+                    failures.push(format!(
+                        "{}: folder run restored {} libraries, per-file pass found {} ({} known-failing)",
+                        app.display(),
+                        summary.unpacked,
+                        protected.len(),
+                        known_failures
+                    ));
+                }
+            }
+            Err(error) => failures.push(format!("{}: folder run failed: {error:#}", app.display())),
+        }
+    }
+
+    for warning in &warnings {
+        eprintln!("WARNING: {warning}");
+    }
+    eprintln!(
+        "android samples: {} app tree(s) — {passed} pass, {} warning(s), {} failure(s)",
+        apps.len(),
+        warnings.len(),
+        failures.len()
+    );
+    assert!(failures.is_empty(), "{}", failures.join("\n"));
+}
@@ -0,0 +1,11 @@
+//! Shared test fixtures.
+#![allow(dead_code)]
+
+use std::path::PathBuf;
+
+/// Path to the workspace-root `samples/` — the user-managed corpus dropped in
+/// by hand. Git-ignored except its README; tests here run against whatever is
+/// present. `CARGO_MANIFEST_DIR` is `senbei-cli/`, so go one level up.
+pub fn samples_dir() -> PathBuf {
+    PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("../samples")
+}
@@ -1,4 +1,4 @@
-use senbei::job::{default_out_root_for_file, out_name};
+use senbei_io::job::{default_out_root_for_file, out_name};
 use std::path::Path;

 #[test]
@@ -1,4 +1,4 @@
-use senbei::logfile::{Log, local_stamp_compact, local_stamp_display};
+use senbei_io::logfile::{Log, local_stamp_compact, local_stamp_display};

 #[test]
 fn local_stamp_compact_matches_shape() {
@@ -1,4 +1,4 @@
-use senbei::job;
+use senbei_io::job;
 use std::path::Path;

 fn list_logs(dir: &Path) -> Vec<std::path::PathBuf> {
@@ -8,7 +8,7 @@
 //! - golden present, bytes differ     -> FAIL (the test fails)
 //! - no golden                        -> WARNING (printed; needs a manual check)
 //!
-//! Inputs go through [`senbei::job::unpack_bytes`], the same routing the CLI
+//! Inputs go through [`senbei_io::job::unpack_bytes`], the same routing the CLI
 //! uses, **not** `unpack_auto` directly. That matters: `unpack_auto` alone
 //! cannot reach the external-companion layout, whose stub is meaningless
 //! without its `<name>._` payload — a corpus wired to `unpack_auto` silently
@@ -17,7 +17,7 @@
 //! samples folder is picked up automatically, exactly as it is on disk.
 //!
 //! An input whose bytes carry the il2cpp metadata magic is routed through
-//! [`senbei::metadata::deobfuscate`] instead, giving the method-token remap
+//! [`senbei_metadata::deobfuscate`] instead, giving the method-token remap
 //! real-world coverage (its unit tests only build synthetic layouts).
 //!
 //! The folder is git-ignored (see `senbei/samples/README.md`), so the set of
@@ -116,10 +116,10 @@ fn samples_unpack_against_goldens() {
            }
        };

-        let got = if senbei::metadata::is_metadata(&bytes) {
+        let got = if senbei_metadata::is_metadata(&bytes) {
            // il2cpp metadata: method-token de-obfuscation, no PE pipeline and
            // no integrity check (the output is not a PE image).
-            match senbei::metadata::deobfuscate(&bytes) {
+            match senbei_metadata::deobfuscate(&bytes) {
                Ok((out, _report)) => out,
                Err(e) => {
                    failures.push(format!("{name}: de-obfuscation failed: {e}"));
@@ -140,7 +140,7 @@ fn samples_unpack_against_goldens() {
                },
                None => None,
            };
-            let image = match senbei::job::unpack_bytes(&bytes, companion.as_deref()) {
+            let image = match senbei_io::job::unpack_bytes(&bytes, companion.as_deref()) {
                Ok(img) => img,
                Err(e) => {
                    failures.push(format!("{name}: unpack failed: {e:?}"));
@@ -0,0 +1,10 @@
+[package]
+name = "senbei-crypto"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+description = "Cryptographic and compression primitives for Senbei"
+
+[dependencies]
+aes.workspace = true
+thiserror.workspace = true
@@ -0,0 +1,8 @@
+//! Android container cryptography and decoding primitives.
+
+mod protector;
+
+pub use protector::{
+    ContainerHeader, EncodedSegment, Error, HuffmanLzDecoder, Module9bConfig, ProtectedDescriptor,
+    decode_container, gf32_mul_fixed, transform_segment,
+};
@@ -0,0 +1,712 @@
+//! Cryptographic and compression primitives used by the Android protector.
+
+use aes::Aes256;
+use aes::cipher::{BlockCipherDecrypt, KeyInit};
+
+const RECORD_SIZE: usize = 0x5c;
+
+/// Errors raised while parsing or decoding protector containers.
+#[derive(Debug, thiserror::Error)]
+pub enum Error {
+    #[error("{0}")]
+    Invalid(String),
+}
+
+type Result<T> = std::result::Result<T, Error>;
+
+fn invalid<T>(message: impl Into<String>) -> Result<T> {
+    Err(Error::Invalid(message.into()))
+}
+
+fn range(data: &[u8], offset: usize, size: usize) -> Result<&[u8]> {
+    let end = offset
+        .checked_add(size)
+        .ok_or_else(|| Error::Invalid("byte range overflow".to_owned()))?;
+    data.get(offset..end).ok_or_else(|| {
+        Error::Invalid(format!(
+            "byte range 0x{offset:x}..0x{end:x} is out of bounds"
+        ))
+    })
+}
+
+fn read_u16(data: &[u8], offset: usize) -> Result<u16> {
+    let bytes: [u8; 2] = range(data, offset, 2)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid u16 range".to_owned()))?;
+    Ok(u16::from_le_bytes(bytes))
+}
+
+fn read_u32(data: &[u8], offset: usize) -> Result<u32> {
+    let bytes: [u8; 4] = range(data, offset, 4)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid u32 range".to_owned()))?;
+    Ok(u32::from_le_bytes(bytes))
+}
+
+fn align_up(value: usize, alignment: usize) -> Result<usize> {
+    let mask = alignment
+        .checked_sub(1)
+        .ok_or_else(|| Error::Invalid("zero alignment".to_owned()))?;
+    value
+        .checked_add(mask)
+        .map(|v| v & !mask)
+        .ok_or_else(|| Error::Invalid("alignment overflow".to_owned()))
+}
+
+/// Multiply by the fixed element used by the native GF(2^32) transform.
+#[must_use]
+pub fn gf32_mul_fixed(mut value: u32) -> u32 {
+    let mut multiplier = 0x9451_1dd2_u32;
+    let mut result = 0_u32;
+    while multiplier != 0 {
+        if multiplier & 1 != 0 {
+            result ^= value;
+        }
+        let carry = value >> 31;
+        value = value.wrapping_shl(1);
+        if carry != 0 {
+            value ^= 0x5793_57eb;
+        }
+        multiplier >>= 1;
+    }
+    result
+}
+
+fn mix_columns(block: [u8; 16]) -> [u8; 16] {
+    const fn xtime(value: u8) -> u8 {
+        (value << 1) ^ if value & 0x80 != 0 { 0x1b } else { 0 }
+    }
+
+    let mut output = [0_u8; 16];
+    for offset in (0..16).step_by(4) {
+        let [a, b, c, d] = block[offset..offset + 4] else {
+            unreachable!("fixed four-byte AES column")
+        };
+        output[offset] = xtime(a) ^ (xtime(b) ^ b) ^ c ^ d;
+        output[offset + 1] = a ^ xtime(b) ^ (xtime(c) ^ c) ^ d;
+        output[offset + 2] = a ^ b ^ xtime(c) ^ (xtime(d) ^ d);
+        output[offset + 3] = (xtime(a) ^ a) ^ b ^ c ^ xtime(d);
+    }
+    output
+}
+
+/// Static configuration recovered from module `0x9B`.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct Module9bConfig {
+    pub header_seed: u32,
+    pub container_seed: u32,
+    pub aes_key: [u8; 32],
+    pub skip_aes: bool,
+    pub schedule_offset: usize,
+}
+
+impl Module9bConfig {
+    /// Parse the unique AES-256 decryption schedule and adjacent configuration.
+    pub fn parse(image: &[u8]) -> Result<Self> {
+        Self::parse_inner(image, true)
+    }
+
+    /// Parse the decoder configuration embedded in the raw Stage 2 image.
+    ///
+    /// The embedded decoder ends before the interpreter-only `skip_aes`
+    /// field, so that flag is definitionally false for this layout.
+    pub fn parse_embedded(image: &[u8]) -> Result<Self> {
+        Self::parse_inner(image, false)
+    }
+
+    fn parse_inner(image: &[u8], has_skip_aes: bool) -> Result<Self> {
+        const MARKER: [u8; 4] = [0x00, 0x01, 0x0e, 0x00];
+        let mut matches = image
+            .windows(MARKER.len())
+            .enumerate()
+            .filter_map(|(offset, bytes)| (bytes == MARKER).then_some(offset));
+        let schedule_offset = matches
+            .next()
+            .ok_or_else(|| Error::Invalid("cannot locate the 0x9B AES-256 schedule".to_owned()))?;
+        if schedule_offset < 8 || matches.next().is_some() {
+            return invalid("cannot uniquely locate the 0x9B AES-256 schedule");
+        }
+
+        let header_seed = read_u32(image, schedule_offset - 8)?;
+        let schedule_size = read_u32(image, schedule_offset - 4)?;
+        if !matches!(schedule_size, 0 | 0xf4) {
+            return invalid(format!(
+                "unexpected 0x9B AES schedule size 0x{schedule_size:x}"
+            ));
+        }
+        let bits = read_u16(image, schedule_offset)?;
+        let rounds = read_u16(image, schedule_offset + 2)?;
+        if (bits, rounds) != (0x100, 14) {
+            return invalid(format!(
+                "unexpected AES schedule header 0x{bits:x}/{rounds}"
+            ));
+        }
+
+        let schedule = range(image, schedule_offset + 4, 15 * 16)?;
+        let mut round_keys = [[0_u8; 16]; 15];
+        for (round, output) in round_keys.iter_mut().enumerate() {
+            let source = &schedule[round * 16..round * 16 + 16];
+            for word in 0..4 {
+                let start = word * 4;
+                for byte in 0..4 {
+                    output[start + byte] = source[start + 3 - byte];
+                }
+            }
+        }
+        let mut aes_key = [0_u8; 32];
+        aes_key[..16].copy_from_slice(&round_keys[14]);
+        aes_key[16..].copy_from_slice(&mix_columns(round_keys[13]));
+
+        let container_seed_offset = schedule_offset
+            .checked_add(0x100)
+            .ok_or_else(|| Error::Invalid("container seed offset overflow".to_owned()))?;
+        let skip_aes = if has_skip_aes {
+            let skip_aes_offset = schedule_offset
+                .checked_add(0x240)
+                .ok_or_else(|| Error::Invalid("skip-AES offset overflow".to_owned()))?;
+            *image.get(skip_aes_offset).ok_or_else(|| {
+                Error::Invalid("module static configuration exceeds its image".to_owned())
+            })? != 0
+        } else {
+            false
+        };
+
+        Ok(Self {
+            header_seed,
+            container_seed: if has_skip_aes {
+                read_u32(image, container_seed_offset)?
+            } else {
+                header_seed
+            },
+            aes_key,
+            skip_aes,
+            schedule_offset,
+        })
+    }
+}
+
+/// Decrypted header at the start of direct-data object `0x9D`.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct ProtectedDescriptor {
+    pub command_id: u32,
+    pub flags: u32,
+    pub outer_offset: u32,
+    pub outer_expected_size: u32,
+    pub auxiliary_offset: u32,
+    pub auxiliary_expected_size: u32,
+}
+
+impl ProtectedDescriptor {
+    /// Decrypt the `0x5c`-byte descriptor with the module header seed.
+    pub fn decrypt(data: &[u8], seed: u32) -> Result<Self> {
+        if data.len() < RECORD_SIZE {
+            return invalid("0x9D descriptor is truncated");
+        }
+        let base0 = seed.wrapping_add(0xd3e8_7144).wrapping_mul(seed);
+        let base1 = base0.wrapping_add(seed.wrapping_mul(0x0bd9_418d));
+        let mut words = [0_u32; RECORD_SIZE / 4];
+        for (index, word) in words.iter_mut().enumerate() {
+            let cipher = read_u32(data, index * 4)?;
+            let subtractor = base0.wrapping_shl(if index & 1 != 0 { 4 } else { 0 });
+            *word = cipher.wrapping_sub(subtractor)
+                ^ base1.wrapping_shr((seed.wrapping_add((index as u32).wrapping_mul(4))) & 7);
+        }
+        if words[6..].iter().any(|&word| word != 0) {
+            return invalid("unexpected nonzero reserved words in the 0x9D descriptor");
+        }
+        let descriptor = Self {
+            command_id: words[0],
+            flags: words[1],
+            outer_offset: words[2],
+            outer_expected_size: words[3],
+            auxiliary_offset: words[4],
+            auxiliary_expected_size: words[5],
+        };
+        if descriptor.command_id != 0x9d || descriptor.outer_offset as usize != RECORD_SIZE {
+            return invalid("unexpected decrypted 0x9D descriptor");
+        }
+        Ok(descriptor)
+    }
+}
+
+/// One encrypted segment in a decoded `0x9D` container header.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct EncodedSegment {
+    pub offset: u32,
+    pub size: u32,
+}
+
+/// Parsed primary or auxiliary `0x9D` container.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ContainerHeader {
+    pub start: usize,
+    pub output_size: u32,
+    pub skip_aes: bool,
+    pub tree: Vec<u8>,
+    pub segments: Vec<EncodedSegment>,
+}
+
+impl ContainerHeader {
+    /// Parse and decrypt a container header, Huffman tree, and segment table.
+    pub fn parse(data: &[u8], start: usize, seed: u32) -> Result<Self> {
+        range(data, start, 12)?;
+        let seed_square = seed.wrapping_mul(seed);
+        let state = seed_square.wrapping_shr(17) ^ seed_square.wrapping_shl(11);
+        let raw0 = read_u32(data, start)?;
+        let raw1 = read_u32(data, start + 4)?;
+        let raw2 = read_u32(data, start + 8)?;
+        let output_size = 0xa21d_fb3a_u32
+            .wrapping_shl(state & 7)
+            .wrapping_add(state.wrapping_mul(0xf87b_337c))
+            .wrapping_add(gf32_mul_fixed(raw0));
+        let flag_word = gf32_mul_fixed(raw1)
+            ^ state
+                .wrapping_add(0xbd19_c63c)
+                .wrapping_add(0x416e_2af2_u32.wrapping_shr(state & 0x0d));
+        let segment_count = (flag_word & 0xff) as usize;
+        let skip_aes = (flag_word >> 8) & 0xff == 1;
+        let tree_size = 0x643a_3a3b_u32
+            .wrapping_shl(state & 0x0b)
+            .wrapping_sub(state ^ 0x3b2b_f538)
+            .wrapping_add(gf32_mul_fixed(raw2)) as usize;
+        if segment_count == 0 || tree_size > 0x1b00 {
+            return invalid(format!(
+                "invalid container fields: segments={segment_count}, tree=0x{tree_size:x}"
+            ));
+        }
+
+        let tree_start = start
+            .checked_add(12)
+            .ok_or_else(|| Error::Invalid("tree offset overflow".to_owned()))?;
+        let mut tree = range(data, tree_start, tree_size)?.to_vec();
+        for offset in (0..tree_size & !3).step_by(4) {
+            let value = read_u32(&tree, offset)?;
+            tree[offset..offset + 4].copy_from_slice(&gf32_mul_fixed(value).to_le_bytes());
+        }
+        let tree_state = state.wrapping_add(0xf1cb_5b81).wrapping_mul(state);
+        let tree_delta = tree_state.wrapping_sub(0x23b3_2203_u32.wrapping_mul(state));
+        for (index, byte) in tree.iter_mut().enumerate() {
+            let shift = u32::try_from(index & 0x1b)
+                .map_err(|_| Error::Invalid("tree shift conversion failed".to_owned()))?;
+            let left = gf32_mul_fixed(tree_state.wrapping_shl(shift));
+            let right = tree_delta.wrapping_shr((index & 0x17) as u32);
+            let adjustment = left.wrapping_sub(right).wrapping_shr((index & 0x1f) as u32);
+            *byte = byte.wrapping_add(adjustment as u8);
+        }
+
+        let table_start = start
+            .checked_add(align_up(12 + tree_size, 4)?)
+            .ok_or_else(|| Error::Invalid("segment table offset overflow".to_owned()))?;
+        let table_size = segment_count
+            .checked_mul(8)
+            .ok_or_else(|| Error::Invalid("segment table size overflow".to_owned()))?;
+        let mut table = range(data, table_start, table_size)?.to_vec();
+        let table_state = state.wrapping_add(0xb31f_451c).wrapping_mul(state);
+        let table_xor = table_state.wrapping_shl(3);
+        let table_add = table_state.wrapping_sub(0x822f_e82d_u32.wrapping_mul(state));
+        for offset in (0..table_size).step_by(4) {
+            let value = read_u32(&table, offset)?;
+            let decoded = gf32_mul_fixed(value ^ table_xor)
+                .wrapping_add(table_add.wrapping_shr(((offset & 7) + 5) as u32));
+            table[offset..offset + 4].copy_from_slice(&decoded.to_le_bytes());
+        }
+        let mut segments = Vec::with_capacity(segment_count);
+        for index in 0..segment_count {
+            let offset = read_u32(&table, index * 8)?;
+            let size = read_u32(&table, index * 8 + 4)?;
+            let absolute = start
+                .checked_add(offset as usize)
+                .and_then(|value| value.checked_add(size as usize));
+            if size == 0 || absolute.is_none_or(|end| end > data.len()) {
+                return invalid(format!("container segment {index} lies outside 0x9D"));
+            }
+            segments.push(EncodedSegment { offset, size });
+        }
+        Ok(Self {
+            start,
+            output_size,
+            skip_aes,
+            tree,
+            segments,
+        })
+    }
+
+    /// End offset of the furthest encrypted segment.
+    pub fn encoded_end(&self) -> Result<usize> {
+        self.segments
+            .iter()
+            .map(|segment| {
+                self.start
+                    .checked_add(segment.offset as usize)
+                    .and_then(|value| value.checked_add(segment.size as usize))
+                    .ok_or_else(|| Error::Invalid("encoded segment end overflow".to_owned()))
+            })
+            .collect::<Result<Vec<_>>>()?
+            .into_iter()
+            .max()
+            .ok_or_else(|| Error::Invalid("container has no encoded segments".to_owned()))
+    }
+}
+
+/// Decoder for the protector's Huffman/LZ writer streams.
+#[derive(Debug, Clone)]
+pub struct HuffmanLzDecoder {
+    tree: Vec<u8>,
+    lookup_symbols: Vec<u16>,
+    lookup_bits: Vec<u8>,
+}
+
+impl HuffmanLzDecoder {
+    /// Build the full 16-bit prefix lookup used by the static decoder.
+    pub fn new(tree: &[u8]) -> Result<Self> {
+        if tree.len() < 256 * 3 || !tree.len().is_multiple_of(3) {
+            return invalid(format!("invalid Huffman tree size 0x{:x}", tree.len()));
+        }
+        let mut result = Self {
+            tree: tree.to_vec(),
+            lookup_symbols: vec![0; 0x1_0000],
+            lookup_bits: vec![0; 0x1_0000],
+        };
+        for word in 0..0x1_0000_u32 {
+            let (symbol, bits) = result.decode_symbol(word)?;
+            if bits <= 16 {
+                result.lookup_symbols[word as usize] = symbol;
+                result.lookup_bits[word as usize] = bits;
+            }
+        }
+        Ok(result)
+    }
+
+    fn entry(&self, index: usize) -> Result<(u16, bool, u8)> {
+        let offset = index
+            .checked_mul(3)
+            .ok_or_else(|| Error::Invalid("Huffman node offset overflow".to_owned()))?;
+        let bytes = range(&self.tree, offset, 3)?;
+        let raw = u16::from(bytes[0]) | (u16::from(bytes[1]) << 8);
+        Ok((raw & 0x7fff, raw & 0x8000 != 0, bytes[2]))
+    }
+
+    fn decode_symbol(&self, word: u32) -> Result<(u16, u8)> {
+        let (mut value, leaf, extra) = self.entry((word & 0xff) as usize)?;
+        if leaf {
+            if extra == 0 {
+                return invalid("zero-width Huffman leaf");
+            }
+            return Ok((value, extra));
+        }
+        let mut bits = extra
+            .checked_add(1)
+            .ok_or_else(|| Error::Invalid("Huffman bit count overflow".to_owned()))?;
+        let mut mask = 1_u32.wrapping_shl(u32::from(extra));
+        loop {
+            let branch = usize::from(word & mask != 0);
+            let (next, is_leaf, _) = self.entry(usize::from(value) + branch)?;
+            value = next;
+            if is_leaf {
+                return Ok((value, bits));
+            }
+            mask = mask.wrapping_shl(1);
+            bits = bits
+                .checked_add(1)
+                .ok_or_else(|| Error::Invalid("Huffman bit count overflow".to_owned()))?;
+            if bits > 31 {
+                return invalid("Huffman code exceeds the native 32-bit window");
+            }
+        }
+    }
+
+    /// Decode one compressed writer payload to its exact expected size.
+    pub fn decode(&self, source: &[u8], output_size: usize) -> Result<Vec<u8>> {
+        let mut output = vec![0_u8; output_size];
+        let mut source_pos = 0_usize;
+        let mut bit_buffer = 0_u64;
+        let mut available = 0_u8;
+        let mut consumed_bits = 0_usize;
+        let mut output_pos = 0_usize;
+        let mut prefix = 0_usize;
+
+        while output_pos < output_size {
+            while available < 24 && source_pos < source.len() {
+                bit_buffer |= u64::from(source[source_pos]) << available;
+                source_pos += 1;
+                available += 8;
+            }
+            let key = (bit_buffer & 0xffff) as usize;
+            let mut bits = self.lookup_bits[key];
+            let symbol = if bits != 0 {
+                self.lookup_symbols[key]
+            } else {
+                let mut value_offset = ((bit_buffer & 0xff) as usize) * 3;
+                let mut node = range(&self.tree, value_offset, 3)?;
+                let mut raw = u16::from(node[0]) | (u16::from(node[1]) << 8);
+                if raw & 0x8000 != 0 {
+                    bits = node[2];
+                    raw & 0x7fff
+                } else {
+                    let extra = node[2];
+                    bits = extra + 1;
+                    let mut mask = 1_u64 << extra;
+                    loop {
+                        let branch = usize::from(bit_buffer & mask != 0);
+                        let index = usize::from(raw & 0x7fff) + branch;
+                        value_offset = index
+                            .checked_mul(3)
+                            .ok_or_else(|| Error::Invalid("Huffman node overflow".to_owned()))?;
+                        node = range(&self.tree, value_offset, 3)?;
+                        raw = u16::from(node[0]) | (u16::from(node[1]) << 8);
+                        if raw & 0x8000 != 0 {
+                            break raw & 0x7fff;
+                        }
+                        mask <<= 1;
+                        bits += 1;
+                    }
+                }
+            };
+            if bits == 0 || bits > available {
+                return invalid("compressed stream ends inside a Huffman code");
+            }
+            bit_buffer >>= bits;
+            available -= bits;
+            consumed_bits = consumed_bits
+                .checked_add(usize::from(bits))
+                .ok_or_else(|| Error::Invalid("consumed bit count overflow".to_owned()))?;
+
+            let kind = symbol & 0x300;
+            let value = usize::from(symbol & 0xff);
+            match kind {
+                0 => {
+                    output[output_pos] = value as u8;
+                    output_pos += 1;
+                }
+                0x100 => {
+                    if prefix > 0xff {
+                        return invalid("compressed prefix exceeds 16 bits");
+                    }
+                    prefix = if prefix == 0 {
+                        value
+                    } else {
+                        value | (prefix << 8)
+                    };
+                }
+                0x200 => {
+                    if prefix == 0 {
+                        prefix = 1;
+                    }
+                    let count = value
+                        .checked_mul(prefix)
+                        .ok_or_else(|| Error::Invalid("repeat count overflow".to_owned()))?;
+                    if !matches!(value, 1 | 2 | 4)
+                        || value > output_pos
+                        || output_pos
+                            .checked_add(count)
+                            .is_none_or(|end| end > output_size)
+                    {
+                        return invalid("invalid compressed repeated-pattern command");
+                    }
+                    let pattern = output[output_pos - value..output_pos].to_vec();
+                    for chunk in output[output_pos..output_pos + count].chunks_exact_mut(value) {
+                        chunk.copy_from_slice(&pattern);
+                    }
+                    output_pos += count;
+                    prefix = 0;
+                }
+                0x300 => {
+                    let length = value;
+                    let distance = prefix.checked_add(length).ok_or_else(|| {
+                        Error::Invalid("back-reference distance overflow".to_owned())
+                    })?;
+                    if distance > output_pos
+                        || output_pos
+                            .checked_add(length)
+                            .is_none_or(|end| end > output_size)
+                    {
+                        return invalid("invalid compressed back-reference");
+                    }
+                    let source_start = output_pos - distance;
+                    output.copy_within(source_start..source_start + length, output_pos);
+                    output_pos += length;
+                    prefix = 0;
+                }
+                _ => unreachable!("masked Huffman symbol kind"),
+            }
+        }
+        if consumed_bits.div_ceil(8) != source.len() {
+            return invalid(format!(
+                "compressed input consumption mismatch: used=0x{:x}, size=0x{:x}",
+                consumed_bits.div_ceil(8),
+                source.len()
+            ));
+        }
+        Ok(output)
+    }
+}
+
+/// Apply the native word transform and optional AES-256-CBC decryption.
+pub fn transform_segment(
+    data: &[u8],
+    seed: u32,
+    aes_key: &[u8; 32],
+    decrypt_aes: bool,
+) -> Result<Vec<u8>> {
+    let mut transformed = data.to_vec();
+    let mut state = seed;
+    let mut left = 0xe34e_ac63_u32;
+    let mut right = 0x07b4_8238_u32;
+    for (index, chunk) in transformed.as_chunks_mut::<4>().0.iter_mut().enumerate() {
+        let index32 = u32::try_from(index)
+            .map_err(|_| Error::Invalid("segment word index exceeds u32".to_owned()))?;
+        left = state
+            .wrapping_add(0x72f6_fcbe)
+            .wrapping_add(left.wrapping_add(0x4f8b_1bca).wrapping_mul(left))
+            .wrapping_shr(index32.wrapping_mul(index32) & 0x0f);
+        right = state
+            .wrapping_sub(0x71b6_a98d)
+            .wrapping_add(right.wrapping_sub(0x1605_a81c).wrapping_mul(right))
+            .wrapping_shl(index32 & 7);
+        state = left ^ right;
+        let mut value = u32::from_le_bytes(*chunk);
+        value = value.wrapping_add(0xb43b_9baf_u32.wrapping_mul(index32 & 0x0d));
+        value ^= 0xaf57_f7fb_u32.wrapping_mul(index32 & 3);
+        value = value.wrapping_sub(state) ^ state;
+        chunk.copy_from_slice(&value.to_le_bytes());
+    }
+
+    if decrypt_aes {
+        let cipher = Aes256::new_from_slice(aes_key)
+            .map_err(|_| Error::Invalid("invalid AES-256 key length".to_owned()))?;
+        let mut previous = [0_u8; 16];
+        for chunk in transformed.as_chunks_mut::<16>().0 {
+            let ciphertext = *chunk;
+            cipher.decrypt_block((&mut *chunk).into());
+            for (byte, prior) in chunk.iter_mut().zip(previous) {
+                *byte ^= prior;
+            }
+            previous = ciphertext;
+        }
+    }
+    Ok(transformed)
+}
+
+/// Decode one complete protector container into its flat output buffer.
+///
+/// This is the static equivalent of the decoder entrypoint embedded in Stage
+/// 2 and in each nested interpreter module.
+pub fn decode_container(
+    data: &[u8],
+    config: &Module9bConfig,
+    expected_size: usize,
+) -> Result<Vec<u8>> {
+    let header = ContainerHeader::parse(data, 0, config.container_seed)?;
+    let header_size = usize::try_from(header.output_size)
+        .map_err(|_| Error::Invalid("container output size exceeds usize".to_owned()))?;
+    if header_size != expected_size {
+        return invalid(format!(
+            "container output size 0x{header_size:x} != expected 0x{expected_size:x}"
+        ));
+    }
+    let decoder = HuffmanLzDecoder::new(&header.tree)?;
+    let decrypt_aes = !(config.skip_aes || header.skip_aes);
+    let mut output = vec![0_u8; expected_size];
+
+    for (segment_index, encoded) in header.segments.iter().enumerate() {
+        let start = header
+            .start
+            .checked_add(encoded.offset as usize)
+            .ok_or_else(|| Error::Invalid("encoded segment start overflow".to_owned()))?;
+        let encoded_data = range(data, start, encoded.size as usize)?;
+        let transformed = transform_segment(
+            encoded_data,
+            config.container_seed,
+            &config.aes_key,
+            decrypt_aes,
+        )?;
+        if transformed.len() < 16 {
+            return invalid(format!(
+                "decoded segment {segment_index} is shorter than its header"
+            ));
+        }
+        let base_offset = read_u32(&transformed, 0)? as usize;
+        let writer_count = read_u32(&transformed, 4)? as usize;
+        let table_offset = read_u32(&transformed, 8)? as usize;
+        let data_offset = read_u32(&transformed, 12)? as usize;
+        let table_size = writer_count
+            .checked_mul(16)
+            .ok_or_else(|| Error::Invalid("writer table size overflow".to_owned()))?;
+        let table_end = table_offset
+            .checked_add(table_size)
+            .ok_or_else(|| Error::Invalid("writer table end overflow".to_owned()))?;
+        if table_end > transformed.len() || data_offset > transformed.len() {
+            return invalid(format!(
+                "decoded segment {segment_index} has invalid writer offsets"
+            ));
+        }
+
+        let mut data_cursor = data_offset;
+        for writer_index in 0..writer_count {
+            let record =
+                table_offset
+                    .checked_add(writer_index.checked_mul(16).ok_or_else(|| {
+                        Error::Invalid("writer record offset overflow".to_owned())
+                    })?)
+                    .ok_or_else(|| Error::Invalid("writer record offset overflow".to_owned()))?;
+            let output_offset = read_u32(&transformed, record)? as usize;
+            let output_size = read_u32(&transformed, record + 4)? as usize;
+            let encoded_size = read_u32(&transformed, record + 8)? as usize;
+            let reserved = read_u32(&transformed, record + 12)?;
+            let encoded_end = data_cursor
+                .checked_add(encoded_size)
+                .ok_or_else(|| Error::Invalid("writer data end overflow".to_owned()))?;
+            if reserved != 0 || encoded_end > transformed.len() {
+                return invalid(format!(
+                    "segment {segment_index} writer {writer_index} has invalid bounds"
+                ));
+            }
+            let source = &transformed[data_cursor..encoded_end];
+            let decoded = if encoded_size == output_size {
+                None
+            } else {
+                Some(decoder.decode(source, output_size)?)
+            };
+            let decoded = decoded.as_deref().unwrap_or(source);
+            let target = base_offset
+                .checked_add(output_offset)
+                .ok_or_else(|| Error::Invalid("writer target offset overflow".to_owned()))?;
+            let target_end = target
+                .checked_add(decoded.len())
+                .ok_or_else(|| Error::Invalid("writer target end overflow".to_owned()))?;
+            let destination = output.get_mut(target..target_end).ok_or_else(|| {
+                Error::Invalid(format!(
+                    "segment {segment_index} writer {writer_index} target is out of range"
+                ))
+            })?;
+            destination.copy_from_slice(decoded);
+            data_cursor = encoded_end;
+        }
+    }
+    Ok(output)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn aes_mix_columns_matches_fips_example() {
+        let input = [
+            0xdb, 0x13, 0x53, 0x45, 0xf2, 0x0a, 0x22, 0x5c, 0x01, 0x01, 0x01, 0x01, 0xc6, 0xc6,
+            0xc6, 0xc6,
+        ];
+        assert_eq!(
+            mix_columns(input),
+            [
+                0x8e, 0x4d, 0xa1, 0xbc, 0x9f, 0xdc, 0x58, 0x9d, 0x01, 0x01, 0x01, 0x01, 0xc6, 0xc6,
+                0xc6, 0xc6,
+            ]
+        );
+    }
+
+    #[test]
+    fn descriptor_rejects_truncated_input() {
+        assert!(ProtectedDescriptor::decrypt(&[0_u8; 16], 1).is_err());
+    }
+}
@@ -0,0 +1,20 @@
+//! Cryptographic and compression primitives for the supported protection
+//! formats.
+
+pub mod android;
+pub mod windows;
+
+// Keep the historical flat paths available to downstream callers while the
+// implementations themselves live under their platform boundary.
+pub use windows::{BufferOperation, DecompressionFailure, Error, MAX_IMAGE_SIZE};
+pub use windows::{bytecode, crc32, primitives};
+
+/// Lowercase hexadecimal representation for digest and diagnostic bytes.
+#[must_use]
+pub fn hex_digest(data: &[u8]) -> String {
+    let mut output = String::with_capacity(data.len() * 2);
+    for byte in data {
+        output.push_str(&format!("{byte:02x}"));
+    }
+    output
+}
@@ -61,8 +61,8 @@ impl OpsLut {
 pub fn generate(data: &[u8], offset: u32) -> Option<Vec<Op>> {
    // Bounds-checked cursor: a corrupt `data_offset` (bad decrypt_data6 / the
    // alignment fallback) must yield `None`, not an out-of-bounds panic — the
-    // panic path would surface as a misleading `UnpackError::Corrupt` instead
-    // of the precise `BytecodeGenFailed`, and any future caller without a
+    // panic path would surface as a misleading `UnpackError::InternalPanic` instead
+    // of the precise `BytecodeGenerationFailed`, and any future caller without a
    // `catch_unwind` wrapper would abort outright.
    let mut pos = offset as usize;
    let mut next = move || {
@@ -0,0 +1,77 @@
+//! Windows PE protection primitives.
+
+pub mod bytecode;
+pub mod crc32;
+pub mod primitives;
+mod tables;
+
+/// Maximum buffer size accepted by allocation-sensitive PE transforms.
+pub const MAX_IMAGE_SIZE: u64 = 1 << 30;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum BufferOperation {
+    Read,
+    CopySource,
+    CopyDestination,
+    ZeroFill,
+}
+
+impl std::fmt::Display for BufferOperation {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::Read => "read",
+            Self::CopySource => "copy source",
+            Self::CopyDestination => "copy destination",
+            Self::ZeroFill => "zero-fill",
+        })
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)]
+pub enum Error {
+    #[error(
+        "{operation} range out of bounds (offset {offset}, size {size}, buffer length {buffer_len})"
+    )]
+    BufferRangeOutOfBounds {
+        operation: BufferOperation,
+        offset: usize,
+        size: usize,
+        buffer_len: usize,
+    },
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, thiserror::Error)]
+#[non_exhaustive]
+pub enum DecompressionFailure {
+    #[error("compressed source size {size} exceeds limit {max}")]
+    SourceTooLarge { size: u32, max: u64 },
+    #[error("Huffman code length {bits} is invalid")]
+    InvalidCodeLength { bits: u8 },
+    #[error("Huffman tree traversal exceeded 64 levels")]
+    HuffmanTraversalLimit,
+    #[error("pending length accumulator overflowed at {pending}")]
+    PendingLengthOverflow { pending: u32 },
+    #[error("output step {step} at byte {written} exceeds expected size {expected}")]
+    OutputOverflow {
+        written: u32,
+        step: u32,
+        expected: u32,
+    },
+    #[error("run-fill width {width} reads before output offset 0x{destination:08X}")]
+    RunFillBeforeOutput { width: u32, destination: u32 },
+    #[error("run-fill width {width} is unsupported")]
+    InvalidRunFillWidth { width: u32 },
+    #[error("back-reference distance {distance} exceeds {written} written bytes")]
+    InvalidBackReference { distance: u32, written: u32 },
+    #[error("Huffman symbol consumed no input and produced no output")]
+    NoProgress,
+    #[error(
+        "output size mismatch (wrote {written}/{expected} bytes after consuming {consumed}/{source_size})"
+    )]
+    OutputSizeMismatch {
+        written: u32,
+        expected: u32,
+        consumed: u32,
+        source_size: u32,
+    },
+}
@@ -146,11 +146,11 @@ mod tests {
    #[test]
    fn generated_tables_match_committed_bytes() {
        assert_eq!(COLUMMIX1.len(), 1024);
-        assert_eq!(super::super::crc32::compute(&COLUMMIX1), 0x7e8d_5d5f);
-        assert_eq!(super::super::crc32::compute(&COLUMMIX2), 0xfcc4_acfc);
-        assert_eq!(super::super::crc32::compute(&COLUMMIX3), 0x637a_f0cd);
-        assert_eq!(super::super::crc32::compute(&COLUMMIX4), 0x1e7b_c381);
-        assert_eq!(super::super::crc32::compute(&SBOX), 0x10fd_6dc1);
+        assert_eq!(crate::crc32::compute(&COLUMMIX1), 0x7e8d_5d5f);
+        assert_eq!(crate::crc32::compute(&COLUMMIX2), 0xfcc4_acfc);
+        assert_eq!(crate::crc32::compute(&COLUMMIX3), 0x637a_f0cd);
+        assert_eq!(crate::crc32::compute(&COLUMMIX4), 0x1e7b_c381);
+        assert_eq!(crate::crc32::compute(&SBOX), 0x10fd_6dc1);
        // Spot-check the first dword of each (matches the original first row).
        assert_eq!(&COLUMMIX1[..4], &[0x50, 0xa7, 0xf4, 0x51]);
        assert_eq!(&COLUMMIX2[..4], &[0xa7, 0xf4, 0x51, 0x50]);
@@ -0,0 +1,14 @@
+[package]
+name = "senbei-elf"
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+description = "ELF format parsing and structural utilities for Senbei"
+
+[dependencies]
+goblin.workspace = true
+thiserror.workspace = true
+
+[lints]
+workspace = true
@@ -0,0 +1,107 @@
+use crate::{Error, Result, invalid};
+
+#[must_use]
+pub fn elf_hash(name: &[u8]) -> u32 {
+    let mut value = 0_u32;
+    for &byte in name {
+        value = value.wrapping_shl(4).wrapping_add(u32::from(byte));
+        let high = value & 0xf000_0000;
+        if high != 0 {
+            value ^= high >> 24;
+            value &= !high;
+        }
+    }
+    value
+}
+
+#[must_use]
+pub fn gnu_hash(name: &[u8]) -> u32 {
+    name.iter().fold(5381_u32, |value, &byte| {
+        value.wrapping_mul(33).wrapping_add(u32::from(byte))
+    })
+}
+
+pub fn build_sysv_hash(names: &[Vec<u8>]) -> Result<Vec<u8>> {
+    if names.len() < 2 {
+        return invalid("dynamic symbol table is unexpectedly empty");
+    }
+    let bucket_count = names.len();
+    let symbol_count = names.len();
+    let mut buckets = vec![0_u32; bucket_count];
+    let mut chains = vec![0_u32; symbol_count];
+    for (symbol_index, name) in names.iter().enumerate().skip(1) {
+        let bucket_index = elf_hash(name) as usize % bucket_count;
+        let symbol_index32 = u32::try_from(symbol_index)
+            .map_err(|_| Error::Invalid("dynamic symbol index exceeds u32".to_owned()))?;
+        if buckets[bucket_index] == 0 {
+            buckets[bucket_index] = symbol_index32;
+            continue;
+        }
+        let mut chain_index = buckets[bucket_index] as usize;
+        while chains[chain_index] != 0 {
+            chain_index = chains[chain_index] as usize;
+        }
+        chains[chain_index] = symbol_index32;
+    }
+    let mut output = Vec::with_capacity((2 + bucket_count + symbol_count) * 4);
+    output.extend_from_slice(
+        &u32::try_from(bucket_count)
+            .map_err(|_| Error::Invalid("SysV bucket count exceeds u32".to_owned()))?
+            .to_le_bytes(),
+    );
+    output.extend_from_slice(
+        &u32::try_from(symbol_count)
+            .map_err(|_| Error::Invalid("SysV symbol count exceeds u32".to_owned()))?
+            .to_le_bytes(),
+    );
+    for value in buckets.into_iter().chain(chains) {
+        output.extend_from_slice(&value.to_le_bytes());
+    }
+    Ok(output)
+}
+
+pub fn build_gnu_hash(names: &[Vec<u8>]) -> Result<Vec<u8>> {
+    let hashes = names
+        .iter()
+        .skip(1)
+        .map(|name| gnu_hash(name))
+        .collect::<Vec<_>>();
+    if hashes.is_empty() {
+        return invalid("GNU hash requires at least one dynamic symbol");
+    }
+    let bloom_shift = 5_u32;
+    let mut bloom_word = 0_u64;
+    for &value in &hashes {
+        bloom_word |= 1_u64 << (value & 63);
+        bloom_word |= 1_u64 << ((value >> bloom_shift) & 63);
+    }
+    let mut chains = hashes
+        .into_iter()
+        .map(|value| value & !1)
+        .collect::<Vec<_>>();
+    let last = chains
+        .last_mut()
+        .ok_or_else(|| Error::Invalid("GNU hash chain is empty".to_owned()))?;
+    *last |= 1;
+    let mut output = Vec::with_capacity(28 + chains.len() * 4);
+    for value in [1_u32, 1, 1, bloom_shift] {
+        output.extend_from_slice(&value.to_le_bytes());
+    }
+    output.extend_from_slice(&bloom_word.to_le_bytes());
+    output.extend_from_slice(&1_u32.to_le_bytes());
+    for value in chains {
+        output.extend_from_slice(&value.to_le_bytes());
+    }
+    Ok(output)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn standard_elf_hash_is_stable() {
+        assert_eq!(elf_hash(b"printf"), 0x0779_05a6);
+        assert_eq!(gnu_hash(b"printf"), 0x156b_2bb8);
+    }
+}
@@ -0,0 +1,636 @@
+use crate::{Error, Result, invalid};
+
+pub const SHT_NOBITS: u32 = 8;
+pub const SHT_STRTAB: u32 = 3;
+pub const SHT_LOUSER: u32 = 0x8000_0000;
+pub const SHF_ALLOC: u64 = 2;
+const PT_LOAD: u32 = 1;
+pub const PF_R: u32 = 4;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct LoadSegment {
+    pub offset: u64,
+    pub virtual_address: u64,
+    pub file_size: u64,
+    pub memory_size: u64,
+    pub flags: u32,
+    pub alignment: u64,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct SectionHeader {
+    pub name: u32,
+    pub section_type: u32,
+    pub flags: u64,
+    pub address: u64,
+    pub offset: u64,
+    pub size: u64,
+    pub link: u32,
+    pub info: u32,
+    pub alignment: u64,
+    pub entry_size: u64,
+}
+
+impl SectionHeader {
+    pub const SIZE: usize = 0x40;
+
+    fn parse(data: &[u8], offset: usize) -> Result<Self> {
+        Ok(Self {
+            name: read_u32(data, offset)?,
+            section_type: read_u32(data, offset + 4)?,
+            flags: read_u64(data, offset + 8)?,
+            address: read_u64(data, offset + 0x10)?,
+            offset: read_u64(data, offset + 0x18)?,
+            size: read_u64(data, offset + 0x20)?,
+            link: read_u32(data, offset + 0x28)?,
+            info: read_u32(data, offset + 0x2c)?,
+            alignment: read_u64(data, offset + 0x30)?,
+            entry_size: read_u64(data, offset + 0x38)?,
+        })
+    }
+
+    pub fn encode(self) -> [u8; Self::SIZE] {
+        let mut output = [0_u8; Self::SIZE];
+        output[0..4].copy_from_slice(&self.name.to_le_bytes());
+        output[4..8].copy_from_slice(&self.section_type.to_le_bytes());
+        output[8..0x10].copy_from_slice(&self.flags.to_le_bytes());
+        output[0x10..0x18].copy_from_slice(&self.address.to_le_bytes());
+        output[0x18..0x20].copy_from_slice(&self.offset.to_le_bytes());
+        output[0x20..0x28].copy_from_slice(&self.size.to_le_bytes());
+        output[0x28..0x2c].copy_from_slice(&self.link.to_le_bytes());
+        output[0x2c..0x30].copy_from_slice(&self.info.to_le_bytes());
+        output[0x30..0x38].copy_from_slice(&self.alignment.to_le_bytes());
+        output[0x38..0x40].copy_from_slice(&self.entry_size.to_le_bytes());
+        output
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ElfLayout {
+    pub entrypoint: u64,
+    pub program_header_offset: usize,
+    pub program_header_size: usize,
+    pub program_header_count: usize,
+    pub program_headers: Vec<LoadSegment>,
+    pub section_headers: Vec<SectionHeader>,
+    pub section_name_index: usize,
+    pub private_section_index: usize,
+}
+
+impl ElfLayout {
+    pub fn parse(data: &[u8], require_private: bool) -> Result<Self> {
+        let ident = slice(data, 0, 6)?;
+        if ident[..4] != *b"\x7fELF" || ident[4] != 2 || ident[5] != 1 {
+            return invalid("input is not a little-endian ELF64 file");
+        }
+        if read_u16(data, 0x12)? != crate::AARCH64_MACHINE {
+            return invalid("input is not an AArch64 ELF");
+        }
+        let entrypoint = read_u64(data, 0x18)?;
+        let program_header_offset = usize_from_u64(read_u64(data, 0x20)?, "program header offset")?;
+        let section_header_offset = usize_from_u64(read_u64(data, 0x28)?, "section header offset")?;
+        let program_header_size = usize::from(read_u16(data, 0x36)?);
+        let program_header_count = usize::from(read_u16(data, 0x38)?);
+        let section_header_size = usize::from(read_u16(data, 0x3a)?);
+        let section_header_count = usize::from(read_u16(data, 0x3c)?);
+        let section_name_index = usize::from(read_u16(data, 0x3e)?);
+        if program_header_size != 0x38 || section_header_size != SectionHeader::SIZE {
+            return invalid("unexpected ELF program/section header size");
+        }
+
+        let mut program_headers = Vec::new();
+        for index in 0..program_header_count {
+            let offset = checked_index(program_header_offset, index, program_header_size)?;
+            if read_u32(data, offset)? != PT_LOAD {
+                continue;
+            }
+            let segment = LoadSegment {
+                flags: read_u32(data, offset + 4)?,
+                offset: read_u64(data, offset + 8)?,
+                virtual_address: read_u64(data, offset + 0x10)?,
+                file_size: read_u64(data, offset + 0x20)?,
+                memory_size: read_u64(data, offset + 0x28)?,
+                alignment: read_u64(data, offset + 0x30)?,
+            };
+            let file_end = segment
+                .offset
+                .checked_add(segment.file_size)
+                .ok_or_else(|| Error::Invalid(format!("PT_LOAD {index} file range overflow")))?;
+            if file_end > data.len() as u64 {
+                return invalid(format!("PT_LOAD {index} exceeds input file"));
+            }
+            program_headers.push(segment);
+        }
+        if program_headers.is_empty() {
+            return invalid("input ELF contains no PT_LOAD segments");
+        }
+
+        let mut section_headers = Vec::with_capacity(section_header_count);
+        for index in 0..section_header_count {
+            let offset = checked_index(section_header_offset, index, section_header_size)?;
+            section_headers.push(SectionHeader::parse(data, offset)?);
+        }
+        if section_name_index >= section_headers.len() {
+            return invalid("ELF section-name index is out of range");
+        }
+        let private = section_headers
+            .iter()
+            .enumerate()
+            .filter_map(|(index, section)| (section.section_type == SHT_LOUSER).then_some(index))
+            .collect::<Vec<_>>();
+        let private_section_index = match private.as_slice() {
+            [index] => *index,
+            [] if !require_private => usize::MAX,
+            _ => {
+                return invalid(format!(
+                    "expected {} SHT_LOUSER section, found {}",
+                    if require_private {
+                        "one"
+                    } else {
+                        "at most one"
+                    },
+                    private.len()
+                ));
+            }
+        };
+        let layout = Self {
+            entrypoint,
+            program_header_offset,
+            program_header_size,
+            program_header_count,
+            program_headers,
+            section_headers,
+            section_name_index,
+            private_section_index,
+        };
+        // Section roles are resolved from the ELF's own string table. Validate
+        // it at the format boundary so callers cannot silently continue with
+        // fabricated or lossy section names.
+        layout.section_names(data)?;
+        Ok(layout)
+    }
+
+    pub fn private_section(&self) -> Result<SectionHeader> {
+        self.section_headers
+            .get(self.private_section_index)
+            .copied()
+            .ok_or_else(|| Error::Invalid("ELF has no private section".to_owned()))
+    }
+
+    pub fn load_end(&self) -> Result<u64> {
+        self.program_headers
+            .iter()
+            .map(|segment| {
+                segment
+                    .virtual_address
+                    .checked_add(segment.memory_size)
+                    .ok_or_else(|| Error::Invalid("PT_LOAD memory end overflow".to_owned()))
+            })
+            .collect::<Result<Vec<_>>>()?
+            .into_iter()
+            .max()
+            .ok_or_else(|| Error::Invalid("ELF has no PT_LOAD memory range".to_owned()))
+    }
+
+    pub fn file_load_end(&self) -> Result<u64> {
+        self.program_headers
+            .iter()
+            .map(|segment| {
+                segment
+                    .offset
+                    .checked_add(segment.file_size)
+                    .ok_or_else(|| Error::Invalid("PT_LOAD file end overflow".to_owned()))
+            })
+            .collect::<Result<Vec<_>>>()?
+            .into_iter()
+            .max()
+            .ok_or_else(|| Error::Invalid("ELF has no PT_LOAD file range".to_owned()))
+    }
+
+    pub fn load_alignment(&self) -> Result<u64> {
+        let alignment = self
+            .program_headers
+            .iter()
+            .map(|segment| segment.alignment)
+            .max()
+            .ok_or_else(|| Error::Invalid("ELF has no PT_LOAD alignment".to_owned()))?;
+        if alignment == 0 || !alignment.is_power_of_two() {
+            return invalid(format!("invalid PT_LOAD alignment 0x{alignment:x}"));
+        }
+        Ok(alignment)
+    }
+
+    pub fn append_load_segment(&self, output: &mut [u8], segment: LoadSegment) -> Result<Self> {
+        if self.program_header_size != 0x38 {
+            return invalid("unexpected ELF program header size");
+        }
+        if segment.file_size == 0 {
+            return invalid("new PT_LOAD has no file contents");
+        }
+        if segment.memory_size < segment.file_size {
+            return invalid("new PT_LOAD memory size is smaller than file size");
+        }
+        if segment.alignment == 0 || !segment.alignment.is_power_of_two() {
+            return invalid(format!(
+                "invalid new PT_LOAD alignment 0x{:x}",
+                segment.alignment
+            ));
+        }
+        if segment.offset % segment.alignment != segment.virtual_address % segment.alignment {
+            return invalid("new PT_LOAD offset and address are misaligned");
+        }
+        let segment_file_end = segment
+            .offset
+            .checked_add(segment.file_size)
+            .ok_or_else(|| Error::Invalid("new PT_LOAD file range overflow".to_owned()))?;
+        let segment_memory_end = segment
+            .virtual_address
+            .checked_add(segment.memory_size)
+            .ok_or_else(|| Error::Invalid("new PT_LOAD memory range overflow".to_owned()))?;
+        if segment_file_end > output.len() as u64 {
+            return invalid("new PT_LOAD exceeds output mapping");
+        }
+        for existing in &self.program_headers {
+            let existing_file_end = existing
+                .offset
+                .checked_add(existing.file_size)
+                .ok_or_else(|| Error::Invalid("PT_LOAD file range overflow".to_owned()))?;
+            if segment.offset < existing_file_end && existing.offset < segment_file_end {
+                return invalid("new PT_LOAD overlaps an existing file range");
+            }
+            let existing_memory_end = existing
+                .virtual_address
+                .checked_add(existing.memory_size)
+                .ok_or_else(|| Error::Invalid("PT_LOAD memory range overflow".to_owned()))?;
+            if segment.virtual_address < existing_memory_end
+                && existing.virtual_address < segment_memory_end
+            {
+                return invalid("new PT_LOAD overlaps an existing memory range");
+            }
+        }
+        let new_count = self
+            .program_header_count
+            .checked_add(1)
+            .ok_or_else(|| Error::Invalid("program header count overflow".to_owned()))?;
+        let new_count_u16 = u16::try_from(new_count)
+            .map_err(|_| Error::Invalid("program header count exceeds u16".to_owned()))?;
+        let header_offset = checked_index(
+            self.program_header_offset,
+            self.program_header_count,
+            self.program_header_size,
+        )?;
+        let header_end = header_offset
+            .checked_add(self.program_header_size)
+            .ok_or_else(|| Error::Invalid("new program header range overflow".to_owned()))?;
+        slice(output, header_offset, self.program_header_size)?;
+        let first_file_section = self
+            .section_headers
+            .iter()
+            .filter(|section| section.section_type != SHT_NOBITS && section.size != 0)
+            .map(|section| section.offset)
+            .min();
+        if first_file_section.is_some_and(|offset| header_end as u64 > offset) {
+            return invalid("no space for an additional program header");
+        }
+
+        let mut header = [0_u8; 0x38];
+        header[0..4].copy_from_slice(&PT_LOAD.to_le_bytes());
+        header[4..8].copy_from_slice(&segment.flags.to_le_bytes());
+        header[8..0x10].copy_from_slice(&segment.offset.to_le_bytes());
+        header[0x10..0x18].copy_from_slice(&segment.virtual_address.to_le_bytes());
+        header[0x18..0x20].copy_from_slice(&segment.virtual_address.to_le_bytes());
+        header[0x20..0x28].copy_from_slice(&segment.file_size.to_le_bytes());
+        header[0x28..0x30].copy_from_slice(&segment.memory_size.to_le_bytes());
+        header[0x30..0x38].copy_from_slice(&segment.alignment.to_le_bytes());
+        output
+            .get_mut(header_offset..header_end)
+            .ok_or_else(|| Error::Invalid("new program header exceeds output".to_owned()))?
+            .copy_from_slice(&header);
+        output
+            .get_mut(0x38..0x3a)
+            .ok_or_else(|| Error::Invalid("ELF header is truncated".to_owned()))?
+            .copy_from_slice(&new_count_u16.to_le_bytes());
+
+        let mut updated = self.clone();
+        updated.program_header_count = new_count;
+        updated.program_headers.push(segment);
+        Ok(updated)
+    }
+
+    /// Resolve every section's name from the ELF `shstrtab` section.
+    ///
+    /// The returned names are source data, not role labels supplied by the
+    /// caller. Any malformed string-table reference is an input error.
+    pub fn section_names(&self, data: &[u8]) -> Result<Vec<String>> {
+        let table = self
+            .section_headers
+            .get(self.section_name_index)
+            .copied()
+            .ok_or_else(|| Error::Invalid("ELF section-name index is out of range".to_owned()))?;
+        if table.section_type != SHT_STRTAB {
+            return invalid(format!(
+                "ELF section-name table has unexpected type 0x{:x}",
+                table.section_type
+            ));
+        }
+        let strings = slice_u64(data, table.offset, table.size)?;
+        if strings.is_empty() || strings[0] != 0 {
+            return invalid("ELF section-name table does not start with NUL");
+        }
+        if strings.last().copied() != Some(0) {
+            return invalid("ELF section-name table is not NUL terminated");
+        }
+        self.section_headers
+            .iter()
+            .enumerate()
+            .map(|(index, section)| {
+                let offset = section.name as usize;
+                if offset >= strings.len() {
+                    return invalid(format!(
+                        "ELF section {index} name offset 0x{offset:x} exceeds section-name table"
+                    ));
+                }
+                let end = strings[offset..]
+                    .iter()
+                    .position(|&byte| byte == 0)
+                    .map(|length| offset + length)
+                    .ok_or_else(|| {
+                        Error::Invalid(format!(
+                            "ELF section {index} name at 0x{offset:x} is unterminated"
+                        ))
+                    })?;
+                let name = std::str::from_utf8(&strings[offset..end]).map_err(|error| {
+                    Error::Invalid(format!(
+                        "ELF section {index} name at 0x{offset:x} is not UTF-8: {error}"
+                    ))
+                })?;
+                if index == 0 && section.name != 0 {
+                    return invalid("ELF null section has a nonzero name offset");
+                }
+                Ok(name.to_owned())
+            })
+            .collect()
+    }
+
+    pub fn file_offset_to_virtual_address(&self, offset: u64, size: u64) -> Result<u64> {
+        let end = offset
+            .checked_add(size)
+            .ok_or_else(|| Error::Invalid("file range overflow".to_owned()))?;
+        for segment in &self.program_headers {
+            let segment_end = segment
+                .offset
+                .checked_add(segment.file_size)
+                .ok_or_else(|| Error::Invalid("PT_LOAD file range overflow".to_owned()))?;
+            if segment.offset <= offset && end <= segment_end {
+                return segment
+                    .virtual_address
+                    .checked_add(offset - segment.offset)
+                    .ok_or_else(|| Error::Invalid("virtual address overflow".to_owned()));
+            }
+        }
+        invalid(format!(
+            "file range 0x{offset:x}..0x{end:x} is not in PT_LOAD"
+        ))
+    }
+}
+
+pub fn slice(data: &[u8], offset: usize, size: usize) -> Result<&[u8]> {
+    let end = offset
+        .checked_add(size)
+        .ok_or_else(|| Error::Invalid("byte range overflow".to_owned()))?;
+    data.get(offset..end).ok_or_else(|| {
+        Error::Invalid(format!(
+            "byte range 0x{offset:x}..0x{end:x} is out of bounds"
+        ))
+    })
+}
+
+pub fn slice_u64(data: &[u8], offset: u64, size: u64) -> Result<&[u8]> {
+    slice(
+        data,
+        usize_from_u64(offset, "file offset")?,
+        usize_from_u64(size, "file size")?,
+    )
+}
+
+pub fn read_u16(data: &[u8], offset: usize) -> Result<u16> {
+    let bytes: [u8; 2] = slice(data, offset, 2)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid u16 range".to_owned()))?;
+    Ok(u16::from_le_bytes(bytes))
+}
+
+pub fn read_u32(data: &[u8], offset: usize) -> Result<u32> {
+    let bytes: [u8; 4] = slice(data, offset, 4)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid u32 range".to_owned()))?;
+    Ok(u32::from_le_bytes(bytes))
+}
+
+pub fn read_u64(data: &[u8], offset: usize) -> Result<u64> {
+    let bytes: [u8; 8] = slice(data, offset, 8)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid u64 range".to_owned()))?;
+    Ok(u64::from_le_bytes(bytes))
+}
+
+pub fn read_i64(data: &[u8], offset: usize) -> Result<i64> {
+    let bytes: [u8; 8] = slice(data, offset, 8)?
+        .try_into()
+        .map_err(|_| Error::Invalid("invalid i64 range".to_owned()))?;
+    Ok(i64::from_le_bytes(bytes))
+}
+
+pub fn usize_from_u64(value: u64, field: &str) -> Result<usize> {
+    usize::try_from(value).map_err(|_| Error::Invalid(format!("{field} 0x{value:x} exceeds usize")))
+}
+
+pub fn checked_index(base: usize, index: usize, stride: usize) -> Result<usize> {
+    index
+        .checked_mul(stride)
+        .and_then(|value| base.checked_add(value))
+        .ok_or_else(|| Error::Invalid("table index overflow".to_owned()))
+}
+
+pub fn align_up(value: u64, alignment: u64) -> Result<u64> {
+    if alignment == 0 || !alignment.is_power_of_two() {
+        return invalid(format!("invalid alignment {alignment}"));
+    }
+    value
+        .checked_add(alignment - 1)
+        .map(|aligned| aligned & !(alignment - 1))
+        .ok_or_else(|| Error::Invalid("alignment overflow".to_owned()))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn layout(name_index: u32) -> ElfLayout {
+        ElfLayout {
+            entrypoint: 0,
+            program_header_offset: 0,
+            program_header_size: 0x38,
+            program_header_count: 0,
+            program_headers: Vec::new(),
+            section_headers: vec![
+                SectionHeader {
+                    name: 0,
+                    section_type: 0,
+                    flags: 0,
+                    address: 0,
+                    offset: 0,
+                    size: 0,
+                    link: 0,
+                    info: 0,
+                    alignment: 0,
+                    entry_size: 0,
+                },
+                SectionHeader {
+                    name: name_index,
+                    section_type: 1,
+                    flags: 0,
+                    address: 0,
+                    offset: 0,
+                    size: 0,
+                    link: 0,
+                    info: 0,
+                    alignment: 0,
+                    entry_size: 0,
+                },
+                SectionHeader {
+                    name: 1,
+                    section_type: SHT_STRTAB,
+                    flags: 0,
+                    address: 0,
+                    offset: 0,
+                    size: 8,
+                    link: 0,
+                    info: 0,
+                    alignment: 1,
+                    entry_size: 0,
+                },
+            ],
+            section_name_index: 2,
+            private_section_index: usize::MAX,
+        }
+    }
+
+    #[test]
+    fn section_names_resolve_from_elf_string_table() {
+        let names = layout(1)
+            .section_names(b"\0text\0\0\0")
+            .expect("valid names");
+        assert_eq!(names, ["", "text", "text"]);
+    }
+
+    #[test]
+    fn section_names_reject_out_of_range_name_offsets() {
+        let error = layout(8)
+            .section_names(b"\0text\0\0\0")
+            .expect_err("invalid offset");
+        assert!(error.to_string().contains("exceeds section-name table"));
+    }
+
+    #[test]
+    fn section_names_reject_invalid_utf8() {
+        let mut elf_layout = layout(1);
+        elf_layout.section_headers[1].name = 1;
+        let error = elf_layout
+            .section_names(b"\0\xff\0\0\0\0\0\0")
+            .expect_err("invalid UTF-8");
+        assert!(error.to_string().contains("is not UTF-8"));
+    }
+
+    #[test]
+    fn section_names_reject_non_string_table() {
+        let mut elf_layout = layout(1);
+        elf_layout.section_headers[2].section_type = 1;
+        let error = elf_layout
+            .section_names(b"\0text\0\0\0")
+            .expect_err("wrong section type");
+        assert!(error.to_string().contains("unexpected type"));
+    }
+
+    #[test]
+    fn section_names_reject_unterminated_table() {
+        let elf_layout = layout(1);
+        let error = elf_layout
+            .section_names(b"\0text\0\x01\x01")
+            .expect_err("unterminated table");
+        assert!(error.to_string().contains("not NUL terminated"));
+    }
+
+    #[test]
+    fn append_load_segment_updates_program_headers() {
+        let elf_layout = ElfLayout {
+            entrypoint: 0,
+            program_header_offset: 0,
+            program_header_size: 0x38,
+            program_header_count: 0,
+            program_headers: Vec::new(),
+            section_headers: Vec::new(),
+            section_name_index: 0,
+            private_section_index: usize::MAX,
+        };
+        let mut output = vec![0_u8; 0x2000];
+        let updated = elf_layout
+            .append_load_segment(
+                &mut output,
+                LoadSegment {
+                    offset: 0x1000,
+                    virtual_address: 0x2000,
+                    file_size: 0x20,
+                    memory_size: 0x20,
+                    flags: PF_R,
+                    alignment: 0x1000,
+                },
+            )
+            .expect("append segment");
+        assert_eq!(updated.program_header_count, 1);
+        assert_eq!(updated.program_headers[0].virtual_address, 0x2000);
+        assert_eq!(&output[0..4], &PT_LOAD.to_le_bytes());
+        assert_eq!(&output[0x38..0x3a], &1_u16.to_le_bytes());
+    }
+
+    #[test]
+    fn append_load_segment_rejects_program_header_overlap() {
+        let mut elf_layout = ElfLayout {
+            entrypoint: 0,
+            program_header_offset: 0,
+            program_header_size: 0x38,
+            program_header_count: 0,
+            program_headers: Vec::new(),
+            section_headers: Vec::new(),
+            section_name_index: 0,
+            private_section_index: usize::MAX,
+        };
+        elf_layout.section_headers.push(SectionHeader {
+            name: 0,
+            section_type: 1,
+            flags: 0,
+            address: 0,
+            offset: 0x20,
+            size: 1,
+            link: 0,
+            info: 0,
+            alignment: 1,
+            entry_size: 0,
+        });
+        let mut output = vec![0_u8; 0x100];
+        let error = elf_layout
+            .append_load_segment(
+                &mut output,
+                LoadSegment {
+                    offset: 0x80,
+                    virtual_address: 0x1080,
+                    file_size: 0x20,
+                    memory_size: 0x20,
+                    flags: PF_R,
+                    alignment: 0x1000,
+                },
+            )
+            .expect_err("overlapping program header");
+        assert!(error.to_string().contains("additional program header"));
+    }
+}
@@ -0,0 +1,129 @@
+//! Basic ELF format parsing shared by the unpacking engine.
+
+use goblin::elf::{Elf, header::EM_AARCH64, program_header::PT_LOAD};
+use thiserror::Error;
+
+pub mod hash;
+pub mod layout;
+
+pub use hash::{build_gnu_hash, build_sysv_hash};
+pub use layout::{
+    ElfLayout, LoadSegment, PF_R, SHF_ALLOC, SHT_LOUSER, SHT_NOBITS, SectionHeader, align_up,
+    checked_index, read_i64, read_u16, read_u32, read_u64, slice, slice_u64, usize_from_u64,
+};
+
+/// ELF machine identifier for AArch64.
+pub const AARCH64_MACHINE: u16 = EM_AARCH64;
+
+/// Dynamic sections required by the restored AArch64 loader image.
+pub const DYNAMIC_SECTION_NAMES: [&str; 8] = [
+    ".dynsym",
+    ".gnu.version",
+    ".gnu.version_r",
+    ".gnu.hash",
+    ".dynstr",
+    ".rela.dyn",
+    ".rela.plt",
+    ".dynamic",
+];
+
+/// Dynamic sections needed to identify a protected image before extraction.
+pub const PROBE_SECTION_NAMES: [&str; 5] = [
+    ".dynsym",
+    ".dynstr",
+    ".gnu.hash",
+    ".gnu.version",
+    ".gnu.version_r",
+];
+
+/// ELF64 dynamic table record sizes.
+pub const ELF64_SYMBOL_SIZE: usize = 0x18;
+pub const ELF64_RELA_SIZE: usize = 0x18;
+
+/// AArch64 relocation kinds used by the dynamic linker.
+pub const R_AARCH64_ABS64: u32 = 0x101;
+pub const R_AARCH64_GLOB_DAT: u32 = 0x401;
+pub const R_AARCH64_JUMP_SLOT: u32 = 0x402;
+pub const R_AARCH64_RELATIVE: u32 = 0x403;
+pub const VER_NDX_GLOBAL: u16 = 1;
+
+/// ELF dynamic-table tag identifiers used by restored images.
+pub const DT_PLTRELSZ: u64 = 2;
+pub const DT_HASH: u64 = 4;
+pub const DT_STRTAB: u64 = 5;
+pub const DT_SYMTAB: u64 = 6;
+pub const DT_RELA: u64 = 7;
+pub const DT_RELASZ: u64 = 8;
+pub const DT_STRSZ: u64 = 10;
+pub const DT_JMPREL: u64 = 23;
+pub const DT_GNU_HASH: u64 = 0x6fff_fef5;
+pub const DT_VERSYM: u64 = 0x6fff_fff0;
+pub const DT_RELACOUNT: u64 = 0x6fff_fff9;
+pub const DT_VERNEED: u64 = 0x6fff_fffe;
+
+#[derive(Debug, Error)]
+pub enum Error {
+    #[error("ELF parse failed: {0}")]
+    Parse(#[from] goblin::error::Error),
+    #[error("input is not an ELF64 little-endian image")]
+    NotElf64,
+    #[error("input is not an AArch64 image")]
+    NotAarch64,
+    #[error("invalid ELF layout: {0}")]
+    Invalid(String),
+}
+
+pub type Result<T> = std::result::Result<T, Error>;
+
+/// Parse an ELF64 little-endian image.
+pub fn parse(data: &[u8]) -> Result<Elf<'_>> {
+    let elf = Elf::parse(data)?;
+    if elf.header.e_ident[4] != 2 || elf.header.e_ident[5] != 1 {
+        return Err(Error::NotElf64);
+    }
+    Ok(elf)
+}
+
+/// Return true when `data` starts with a valid AArch64 ELF64 image.
+pub fn is_aarch64(data: &[u8]) -> bool {
+    parse(data)
+        .map(|elf| elf.header.e_machine == EM_AARCH64)
+        .unwrap_or(false)
+}
+
+/// Return whether a short prefix identifies an ELF64 little-endian AArch64
+/// image. This is intentionally a prefix-only check for filesystem scanners;
+/// callers that need structural guarantees must use [`parse`].
+#[must_use]
+pub fn is_aarch64_prefix(data: &[u8]) -> bool {
+    data.get(0..6) == Some(b"\x7fELF\x02\x01")
+        && data
+            .get(18..20)
+            .is_some_and(|bytes| u16::from_le_bytes([bytes[0], bytes[1]]) == EM_AARCH64)
+}
+
+/// Return the maximum file end among PT_LOAD segments.
+pub fn load_file_end(data: &[u8]) -> Result<u64> {
+    let elf = parse(data)?;
+    Ok(elf
+        .program_headers
+        .iter()
+        .filter(|ph| ph.p_type == PT_LOAD)
+        .map(|ph| ph.p_offset.saturating_add(ph.p_filesz))
+        .max()
+        .unwrap_or(0))
+}
+
+pub(crate) fn invalid<T>(message: impl Into<String>) -> Result<T> {
+    Err(Error::Invalid(message.into()))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn rejects_non_elf() {
+        assert!(matches!(parse(b"not elf"), Err(Error::Parse(_))));
+    }
+}
@@ -0,0 +1,21 @@
+[package]
+name = "senbei-engine"
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+description = "Platform unpacking engines for Senbei"
+
+[dependencies]
+memmap2.workspace = true
+serde.workspace = true
+serde_json.workspace = true
+sha2.workspace = true
+tempfile.workspace = true
+thiserror.workspace = true
+senbei-crypto.workspace = true
+senbei-elf.workspace = true
+senbei-pe.workspace = true
+
+[lints]
+workspace = true
@@ -0,0 +1,32 @@
+//! Shared Android engine filesystem and digest helpers.
+
+use std::io::Write;
+use std::path::{Path, PathBuf};
+
+use sha2::{Digest, Sha256};
+use tempfile::NamedTempFile;
+
+pub(crate) fn absolute(path: &Path) -> std::io::Result<PathBuf> {
+    if path.is_absolute() {
+        Ok(path.to_path_buf())
+    } else {
+        std::env::current_dir().map(|current| current.join(path))
+    }
+}
+
+pub(crate) fn write_atomic(path: &Path, data: &[u8]) -> std::io::Result<()> {
+    let parent = path.parent().unwrap_or_else(|| Path::new("."));
+    std::fs::create_dir_all(parent)?;
+    let mut temporary = NamedTempFile::new_in(parent)?;
+    temporary.write_all(data)?;
+    temporary.as_file().sync_all()?;
+    temporary.persist(path).map_err(|error| error.error)?;
+    Ok(())
+}
+
+#[must_use]
+pub(crate) fn sha256(data: &[u8]) -> String {
+    let mut digest = Sha256::new();
+    digest.update(data);
+    senbei_crypto::hex_digest(&digest.finalize())
+}
@@ -0,0 +1,63 @@
+use std::path::{Path, PathBuf};
+
+/// Stage 1 or Stage 2 extraction failure.
+#[derive(Debug, thiserror::Error)]
+pub enum Error {
+    #[error("{action} `{path}`: {source}")]
+    Io {
+        action: &'static str,
+        path: PathBuf,
+        #[source]
+        source: std::io::Error,
+    },
+    #[error("parse ELF `{path}`: {source}")]
+    Elf {
+        path: PathBuf,
+        #[source]
+        source: senbei_elf::Error,
+    },
+    #[error("serialize extraction index: {0}")]
+    Json(#[from] serde_json::Error),
+    #[error("embedded Stage 2 decoder configuration: {0}")]
+    EmbeddedConfig(#[source] senbei_crypto::android::Error),
+    #[error(
+        "depth {depth} stream 0x{stream_id:02X} interpreter 0x{interpreter_id:02X} configuration: {source}"
+    )]
+    InterpreterConfig {
+        depth: usize,
+        stream_id: u32,
+        interpreter_id: u32,
+        #[source]
+        source: senbei_crypto::android::Error,
+    },
+    #[error(
+        "depth {depth} stream 0x{stream_id:02X} record {record_index} command 0x{command_id:02X} {part}: {source}"
+    )]
+    RecordDecode {
+        depth: usize,
+        stream_id: u32,
+        record_index: usize,
+        command_id: u32,
+        part: &'static str,
+        #[source]
+        source: senbei_crypto::android::Error,
+    },
+    #[error("{0}")]
+    Invalid(String),
+}
+
+impl Error {
+    pub(crate) fn io(action: &'static str, path: &Path, source: std::io::Error) -> Self {
+        Self::Io {
+            action,
+            path: path.to_path_buf(),
+            source,
+        }
+    }
+}
+
+pub(crate) type Result<T> = std::result::Result<T, Error>;
+
+pub(crate) fn invalid<T>(message: impl Into<String>) -> Result<T> {
+    Err(Error::Invalid(message.into()))
+}
@@ -0,0 +1,12 @@
+mod error;
+mod pipeline;
+mod probe;
+mod report;
+mod stage1;
+mod stream;
+
+pub use error::Error;
+pub use pipeline::{ExtractOptions, extract_stage2};
+pub use probe::is_protected_libil2cpp;
+pub use report::ExtractionReport;
+pub use stage1::{DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE};
@@ -0,0 +1,508 @@
+use std::collections::{BTreeMap, BTreeSet, HashSet};
+use std::fs::File;
+use std::path::{Path, PathBuf};
+
+use memmap2::MmapOptions;
+use senbei_crypto::android::{Module9bConfig, decode_container};
+use serde_json::to_vec_pretty;
+
+use super::super::common;
+use super::error::{Error, Result, invalid};
+use super::report::{
+    ArtifactReport, DecoderReport, ExtractionReport, ModuleRegistryEntry, RecordReport,
+    Stage1Report, StreamParent, StreamReport,
+};
+use super::stage1::{
+    DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE, SHT_LOUSER, Stage1Result, inspect,
+};
+use super::stream::{DIRECT_FLAG, Record, parse_record_stream};
+
+/// Inputs and output locations for one complete static Stage 2 extraction.
+#[derive(Debug, Clone)]
+pub struct ExtractOptions {
+    pub input: PathBuf,
+    pub output_dir: PathBuf,
+    pub stage2_output: Option<PathBuf>,
+    pub outer_size: usize,
+    pub cipher_constant: u32,
+}
+
+impl ExtractOptions {
+    #[must_use]
+    pub fn with_defaults(input: PathBuf, output_dir: PathBuf) -> Self {
+        Self {
+            input,
+            output_dir,
+            stage2_output: None,
+            outer_size: DEFAULT_OUTER_SIZE,
+            cipher_constant: DEFAULT_CIPHER_CONSTANT,
+        }
+    }
+}
+
+#[derive(Debug)]
+struct LoadedModule {
+    image: Vec<u8>,
+    metadata: Option<Vec<u8>>,
+    image_path: String,
+    metadata_path: Option<String>,
+    sha256: String,
+    depth: usize,
+    record_index: usize,
+    command_id: u32,
+    init_offset: u32,
+    entry_offset: u32,
+}
+
+#[derive(Debug, Clone, Copy)]
+struct ArtifactSpec<'a> {
+    suffix: &'a str,
+    kind: &'a str,
+    classification: &'a str,
+}
+
+struct Extractor {
+    output_dir: PathBuf,
+    streams: Vec<StreamReport>,
+    artifacts: Vec<ArtifactReport>,
+    registry: BTreeMap<u32, LoadedModule>,
+    seen_streams: HashSet<(u32, String)>,
+}
+
+pub fn extract_stage2(options: &ExtractOptions) -> Result<ExtractionReport> {
+    let input_path = absolute(&options.input)?;
+    let output_dir = absolute(&options.output_dir)?;
+    if !input_path.is_file() {
+        return invalid(format!(
+            "protected ELF does not exist: {}",
+            input_path.display()
+        ));
+    }
+    if let Some(stage2_output) = &options.stage2_output {
+        let stage2_output = absolute(stage2_output)?;
+        if stage2_output == input_path {
+            return invalid("refusing to overwrite the protected ELF with Stage 2 output");
+        }
+    }
+    std::fs::create_dir_all(&output_dir)
+        .map_err(|source| Error::io("create Stage 2 output directory", &output_dir, source))?;
+
+    let file = File::open(&input_path)
+        .map_err(|source| Error::io("open protected ELF", &input_path, source))?;
+    // SAFETY: the mapping is read-only, the file remains open for the mapping
+    // lifetime, and extraction never mutates or truncates the source.
+    let source = unsafe { MmapOptions::new().map(&file) }
+        .map_err(|source| Error::io("map protected ELF", &input_path, source))?;
+    let stage1 = inspect(
+        &source,
+        &input_path,
+        options.outer_size,
+        options.cipher_constant,
+    )?;
+    if let Some(stage2_output) = &options.stage2_output {
+        write_atomic(&absolute(stage2_output)?, &stage1.plaintext)?;
+    }
+
+    let core_config =
+        Module9bConfig::parse_embedded(&stage1.plaintext).map_err(Error::EmbeddedConfig)?;
+    let bootstrap_end = stage1
+        .remaining_file_offset
+        .checked_add(stage1.remaining_size)
+        .ok_or_else(|| Error::Invalid("Stage 2 bootstrap range overflow".to_owned()))?;
+    let bootstrap = source
+        .get(stage1.remaining_file_offset..bootstrap_end)
+        .ok_or_else(|| Error::Invalid("Stage 2 bootstrap range is outside the ELF".to_owned()))?;
+    let mut extractor = Extractor {
+        output_dir: output_dir.clone(),
+        streams: Vec::new(),
+        artifacts: Vec::new(),
+        registry: BTreeMap::new(),
+        seen_streams: HashSet::new(),
+    };
+    extractor.extract_stream(
+        bootstrap,
+        0xe2,
+        0,
+        None,
+        Some(stage1.remaining_file_offset),
+        core_config,
+    )?;
+
+    let module_registry = extractor
+        .registry
+        .values()
+        .map(|module| ModuleRegistryEntry {
+            command_id: module.command_id,
+            size: module.image.len(),
+            sha256: module.sha256.clone(),
+            depth: module.depth,
+            record_index: module.record_index,
+            image_path: module.image_path.clone(),
+            metadata_path: module.metadata_path.clone(),
+            init_offset: module.init_offset,
+            entry_offset: module.entry_offset,
+            classification: if module.metadata.is_some() {
+                "module_image".to_owned()
+            } else {
+                "decoded_data".to_owned()
+            },
+        })
+        .collect::<Vec<_>>();
+    let report = ExtractionReport {
+        format_version: 4,
+        protected_elf: input_path.display().to_string(),
+        output_dir: output_dir.display().to_string(),
+        stage1: stage1_report(&stage1, options.outer_size),
+        streams: extractor.streams,
+        artifacts: extractor.artifacts,
+        errors: Vec::new(),
+        module_registry,
+    };
+    write_json_atomic(&output_dir.join("index.json"), &report)?;
+    Ok(report)
+}
+
+impl Extractor {
+    fn extract_stream(
+        &mut self,
+        stream: &[u8],
+        stream_id: u32,
+        depth: usize,
+        parent: Option<StreamParent>,
+        source_file_offset: Option<usize>,
+        config: Module9bConfig,
+    ) -> Result<()> {
+        let digest = sha256(stream);
+        if !self.seen_streams.insert((stream_id, digest.clone())) {
+            return Ok(());
+        }
+        let (header, records, table_size) =
+            parse_record_stream(stream, stream_id).map_err(|source| {
+                Error::Invalid(format!(
+                    "depth {depth} stream 0x{stream_id:02X} record table: {source}"
+                ))
+            })?;
+        let mut stream_report = StreamReport {
+            depth,
+            stream_id,
+            parent,
+            source_file_offset,
+            available_size: stream.len(),
+            descriptor_table_size: table_size,
+            encrypted_header_words: header.encrypted_words,
+            decrypted_header_words: header.decrypted_words,
+            record_state: header.record_state,
+            sha256: digest,
+            decoder: decoder_report(
+                if depth == 0 {
+                    "embedded_stage2"
+                } else {
+                    "decoded_interpreter"
+                },
+                (depth != 0).then_some(stream_id),
+                &config,
+            ),
+            records: Vec::with_capacity(records.len()),
+        };
+        let mut direct_records = Vec::new();
+        let mut modules_at_level = BTreeSet::new();
+
+        for record in records {
+            let mut result = record_report(record);
+            let mut image_data = None;
+            let mut metadata_data = None;
+
+            if !record.direct() && record.image_size != 0 {
+                let image_source = record_tail(stream, record.image_offset)?;
+                let image = decode_container(image_source, &config, record.image_size as usize)
+                    .map_err(|source| Error::RecordDecode {
+                        depth,
+                        stream_id,
+                        record_index: record.index,
+                        command_id: record.command_id,
+                        part: "image decode",
+                        source,
+                    })?;
+                let classification = if record.metadata_size != 0 {
+                    "module_image"
+                } else {
+                    "decoded_data"
+                };
+                let artifact = self.write_artifact(
+                    &record,
+                    depth,
+                    stream_id,
+                    ArtifactSpec {
+                        suffix: "module.bin",
+                        kind: "decoded_container",
+                        classification,
+                    },
+                    &image,
+                )?;
+                result.image = Some(artifact.clone());
+                image_data = Some((image, artifact));
+            }
+            if record.metadata_size != 0 {
+                let metadata_source = record_tail(stream, record.metadata_offset)?;
+                let metadata =
+                    decode_container(metadata_source, &config, record.metadata_size as usize)
+                        .map_err(|source| Error::RecordDecode {
+                            depth,
+                            stream_id,
+                            record_index: record.index,
+                            command_id: record.command_id,
+                            part: "metadata decode",
+                            source,
+                        })?;
+                let artifact = self.write_artifact(
+                    &record,
+                    depth,
+                    stream_id,
+                    ArtifactSpec {
+                        suffix: "metadata.bin",
+                        kind: "decoded_metadata",
+                        classification: "decoded_metadata",
+                    },
+                    &metadata,
+                )?;
+                result.metadata = Some(artifact.clone());
+                metadata_data = Some((metadata, artifact));
+            }
+            if let Some((image, image_artifact)) = image_data {
+                let (metadata, metadata_path) = if let Some((data, artifact)) = metadata_data {
+                    (Some(data), Some(artifact.path))
+                } else {
+                    (None, None)
+                };
+                self.register_module(LoadedModule {
+                    sha256: image_artifact.sha256.clone(),
+                    image_path: image_artifact.path.clone(),
+                    metadata_path,
+                    image,
+                    metadata,
+                    depth,
+                    record_index: record.index,
+                    command_id: record.command_id,
+                    init_offset: record.init_offset,
+                    entry_offset: record.entry_offset,
+                })?;
+                modules_at_level.insert(record.command_id);
+            }
+            if record.direct() && record.image_size != 0 {
+                direct_records.push((record, stream_report.records.len()));
+            }
+            stream_report.records.push(result);
+        }
+
+        let mut children = Vec::new();
+        for (record, report_index) in direct_records {
+            let next_stream_id = record.command_id.wrapping_sub(0x10);
+            if modules_at_level.contains(&next_stream_id) {
+                stream_report.records[report_index].nested_stream_id = Some(next_stream_id);
+                children.push((record, next_stream_id));
+                continue;
+            }
+            let direct_data = record_slice(stream, record.image_offset, record.image_size)?;
+            let artifact = self.write_artifact(
+                &record,
+                depth,
+                stream_id,
+                ArtifactSpec {
+                    suffix: "direct.bin",
+                    kind: "direct",
+                    classification: "direct_data",
+                },
+                direct_data,
+            )?;
+            stream_report.records[report_index].image = Some(artifact);
+        }
+
+        self.streams.push(stream_report);
+        for (record, next_stream_id) in children {
+            let child_data = record_slice(stream, record.image_offset, record.image_size)?;
+            let parent = StreamParent {
+                stream_id,
+                record_index: record.index,
+                command_id: record.command_id,
+            };
+            let interpreter = self.registry.get(&next_stream_id).ok_or_else(|| {
+                Error::Invalid(format!(
+                    "depth {depth} stream 0x{stream_id:02X} child 0x{next_stream_id:02X} has no interpreter module"
+                ))
+            })?;
+            let interpreter_config =
+                Module9bConfig::parse(&interpreter.image).map_err(|source| {
+                    Error::InterpreterConfig {
+                        depth: depth + 1,
+                        stream_id: next_stream_id,
+                        interpreter_id: next_stream_id,
+                        source,
+                    }
+                })?;
+            self.extract_stream(
+                child_data,
+                next_stream_id,
+                depth + 1,
+                Some(parent),
+                None,
+                interpreter_config,
+            )?;
+        }
+        Ok(())
+    }
+
+    fn register_module(&mut self, module: LoadedModule) -> Result<()> {
+        if let Some(previous) = self.registry.get(&module.command_id) {
+            if previous.sha256 != module.sha256 {
+                return invalid(format!(
+                    "module 0x{:02X} produced conflicting images: {} and {}",
+                    module.command_id, previous.sha256, module.sha256
+                ));
+            }
+            return Ok(());
+        }
+        self.registry.insert(module.command_id, module);
+        Ok(())
+    }
+
+    fn write_artifact(
+        &mut self,
+        record: &Record,
+        depth: usize,
+        stream_id: u32,
+        spec: ArtifactSpec<'_>,
+        data: &[u8],
+    ) -> Result<ArtifactReport> {
+        let digest = sha256(data);
+        let filename = format!(
+            "d{depth:02}_s{stream_id:02X}_r{:03}_id{:08X}_{}.{}",
+            record.index,
+            record.command_id,
+            &digest[..12],
+            spec.suffix
+        );
+        let path = self.output_dir.join(filename);
+        write_atomic(&path, data)?;
+        let artifact = ArtifactReport {
+            kind: spec.kind.to_owned(),
+            path: path
+                .file_name()
+                .ok_or_else(|| Error::Invalid("artifact path has no file name".to_owned()))?
+                .to_string_lossy()
+                .into_owned(),
+            size: data.len(),
+            sha256: digest,
+            depth,
+            stream_id,
+            record_index: Some(record.index),
+            command_id: Some(record.command_id),
+            classification: spec.classification.to_owned(),
+        };
+        self.artifacts.push(artifact.clone());
+        Ok(artifact)
+    }
+}
+
+fn record_report(record: Record) -> RecordReport {
+    RecordReport {
+        index: record.index,
+        command_id: record.command_id,
+        flags: record.flags,
+        image_offset: record.image_offset,
+        image_size: record.image_size,
+        metadata_offset: record.metadata_offset,
+        metadata_size: record.metadata_size,
+        id_copy: record.id_copy,
+        entry_offset: record.entry_offset,
+        init_offset: record.init_offset,
+        direct: record.flags & DIRECT_FLAG != 0,
+        extraction_status: "complete".to_owned(),
+        image: None,
+        metadata: None,
+        nested_stream_id: None,
+    }
+}
+
+fn decoder_report(
+    kind: &str,
+    interpreter_id: Option<u32>,
+    config: &Module9bConfig,
+) -> DecoderReport {
+    DecoderReport {
+        kind: kind.to_owned(),
+        interpreter_id,
+        header_seed: config.header_seed,
+        container_seed: config.container_seed,
+        schedule_offset: config.schedule_offset,
+        aes_key_sha256: sha256(&config.aes_key),
+        skip_aes: config.skip_aes,
+    }
+}
+
+fn record_slice(stream: &[u8], offset: u32, size: u32) -> Result<&[u8]> {
+    let offset = usize::try_from(offset)
+        .map_err(|_| Error::Invalid("record payload offset exceeds usize".to_owned()))?;
+    let size = usize::try_from(size)
+        .map_err(|_| Error::Invalid("record payload size exceeds usize".to_owned()))?;
+    let end = offset
+        .checked_add(size)
+        .ok_or_else(|| Error::Invalid("record payload range overflows usize".to_owned()))?;
+    stream.get(offset..end).ok_or_else(|| {
+        Error::Invalid(format!(
+            "record payload range 0x{offset:x}..0x{end:x} exceeds stream 0x{:x}",
+            stream.len()
+        ))
+    })
+}
+
+fn record_tail(stream: &[u8], offset: u32) -> Result<&[u8]> {
+    let offset = usize::try_from(offset)
+        .map_err(|_| Error::Invalid("record container offset exceeds usize".to_owned()))?;
+    stream.get(offset..).ok_or_else(|| {
+        Error::Invalid(format!(
+            "record container offset 0x{offset:x} exceeds stream 0x{:x}",
+            stream.len()
+        ))
+    })
+}
+
+fn stage1_report(stage1: &Stage1Result, outer_size: usize) -> Stage1Report {
+    Stage1Report {
+        section_index: stage1.section_index,
+        section_type: SHT_LOUSER,
+        section_offset: stage1.section_offset,
+        section_size: stage1.section_size,
+        outer_size,
+        header_offset: stage1.header_offset,
+        header_key: stage1.header.key,
+        payload_offset: stage1.header.payload_offset,
+        payload_size: stage1.header.payload_size,
+        payload_key: stage1.header.payload_key,
+        entry_offset: stage1.header.entry_offset,
+        protect_size: stage1.header.protect_size,
+        stage2_file_offset: stage1.payload_file_offset,
+        stage2_size: stage1.plaintext.len(),
+        stage2_sha256: sha256(&stage1.plaintext),
+        remaining_file_offset: stage1.remaining_file_offset,
+        remaining_size: stage1.remaining_size,
+    }
+}
+
+fn write_json_atomic(path: &Path, value: &impl serde::Serialize) -> Result<()> {
+    let mut bytes = to_vec_pretty(value)?;
+    bytes.push(b'\n');
+    write_atomic(path, &bytes)
+}
+
+fn write_atomic(path: &Path, data: &[u8]) -> Result<()> {
+    common::write_atomic(path, data)
+        .map_err(|source| Error::io("write temporary output", path, source))
+}
+
+fn absolute(path: &Path) -> Result<PathBuf> {
+    common::absolute(path).map_err(|source| Error::io("query current directory", path, source))
+}
+
+fn sha256(data: &[u8]) -> String {
+    common::sha256(data)
+}
@@ -0,0 +1,19 @@
+use std::path::Path;
+
+use senbei_crypto::android::Module9bConfig;
+
+use super::stage1::{self, DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE};
+
+/// Return whether `data` has a supported protected AArch64 IL2CPP layout.
+#[must_use]
+pub fn is_protected_libil2cpp(data: &[u8]) -> bool {
+    let Ok(stage1) = stage1::inspect(
+        data,
+        Path::new("<probe>"),
+        DEFAULT_OUTER_SIZE,
+        DEFAULT_CIPHER_CONSTANT,
+    ) else {
+        return false;
+    };
+    Module9bConfig::parse_embedded(&stage1.plaintext).is_ok()
+}
@@ -0,0 +1,115 @@
+use serde::Serialize;
+
+#[derive(Debug, Clone, Serialize)]
+pub struct Stage1Report {
+    pub section_index: usize,
+    pub section_type: u32,
+    pub section_offset: usize,
+    pub section_size: usize,
+    pub outer_size: usize,
+    pub header_offset: usize,
+    pub header_key: u32,
+    pub payload_offset: u32,
+    pub payload_size: u32,
+    pub payload_key: u32,
+    pub entry_offset: u32,
+    pub protect_size: u32,
+    pub stage2_file_offset: usize,
+    pub stage2_size: usize,
+    pub stage2_sha256: String,
+    pub remaining_file_offset: usize,
+    pub remaining_size: usize,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct DecoderReport {
+    pub kind: String,
+    pub interpreter_id: Option<u32>,
+    pub header_seed: u32,
+    pub container_seed: u32,
+    pub schedule_offset: usize,
+    pub aes_key_sha256: String,
+    pub skip_aes: bool,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct ArtifactReport {
+    pub kind: String,
+    pub path: String,
+    pub size: usize,
+    pub sha256: String,
+    pub depth: usize,
+    pub stream_id: u32,
+    pub record_index: Option<usize>,
+    pub command_id: Option<u32>,
+    pub classification: String,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct RecordReport {
+    pub index: usize,
+    pub command_id: u32,
+    pub flags: u32,
+    pub image_offset: u32,
+    pub image_size: u32,
+    pub metadata_offset: u32,
+    pub metadata_size: u32,
+    pub id_copy: u32,
+    pub entry_offset: u32,
+    pub init_offset: u32,
+    pub direct: bool,
+    pub extraction_status: String,
+    pub image: Option<ArtifactReport>,
+    pub metadata: Option<ArtifactReport>,
+    pub nested_stream_id: Option<u32>,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct StreamParent {
+    pub stream_id: u32,
+    pub record_index: usize,
+    pub command_id: u32,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct StreamReport {
+    pub depth: usize,
+    pub stream_id: u32,
+    pub parent: Option<StreamParent>,
+    pub source_file_offset: Option<usize>,
+    pub available_size: usize,
+    pub descriptor_table_size: usize,
+    pub encrypted_header_words: [u32; 2],
+    pub decrypted_header_words: [u32; 2],
+    pub record_state: u32,
+    pub sha256: String,
+    pub decoder: DecoderReport,
+    pub records: Vec<RecordReport>,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct ModuleRegistryEntry {
+    pub command_id: u32,
+    pub size: usize,
+    pub sha256: String,
+    pub depth: usize,
+    pub record_index: usize,
+    pub image_path: String,
+    pub metadata_path: Option<String>,
+    pub init_offset: u32,
+    pub entry_offset: u32,
+    pub classification: String,
+}
+
+/// Machine-readable output of one complete static Stage 2 extraction.
+#[derive(Debug, Clone, Serialize)]
+pub struct ExtractionReport {
+    pub format_version: u32,
+    pub protected_elf: String,
+    pub output_dir: String,
+    pub stage1: Stage1Report,
+    pub streams: Vec<StreamReport>,
+    pub artifacts: Vec<ArtifactReport>,
+    pub errors: Vec<String>,
+    pub module_registry: Vec<ModuleRegistryEntry>,
+}
@@ -0,0 +1,236 @@
+use std::path::Path;
+
+use senbei_elf::{AARCH64_MACHINE, Error as ElfError, parse};
+
+use super::error::{Error, Result, invalid};
+
+pub(crate) use senbei_elf::SHT_LOUSER;
+pub const DEFAULT_CIPHER_CONSTANT: u32 = 0xbf20_165d;
+pub const DEFAULT_OUTER_SIZE: usize = 0x23c;
+
+#[derive(Debug, Clone, Copy)]
+pub(crate) struct Stage1Header {
+    pub key: u32,
+    pub reserved: u32,
+    pub payload_offset: u32,
+    pub payload_size: u32,
+    pub payload_key: u32,
+    pub entry_offset: u32,
+    pub protect_size: u32,
+    pub size_copy: u32,
+}
+
+#[derive(Debug)]
+pub(crate) struct Stage1Result {
+    pub section_index: usize,
+    pub section_offset: usize,
+    pub section_size: usize,
+    pub header_offset: usize,
+    pub payload_file_offset: usize,
+    pub remaining_file_offset: usize,
+    pub remaining_size: usize,
+    pub header: Stage1Header,
+    pub plaintext: Vec<u8>,
+}
+
+pub(crate) fn inspect(
+    data: &[u8],
+    path: &Path,
+    outer_size: usize,
+    cipher_constant: u32,
+) -> Result<Stage1Result> {
+    let elf = parse(data).map_err(|source: ElfError| Error::Elf {
+        path: path.to_path_buf(),
+        source,
+    })?;
+    if elf.header.e_machine != AARCH64_MACHINE {
+        return invalid(format!(
+            "expected AArch64 ELF (machine 0x{AARCH64_MACHINE:X}), got 0x{:X}",
+            elf.header.e_machine
+        ));
+    }
+    let matches = elf
+        .section_headers
+        .iter()
+        .enumerate()
+        .filter(|(_, section)| section.sh_type == SHT_LOUSER)
+        .collect::<Vec<_>>();
+    if matches.len() != 1 {
+        return invalid(format!(
+            "expected exactly one SHT_LOUSER section, found {}",
+            matches.len()
+        ));
+    }
+    for wanted in senbei_elf::PROBE_SECTION_NAMES {
+        if !elf.section_headers.iter().any(|section| {
+            elf.shdr_strtab
+                .get_at(section.sh_name)
+                .is_some_and(|name| name == wanted)
+        }) {
+            return invalid(format!("protected ELF lacks required section {wanted}"));
+        }
+    }
+    let (section_index, section) = matches[0];
+    let section_offset = usize::try_from(section.sh_offset)
+        .map_err(|_| Error::Invalid("SHT_LOUSER offset exceeds usize".to_owned()))?;
+    let section_size = usize::try_from(section.sh_size)
+        .map_err(|_| Error::Invalid("SHT_LOUSER size exceeds usize".to_owned()))?;
+    let section_end = section_offset
+        .checked_add(section_size)
+        .ok_or_else(|| Error::Invalid("SHT_LOUSER range overflows usize".to_owned()))?;
+    if section_end > data.len() {
+        return invalid("SHT_LOUSER range extends beyond the input file");
+    }
+    let header_relative = outer_size;
+    if outer_size
+        .checked_add(0x1000)
+        .is_none_or(|end| end > section_size)
+    {
+        return invalid("Stage 1 outer header leaves no complete parameter area");
+    }
+    let header_offset = section_offset
+        .checked_add(header_relative)
+        .ok_or_else(|| Error::Invalid("Stage 1 header offset overflow".to_owned()))?;
+    let header_raw = bytes(data, header_offset, 0x1000)?;
+    let header = decrypt_header(header_raw, cipher_constant)?;
+    if header.reserved != 0 {
+        return invalid(format!(
+            "Stage 1 header reserved word is nonzero: 0x{:x}",
+            header.reserved
+        ));
+    }
+    if header.size_copy != header.payload_size {
+        return invalid(format!(
+            "Stage 1 payload size copy 0x{:x} != size 0x{:x}",
+            header.size_copy, header.payload_size
+        ));
+    }
+    let private_size = section_size - outer_size;
+    let payload_offset = usize::try_from(header.payload_offset)
+        .map_err(|_| Error::Invalid("Stage 1 payload offset exceeds usize".to_owned()))?;
+    let payload_size = usize::try_from(header.payload_size)
+        .map_err(|_| Error::Invalid("Stage 1 payload size exceeds usize".to_owned()))?;
+    let payload_end = payload_offset
+        .checked_add(payload_size)
+        .ok_or_else(|| Error::Invalid("Stage 1 payload range overflow".to_owned()))?;
+    if payload_offset < 0x20 || payload_end > private_size {
+        return invalid(format!(
+            "Stage 1 payload range 0x{payload_offset:x}..0x{payload_end:x} exceeds private size 0x{private_size:x}"
+        ));
+    }
+    if payload_size == 0 || payload_size % 4 != 0 {
+        return invalid(format!(
+            "Stage 1 payload size must be nonzero and word aligned: 0x{payload_size:x}"
+        ));
+    }
+    let entry_offset = usize::try_from(header.entry_offset)
+        .map_err(|_| Error::Invalid("Stage 1 entry offset exceeds usize".to_owned()))?;
+    if entry_offset >= payload_size {
+        return invalid("Stage 1 entry offset is outside the payload");
+    }
+    let protect_size = usize::try_from(header.protect_size)
+        .map_err(|_| Error::Invalid("Stage 1 protect size exceeds usize".to_owned()))?;
+    if protect_size > payload_size {
+        return invalid("Stage 1 mprotect length exceeds the payload");
+    }
+    let payload_file_offset = header_offset
+        .checked_add(payload_offset)
+        .ok_or_else(|| Error::Invalid("Stage 1 payload file offset overflow".to_owned()))?;
+    let encrypted = bytes(data, payload_file_offset, payload_size)?;
+    let plaintext = decrypt_words(encrypted, header.payload_key, cipher_constant)?;
+    let aligned_payload_end = (payload_end + 3) & !3;
+    let remaining_relative = aligned_payload_end;
+    if remaining_relative > private_size {
+        return invalid("aligned Stage 2 cursor exceeds SHT_LOUSER");
+    }
+    let remaining_file_offset = section_offset
+        .checked_add(outer_size)
+        .and_then(|value| value.checked_add(remaining_relative))
+        .ok_or_else(|| Error::Invalid("Stage 2 stream offset overflow".to_owned()))?;
+    Ok(Stage1Result {
+        section_index,
+        section_offset,
+        section_size,
+        header_offset,
+        payload_file_offset,
+        remaining_file_offset,
+        remaining_size: private_size - remaining_relative,
+        header,
+        plaintext,
+    })
+}
+
+fn decrypt_header(raw: &[u8], constant: u32) -> Result<Stage1Header> {
+    let key = read_u32(raw, 0)?;
+    let mut decoded = decrypt_words(&raw[..0x20], key, constant)?;
+    decoded[..4].copy_from_slice(&key.to_le_bytes());
+    Ok(Stage1Header {
+        key,
+        reserved: read_u32(&decoded, 4)?,
+        payload_offset: read_u32(&decoded, 8)?,
+        payload_size: read_u32(&decoded, 12)?,
+        payload_key: read_u32(&decoded, 16)?,
+        entry_offset: read_u32(&decoded, 20)?,
+        protect_size: read_u32(&decoded, 24)?,
+        size_copy: read_u32(&decoded, 28)?,
+    })
+}
+
+fn decrypt_words(ciphertext: &[u8], key: u32, constant: u32) -> Result<Vec<u8>> {
+    if !ciphertext.len().is_multiple_of(4) {
+        return invalid("Stage 1 word cipher input is not 4-byte aligned");
+    }
+    let mut plaintext = ciphertext.to_vec();
+    for (index, chunk) in plaintext.as_chunks_mut::<4>().0.iter_mut().enumerate() {
+        let index = u32::try_from(index)
+            .map_err(|_| Error::Invalid("Stage 1 word index exceeds u32".to_owned()))?;
+        let mut word = u32::from_le_bytes(*chunk);
+        word = word.wrapping_add(index.wrapping_add(3).wrapping_mul(key));
+        word ^= constant.wrapping_mul(index.wrapping_add(1));
+        chunk.copy_from_slice(&word.to_le_bytes());
+    }
+    Ok(plaintext)
+}
+
+fn bytes(data: &[u8], offset: usize, size: usize) -> Result<&[u8]> {
+    let end = offset
+        .checked_add(size)
+        .ok_or_else(|| Error::Invalid("byte range overflow".to_owned()))?;
+    data.get(offset..end).ok_or_else(|| {
+        Error::Invalid(format!(
+            "byte range 0x{offset:x}..0x{end:x} is outside the input"
+        ))
+    })
+}
+
+fn read_u32(data: &[u8], offset: usize) -> Result<u32> {
+    let bytes = bytes(data, offset, 4)?;
+    Ok(u32::from_le_bytes(bytes.try_into().map_err(|_| {
+        Error::Invalid("invalid u32 byte range".to_owned())
+    })?))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn stage1_word_transform_round_trips() {
+        let key = 0x1234_5678;
+        let constant = DEFAULT_CIPHER_CONSTANT;
+        let plain = [0x1122_3344_u32, 0xaabb_ccdd, 0x0102_0304];
+        let mut cipher = Vec::new();
+        for (index, value) in plain.into_iter().enumerate() {
+            let index = index as u32;
+            let word = (value ^ constant.wrapping_mul(index + 1))
+                .wrapping_sub((index + 3).wrapping_mul(key));
+            cipher.extend_from_slice(&word.to_le_bytes());
+        }
+        let decoded = decrypt_words(&cipher, key, constant).unwrap();
+        let expected = plain
+            .into_iter()
+            .flat_map(u32::to_le_bytes)
+            .collect::<Vec<_>>();
+        assert_eq!(decoded, expected);
+    }
+}
@@ -0,0 +1,164 @@
+use senbei_crypto::android::gf32_mul_fixed;
+
+use super::error::{Error, Result, invalid};
+
+pub(crate) const RECORD_SIZE: usize = 0x5c;
+pub(crate) const DIRECT_FLAG: u32 = 2;
+
+#[derive(Debug, Clone, Copy)]
+pub(crate) struct Record {
+    pub index: usize,
+    pub command_id: u32,
+    pub flags: u32,
+    pub image_offset: u32,
+    pub image_size: u32,
+    pub metadata_offset: u32,
+    pub metadata_size: u32,
+    pub id_copy: u32,
+    pub entry_offset: u32,
+    pub init_offset: u32,
+}
+
+impl Record {
+    pub(crate) fn direct(self) -> bool {
+        self.flags & DIRECT_FLAG != 0
+    }
+}
+
+#[derive(Debug, Clone, Copy)]
+pub(crate) struct StreamHeader {
+    pub encrypted_words: [u32; 2],
+    pub decrypted_words: [u32; 2],
+    pub record_state: u32,
+}
+
+pub(crate) fn parse_record_stream(
+    stream: &[u8],
+    stream_id: u32,
+) -> Result<(StreamHeader, Vec<Record>, usize)> {
+    if stream.len() < 8 {
+        return invalid(format!(
+            "stream 0x{stream_id:02X} is shorter than its 8-byte header"
+        ));
+    }
+    let cipher0 = read_u32(stream, 0)?;
+    let cipher1 = read_u32(stream, 4)?;
+    let key = stream_id.wrapping_mul(0x9d32_3cd7);
+    let shift = stream_id & 7;
+    let base = (key >> shift)
+        .wrapping_add(0x5e72_7d74)
+        .wrapping_add(key.wrapping_shl(stream_id & 0xb))
+        .wrapping_add(0xf71e_3005);
+    let plain0 =
+        gf32_mul_fixed(cipher0.wrapping_add(0xcbf0_c1d8)) ^ 0xeb_e81dba_u32.wrapping_add(base);
+    let plain1 = gf32_mul_fixed(cipher1.wrapping_add(cipher0))
+        ^ 0xeb_e81dba_u32.wrapping_mul(5).wrapping_add(base);
+    let header = StreamHeader {
+        encrypted_words: [cipher0, cipher1],
+        decrypted_words: [plain0, plain1],
+        record_state: plain1.wrapping_add(base),
+    };
+
+    let mut records = Vec::new();
+    let mut first_payload = stream.len();
+    for index in 0..256_usize {
+        let start =
+            8_usize
+                .checked_add(index.checked_mul(RECORD_SIZE).ok_or_else(|| {
+                    Error::Invalid("record descriptor offset overflow".to_owned())
+                })?)
+                .ok_or_else(|| Error::Invalid("record descriptor offset overflow".to_owned()))?;
+        let end = start
+            .checked_add(RECORD_SIZE)
+            .ok_or_else(|| Error::Invalid("record descriptor end overflow".to_owned()))?;
+        if end > stream.len() {
+            return invalid(format!(
+                "stream 0x{stream_id:02X} descriptor table is truncated at record {index}"
+            ));
+        }
+        let record = decrypt_record(&stream[start..end], index, header.record_state)?;
+        if record.id_copy != 0 && record.command_id != record.id_copy {
+            return invalid(format!(
+                "stream 0x{stream_id:02X} record {index} command/id mismatch: 0x{:X} != 0x{:X}",
+                record.command_id, record.id_copy
+            ));
+        }
+        for (offset, size) in [
+            (record.image_offset, record.image_size),
+            (record.metadata_offset, record.metadata_size),
+        ] {
+            if offset != 0 && size != 0 {
+                let offset = usize::try_from(offset).map_err(|_| {
+                    Error::Invalid(format!(
+                        "stream 0x{stream_id:02X} record {index} payload offset exceeds usize"
+                    ))
+                })?;
+                if offset >= stream.len() {
+                    return invalid(format!(
+                        "stream 0x{stream_id:02X} record {index} payload offset 0x{offset:x} exceeds stream 0x{:x}",
+                        stream.len()
+                    ));
+                }
+                first_payload = first_payload.min(offset);
+            }
+        }
+        records.push(record);
+        if end == first_payload {
+            return Ok((header, records, first_payload));
+        }
+        if end > first_payload {
+            return invalid(format!(
+                "stream 0x{stream_id:02X} descriptor table crosses first payload at 0x{first_payload:x}"
+            ));
+        }
+    }
+    invalid(format!(
+        "stream 0x{stream_id:02X} has no descriptor boundary in 256 records"
+    ))
+}
+
+fn decrypt_record(raw: &[u8], index: usize, state: u32) -> Result<Record> {
+    if raw.len() != RECORD_SIZE {
+        return invalid(format!(
+            "record {index} has size 0x{:x}, expected 0x{RECORD_SIZE:x}",
+            raw.len()
+        ));
+    }
+    let product = state.wrapping_add(0x96f6_0b71).wrapping_mul(state);
+    let index_mask = product.wrapping_shl(((index + 1) & 3) as u32);
+    let mix = state.wrapping_mul(0x06a5_5bcc).wrapping_add(product);
+    let mut accumulator = 0x7993_4cf6_u32;
+    let mut feedback = 0xf02f_7685_u32;
+    let mut words = [0_u32; RECORD_SIZE / 4];
+    for (word_index, chunk) in raw.as_chunks::<4>().0.iter().enumerate() {
+        feedback = feedback.wrapping_mul(feedback);
+        let cipher = u32::from_le_bytes(*chunk);
+        let mut value = gf32_mul_fixed(cipher ^ (feedback >> 3)) ^ index_mask;
+        value = value.wrapping_add(accumulator).wrapping_add(state);
+        value = value.wrapping_sub(mix >> ((word_index * 4 + 3) & 5));
+        words[word_index] = value;
+        accumulator = accumulator.wrapping_add(0xe64d_33d8);
+        feedback = cipher;
+    }
+    Ok(Record {
+        index,
+        command_id: words[0],
+        flags: words[1],
+        image_offset: words[2],
+        image_size: words[3],
+        metadata_offset: words[4],
+        metadata_size: words[5],
+        id_copy: words[6],
+        entry_offset: words[7],
+        init_offset: words[8],
+    })
+}
+
+fn read_u32(data: &[u8], offset: usize) -> Result<u32> {
+    let bytes = data.get(offset..offset + 4).ok_or_else(|| {
+        Error::Invalid(format!("record header range 0x{offset:x} is out of bounds"))
+    })?;
+    Ok(u32::from_le_bytes(bytes.try_into().map_err(|_| {
+        Error::Invalid("invalid record u32 range".to_owned())
+    })?))
+}
@@ -0,0 +1,11 @@
+//! Android AArch64 extraction and ELF restoration.
+
+mod common;
+mod extract;
+mod restore;
+
+pub use extract::{
+    DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE, Error as ExtractionError, ExtractOptions,
+    ExtractionReport, extract_stage2, is_protected_libil2cpp,
+};
+pub use restore::{Error as RestoreError, RestoreOptions, RestoreReport, restore_libil2cpp};
@@ -0,0 +1,105 @@
+use std::collections::BTreeMap;
+use std::path::{Path, PathBuf};
+
+use serde_json::Value;
+
+use super::error::{Error, Result, invalid};
+
+const REQUIRED_IDS: [u32; 3] = [0x9b, 0x9d, 0x9e];
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub(crate) struct Artifact {
+    pub path: PathBuf,
+    pub size: u64,
+}
+
+pub(crate) fn load_artifacts(index_path: &Path) -> Result<BTreeMap<u32, Artifact>> {
+    let text = std::fs::read_to_string(index_path)
+        .map_err(|error| Error::io("read module index", index_path, error))?;
+    let document: Value = serde_json::from_str(&text)?;
+    let root = index_path.parent().unwrap_or_else(|| Path::new("."));
+    let mut result = BTreeMap::new();
+
+    if let Some(items) = document.get("module_registry").and_then(Value::as_array) {
+        for item in items {
+            let Some(command_id) = item.get("command_id").and_then(Value::as_u64) else {
+                continue;
+            };
+            let command_id = u32::try_from(command_id)
+                .map_err(|_| Error::Invalid("module command ID exceeds u32".to_owned()))?;
+            if !REQUIRED_IDS.contains(&command_id) {
+                continue;
+            }
+            let Some(path) = item.get("image_path").and_then(Value::as_str) else {
+                continue;
+            };
+            let size = item
+                .get("size")
+                .and_then(Value::as_u64)
+                .ok_or_else(|| Error::Invalid(format!("module 0x{command_id:02X} lacks size")))?;
+            result.insert(
+                command_id,
+                Artifact {
+                    path: root.join(path),
+                    size,
+                },
+            );
+        }
+    }
+    if let Some(streams) = document.get("streams").and_then(Value::as_array) {
+        for stream in streams {
+            let Some(records) = stream.get("records").and_then(Value::as_array) else {
+                continue;
+            };
+            for record in records {
+                let Some(command_id) = record.get("command_id").and_then(Value::as_u64) else {
+                    continue;
+                };
+                let command_id = u32::try_from(command_id)
+                    .map_err(|_| Error::Invalid("record command ID exceeds u32".to_owned()))?;
+                if !REQUIRED_IDS.contains(&command_id) {
+                    continue;
+                }
+                let Some(image) = record.get("image") else {
+                    continue;
+                };
+                let Some(path) = image.get("path").and_then(Value::as_str) else {
+                    continue;
+                };
+                let size = image.get("size").and_then(Value::as_u64).ok_or_else(|| {
+                    Error::Invalid(format!("record 0x{command_id:02X} lacks image size"))
+                })?;
+                result.insert(
+                    command_id,
+                    Artifact {
+                        path: root.join(path),
+                        size,
+                    },
+                );
+            }
+        }
+    }
+
+    let missing = REQUIRED_IDS
+        .iter()
+        .filter(|id| !result.contains_key(id))
+        .map(|id| format!("0x{id:02X}"))
+        .collect::<Vec<_>>();
+    if !missing.is_empty() {
+        return invalid(format!(
+            "module index lacks required IDs: {}",
+            missing.join(", ")
+        ));
+    }
+    for (&command_id, artifact) in &result {
+        let metadata = std::fs::metadata(&artifact.path)
+            .map_err(|error| Error::io("inspect artifact", &artifact.path, error))?;
+        if !metadata.is_file() || metadata.len() != artifact.size {
+            return invalid(format!(
+                "invalid artifact for module 0x{command_id:02X}: {}",
+                artifact.path.display()
+            ));
+        }
+    }
+    Ok(result)
+}
@@ -0,0 +1,37 @@
+use std::path::{Path, PathBuf};
+
+/// ELF restoration failure.
+#[derive(Debug, thiserror::Error)]
+pub enum Error {
+    #[error("{action} `{path}`: {source}")]
+    Io {
+        action: &'static str,
+        path: PathBuf,
+        #[source]
+        source: std::io::Error,
+    },
+    #[error("cannot parse module index: {0}")]
+    Json(#[from] serde_json::Error),
+    #[error(transparent)]
+    Crypto(#[from] senbei_crypto::android::Error),
+    #[error(transparent)]
+    Elf(#[from] senbei_elf::Error),
+    #[error("{0}")]
+    Invalid(String),
+}
+
+impl Error {
+    pub(crate) fn io(action: &'static str, path: &Path, source: std::io::Error) -> Self {
+        Self::Io {
+            action,
+            path: path.to_path_buf(),
+            source,
+        }
+    }
+}
+
+pub(crate) type Result<T> = std::result::Result<T, Error>;
+
+pub(crate) fn invalid<T>(message: impl Into<String>) -> Result<T> {
+    Err(Error::Invalid(message.into()))
+}
@@ -0,0 +1,6 @@
+mod artifact;
+mod error;
+mod pipeline;
+
+pub use error::Error;
+pub use pipeline::{RestoreOptions, RestoreReport, restore_libil2cpp};
@@ -0,0 +1,22 @@
+//! Platform-specific unpacking engines.
+
+pub mod android;
+pub mod windows;
+
+pub use windows::{
+    Detected, IntegrityReport, Kind, UnpackError, check_integrity, detect, unpack_auto,
+    unpack_auto_v, unpack_dll, unpack_dll_v, unpack_exe, unpack_exe_v,
+};
+
+/// Deterministic worker-thread cap shared by filesystem scanning and engines.
+pub fn thread_cap() -> usize {
+    if let Ok(value) = std::env::var("SENBEI_THREADS")
+        && let Ok(count) = value.trim().parse::<usize>()
+        && count >= 1
+    {
+        return count;
+    }
+    std::thread::available_parallelism()
+        .map(|count| count.get())
+        .unwrap_or(1)
+}
@@ -0,0 +1,3 @@
+mod pipeline;
+
+pub use pipeline::*;
@@ -13,9 +13,12 @@
 //!   CalculateChecksumWithSizeXor        -> primitives::calculate_checksum
 //!   CalculateCrc32                      -> crc32::compute (via above)

-use super::UnpackError;
-use super::bytecode::{Op, OpsLut, generate};
-use super::primitives::{self, *};
+use super::super::{
+    BufferOperation, BytecodeStage, DecompressionStage, DescriptorTable, SectionPipeline,
+    UnpackError,
+};
+use senbei_crypto::bytecode::{Op, OpsLut, generate};
+use senbei_crypto::primitives::{self, *};

 /// Read a signed 32-bit little-endian value.
 fn get_i32(d: &[u8], offset: i32) -> i32 {
@@ -68,6 +71,7 @@ fn decrypt_data4(
    key: i32,
    decomp_params: &[i32; 4],
    transform: Option<&[Op]>,
+    stage: DecompressionStage,
 ) -> Result<(), UnpackError> {
    let addr = get_i32(d, offset);
    let size = get_i32(d, offset + 4);
@@ -84,19 +88,17 @@ fn decrypt_data4(
        OpsLut::new(ops).map_region(d, addr as usize, size as usize);
    }

-    if size != decompressed_size {
-        // decompress reports corruption (after partial writes) via its bool;
-        // surface it instead of shipping a garbage block.
-        if !decompress(
+    if size != decompressed_size
+        && let Err(reason) = primitives::decompress_detailed(
            d,
            addr as u32,
            compressed_addr as u32,
            decomp_params[1] as u32,
            size as u32,
            decompressed_size as u32,
-        ) {
-            return Err(UnpackError::DecompressFailed);
-        }
+        )
+    {
+        return Err(UnpackError::StageDecompressionFailed { stage, reason });
    }
    Ok(())
 }
@@ -218,7 +220,11 @@ fn decrypt_and_decompress_data(
        // Guard: need 16 bytes at section_data_offset in `d`
        let off = section_data_offset as usize;
        if off.saturating_add(16) > d.len() {
-            return Err(UnpackError::OutOfBounds(off));
+            return Err(UnpackError::DescriptorOutOfBounds {
+                table: DescriptorTable::DllSectionBlocks,
+                offset: off,
+                image_len: d.len(),
+            });
        }
        decrypt_data6_shift6(d, section_data_offset, 16);
        let dest_offset = get_i32(d, section_data_offset);
@@ -246,10 +252,10 @@ fn decrypt_and_decompress_data(
        let lut = OpsLut::new(decrypt_func);
        let ko0 = decomp_params[0];
        let ko2 = decomp_params[2];
-        let ks_snap =
-            primitives::aes_schedule_snapshot(d, ko2 as u32).ok_or(UnpackError::Corrupt)?;
+        let ks_snap = primitives::aes_schedule_snapshot(d, ko2 as u32)
+            .ok_or(UnpackError::InvalidAesKeySchedule { offset: ko2 as u32 })?;
        let tab_snap = primitives::huffman_table_snapshot(d, ko0 as u32)
-            .ok_or(UnpackError::DecompressFailed)?;
+            .ok_or(UnpackError::InvalidHuffmanTable { offset: ko0 as u32 })?;
        let spans: Vec<(usize, usize)> = blocks
            .iter()
            .map(|b| {
@@ -277,12 +283,15 @@ fn decrypt_and_decompress_data(
                    b.size as u32,
                    b.expected_crc as u32,
                ) {
-                    return Err(UnpackError::DecompressFailed);
+                    return Err(UnpackError::SectionDecompressionFailed {
+                        pipeline: SectionPipeline::Dll,
+                        block: i,
+                    });
                }
            }
            Ok(())
        };
-        super::parallel::parallel_for(d, &spans, 1, do_block)?;
+        super::super::parallel::parallel_for(d, &spans, 1, do_block)?;
    }

    // Zero-fill loop.
@@ -292,7 +301,11 @@ fn decrypt_and_decompress_data(
        // decrypts 16 too, so guard 16 (an 8-byte guard would let
        // decrypt_data6_shift6 index past the end of a truncated descriptor).
        if off.saturating_add(16) > d.len() {
-            return Err(UnpackError::OutOfBounds(off));
+            return Err(UnpackError::DescriptorOutOfBounds {
+                table: DescriptorTable::DllZeroFill,
+                offset: off,
+                image_len: d.len(),
+            });
        }
        decrypt_data6_shift6(d, section_data_offset, 16);
        let zero_offset = get_i32(d, section_data_offset);
@@ -305,7 +318,12 @@ fn decrypt_and_decompress_data(
        for i in 0..zero_size {
            let idx = (zero_offset + i) as usize;
            if idx >= d.len() {
-                return Err(UnpackError::OutOfBounds(idx));
+                return Err(UnpackError::BufferRangeOutOfBounds {
+                    operation: BufferOperation::ZeroFill,
+                    offset: idx,
+                    size: 1,
+                    buffer_len: d.len(),
+                });
            }
            d[idx] = 0;
        }
@@ -324,12 +342,16 @@ pub fn unpack_dll(input: &[u8]) -> Result<Vec<u8>, UnpackError> {
 pub fn unpack_dll_v(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError> {
    // Trap any out-of-bounds panic from a truncated/garbled file and report it
    // as a clean error so the public API stays panic-free.
-    super::catch_unpack(move || unpack_dll_inner(input, verbose))
+    super::super::catch_unpack(move || unpack_dll_inner(input, verbose))
 }

 fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError> {
-    if input.len() < 4096 {
-        return Err(UnpackError::InputTooShort(input.len()));
+    const HEADER_LEN: usize = 4128;
+    if input.len() < HEADER_LEN {
+        return Err(UnpackError::InputTooShort {
+            actual: input.len(),
+            required: HEADER_LEN,
+        });
    }

    // `file_data` and `original_file_data` both borrow the same protected input.
@@ -347,15 +369,18 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        println!("  keys[6] anchor    = 0x{:08X}", keys[6] as u32);
    }

-    if !super::is_supported_magic(keys[1] as u32) {
-        return Err(UnpackError::DllUnpack(
-            "Not a Crackproof protected file (KONN magic mismatch)".into(),
-        ));
+    if !super::super::is_supported_magic(keys[1] as u32) {
+        return Err(UnpackError::HeaderMagicMismatch {
+            found: keys[1] as u32,
+        });
    }

    let pe_offset = get_i32(file_data, 60);
    if pe_offset < 0 || (pe_offset as usize).saturating_add(84) > file_data.len() {
-        return Err(UnpackError::DllUnpack("implausible PE offset".into()));
+        return Err(UnpackError::InvalidPeOffset {
+            offset: i64::from(pe_offset),
+            input_len: file_data.len(),
+        });
    }
    // This pipeline is PE32+-only: its header fixups write the data
    // directories at PE32+ offsets (pe+144..180, pe+136 for the DD blob). On a
@@ -363,14 +388,18 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
    // structurally plausible but unloadable file. Reject early with a clear
    // error so `unpack_auto`'s EXE-pipeline fallback handles PE32 DLLs (that
    // path is PE32-aware — see run_pe32), instead of us mangling them here.
-    if get_i32(file_data, pe_offset + 24) & 0xFFFF != 0x20B {
-        return Err(UnpackError::DllUnpack(
-            "not a PE32+ image (the DLL pipeline handles 64-bit only)".into(),
-        ));
+    let optional_magic = get_u16(file_data, (pe_offset + 24) as u32);
+    if optional_magic != 0x20B {
+        return Err(UnpackError::UnsupportedDllPeMagic {
+            found: optional_magic,
+        });
    }
    let size_of_image = get_i32(file_data, pe_offset + 80);
-    if size_of_image <= 0 || size_of_image as u64 > super::MAX_IMAGE_SIZE {
-        return Err(UnpackError::DllUnpack("implausible SizeOfImage".into()));
+    if size_of_image <= 0 || size_of_image as u64 > super::super::MAX_IMAGE_SIZE {
+        return Err(UnpackError::InvalidImageSize {
+            size: i64::from(size_of_image),
+            max: super::super::MAX_IMAGE_SIZE,
+        });
    }
    let mut out = vec![0u8; size_of_image as usize];
    let base_offset = keys[6] - keys[3] + 0x2000;
@@ -438,6 +467,16 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        println!("  checksum1       = 0x{:08X}", checksum1 as u32);
        println!("  decrypted_addr1 = 0x{:08X}", decrypted_addr1 as u32);
    }
+    let primary_end = decrypted_addr1.checked_add(3856);
+    if decrypted_addr1 < keys[3]
+        || primary_end.is_none_or(|end| end < 0 || end as usize > out.len())
+    {
+        return Err(UnpackError::InvalidDllPrimaryDescriptor {
+            address: decrypted_addr1 as u32,
+            minimum: keys[3] as u32,
+            image_len: out.len(),
+        });
+    }
    let import_offset = get_i32(&out, decrypted_addr1 + 3444);
    let decrypted_addr2_size = get_i32(&out, decrypted_addr1 + 3632);
    decrypt_data3(
@@ -512,6 +551,7 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        table_val ^ checksum2 ^ (xor_accumulator as i32),
        &decomp_params,
        None,
+        DecompressionStage::DllCodeBlock1,
    )?;

    let addr3b = get_i32(&out, decrypted_addr1 + 3728);
@@ -527,7 +567,7 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
    let crc_val = {
        let a = crc_data_addr as usize;
        let n = crc_data_size as usize;
-        super::crc32::compute(&out[a..a + n]) as i32
+        senbei_crypto::crc32::compute(&out[a..a + n]) as i32
    };
    let crc_xored = crc_data_size ^ crc_val;
    let trailing_val = get_i32(&out, crc_data_addr + crc_data_size - 4);
@@ -537,6 +577,7 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        crc_xored ^ (xor_accumulator as i32) ^ trailing_val,
        &decomp_params,
        None,
+        DecompressionStage::DllCodeBlock2,
    )?;

    let checksum3 = calculate_checksum(&out, (decrypted_addr1 + 3480) as u32) as i32;
@@ -549,6 +590,7 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        (not_val ^ (xor_key as u32)) as i32,
        &decomp_params,
        None,
+        DecompressionStage::DllCodeBlock3,
    )?;

    let addr4 = get_i32(&out, addr4_offset);
@@ -574,8 +616,9 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        lfsr_seed_val = lfsr_seed_val.wrapping_add(k);
    }

-    let decrypt_func = generate(&out, lfsr as u32)
-        .ok_or_else(|| UnpackError::DllUnpack("Failed to build decryption expression".into()))?;
+    let decrypt_func = generate(&out, lfsr as u32).ok_or(UnpackError::BytecodeGenerationFailed(
+        BytecodeStage::DllPrimaryDecryptor,
+    ))?;

    let addr5_offset = decrypted_addr1 + 3840;
    let addr5 = get_i32(&out, addr5_offset);
@@ -585,6 +628,7 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
        lfsr_seed_val ^ xor_key ^ checksum4,
        &decomp_params,
        Some(&decrypt_func),
+        DecompressionStage::DllCodeBlock4,
    )?;
    if verbose {
        println!("[7/9] Decrypting code block 4 (addr5)...");
@@ -603,9 +647,9 @@ fn unpack_dll_inner(input: &[u8], verbose: bool) -> Result<Vec<u8>, UnpackError>
    let lfsr2 = metadata_offset + 88;
    decrypt_data6(&mut out, lfsr2 as u32);

-    let decrypt_func2 = generate(&out, lfsr2 as u32).ok_or_else(|| {
-        UnpackError::DllUnpack("Failed to build second decryption expression".into())
-    })?;
+    let decrypt_func2 = generate(&out, lfsr2 as u32).ok_or(
+        UnpackError::BytecodeGenerationFailed(BytecodeStage::DllSectionDecryptor),
+    )?;

    let section_image_base = 4095 - get_i32(original_file_data, 4224);
    let section_data_offset = get_i32(&out, addr5 + 11976);
@@ -0,0 +1,260 @@
+pub use senbei_crypto::{BufferOperation, DecompressionFailure};
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum DecompressionStage {
+    ExeStage3,
+    ExeStage3Secondary,
+    ExeStage4,
+    ExeStage5,
+    Pe32FourthStage,
+    Pe32FifthStage,
+    Pe32SeventhStage,
+    DllCodeBlock1,
+    DllCodeBlock2,
+    DllCodeBlock3,
+    DllCodeBlock4,
+}
+
+impl std::fmt::Display for DecompressionStage {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::ExeStage3 => "EXE stage3",
+            Self::ExeStage3Secondary => "EXE secondary stage3",
+            Self::ExeStage4 => "EXE stage4",
+            Self::ExeStage5 => "EXE stage5",
+            Self::Pe32FourthStage => "PE32 fourth stage",
+            Self::Pe32FifthStage => "PE32 fifth stage",
+            Self::Pe32SeventhStage => "PE32 seventh stage",
+            Self::DllCodeBlock1 => "DLL code block 1",
+            Self::DllCodeBlock2 => "DLL code block 2",
+            Self::DllCodeBlock3 => "DLL code block 3",
+            Self::DllCodeBlock4 => "DLL code block 4",
+        })
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum BytecodeStage {
+    ExeStage4,
+    ExeStage5,
+    Pe32CustomDecryptor,
+    Pe32FileDecryptor,
+    DllPrimaryDecryptor,
+    DllSectionDecryptor,
+}
+
+impl std::fmt::Display for BytecodeStage {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::ExeStage4 => "EXE stage4",
+            Self::ExeStage5 => "EXE stage5",
+            Self::Pe32CustomDecryptor => "PE32 custom decryptor",
+            Self::Pe32FileDecryptor => "PE32 file decryptor",
+            Self::DllPrimaryDecryptor => "DLL primary decryptor",
+            Self::DllSectionDecryptor => "DLL section decryptor",
+        })
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum SectionPipeline {
+    ExePe32Plus,
+    ExePe32,
+    Dll,
+}
+
+impl std::fmt::Display for SectionPipeline {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::ExePe32Plus => "PE32+ EXE",
+            Self::ExePe32 => "PE32 EXE",
+            Self::Dll => "DLL",
+        })
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum DescriptorTable {
+    DllSectionBlocks,
+    DllZeroFill,
+}
+
+impl std::fmt::Display for DescriptorTable {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::DllSectionBlocks => "DLL section-block",
+            Self::DllZeroFill => "DLL zero-fill",
+        })
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)]
+#[non_exhaustive]
+pub enum UnpackError {
+    #[error("input too short (need at least {required} bytes, got {actual})")]
+    InputTooShort { actual: usize, required: usize },
+
+    #[error("decrypted header magic mismatch (got 0x{found:08X})")]
+    HeaderMagicMismatch { found: u32 },
+
+    #[error("anchor field not found — corrupt data or wrong offset")]
+    AnchorNotFound,
+
+    #[error("stage1 descriptor not found near anchor 0x{anchor:08X}")]
+    Stage1DescriptorNotFound { anchor: u32 },
+
+    #[error("stage2 field not found — corrupt data or wrong offset")]
+    Stage2NotFound,
+
+    #[error("chk_src_start not found — corrupt data or wrong offset")]
+    ChkSrcStartNotFound,
+
+    #[error("table_start not found — corrupt data or wrong offset")]
+    TableStartNotFound,
+
+    #[error("{0} bytecode generation failed — corrupt data or wrong offset")]
+    BytecodeGenerationFailed(BytecodeStage),
+
+    #[error("stage5 marker not found — this build's layout is not supported by this unpacker")]
+    Stage5MarkerNotFound,
+
+    #[error("not a Crackproof-protected file")]
+    NotCrackproof,
+
+    #[error("invalid PE header offset {offset} for {input_len}-byte input")]
+    InvalidPeOffset { offset: i64, input_len: usize },
+
+    #[error("DLL pipeline requires PE32+ optional-header magic, got 0x{found:04X}")]
+    UnsupportedDllPeMagic { found: u16 },
+
+    #[error(
+        "DLL primary descriptor address 0x{address:08X} is below layout base 0x{minimum:08X} or outside {image_len}-byte image"
+    )]
+    InvalidDllPrimaryDescriptor {
+        address: u32,
+        minimum: u32,
+        image_len: usize,
+    },
+
+    #[error("invalid SizeOfImage {size}; expected 1..={max}")]
+    InvalidImageSize { size: i64, max: u64 },
+
+    #[error(
+        "{operation} range out of bounds (offset {offset}, size {size}, buffer length {buffer_len})"
+    )]
+    BufferRangeOutOfBounds {
+        operation: BufferOperation,
+        offset: usize,
+        size: usize,
+        buffer_len: usize,
+    },
+
+    #[error("managed stub {region} restoration failed: {source}")]
+    ManagedStubRestoreFailed {
+        region: &'static str,
+        #[source]
+        source: senbei_pe::Error,
+    },
+
+    #[error(
+        "EXE checksum descriptor at 0x{descriptor:08X} points outside input (offset {offset}, size {size}, input length {image_len})"
+    )]
+    ExeChecksumRangeOutOfBounds {
+        descriptor: u32,
+        offset: usize,
+        size: usize,
+        image_len: usize,
+    },
+
+    #[error(
+        "{table} descriptor out of bounds (offset {offset}, size 16, image length {image_len})"
+    )]
+    DescriptorOutOfBounds {
+        table: DescriptorTable,
+        offset: usize,
+        image_len: usize,
+    },
+
+    #[error("PE32 tbl not found — corrupt data or wrong offset")]
+    Pe32TblNotFound,
+
+    #[error("PE32 thirdStage decrypt failed — corrupt data or wrong offset")]
+    Pe32ThirdStageFailed,
+
+    #[error("PE32 customDecryptor not found in sevenStage")]
+    Pe32CustomDecryptorNotFound,
+
+    #[error("PE32 eighthStageKey not found")]
+    Pe32EighthKeyNotFound,
+
+    #[error("PE32 file LFSR not found in eighthStage")]
+    Pe32FileLfsrNotFound,
+
+    #[error("{stage} decompression failed: {reason}")]
+    StageDecompressionFailed {
+        stage: DecompressionStage,
+        reason: DecompressionFailure,
+    },
+
+    #[error("{pipeline} section block {block} decompression failed")]
+    SectionDecompressionFailed {
+        pipeline: SectionPipeline,
+        block: usize,
+    },
+
+    #[error("AES key schedule is outside the image at offset {offset}")]
+    InvalidAesKeySchedule { offset: u32 },
+
+    #[error("Huffman table is outside the image at offset {offset}")]
+    InvalidHuffmanTable { offset: u32 },
+
+    #[error("DLL pipeline failed: {dll}; EXE fallback failed: {exe}")]
+    PipelineFallbackFailed {
+        dll: Box<UnpackError>,
+        exe: Box<UnpackError>,
+    },
+
+    #[error(
+        "PE32 second-stage range is invalid (offset {offset}, size {size}, image length {image_len})"
+    )]
+    Pe32SecondStageRangeInvalid {
+        offset: u32,
+        size: u32,
+        image_len: usize,
+    },
+
+    #[error("PE32 relocation-data descriptor not found")]
+    Pe32RelocationDataNotFound,
+
+    #[error("file decryptor candidate failed structural validation")]
+    FileDecryptorValidationFailed,
+
+    #[error("PE32 memory image could not be rebuilt as a file-layout PE")]
+    Pe32OutputLayoutInvalid,
+
+    #[error("internal panic at {file}:{line}:{column}: {message}")]
+    InternalPanic {
+        message: String,
+        file: String,
+        line: u32,
+        column: u32,
+    },
+}
+
+impl From<senbei_crypto::Error> for UnpackError {
+    fn from(error: senbei_crypto::Error) -> Self {
+        match error {
+            senbei_crypto::Error::BufferRangeOutOfBounds {
+                operation,
+                offset,
+                size,
+                buffer_len,
+            } => Self::BufferRangeOutOfBounds {
+                operation,
+                offset,
+                size,
+                buffer_len,
+            },
+        }
+    }
+}
@@ -0,0 +1,3 @@
+mod pipeline;
+
+pub use pipeline::*;
@@ -42,39 +42,61 @@ fn rd_u32(d: &[u8], off: u32) -> Option<u32> {
        .map(|s| u32::from_le_bytes([s[0], s[1], s[2], s[3]]))
 }

-/// A parsed section-table entry (only the fields we translate against).
-struct Section {
-    va: u32,
-    vsize: u32,
-    raw_ptr: u32,
-    raw_size: u32,
-    chars: u32,
-}
+/// PE format section data used by the integrity policy.
+type Section = senbei_pe::Section;

 /// Walk the output's own section table and translate an RVA to a file offset.
 /// Works for both memory-image output (raw_ptr == va) and compacted disk
 /// output (real raw pointers), because it consults whatever the output declares.
 /// Returns the offset only if the translated range `[off, off+need)` lies inside
 /// the file.
-fn rva_to_off(secs: &[Section], file_len: usize, rva: u32, need: u32) -> Option<u32> {
-    for s in secs {
-        // The mapped span is the larger of virtual and raw size, so an RVA that
-        // falls in the virtual tail of a section still resolves.
-        let span = s.vsize.max(s.raw_size);
-        if span == 0 {
-            continue;
-        }
-        if rva >= s.va && rva < s.va.wrapping_add(span) {
-            let delta = rva - s.va;
-            let off = s.raw_ptr.checked_add(delta)?;
-            let end = off.checked_add(need)?;
-            if (end as usize) <= file_len {
-                return Some(off);
-            }
-            return None;
+fn rva_to_off(data: &[u8], headers: senbei_pe::Headers, rva: u32, need: u32) -> Option<u32> {
+    let offset = u32::try_from(senbei_pe::rva_to_offset(data, headers, rva).ok()?).ok()?;
+    let end = offset.checked_add(need)?;
+    (usize::try_from(end).ok()? <= data.len()).then_some(offset)
+}
+
+fn is_executable_rva(secs: &[Section], rva: u32) -> bool {
+    secs.iter().any(|section| {
+        let span = section.virtual_size.max(section.raw_size);
+        rva >= section.virtual_address
+            && rva < section.virtual_address.wrapping_add(span)
+            && (section.characteristics & 0x2000_0000) != 0
+    })
+}
+
+fn check_common_entry_branches(
+    stub: &[u8],
+    ep: u32,
+    secs: &[Section],
+    report: &mut IntegrityReport,
+) {
+    if stub.len() < 18
+        || stub[0..3] != [0x48, 0x83, 0xEC]
+        || stub[4] != 0xE8
+        || stub[9..12] != [0x48, 0x83, 0xC4]
+        || stub[12] != stub[3]
+        || stub[13] != 0xE9
+    {
+        return;
+    }
+    for (name, rel_off, instruction_len) in [("call", 5usize, 9i64), ("jump", 14usize, 18i64)] {
+        let rel = i32::from_le_bytes([
+            stub[rel_off],
+            stub[rel_off + 1],
+            stub[rel_off + 2],
+            stub[rel_off + 3],
+        ]) as i64;
+        let target = i64::from(ep) + instruction_len + rel;
+        let valid = u32::try_from(target)
+            .ok()
+            .is_some_and(|rva| is_executable_rva(secs, rva));
+        if !valid {
+            report.issues.push(format!(
+                "entry point {name} target 0x{target:X} is outside executable sections (DD8 selection is likely wrong)"
+            ));
        }
    }
-    None
 }

 /// Inspect an unpacked PE image and report any defect that would make the OS
@@ -108,7 +130,6 @@ pub fn check(out: &[u8]) -> IntegrityReport {
            return r;
        }
    };
-    let opt_hdr_size = rd_u16(out, pe_off.wrapping_add(20)).unwrap_or(0) as u32;
    let opt = pe_off.wrapping_add(24);
    let magic = match rd_u16(out, opt) {
        Some(v) => v,
@@ -138,42 +159,38 @@ pub fn check(out: &[u8]) -> IntegrityReport {
    }

    // --- Section table ------------------------------------------------------
-    let sec_table = opt.wrapping_add(opt_hdr_size);
-    let mut secs: Vec<Section> = Vec::new();
-    for i in 0..num_sections {
-        let base = sec_table.wrapping_add(i * 40);
-        // If the table runs past EOF the image is structurally broken.
-        let (vsize, va, raw_size, raw_ptr, chars) = match (
-            rd_u32(out, base.wrapping_add(8)),
-            rd_u32(out, base.wrapping_add(12)),
-            rd_u32(out, base.wrapping_add(16)),
-            rd_u32(out, base.wrapping_add(20)),
-            rd_u32(out, base.wrapping_add(36)),
-        ) {
-            (Some(a), Some(b), Some(c), Some(d), Some(e)) => (a, b, c, d, e),
-            _ => {
-                r.issues
-                    .push("section table extends past end of file".into());
-                return r;
-            }
-        };
-        // Raw data must lie within the file for compacted (disk-layout) output.
-        if raw_size != 0 {
-            let end = raw_ptr.wrapping_add(raw_size) as usize;
-            if end > file_len {
-                r.issues.push(format!(
-                    "section #{i} raw data [0x{raw_ptr:X}..0x{end:X}] exceeds file size 0x{file_len:X}"
-                ));
-            }
+    let headers = match senbei_pe::parse(out) {
+        Ok(headers) => headers,
+        Err(_) => {
+            r.issues
+                .push("section table extends past end of file".into());
+            return r;
        }
-        secs.push(Section {
-            va,
-            vsize,
-            raw_ptr,
-            raw_size,
-            chars,
-        });
-    }
+    };
+    let parsed_sections = match senbei_pe::sections(out, headers) {
+        Ok(sections) => sections,
+        Err(_) => {
+            r.issues
+                .push("section table extends past end of file".into());
+            return r;
+        }
+    };
+    let secs: Vec<Section> = parsed_sections
+        .into_iter()
+        .enumerate()
+        .map(|(i, section)| {
+            if section.raw_size != 0 {
+                let end = section.raw_offset.wrapping_add(section.raw_size) as usize;
+                if end > file_len {
+                    r.issues.push(format!(
+                        "section #{i} raw data [0x{:X}..0x{end:X}] exceeds file size 0x{file_len:X}",
+                        section.raw_offset
+                    ));
+                }
+            }
+            section
+        })
+        .collect();

    // --- Managed (CLR) detection ------------------------------------------
    // The COR20 (CLR) data directory, when present and non-zero, marks a managed
@@ -223,7 +240,7 @@ pub fn check(out: &[u8]) -> IntegrityReport {
            r.issues.push("entry point RVA is zero".into());
        }
    } else if !is_managed {
-        match rva_to_off(&secs, file_len, ep, 16) {
+        match rva_to_off(out, headers, ep, 16) {
            None => {
                r.issues.push(format!(
                    "entry point RVA 0x{ep:X} does not map into any section"
@@ -245,14 +262,19 @@ pub fn check(out: &[u8]) -> IntegrityReport {
                }
                // The entry must live in an executable section.
                let exec = secs.iter().any(|s| {
-                    let span = s.vsize.max(s.raw_size);
-                    ep >= s.va && ep < s.va.wrapping_add(span) && (s.chars & 0x2000_0000) != 0
+                    let span = s.virtual_size.max(s.raw_size);
+                    ep >= s.virtual_address
+                        && ep < s.virtual_address.wrapping_add(span)
+                        && (s.characteristics & 0x2000_0000) != 0
                });
                if !exec {
                    r.issues.push(format!(
                        "entry point RVA 0x{ep:X} is not in an executable section"
                    ));
                }
+                if let Some(entry_stub) = out.get(off as usize..off as usize + 18) {
+                    check_common_entry_branches(entry_stub, ep, &secs, &mut r);
+                }
            }
        }
    }
@@ -270,7 +292,7 @@ pub fn check(out: &[u8]) -> IntegrityReport {
    if !is_managed {
        let imp_rva = rd_u32(out, dd_base.wrapping_add(8)).unwrap_or(0);
        if imp_rva != 0 {
-            match rva_to_off(&secs, file_len, imp_rva, 20) {
+            match rva_to_off(out, headers, imp_rva, 20) {
                None => r.issues.push(format!(
                    "import directory RVA 0x{imp_rva:X} does not map into any section"
                )),
@@ -285,7 +307,7 @@ pub fn check(out: &[u8]) -> IntegrityReport {
                        if name_rva == 0 {
                            break;
                        }
-                        match rva_to_off(&secs, file_len, name_rva, 1) {
+                        match rva_to_off(out, headers, name_rva, 1) {
                            None => r.issues.push(format!(
                                "import descriptor {i} DLL name RVA 0x{name_rva:X} does not map into any section"
                            )),
@@ -313,7 +335,7 @@ pub fn check(out: &[u8]) -> IntegrityReport {
    // still refuses to load. Validate: COR20 cb == 0x48, and the MetaData stream
    // begins with the "BSJB" signature.
    if is_managed {
-        match rva_to_off(&secs, file_len, clr_rva, 0x48) {
+        match rva_to_off(out, headers, clr_rva, 0x48) {
            None => r.issues.push(format!(
                "CLR (COR20) directory RVA 0x{clr_rva:X} does not map into any section"
            )),
@@ -327,7 +349,7 @@ pub fn check(out: &[u8]) -> IntegrityReport {
                    // MetaData RVA/size live at COR20 + 0x08 / + 0x0C.
                    let md_rva = rd_u32(out, coff.wrapping_add(8)).unwrap_or(0);
                    if md_rva != 0 {
-                        match rva_to_off(&secs, file_len, md_rva, 4) {
+                        match rva_to_off(out, headers, md_rva, 4) {
                            None => r.issues.push(format!(
                                "CLR MetaData RVA 0x{md_rva:X} does not map into any section"
                            )),
@@ -363,3 +385,40 @@ fn looks_like_dll_name(d: &[u8], off: u32) -> bool {
    }
    d[start..end].iter().all(|&b| (0x20..0x7F).contains(&b))
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn executable_text() -> Vec<Section> {
+        vec![Section {
+            virtual_address: 0x1000,
+            virtual_size: 0x4000,
+            raw_offset: 0x1000,
+            raw_size: 0x4000,
+            characteristics: 0x6000_0020,
+        }]
+    }
+
+    #[test]
+    fn common_entry_stub_rejects_out_of_image_branches() {
+        let stub = [
+            0x48, 0x83, 0xEC, 0x28, 0xE8, 0x5B, 0x02, 0x41, 0x00, 0x48, 0x83, 0xC4, 0x28, 0xE9,
+            0x7A, 0xFE, 0x54, 0xFF,
+        ];
+        let mut report = IntegrityReport::default();
+        check_common_entry_branches(&stub, 0x1264, &executable_text(), &mut report);
+        assert_eq!(report.issues.len(), 2);
+    }
+
+    #[test]
+    fn common_entry_stub_accepts_executable_branches() {
+        let stub = [
+            0x48, 0x83, 0xEC, 0x28, 0xE8, 0x5B, 0x02, 0x00, 0x00, 0x48, 0x83, 0xC4, 0x28, 0xE9,
+            0x7A, 0xFE, 0xFF, 0xFF,
+        ];
+        let mut report = IntegrityReport::default();
+        check_common_entry_branches(&stub, 0x1264, &executable_text(), &mut report);
+        assert!(report.ok());
+    }
+}
@@ -0,0 +1,14 @@
+//! Internal PE layout discovery and image reconstruction.
+
+mod dd8;
+mod discovery;
+mod image;
+
+pub(super) use dd8::{select_dd8_formula_pe32, select_dd8_shift};
+pub(super) use discovery::{
+    discover_eighth_slots, find_bytecode_offset, find_lfsr_block, find_str_pos, find_tbl_pe32,
+    find_v_after_pad, find_v4_offset, get_string_to_null, section_name, trial_decrypt5_u32,
+};
+pub(super) use image::{
+    compact_memory_image_to_pe, move_pe32_imports_to_kmiat, pe32_imports_already_match_idata_layout,
+};
@@ -0,0 +1,609 @@
+//! Validation-driven selection for per-page text transforms.
+
+use super::discovery::trial_decrypt5_u32;
+
+/// PE32 `.text` dd8 key-formula selection with a skip decision. The packer keys
+/// the per-page XOR either with `page+1` or `0x8000*(page+1)`; the formula is
+/// not recorded. Replays the dd8 page pass on a scratch copy of sample pages
+/// (25/50/75% of `.text`) under each formula and counts how many positions
+/// decode to `0xCC` (int3 padding).
+///
+/// Returns `Some(true)` for the `0x8000*(page+1)` formula, `Some(false)` for
+/// `page+1`, or `None` when `.text` must NOT be dd8-decrypted at all. The packer
+/// dd8-encrypts `.text` on EXEs (so unpacking must replay it) but leaves a native
+/// DLL's `.text` plaintext; replaying dd8 there scrambles ~1 byte per 16-byte
+/// block. The decision: dd8 only *restores* int3 padding when `.text` was
+/// genuinely encrypted, so apply it only when the chosen formula's whole-page
+/// 0xCC count rises *clearly* above the no-dd8 baseline; otherwise skip.
+///
+/// "Clearly" matters: dd8 XORs 255 positions per page with pseudo-random bytes,
+/// so on an already-plaintext `.text` it manufactures ~1 spurious `0xCC` per
+/// sampled page for free (255/256 expected). A bare `best > baseline` test is
+/// therefore biased towards *applying* dd8 on exactly the inputs that must skip
+/// it — and a wrongly-applied dd8 is silent: it scrambles ~1 byte per 16 with no
+/// error and nothing downstream (not even `integrity::check`, which only reads
+/// 16 bytes at the entry point) notices. The [`MIN_DD8_NET_GAIN`] floor below is
+/// the PE32 counterpart of the margin+floor `select_dd8_shift` already applies
+/// on PE32+ for the same failure mode.
+pub fn select_dd8_formula_pe32(data: &[u8], text_off: u32, text_size: u32) -> Option<bool> {
+    let num_pages_total = text_size / 0x1000;
+    let mut sample_pages: Vec<u32> = Vec::new();
+    for frac in [0.25f64, 0.5, 0.75] {
+        let pg = (num_pages_total as f64 * frac) as u32;
+        if pg > 0 && pg < num_pages_total {
+            sample_pages.push(pg);
+        }
+    }
+    if sample_pages.is_empty() && num_pages_total > 1 {
+        sample_pages.push(num_pages_total / 2);
+    }
+    let score = |big: bool| -> i64 {
+        let mut total = 0i64;
+        for &sp in &sample_pages {
+            let pg_off = (text_off + sp * 0x1000) as usize;
+            if pg_off + 0x1000 > data.len() {
+                continue;
+            }
+            let mut buf = [0u8; 0x1000];
+            buf.copy_from_slice(&data[pg_off..pg_off + 0x1000]);
+            let pk = if big {
+                0x8000u32.wrapping_mul(sp.wrapping_add(1))
+            } else {
+                sp.wrapping_add(1)
+            };
+            let mut k = pk;
+            let rk = k.rotate_right(15);
+            k = rk;
+            for bi in 1..256u32 {
+                let rk = k.rotate_right(15);
+                let ri = rk.wrapping_add(bi);
+                k = ri.wrapping_add(bi);
+                let tidx = (bi.wrapping_mul(16).wrapping_add(ri & 0xF)) as usize;
+                if tidx < buf.len() {
+                    buf[tidx] ^= k as u8;
+                }
+            }
+            total += buf.iter().filter(|&&b| b == 0xCC).count() as i64;
+        }
+        total
+    };
+    let s_small = score(false);
+    let s_big = score(true);
+    // Baseline: whole-page 0xCC over the same sample pages with NO dd8. dd8 only
+    // rewrites 255 bytes per page, so comparing the chosen formula's whole-page
+    // 0xCC against this baseline reveals whether dd8 *restores* int3 padding
+    // (count rises -> .text was packer-encrypted, apply) or merely scrambles
+    // already-plaintext code (count falls -> native-DLL .text left intact, skip).
+    let mut baseline: i64 = 0;
+    for &sp in &sample_pages {
+        let pg_off = (text_off + sp * 0x1000) as usize;
+        if pg_off + 0x1000 > data.len() {
+            continue;
+        }
+        baseline += data[pg_off..pg_off + 0x1000]
+            .iter()
+            .filter(|&&b| b == 0xCC)
+            .count() as i64;
+    }
+    let big = s_big > s_small;
+    let best = s_small.max(s_big);
+    // Minimum net 0xCC gain over the baseline before dd8 is applied. Noise on an
+    // already-plaintext `.text` is ~1 manufactured 0xCC per sampled page (3 pages
+    // -> ~3); every corpus build that genuinely needs dd8 gains +154 or more
+    // (observed +154 and +312), and the one native DLL that must skip scores -18.
+    // A floor of 32 sits ~10x above the noise and ~5x below the smallest true
+    // positive, so it changes no existing decision.
+    const MIN_DD8_NET_GAIN: i64 = 32;
+    let apply = best.saturating_sub(baseline) >= MIN_DD8_NET_GAIN;
+    if std::env::var("SEL_DIAG").is_ok() {
+        eprintln!(
+            "SEL pe32 dd8 s_small={} s_big={} baseline={} gain={} big={} apply={}",
+            s_small,
+            s_big,
+            baseline,
+            best - baseline,
+            big,
+            apply
+        );
+    }
+    // When no interior pages could be sampled (tiny .text) we cannot measure the
+    // effect; preserve the historical behavior of applying dd8.
+    if sample_pages.is_empty() || apply {
+        Some(big)
+    } else {
+        None
+    }
+}
+
+// ---------------------------------------------------------------------------
+// dd8 page-XOR shift selection.
+//
+// The packer scrambles ~1 byte per 16-byte block of .text via decrypt_data8,
+// keyed by `page_idx << shift` (absolute page index = text_va >> 12). Observed
+// shifts are 0 and 15. The shift is NOT stored in any header/config field, so
+// the decision must be validated against the resulting .text content.
+//
+// A recognised CRT entry stub is the strongest oracle: decode skip/0/15 and
+// require both of its direct rel32 branches to land in executable .text. This
+// includes the call/jump displacement bytes themselves; an older entry oracle
+// wildcarded those bytes and could accept a stub whose opcodes looked right but
+// whose branch targets were outside the image.
+//
+// Other entry shapes fall back to padding statistics over a few sample pages
+// (head/tail margin skipped: entry/exit regions have atypical padding density).
+// The primary signal is a *structural* fingerprint: the MSVC function-end
+// padding pattern, a 0xC3 RET opcode followed by a run of >= 4 0xCC int3 bytes.
+// dd8 XORs one pseudo-random byte per 16-byte block, so an already-plaintext
+// page keeps its padding runs only under "no dd8", while a packer-encrypted
+// page restores them only under the correct shift — a wrong candidate destroys
+// every run it touches and essentially never manufactures a RET followed by a
+// long int3 run by chance. This separates the states far more cleanly than a
+// bare 0xCC count, which a wrong candidate inflates for free (~255 coincidences
+// per page at p=1/256).
+//
+// When no candidate produces any RET-anchored padding (sampled pages with
+// dense code and no padded epilogues), the fingerprint is silent, so the
+// decision falls back to the older mutated-position 0xCC count. Both signals
+// use the same decision rule: a candidate must beat the no-dd8 baseline by a
+// clear margin AND an absolute floor, otherwise dd8 is skipped — a wrongly
+// applied dd8 scrambles ~1 byte per 16 with no error surfaced downstream.
+// ---------------------------------------------------------------------------
+pub fn select_dd8_shift(data: &[u8], text_va: u32, text_size: u32, info3: u32) -> u32 {
+    if let Some((shift, scores)) = select_dd8_by_entry_stub(data, text_va, text_size, info3) {
+        if std::env::var("SEL_DIAG").is_ok() {
+            eprintln!(
+                "SEL dd8 entry best_shift={} none={} s0={} s15={}",
+                shift, scores[0], scores[1], scores[2]
+            );
+        }
+        return shift;
+    }
+    let num_pages_total = text_size >> 12;
+    // Fewer than two pages: nothing meaningful to sample; preserve the
+    // historical behavior (shift 0 — the dd8 loop is empty or single-page).
+    if num_pages_total < 2 {
+        return 0;
+    }
+    let text_off = text_va as usize;
+
+    // Sample up to 4 pages, skipping a head/tail margin (entry/exit regions
+    // have atypical padding density). Small .text: sample every page.
+    let mut sample_pages: Vec<u32> = Vec::new();
+    if num_pages_total <= 4 {
+        sample_pages.extend(0..num_pages_total);
+    } else {
+        let margin = (num_pages_total / 8).max(1);
+        let lo = margin;
+        let hi = num_pages_total - margin;
+        if hi <= lo {
+            sample_pages.extend(0..num_pages_total);
+        } else {
+            let step = ((hi - lo) / 4).max(1);
+            let mut i = 0;
+            while i < 4 {
+                let p = lo + i * step;
+                if p < num_pages_total {
+                    sample_pages.push(p);
+                }
+                i += 1;
+            }
+        }
+    }
+    if sample_pages.is_empty() {
+        return 0;
+    }
+
+    let abs_base = text_va >> 12;
+    // Require a clear 2x margin over the already-plaintext baseline AND an
+    // absolute floor. The 2x test alone trips on noise when the counts are
+    // tiny: an external-companion DLL whose .text is already plaintext scores
+    // s15=4 vs none=1 — a spurious 4x — and gets dd8 wrongly applied,
+    // corrupting ~1 byte per 16. The floor rejects that noise while sitting
+    // far below every genuinely-encrypted build's score.
+    const MIN_DD8_HITS: u32 = 8;
+    let margin_pick = |none: u32, s0: u32, s15: u32| -> u32 {
+        let mut best_score = none;
+        let mut best_shift = 99u32; // 99 == skip dd8
+        for (shift, hits) in [(0u32, s0), (15u32, s15)] {
+            if hits > best_score {
+                best_score = hits;
+                best_shift = shift;
+            }
+        }
+        if best_shift != 99 && (best_score < none * 2 || best_score < MIN_DD8_HITS) {
+            best_shift = 99;
+        }
+        best_shift
+    };
+
+    // Primary: RET+int3 padding fingerprint. The fingerprint is diluted across
+    // the whole page (dd8 touches only 255 of 4096 bytes, so even an encrypted
+    // page keeps most of its padding runs), so instead of the fallback's 2x
+    // margin the gate is a *positive delta* over the no-dd8 baseline: on an
+    // already-plaintext .text each wrong shift destroys runs (scores below the
+    // baseline), while the correct shift on an encrypted page restores them
+    // (scores above it). The floor on the delta rejects noise-level gains.
+    let r_none = fingerprint_score(data, text_off, abs_base, &sample_pages, None);
+    let r0 = fingerprint_score(data, text_off, abs_base, &sample_pages, Some(0));
+    let r15 = fingerprint_score(data, text_off, abs_base, &sample_pages, Some(15));
+    // Fallback: mutated-position 0xCC count, for pages whose code has no
+    // RET-anchored padding at all (the fingerprint is silent there).
+    let (none_hits, s0, s15);
+    let best_shift = if r_none != 0 || r0 != 0 || r15 != 0 {
+        none_hits = 0;
+        s0 = 0;
+        s15 = 0;
+        let mut best_score = r_none;
+        let mut shift = 99u32;
+        for (s, score) in [(0u32, r0), (15u32, r15)] {
+            if score > best_score {
+                best_score = score;
+                shift = s;
+            }
+        }
+        if shift != 99 && best_score.saturating_sub(r_none) < MIN_DD8_HITS {
+            shift = 99;
+        }
+        shift
+    } else {
+        none_hits = score_dd8_baseline(data, text_off, &sample_pages);
+        s0 = score_dd8_shift(data, text_off, text_va, &sample_pages, 0);
+        s15 = score_dd8_shift(data, text_off, text_va, &sample_pages, 15);
+        margin_pick(none_hits, s0, s15)
+    };
+    if std::env::var("SEL_DIAG").is_ok() {
+        eprintln!(
+            "SEL dd8 best_shift={} fp=({},{},{}) cc=({},{},{}) samples={:?}",
+            best_shift, r_none, r0, r15, none_hits, s0, s15, sample_pages
+        );
+    }
+    best_shift
+}
+
+/// Minimum 0xCC run length after a RET for the run to count as MSVC
+/// function-end padding.
+const MIN_CC_RUN: u32 = 4;
+
+/// Total length of MSVC function-end padding runs in a page: each 0xC3 byte
+/// followed by >= [`MIN_CC_RUN`] 0xCC bytes contributes the run length.
+fn ret_int3_score(page: &[u8]) -> u32 {
+    let mut total = 0u32;
+    let mut i = 0;
+    while i < page.len() {
+        if page[i] == 0xC3 {
+            let mut j = i + 1;
+            while j < page.len() && page[j] == 0xCC {
+                j += 1;
+            }
+            let run = (j - i - 1) as u32;
+            if run >= MIN_CC_RUN {
+                total += run;
+            }
+            i = j;
+        } else {
+            i += 1;
+        }
+    }
+    total
+}
+
+/// Replay the dd8 page-XOR in place on one sample page.
+fn dd8_apply(buf: &mut [u8; 0x1000], abs_page: u32, shift: u32) {
+    let mut key = abs_page << shift;
+    for bi in 0..256u32 {
+        let mixed = key.rotate_right(15).wrapping_add(bi);
+        key = mixed.wrapping_add(bi);
+        // The packer's dd8 loop does not XOR block i=0 (see decrypt_data8).
+        if bi == 0 {
+            continue;
+        }
+        let tidx = (bi.wrapping_mul(16).wrapping_add(mixed & 0xF)) as usize;
+        buf[tidx] ^= key as u8;
+    }
+}
+
+/// Sum the RET+int3 fingerprint over the sample pages for one candidate
+/// (`None` = the no-dd8 baseline, page as-is).
+fn fingerprint_score(
+    data: &[u8],
+    text_off: usize,
+    abs_base: u32,
+    sample_pages: &[u32],
+    shift: Option<u32>,
+) -> u32 {
+    let mut total = 0u32;
+    for &sp in sample_pages {
+        let pg_off = text_off + (sp as usize) * 0x1000;
+        if pg_off + 0x1000 > data.len() {
+            continue;
+        }
+        let mut page = [0u8; 0x1000];
+        page.copy_from_slice(&data[pg_off..pg_off + 0x1000]);
+        if let Some(sh) = shift {
+            dd8_apply(&mut page, abs_base.wrapping_add(sp), sh);
+        }
+        total += ret_int3_score(&page);
+    }
+    total
+}
+
+/// Select DD8 from the common CRT entry stub when its direct call and jump
+/// provide a stronger oracle than sparse padding statistics. The candidate is
+/// accepted only when it is the sole one whose two branch targets stay inside
+/// `.text`; unrecognised entry code falls through to the padding selector.
+fn select_dd8_by_entry_stub(
+    data: &[u8],
+    text_va: u32,
+    text_size: u32,
+    info3: u32,
+) -> Option<(u32, [u8; 3])> {
+    for entry in entry_candidates(data, text_va, text_size, info3) {
+        let [Some(none), Some(s0), Some(s15)] = [None, Some(0), Some(15)]
+            .map(|shift| entry_stub_branch_score(data, text_va, text_size, entry, shift))
+        else {
+            continue;
+        };
+        let scores = [none, s0, s15];
+        let best = scores.iter().copied().max()?;
+        if best == 2 && scores.iter().filter(|&&score| score == best).count() == 1 {
+            let index = scores.iter().position(|&score| score == best)?;
+            return Some(([99, 0, 15][index], scores));
+        }
+    }
+    None
+}
+
+fn entry_candidates(data: &[u8], text_va: u32, text_size: u32, info3: u32) -> Vec<u32> {
+    let text_end = text_va.saturating_add(text_size);
+    let mut entries = Vec::with_capacity(3);
+    if let Some(pe) = read_u32(data, 0x3C)
+        && let Some(entry) = pe.checked_add(40).and_then(|offset| read_u32(data, offset))
+        && (text_va..text_end).contains(&entry)
+    {
+        entries.push(entry);
+    }
+    for metadata_off in [32u32, 64] {
+        let Some(end) = info3
+            .checked_add(metadata_off)
+            .and_then(|offset| offset.checked_add(8))
+        else {
+            continue;
+        };
+        if end as usize > data.len() {
+            continue;
+        }
+        let entry = trial_decrypt5_u32(data, info3 + metadata_off);
+        let image_base = trial_decrypt5_u32(data, info3 + metadata_off + 4);
+        if image_base == info3 && (text_va..text_end).contains(&entry) && !entries.contains(&entry)
+        {
+            entries.push(entry);
+        }
+    }
+    entries
+}
+
+fn entry_stub_branch_score(
+    data: &[u8],
+    text_va: u32,
+    text_size: u32,
+    entry: u32,
+    shift: Option<u32>,
+) -> Option<u8> {
+    let text_end = text_va.checked_add(text_size)?;
+    if entry < text_va || entry.checked_add(18)? > text_end {
+        return None;
+    }
+
+    let mut stub = [0u8; 18];
+    for (offset, byte) in stub.iter_mut().enumerate() {
+        *byte = dd8_candidate_byte(data, entry + offset as u32, shift)?;
+    }
+    if stub[0..3] != [0x48, 0x83, 0xEC]
+        || stub[4] != 0xE8
+        || stub[9..12] != [0x48, 0x83, 0xC4]
+        || stub[12] != stub[3]
+        || stub[13] != 0xE9
+    {
+        return None;
+    }
+
+    let call_rel = i32::from_le_bytes(stub[5..9].try_into().ok()?) as i64;
+    let jump_rel = i32::from_le_bytes(stub[14..18].try_into().ok()?) as i64;
+    let call_target = i64::from(entry) + 9 + call_rel;
+    let jump_target = i64::from(entry) + 18 + jump_rel;
+    let in_text = |target: i64| target >= i64::from(text_va) && target < i64::from(text_end);
+    Some(u8::from(in_text(call_target)) + u8::from(in_text(jump_target)))
+}
+
+fn dd8_candidate_byte(data: &[u8], rva: u32, shift: Option<u32>) -> Option<u8> {
+    let mut byte = *data.get(rva as usize)?;
+    let Some(shift) = shift else {
+        return Some(byte);
+    };
+    let page = rva >> 12;
+    let block = (rva & 0xFFF) >> 4;
+    let mut key = page << shift;
+    for index in 0..=block {
+        let mixed = key.rotate_right(15).wrapping_add(index);
+        key = mixed.wrapping_add(index);
+        if index != 0 {
+            let target = (page << 12)
+                .wrapping_add(index << 4)
+                .wrapping_add(mixed & 0xF);
+            if target == rva {
+                byte ^= key as u8;
+            }
+        }
+    }
+    Some(byte)
+}
+
+fn read_u32(data: &[u8], offset: u32) -> Option<u32> {
+    let start = offset as usize;
+    let bytes = data.get(start..start.checked_add(4)?)?;
+    Some(u32::from_le_bytes(bytes.try_into().ok()?))
+}
+
+// Baseline: count int3 pads already present at the first byte of each 16-byte
+// block, i.e. the positions dd8 would target if its in-block offset were 0.
+fn score_dd8_baseline(data: &[u8], text_off: usize, sample_pages: &[u32]) -> u32 {
+    let mut hits = 0u32;
+    for &sp in sample_pages {
+        let pg_off = text_off + (sp as usize) * 0x1000;
+        if pg_off + 0x1000 > data.len() {
+            continue;
+        }
+        for bi in 1..256usize {
+            if data[pg_off + bi * 16] == 0xCC {
+                hits += 1;
+            }
+        }
+    }
+    hits
+}
+
+// Replay decrypt_data8 on each sample page under `shift` and count how many of
+// the 255 mutated positions decode to 0xCC.
+fn score_dd8_shift(
+    data: &[u8],
+    text_off: usize,
+    text_va: u32,
+    sample_pages: &[u32],
+    shift: u32,
+) -> u32 {
+    let abs_base = text_va >> 12;
+    let mut hits = 0u32;
+    for &sp in sample_pages {
+        let pg_off = text_off + (sp as usize) * 0x1000;
+        if pg_off + 0x1000 > data.len() {
+            continue;
+        }
+        let abs_page = abs_base.wrapping_add(sp);
+        let mut key = abs_page << shift;
+        for bi in 0..256u32 {
+            let mixed = key.rotate_right(15).wrapping_add(bi);
+            key = mixed.wrapping_add(bi);
+            if bi == 0 {
+                continue;
+            }
+            let tidx = (bi.wrapping_mul(16).wrapping_add(mixed & 0xF)) as usize;
+            if tidx < 0x1000 {
+                let mutated = data[pg_off + tidx] ^ (key as u8);
+                if mutated == 0xCC {
+                    hits += 1;
+                }
+            }
+        }
+    }
+    hits
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn entry_stub_fixture() -> Vec<u8> {
+        let mut data = vec![0u8; 0x5000];
+        data[0x3C..0x40].copy_from_slice(&0x100u32.to_le_bytes());
+        data[0x128..0x12C].copy_from_slice(&0x1264u32.to_le_bytes());
+        data[0x1264..0x1276].copy_from_slice(&[
+            0x48, 0x83, 0xEC, 0x28, 0xE8, 0x5B, 0x02, 0x00, 0x00, 0x48, 0x83, 0xC4, 0x28, 0xE9,
+            0x7A, 0xFE, 0xFF, 0xFF,
+        ]);
+        data
+    }
+
+    fn apply_dd8_page(data: &mut [u8], page_rva: u32, shift: u32) {
+        let mut key = (page_rva >> 12) << shift;
+        for index in 0..256u32 {
+            let mixed = key.rotate_right(15).wrapping_add(index);
+            key = mixed.wrapping_add(index);
+            if index == 0 {
+                continue;
+            }
+            let target = page_rva.wrapping_add(index << 4).wrapping_add(mixed & 0xF) as usize;
+            data[target] ^= key as u8;
+        }
+    }
+
+    #[test]
+    fn entry_stub_selects_plaintext_and_both_dd8_shifts() {
+        let plain = entry_stub_fixture();
+        assert_eq!(select_dd8_shift(&plain, 0x1000, 0x4000, 0), 99);
+
+        for expected in [0u32, 15] {
+            let mut encrypted = plain.clone();
+            apply_dd8_page(&mut encrypted, 0x1000, expected);
+            assert_eq!(select_dd8_shift(&encrypted, 0x1000, 0x4000, 0), expected);
+        }
+    }
+
+    /// Seed the first `count` dd8-targeted positions of each sampled page with
+    /// the byte that decodes to `0xCC` under the `page+1` formula — i.e. an
+    /// encrypted `.text` whose plaintext is int3 padding. Positions whose key
+    /// byte would make the *ciphertext* itself `0xCC` are skipped so the
+    /// fixture contains no `0xCC` at all and every post-dd8 `0xCC` is a genuine
+    /// gain over a zero baseline.
+    fn seed_dd8_int3(data: &mut [u8], text_off: u32, pages: &[u32], count: u32) {
+        for &sp in pages {
+            let pg_off = (text_off + sp * 0x1000) as usize;
+            let mut k = sp.wrapping_add(1);
+            k = k.rotate_right(15);
+            let mut planted = 0u32;
+            for bi in 1..256u32 {
+                let ri = k.rotate_right(15).wrapping_add(bi);
+                k = ri.wrapping_add(bi);
+                if planted >= count {
+                    continue;
+                }
+                let ct = 0xCCu8 ^ (k as u8);
+                if ct == 0xCC {
+                    continue;
+                }
+                let tidx = (bi.wrapping_mul(16).wrapping_add(ri & 0xF)) as usize;
+                data[pg_off + tidx] = ct;
+                planted += 1;
+            }
+        }
+    }
+
+    /// Review regression: a near-plaintext `.text` must NOT be dd8-decrypted.
+    /// dd8 XORs 255 positions per page with pseudo-random bytes, so it
+    /// manufactures a few `0xCC` for free — under the old bare
+    /// `best > baseline` test any positive gain was enough to "apply" dd8 and
+    /// scramble ~1 byte per 16 of a native DLL's already-plaintext code,
+    /// silently (nothing downstream, including the integrity check, notices).
+    /// Here the gain is real but small; the floor must still reject it.
+    #[test]
+    fn pe32_dd8_skips_text_whose_gain_is_only_noise_sized() {
+        let text_off: u32 = 0x1000;
+        let text_size: u32 = 8 * 0x1000;
+        let mut data = vec![0u8; (text_off + text_size) as usize];
+        seed_dd8_int3(&mut data, text_off, &[2, 4, 6], 5);
+        assert!(
+            !data.contains(&0xCC),
+            "fixture must have a zero 0xCC baseline"
+        );
+        assert_eq!(
+            select_dd8_formula_pe32(&data, text_off, text_size),
+            None,
+            "a gain this small is indistinguishable from dd8's own noise"
+        );
+    }
+
+    /// Control for the above: a `.text` whose dd8 pass restores a large amount
+    /// of int3 padding clears the floor and is decrypted. Same fixture shape,
+    /// only the amount of restored padding differs.
+    #[test]
+    fn pe32_dd8_applies_when_padding_is_restored() {
+        let text_off: u32 = 0x1000;
+        let text_size: u32 = 8 * 0x1000;
+        let mut data = vec![0u8; (text_off + text_size) as usize];
+        seed_dd8_int3(&mut data, text_off, &[2, 4, 6], 255);
+        assert_eq!(
+            select_dd8_formula_pe32(&data, text_off, text_size),
+            Some(false),
+            "encrypted .text must be decrypted with the page+1 formula"
+        );
+    }
+}
@@ -0,0 +1,507 @@
+//! Structural locators for protected PE stages.
+
+use senbei_crypto::primitives::{get_u32, lfsr_keystream};
+
+/// Find the 4-byte v_val that follows the LAST occurrence of `48 EB 01 B9`
+/// (REX.W jmp+1; mov ecx,imm32) plus any 0xCC padding. Used to locate
+/// stage4's accum2 seed. Works across builds even when API-name anchors are
+/// absent.
+pub fn find_v_after_pad(data: &[u8], base: u32, len: u32) -> Option<u32> {
+    let start = base as usize;
+    let end = (base.saturating_add(len)) as usize;
+    if end > data.len() {
+        return None;
+    }
+    let sig = [0x48u8, 0xEB, 0x01, 0xB9];
+    let slice = &data[start..end];
+    // last occurrence
+    let mut last = None;
+    let mut i = 0usize;
+    while i + sig.len() <= slice.len() {
+        if slice[i..i + sig.len()] == sig {
+            last = Some(i);
+        }
+        i += 1;
+    }
+    let pos = last?;
+    // skip CCs after the `48 EB 01 B9`
+    let mut after = pos + sig.len();
+    while after < slice.len() && slice[after] == 0xCC {
+        after += 1;
+    }
+    if after + 4 > slice.len() {
+        return None;
+    }
+    Some((start + after) as u32)
+}
+
+/// Predict the 4 bytes that DecryptData5(va, size) would produce at va+0..va+4
+/// without mutating the buffer. The cipher's per-byte transform depends only
+/// on the byte itself and the low 8 bits of (va+i), with no cross-byte state,
+/// so each byte can be decrypted in isolation. Used to detect the EP/DD layout
+/// offset before committing to the actual call.
+pub fn trial_decrypt5_u32(data: &[u8], va: u32) -> u32 {
+    let mut out = [0u8; 4];
+    for i in 0..4u32 {
+        let b3 = data[(va + i) as usize];
+        let b = (va + i) as u8;
+        let b2 = b.wrapping_add(1);
+        let b4 = b3.rotate_left(2) ^ b2;
+        let b5 = b4.rotate_left(2) ^ b;
+        out[i as usize] = b5.rotate_left(2);
+    }
+    u32::from_le_bytes(out)
+}
+
+/// Scan stage4/stage5 for the encrypted custom-decryptor bytecode block. The
+/// raw byte at p+95 is used by decrypt_data6 as the iteration count. We trial-
+/// decrypt that many bytes with the LFSR keystream and accept the first
+/// position where the byte stream parses as a valid opcode sequence ending in
+/// 195 (ret).
+pub fn find_bytecode_offset(data: &[u8], base: u32, len: u32) -> Option<u32> {
+    let start = base as usize;
+    let end = (base.saturating_add(len)) as usize;
+    if end > data.len() {
+        return None;
+    }
+    let mut ks = [0u8; 256];
+    lfsr_keystream(&mut ks);
+    // Scan forward from `start+16` on 16-byte boundaries relative to `start`.
+    // The bytecode block is positioned a fixed offset into stage4/stage5; the
+    // lowest parseable candidate is the real one (later ones are coincidental
+    // parses of trailing filler bytes that happen to map to valid opcodes).
+    // The enclosing buffer isn't necessarily 16-aligned to its absolute
+    // address in newer builds, so we anchor the stride to `start`.
+    let mut p = start + 16;
+    while p + 96 <= end {
+        let count = data[p + 95] as usize;
+        if count >= 8 && p + count <= end {
+            let mut buf = [0u8; 256];
+            let take = count.min(256);
+            for i in 0..take {
+                buf[i] = data[p + i] ^ ks[i];
+            }
+            if let Some(nops) = parse_bytecode_check(&buf[..take])
+                && nops >= 4
+            {
+                return Some(p as u32);
+            }
+        }
+        p += 16;
+    }
+    None
+}
+
+/// Validate bytecode structure without allocating a `Vec` of ops. Returns
+/// `Some(non_nop_op_count)` if the byte stream parses successfully as a valid
+/// opcode sequence ending in 195 (ret), `None` otherwise. Allows non-trivial
+/// bytecode filtering by op count.
+pub fn parse_bytecode_check(buf: &[u8]) -> Option<usize> {
+    let mut i = 0usize;
+    let mut nops: usize = 0;
+    while i < buf.len() {
+        let b = buf[i];
+        i += 1;
+        match b {
+            4 | 44 | 52 => {
+                if i >= buf.len() {
+                    return None;
+                }
+                i += 1;
+                nops += 1;
+            }
+            144 => {}
+            192 | 254 => {
+                if i >= buf.len() {
+                    return None;
+                }
+                let mb = buf[i];
+                i += 1;
+                let rm = mb & 7;
+                let mod_ = (mb >> 6) & 3;
+                let reg = (mb >> 3) & 7;
+                if mod_ != 3 || rm != 0 {
+                    return None;
+                }
+                if reg > 1 {
+                    return None;
+                }
+                if b == 192 {
+                    if i >= buf.len() {
+                        return None;
+                    }
+                    i += 1;
+                }
+                nops += 1;
+            }
+            195 => return Some(nops),
+            _ => return None,
+        }
+    }
+    None
+}
+
+/// Locate stage3's v4_val: the last non-zero dword in the buffer, anchored
+/// by the `C3 CC CC CC` (ret + 3 int3) immediately before it.
+pub fn find_v4_offset(data: &[u8], base: u32, len: u32) -> Option<u32> {
+    let start = base as usize;
+    let end = (base.saturating_add(len)) as usize;
+    if end > data.len() || end < start + 4 {
+        return None;
+    }
+    // walk backwards looking for the first non-zero byte
+    let mut i = end;
+    while i > start && data[i - 1] == 0 {
+        i -= 1;
+    }
+    if i < start + 4 {
+        return None;
+    }
+    // v_val occupies the 4 bytes ending at i (rounded up to dword boundary)
+    let v_end = i;
+    let v_start = ((v_end + 3) & !3).saturating_sub(4);
+    // require that the 4 bytes preceding v_val match `C3 CC CC CC`
+    if v_start < start + 4 || data[v_start - 4..v_start] != [0xC3, 0xCC, 0xCC, 0xCC] {
+        return None;
+    }
+    Some(v_start as u32)
+}
+
+/// Scan a sub-buffer for an ASCII needle; return its absolute position.
+pub fn find_str_pos(data: &[u8], base: u32, len: u32, needle: &[u8]) -> Option<u32> {
+    let start = base as usize;
+    let end = (base.saturating_add(len)) as usize;
+    if end > data.len() || needle.is_empty() {
+        return None;
+    }
+    data[start..end]
+        .windows(needle.len())
+        .position(|w| w == needle)
+        .map(|rel| (start + rel) as u32)
+}
+
+pub fn get_string_to_null(data: &[u8], offset: u32) -> String {
+    let start = offset as usize;
+    if start >= data.len() {
+        return String::new();
+    }
+    // Bounded: an unterminated run must never walk off the end of the buffer
+    // (panic) or scan unboundedly into unrelated data.
+    let limit = start.saturating_add(4096).min(data.len());
+    let mut i = start;
+    while i < limit && data[i] != 0 {
+        i += 1;
+    }
+    String::from_utf8_lossy(&data[start..i]).into_owned()
+}
+
+/// Read a PE section-name field: exactly 8 bytes, NOT necessarily
+/// NUL-terminated (a full-width name like `.textbss` has no NUL at all).
+/// Returns the name with trailing NULs stripped. Using `get_string_to_null`
+/// here would run past the field into the VirtualSize/VirtualAddress dwords.
+pub fn section_name(data: &[u8], offset: u32) -> String {
+    let start = offset as usize;
+    let Some(field) = data.get(start..start + 8) else {
+        return String::new();
+    };
+    let end = field.iter().position(|&b| b == 0).unwrap_or(8);
+    String::from_utf8_lossy(&field[..end]).into_owned()
+}
+
+// ---------------------------------------------------------------------------
+// PE32 (32-bit) helpers
+// ---------------------------------------------------------------------------
+
+/// PE32 shell-table locator. Walks the shell region (`info[6]`) for a dword
+/// equal to `info[6]` followed by a plausible shell size, returning the table
+/// base (`candidate = off - 0x88`) when `candidate+0x58` holds a valid pointer.
+pub fn find_tbl_pe32(data: &[u8], info: &[u32; 8]) -> Option<u32> {
+    let shell = info[6];
+    if (data.len() as u64) < 0x100 {
+        return None;
+    }
+    let hi = (shell as u64)
+        .saturating_add(0x3000)
+        .min(data.len() as u64 - 0x100) as u32;
+    let mut off = shell;
+    while off < hi {
+        if off as usize + 8 <= data.len() {
+            let candidate = off.wrapping_sub(0x88);
+            if candidate >= shell && get_u32(data, off) == info[6] {
+                let shell_size_val = get_u32(data, off.wrapping_add(4));
+                if shell_size_val > 0x1000 && shell_size_val < 0x100000 {
+                    let v58_off = candidate.wrapping_add(0x58);
+                    if (v58_off as usize + 4) <= data.len() {
+                        let v58 = get_u32(data, v58_off);
+                        if v58 > 0 && (v58 as usize) < data.len() {
+                            return Some(candidate);
+                        }
+                    }
+                }
+            }
+        }
+        off = off.wrapping_add(4);
+    }
+    None
+}
+
+/// Locate an LFSR-encrypted bytecode block (decrypt_data6 form) in a region.
+/// `start_off` is the byte offset to begin scanning at, `scan_backward`
+/// controls direction. Returns the relative offset of the block. Includes full
+/// opcode-walk validation of candidate blocks.
+pub fn find_lfsr_block(
+    data: &[u8],
+    base: u32,
+    size: u32,
+    start_off: u32,
+    scan_backward: bool,
+) -> Option<u32> {
+    if size < 96 {
+        return None;
+    }
+    let mut ks = [0u8; 128];
+    lfsr_keystream(&mut ks);
+    let check = |scan_off: u32| -> bool {
+        let abs_off = base.wrapping_add(scan_off) as usize;
+        if abs_off + 96 > data.len() {
+            return false;
+        }
+        let sz = data[abs_off + 95] as usize;
+        if !(10..=95).contains(&sz) {
+            return false;
+        }
+        let mut decoded = [0u8; 95];
+        for bi in 0..sz {
+            decoded[bi] = data[abs_off + bi] ^ ks[bi];
+        }
+        // Full bytecode validation (shared with the stage4/5 locator): every
+        // opcode must decode with a valid ModR/M and the stream must REACH a
+        // RET (0xC3) as an opcode. The previous check only required a 0xC3
+        // byte *anywhere* in the window and accepted a walk that ran off the
+        // end without hitting RET — a `0x04 0xC3` (ADD 0xC3) tail passed, so
+        // coincidental LFSR-shaped garbage was accepted as a decryptor block.
+        parse_bytecode_check(&decoded[..sz]).is_some()
+    };
+    if scan_backward {
+        let hi = size - 96;
+        if hi >= start_off {
+            let mut scan_off = hi;
+            loop {
+                if check(scan_off) {
+                    return Some(scan_off);
+                }
+                if scan_off == start_off {
+                    break;
+                }
+                scan_off -= 1;
+            }
+        }
+    } else {
+        let hi = size - 95;
+        let mut scan_off = start_off;
+        while scan_off < hi {
+            if check(scan_off) {
+                return Some(scan_off);
+            }
+            scan_off += 1;
+        }
+    }
+    None
+}
+
+/// Slots discovered in the eighthStage for the marker-less layout.
+pub struct EighthSlots {
+    /// Absolute address of the file-data decryptor LFSR bytecode block. The
+    /// fileCS chain pointer is derived downstream as `file_lfsr - 0x58`.
+    pub file_lfsr: u32,
+    /// Absolute address of the compressedInfo (ptr,size) table pointer slot.
+    pub compressed_info_ptr: u32,
+}
+
+/// Marker-independent eighthStage slot discovery (PE32+ branch).
+///
+/// Newer Crackproof builds (e.g. some native/managed DLLs) omit the
+/// `pm\0\0cm\0\0` and `00 00 00 40 01 00 00 00` markers that the older layout's
+/// walk3/walk4/walk5 slot derivation relies on. Instead this discovers the
+/// slots structurally:
+///   * Scan the eighthStage for every LFSR (decrypt_data6) bytecode block.
+///   * The file decryptor is the LFSR block whose `fileCS = lfsr - 0x58` holds
+///     a pointer sitting just past `info[3]` (smallest positive distance).
+///   * `compressedInfo` is the pointer slot whose 16-byte target, after a
+///     trial `decrypt_data5`, parses as a plausible (src,sSize,dst,dSize)
+///     descriptor.
+///
+/// Returns `None` if no plausible file LFSR is found. `eighth_start`/`eighth_dsz`
+/// bound the search region; `info3` is `info[3]`; `compress_data_offset` is
+/// `(!u32(file_data,0x1080)) + 0x1000`; `file_data_len` is the protected file
+/// length.
+#[allow(clippy::too_many_arguments)]
+pub fn discover_eighth_slots(
+    data: &[u8],
+    eighth_start: u32,
+    eighth_dsz: u32,
+    info3: u32,
+    compress_data_offset: u32,
+    file_data_len: u32,
+) -> Option<EighthSlots> {
+    // Collect all LFSR candidates (forward scan).
+    //
+    // Advance by 1 after each hit, NOT by 96. A false-positive LFSR match can sit
+    // just before the real file-decryptor block (observed on an il2cpp game
+    // assembly build, 2026-07-13: junk at rel=0x31C1, real block at 0x3210).
+    // Stepping by the LFSR body size then skips the real block and discovery
+    // fails. Byte-stepping is cheap: eighthStage is only a few KB.
+    let mut all_lfsrs: Vec<u32> = Vec::new();
+    let mut scan_off: u32 = 0;
+    while scan_off + 95 < eighth_dsz {
+        match find_lfsr_block(data, eighth_start, eighth_dsz, scan_off, false) {
+            Some(found) => {
+                all_lfsrs.push(found);
+                scan_off = found + 1;
+            }
+            None => break,
+        }
+    }
+
+    // Pick the file LFSR: prefer the candidate whose fileCS pointer sits the
+    // smallest positive distance past info[3].
+    let mut off_file_lfsr: Option<u32> = None;
+    let mut best_dist: Option<u32> = None;
+    for &lfsr_off in &all_lfsrs {
+        if lfsr_off < 0x58 {
+            continue;
+        }
+        let cs_off = lfsr_off - 0x58;
+        let cs_val = get_u32(data, eighth_start.wrapping_add(cs_off));
+        if !(0x1000 < cs_val && (cs_val as usize) < data.len()) {
+            continue;
+        }
+        if cs_val < info3 {
+            continue;
+        }
+        let dist = cs_val - info3;
+        if best_dist.is_none_or(|b| dist < b) {
+            best_dist = Some(dist);
+            off_file_lfsr = Some(lfsr_off);
+        }
+    }
+    // Fallback: last LFSR with any in-image fileCS pointer.
+    if off_file_lfsr.is_none() {
+        for &lfsr_off in all_lfsrs.iter().rev() {
+            if lfsr_off < 0x58 {
+                continue;
+            }
+            let cs_val = get_u32(data, eighth_start.wrapping_add(lfsr_off - 0x58));
+            if 0x1000 < cs_val && (cs_val as usize) < data.len() {
+                off_file_lfsr = Some(lfsr_off);
+                break;
+            }
+        }
+    }
+    let off_file_lfsr = off_file_lfsr?;
+    let off_file_cs = off_file_lfsr - 0x58;
+
+    // Trial-decrypt to find compressedInfo: the pointer slot in the data area
+    // (between fileCS region start and the LFSR) whose target parses as a valid
+    // (src,sSize,dst,dSize) descriptor after a transient decrypt_data5.
+    let scan_from = off_file_lfsr.saturating_sub(0x400);
+    let mut off_compressed_info: Option<u32> = None;
+    let mut doff = scan_from;
+    while doff < off_file_lfsr {
+        if doff == off_file_cs {
+            doff += 4;
+            continue;
+        }
+        let ptr_val = get_u32(data, eighth_start.wrapping_add(doff));
+        if !(0x1000 < ptr_val && (ptr_val as usize) < data.len().saturating_sub(16)) {
+            doff += 4;
+            continue;
+        }
+        // Predict decrypt_data5(ptr_val, 16) without mutating: each dword is
+        // position-keyed and independent, so trial_decrypt5_u32 per dword.
+        let src2 = trial_decrypt5_u32(data, ptr_val);
+        let s_sz2 = trial_decrypt5_u32(data, ptr_val + 4);
+        let dst2 = trial_decrypt5_u32(data, ptr_val + 8);
+        let d_sz2 = trial_decrypt5_u32(data, ptr_val + 12);
+        let src_file_off = src2.wrapping_add(compress_data_offset);
+        let valid = s_sz2 > 0
+            && s_sz2 < 0x200000
+            && (src_file_off as u64 + s_sz2 as u64) <= file_data_len as u64
+            && dst2 >= 0x1000
+            && (dst2 as u64 + d_sz2 as u64) <= data.len() as u64
+            && d_sz2 >= s_sz2
+            && d_sz2 < 0x200000;
+        if valid {
+            off_compressed_info = Some(doff);
+            break;
+        }
+        doff += 4;
+    }
+    let off_compressed_info = off_compressed_info?;
+
+    Some(EighthSlots {
+        file_lfsr: eighth_start.wrapping_add(off_file_lfsr),
+        compressed_info_ptr: eighth_start.wrapping_add(off_compressed_info),
+    })
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    /// Task 4.1 regression: build a synthetic buffer whose valid bytecode block
+    /// sits PAST `len` but within `len*2`. Assert that the smaller window misses
+    /// it and the doubled window finds it.
+    #[test]
+    fn bytecode_locate_double_window_retry() {
+        // We place the block at offset (base + len + 16) which is inside
+        // the len*2 window but outside the len window.
+        let base: u32 = 0;
+        let len: u32 = 256;
+        // Block sits at base + len + 16 = 272, aligned to 16.
+        let block_pos: usize = (base + len + 16) as usize; // 272
+
+        // The buffer must be large enough for the block (block_pos + 96 bytes).
+        let buf_len = block_pos + 256;
+        let mut buf = vec![0u8; buf_len];
+
+        // Build a valid plaintext op stream:
+        //   [4, 0, 4, 0, 4, 0, 4, 0, 195]  (4 ADD-AL ops then RET)
+        // Padded to 10 bytes total; count >= 8.
+        let count: usize = 10;
+        let mut plain = [0u8; 256];
+        plain[0] = 4;
+        plain[1] = 0;
+        plain[2] = 4;
+        plain[3] = 0;
+        plain[4] = 4;
+        plain[5] = 0;
+        plain[6] = 4;
+        plain[7] = 0;
+        plain[8] = 195; // ret
+
+        // Compute the LFSR keystream and XOR the first `count` bytes to get the
+        // encrypted representation that the scanner would decrypt back.
+        let mut ks = [0u8; 256];
+        lfsr_keystream(&mut ks);
+        for i in 0..count {
+            buf[block_pos + i] = plain[i] ^ ks[i];
+        }
+        // Raw count byte at block_pos+95 (outside the XOR range since count=10 < 95).
+        buf[block_pos + 95] = count as u8;
+
+        // Verify our construction: find_bytecode_offset with len should NOT find it.
+        assert_eq!(
+            find_bytecode_offset(&buf, base, len),
+            None,
+            "smaller window should not find the block"
+        );
+
+        // The doubled window should find it at block_pos.
+        assert_eq!(
+            find_bytecode_offset(&buf, base, len.saturating_mul(2)),
+            Some(block_pos as u32),
+            "doubled window should locate the block"
+        );
+    }
+}
@@ -0,0 +1,481 @@
+//! PE import reconstruction and memory-image compaction.
+
+use senbei_crypto::primitives::{get_u16, get_u32, write_u16, write_u32};
+
+use super::super::MAX_IMAGE_SIZE;
+
+/// Read a NUL-terminated byte string starting at `off`, bounded to 512 bytes.
+/// Returns the raw bytes up to the terminator (excluding it).
+fn read_cstr_bounded(data: &[u8], off: u32) -> Vec<u8> {
+    let start = off as usize;
+    if start >= data.len() {
+        return Vec::new();
+    }
+    let limit = (start + 512).min(data.len());
+    let mut end = start;
+    while end < limit && data[end] != 0 {
+        end += 1;
+    }
+    data[start..end].to_vec()
+}
+
+fn align_up_u32(value: u32, alignment: u32) -> u32 {
+    ((value.wrapping_add(alignment - 1)) / alignment).wrapping_mul(alignment)
+}
+
+fn align_up_u64(value: u64, alignment: u64) -> u64 {
+    value.div_ceil(alignment) * alignment
+}
+
+#[derive(Clone)]
+enum ImportFunc {
+    Ordinal(u32),
+    Name(u16, Vec<u8>),
+}
+
+struct ImportDesc {
+    time_date: u32,
+    fwd_chain: u32,
+    dll_name: Vec<u8>,
+    iat_rva: u32,
+    functions: Vec<ImportFunc>,
+}
+
+/// Return true when PE32 imports already sit in the original `.idata` layout
+/// (so no relocation to `.kmiat` is needed). May write the IAT data directory
+/// (pe+0xD8).
+pub fn pe32_imports_already_match_idata_layout(data: &mut [u8], pe_header: u32) -> bool {
+    let opt_hdr_size = get_u16(data, pe_header.wrapping_add(20)) as u32;
+    let sec_table = pe_header.wrapping_add(24).wrapping_add(opt_hdr_size);
+    let num_sections = get_u16(data, pe_header.wrapping_add(6)) as u32;
+    let import_rva = get_u32(data, pe_header.wrapping_add(0x80));
+    let import_size = get_u32(data, pe_header.wrapping_add(0x84));
+    let len = data.len() as u32;
+    if !(import_rva > 0 && import_size > 0) {
+        return false;
+    }
+    for idx in 0..num_sections {
+        let sec_off = sec_table.wrapping_add(idx * 40);
+        if (sec_off as usize + 40) > data.len() {
+            return false;
+        }
+        if &data[sec_off as usize..sec_off as usize + 6] != b".idata" {
+            continue;
+        }
+        let sec_va = get_u32(data, sec_off.wrapping_add(12));
+        let sec_size =
+            get_u32(data, sec_off.wrapping_add(8)).max(get_u32(data, sec_off.wrapping_add(16)));
+        let sec_end = sec_va.wrapping_add(sec_size);
+        if !(sec_va <= import_rva
+            && import_rva < sec_end
+            && import_rva.wrapping_add(import_size) <= sec_end)
+        {
+            continue;
+        }
+        let first_oft = get_u32(data, import_rva);
+        let first_name = get_u32(data, import_rva.wrapping_add(12));
+        let first_iat = get_u32(data, import_rva.wrapping_add(16));
+        if !(sec_va <= first_oft
+            && first_oft < sec_end
+            && sec_va <= first_iat
+            && first_iat < sec_end)
+        {
+            return false;
+        }
+        if !(0x1000 < first_name && first_name < len) {
+            return false;
+        }
+        let dll_name = read_cstr_bounded(data, first_name);
+        let lower: Vec<u8> = dll_name.iter().map(|b| b.to_ascii_lowercase()).collect();
+        if !lower.ends_with(b".dll") {
+            return false;
+        }
+        let mut iat_min = first_iat;
+        let mut iat_max = first_iat;
+        let mut idt_pos = import_rva;
+        while idt_pos.wrapping_add(20) <= len {
+            let oft_rva = get_u32(data, idt_pos);
+            let name_rva = get_u32(data, idt_pos.wrapping_add(12));
+            let iat_rva = get_u32(data, idt_pos.wrapping_add(16));
+            if oft_rva == 0 && name_rva == 0 && iat_rva == 0 {
+                break;
+            }
+            if !(sec_va <= oft_rva && oft_rva < sec_end && sec_va <= iat_rva && iat_rva < sec_end) {
+                return false;
+            }
+            let mut thunk = iat_rva;
+            while thunk.wrapping_add(4) <= sec_end {
+                let tv = get_u32(data, thunk);
+                thunk = thunk.wrapping_add(4);
+                if tv == 0 {
+                    break;
+                }
+            }
+            iat_min = iat_min.min(iat_rva);
+            iat_max = iat_max.max(thunk);
+            idt_pos = idt_pos.wrapping_add(20);
+        }
+        if iat_max > iat_min {
+            write_u32(data, pe_header.wrapping_add(0xD8), iat_min);
+            write_u32(data, pe_header.wrapping_add(0xDC), iat_max - iat_min);
+        }
+        return true;
+    }
+    false
+}
+
+/// Rebuild PE32 import metadata (descriptors, lookup tables, names) into the
+/// last section as `.kmiat`, leaving the loader-written IAT in place. Mutates
+/// `data` (may grow it).
+pub fn move_pe32_imports_to_kmiat(data: &mut Vec<u8>, pe_header: u32) {
+    const SECTION_SIZE: u32 = 0x7000;
+    let opt_hdr_size = get_u16(data, pe_header.wrapping_add(20)) as u32;
+    let opt_hdr = pe_header.wrapping_add(24);
+    let sec_table = opt_hdr.wrapping_add(opt_hdr_size);
+    let num_sections = get_u16(data, pe_header.wrapping_add(6)) as u32;
+    if num_sections == 0 {
+        return;
+    }
+    let import_rva = get_u32(data, pe_header.wrapping_add(0x80));
+    let import_size = get_u32(data, pe_header.wrapping_add(0x84));
+    let len = data.len() as u32;
+    if !(0x1000 < import_rva && import_rva < len && import_size > 0 && import_size < SECTION_SIZE) {
+        return;
+    }
+
+    let mut descriptors: Vec<ImportDesc> = Vec::new();
+    let mut idt_pos = import_rva;
+    while idt_pos.wrapping_add(20) <= len {
+        let oft_rva = get_u32(data, idt_pos);
+        let time_date = get_u32(data, idt_pos.wrapping_add(4));
+        let fwd_chain = get_u32(data, idt_pos.wrapping_add(8));
+        let name_rva = get_u32(data, idt_pos.wrapping_add(12));
+        let iat_rva = get_u32(data, idt_pos.wrapping_add(16));
+        if oft_rva == 0 && name_rva == 0 && iat_rva == 0 {
+            break;
+        }
+        if !(0x1000 < name_rva && name_rva < len) {
+            break;
+        }
+        let dll_name = read_cstr_bounded(data, name_rva);
+        let thunk_rva = if 0x1000 < oft_rva && oft_rva < len {
+            oft_rva
+        } else {
+            iat_rva
+        };
+        let mut functions: Vec<ImportFunc> = Vec::new();
+        let mut thunk_pos = thunk_rva;
+        while 0x1000 < thunk_pos.wrapping_add(4) && thunk_pos.wrapping_add(4) <= len {
+            let thunk_val = get_u32(data, thunk_pos);
+            if thunk_val == 0 {
+                break;
+            }
+            if thunk_val & 0x8000_0000 != 0 {
+                functions.push(ImportFunc::Ordinal(thunk_val & 0xFFFF));
+            } else {
+                let hint = if thunk_val.wrapping_add(2) <= len {
+                    get_u16(data, thunk_val)
+                } else {
+                    0
+                };
+                let func_name = if thunk_val.wrapping_add(2) < len {
+                    read_cstr_bounded(data, thunk_val.wrapping_add(2))
+                } else {
+                    Vec::new()
+                };
+                functions.push(ImportFunc::Name(hint, func_name));
+            }
+            thunk_pos = thunk_pos.wrapping_add(4);
+        }
+        descriptors.push(ImportDesc {
+            time_date,
+            fwd_chain,
+            dll_name,
+            iat_rva,
+            functions,
+        });
+        idt_pos = idt_pos.wrapping_add(20);
+    }
+    if descriptors.is_empty() {
+        return;
+    }
+
+    for desc in &mut descriptors {
+        let lower: Vec<u8> = desc
+            .dll_name
+            .iter()
+            .map(|b| b.to_ascii_lowercase())
+            .collect();
+        if lower.starts_with(b"api-ms-win-crt-") {
+            desc.dll_name = b"ucrtbase.dll".to_vec();
+        } else {
+            desc.dll_name = lower;
+        }
+    }
+    descriptors.sort_by_key(|d| d.iat_rva);
+
+    let last_sec = sec_table.wrapping_add((num_sections - 1) * 40);
+    let kmiat_rva = get_u32(data, last_sec.wrapping_add(12));
+    // A zero last-section VA means a corrupt section table: building .kmiat at
+    // RVA 0 would zero the DOS/PE headers and emit a structurally broken image
+    // with no error. Bail and keep the original import table.
+    if kmiat_rva == 0 {
+        return;
+    }
+    // Grow the image when .kmiat overruns it, but cap the growth: a corrupt VA
+    // could otherwise request a multi-gigabyte allocation, which aborts the
+    // process (uncatchable). Use u64 math so a near-u32::MAX VA cannot wrap the
+    // end calculation the way the previous wrapping/plain-add mix could.
+    let kmiat_end = kmiat_rva as u64 + SECTION_SIZE as u64;
+    if kmiat_end > MAX_IMAGE_SIZE {
+        return;
+    }
+    if kmiat_end > data.len() as u64 {
+        data.resize(kmiat_end as usize, 0);
+    }
+    // Zero the .kmiat region.
+    for b in &mut data[kmiat_rva as usize..kmiat_end as usize] {
+        *b = 0;
+    }
+
+    let idt_size = (descriptors.len() as u32 + 1) * 20;
+    let oft_start = kmiat_rva;
+    let mut idt_rva = oft_start;
+    for desc in &descriptors {
+        idt_rva = idt_rva.wrapping_add((desc.functions.len() as u32 + 1) * 4);
+    }
+    idt_rva = align_up_u32(idt_rva.wrapping_add(0x2C), 4);
+
+    // Size check: compute the final name_pos and bail if it overruns .kmiat.
+    let mut name_pos_check = idt_rva.wrapping_add(idt_size);
+    for desc in &descriptors {
+        name_pos_check = name_pos_check.wrapping_add(desc.dll_name.len() as u32 + 1);
+        for func in &desc.functions {
+            if let ImportFunc::Name(_, fname) = func {
+                name_pos_check = name_pos_check.wrapping_add(2 + fname.len() as u32 + 1);
+            }
+        }
+    }
+    if name_pos_check > kmiat_rva.wrapping_add(SECTION_SIZE) {
+        // Section too small; keep existing import table untouched.
+        return;
+    }
+
+    let mut oft_pos = oft_start;
+    let mut name_pos = idt_rva.wrapping_add(idt_size);
+    for (idx, desc) in descriptors.iter().enumerate() {
+        let idt_entry = idt_rva.wrapping_add(idx as u32 * 20);
+        let current_oft = oft_pos;
+        write_u32(data, idt_entry, current_oft);
+        write_u32(data, idt_entry.wrapping_add(4), desc.time_date);
+        write_u32(data, idt_entry.wrapping_add(8), desc.fwd_chain);
+        let dll_name_pos = name_pos;
+        write_u32(data, idt_entry.wrapping_add(12), dll_name_pos);
+        write_u32(data, idt_entry.wrapping_add(16), desc.iat_rva);
+
+        let dnp = dll_name_pos as usize;
+        data[dnp..dnp + desc.dll_name.len()].copy_from_slice(&desc.dll_name);
+        data[dnp + desc.dll_name.len()] = 0;
+        name_pos = name_pos.wrapping_add(desc.dll_name.len() as u32 + 1);
+
+        for func in &desc.functions {
+            match func {
+                ImportFunc::Ordinal(ord) => {
+                    write_u32(data, oft_pos, 0x8000_0000 | ord);
+                }
+                ImportFunc::Name(hint, fname) => {
+                    let hint_name_rva = name_pos;
+                    write_u32(data, oft_pos, hint_name_rva);
+                    write_u16(data, hint_name_rva, *hint as u32);
+                    let fp = (hint_name_rva + 2) as usize;
+                    data[fp..fp + fname.len()].copy_from_slice(fname);
+                    data[fp + fname.len()] = 0;
+                    name_pos = name_pos.wrapping_add(2 + fname.len() as u32 + 1);
+                }
+            }
+            oft_pos = oft_pos.wrapping_add(4);
+        }
+        write_u32(data, oft_pos, 0);
+        oft_pos = oft_pos.wrapping_add(4);
+    }
+    // Null-terminator IDT entry (20 zero bytes) after the last descriptor.
+    let term = idt_rva.wrapping_add(descriptors.len() as u32 * 20) as usize;
+    for b in &mut data[term..term + 20] {
+        *b = 0;
+    }
+
+    let ls = last_sec as usize;
+    data[ls..ls + 8].copy_from_slice(b".kmiat\x00\x00");
+    write_u32(data, last_sec.wrapping_add(8), SECTION_SIZE);
+    write_u32(data, last_sec.wrapping_add(16), SECTION_SIZE);
+    write_u32(data, last_sec.wrapping_add(36), 0xE000_0060);
+    write_u32(data, pe_header.wrapping_add(0x80), idt_rva);
+    write_u32(data, pe_header.wrapping_add(0x84), idt_size);
+    write_u32(
+        data,
+        pe_header.wrapping_add(80),
+        kmiat_rva.wrapping_add(SECTION_SIZE),
+    );
+}
+
+/// Convert the unpacked RVA-addressed image back to a compact PE file layout
+/// (headers at 0x400, sections packed consecutively, FileAlignment 0x200).
+/// Returns `None` if the accumulated output size wraps or exceeds
+/// [`MAX_IMAGE_SIZE`]: the final allocation is sized from header-derived
+/// section data, and an uncapped `vec![0; n]` from a corrupt header would abort
+/// the process (which `catch_unpack` cannot trap).
+pub fn compact_memory_image_to_pe(data: &[u8], pe_header: u32) -> Option<Vec<u8>> {
+    const FILE_ALIGNMENT: u32 = 0x200;
+    const HEADER_SIZE: u32 = 0x400;
+    let opt_hdr_size = get_u16(data, pe_header.wrapping_add(20)) as u32;
+    let opt_hdr = pe_header.wrapping_add(24);
+    let sec_table = opt_hdr.wrapping_add(opt_hdr_size);
+    let num_sections = get_u16(data, pe_header.wrapping_add(6)) as u32;
+
+    struct SecLayout {
+        sec_off: u32,
+        va: u32,
+        vsize: u32,
+        raw_ptr: u32,
+        raw_size: u32,
+    }
+
+    let mut raw_cursor: u64 = HEADER_SIZE as u64;
+    let mut raw_layout: Vec<SecLayout> = Vec::new();
+    for idx in 0..num_sections {
+        let sec_off = sec_table.wrapping_add(idx * 40);
+        let vsize = get_u32(data, sec_off.wrapping_add(8));
+        let va = get_u32(data, sec_off.wrapping_add(12));
+        let sd_start = va as usize;
+        let sd_end = if (va.wrapping_add(vsize) as usize) <= data.len() {
+            va.wrapping_add(vsize) as usize
+        } else {
+            data.len()
+        };
+        let section_data: &[u8] = if sd_start <= sd_end && sd_start <= data.len() {
+            &data[sd_start..sd_end]
+        } else {
+            &[]
+        };
+
+        let mut last_nonzero: i64 = -1;
+        for pos in (0..section_data.len()).rev() {
+            if section_data[pos] != 0 {
+                last_nonzero = pos as i64;
+                break;
+            }
+        }
+        let meaningful = if last_nonzero >= 0 {
+            (last_nonzero + 1) as u32
+        } else {
+            0
+        };
+        let mut raw_size = if meaningful != 0 {
+            align_up_u32(meaningful, FILE_ALIGNMENT)
+        } else {
+            0
+        };
+        if vsize != 0 && raw_size == 0 {
+            raw_size = FILE_ALIGNMENT;
+        }
+        raw_size = raw_size.min(align_up_u32(section_data.len() as u32, FILE_ALIGNMENT));
+
+        let raw_ptr = if raw_size != 0 { raw_cursor as u32 } else { 0 };
+        raw_layout.push(SecLayout {
+            sec_off,
+            va,
+            vsize,
+            raw_ptr,
+            raw_size,
+        });
+        if raw_size != 0 {
+            // Accumulate in u64 and cap: section sizes are header-derived, and
+            // a corrupt table could otherwise wrap raw_cursor (small alloc,
+            // huge recorded raw_ptrs → OOB panic) or request an abort-sized
+            // allocation.
+            raw_cursor = align_up_u64(raw_cursor + raw_size as u64, FILE_ALIGNMENT as u64);
+            if raw_cursor > MAX_IMAGE_SIZE {
+                return None;
+            }
+        }
+    }
+
+    let mut compact = vec![0u8; raw_cursor as usize];
+    let hdr_copy = (HEADER_SIZE as usize).min(data.len());
+    compact[..hdr_copy].copy_from_slice(&data[..hdr_copy]);
+    write_u32(&mut compact, opt_hdr.wrapping_add(36), FILE_ALIGNMENT);
+    write_u32(&mut compact, opt_hdr.wrapping_add(60), HEADER_SIZE);
+
+    for sl in &raw_layout {
+        write_u32(&mut compact, sl.sec_off.wrapping_add(16), sl.raw_size);
+        write_u32(&mut compact, sl.sec_off.wrapping_add(20), sl.raw_ptr);
+        if sl.raw_size != 0 {
+            let sd_start = sl.va as usize;
+            let sd_end = if (sl.va.wrapping_add(sl.vsize) as usize) <= data.len() {
+                sl.va.wrapping_add(sl.vsize) as usize
+            } else {
+                data.len()
+            };
+            let section_data: &[u8] = if sd_start <= sd_end {
+                &data[sd_start..sd_end]
+            } else {
+                &[]
+            };
+            let copy_size = (sl.raw_size as usize).min(section_data.len());
+            let rp = sl.raw_ptr as usize;
+            compact[rp..rp + copy_size].copy_from_slice(&section_data[..copy_size]);
+        }
+    }
+    Some(compact)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    /// Review regression: a zero last-section VA (corrupt section table) must
+    /// bail instead of building .kmiat at RVA 0 — the old code zeroed
+    /// `[0, 0x7000)`, wiping the DOS/PE headers, and returned the broken image
+    /// as a success. A near-2 GiB VA must likewise refuse to grow the image
+    /// past [`MAX_IMAGE_SIZE`].
+    #[test]
+    fn kmiat_bogus_section_va_bails_without_wiping_headers() {
+        for last_sec_va in [0u32, 0x5000_0000] {
+            let pe: u32 = 0x80;
+            let mut data = vec![0xAAu8; 0x8000];
+            // COFF header: 1 section, optional header size 0xE0 (PE32).
+            write_u16(&mut data, pe + 6, 1);
+            write_u16(&mut data, pe + 20, 0xE0);
+            // Import directory at pe+0x80: one descriptor + null terminator.
+            write_u32(&mut data, pe + 0x80, 0x1100);
+            write_u32(&mut data, pe + 0x84, 0x28);
+            write_u32(&mut data, 0x1100, 0x1200); // OFT rva
+            write_u32(&mut data, 0x1100 + 12, 0x1300); // name rva
+            write_u32(&mut data, 0x1100 + 16, 0x1400); // IAT rva
+            for b in &mut data[0x1100 + 20..0x1100 + 40] {
+                *b = 0; // null terminator descriptor
+            }
+            data[0x1300..0x1300 + 13].copy_from_slice(b"KERNEL32.dll\0");
+            write_u32(&mut data, 0x1200, 0x1500); // thunk -> hint/name
+            write_u32(&mut data, 0x1204, 0); // thunk terminator
+            data[0x1500..0x1502].copy_from_slice(&0u16.to_le_bytes());
+            data[0x1502..0x1502 + 12].copy_from_slice(b"ExitProcess\0");
+            // Section table at pe+24+0xE0 = 0x178; VA field at +12.
+            write_u32(&mut data, 0x178 + 12, last_sec_va);
+
+            let head_before: Vec<u8> = data[..0x400].to_vec();
+            let len_before = data.len();
+            move_pe32_imports_to_kmiat(&mut data, pe);
+            assert_eq!(
+                data.len(),
+                len_before,
+                "VA 0x{last_sec_va:08X}: image must not grow"
+            );
+            assert_eq!(
+                &data[..0x400],
+                &head_before[..],
+                "VA 0x{last_sec_va:08X}: headers must be untouched"
+            );
+        }
+    }
+}
@@ -1,16 +1,20 @@
-//! Pure, panic-free Crackproof unpacker core. No file I/O lives here.
+//! PE detection, unpacking, and structural validation.

-mod bytecode;
-mod crc32;
 pub mod dll;
+mod error;
 pub mod exe;
 pub mod integrity;
+mod layout;
 pub(crate) mod parallel;
-pub(crate) mod primitives;
-mod tables;

+use senbei_crypto::primitives;
+use std::cell::RefCell;
+use std::sync::{Arc, Mutex};
+
+pub use crate::thread_cap;
 pub use dll::{unpack_dll, unpack_dll_v};
-pub use exe::{UnpackError, unpack as unpack_exe, unpack_v as unpack_exe_v};
+pub use error::*;
+pub use exe::{unpack as unpack_exe, unpack_v as unpack_exe_v};
 pub use integrity::{IntegrityReport, check as check_integrity};

 /// Maximum plausible PE `SizeOfImage` we are willing to allocate a zero buffer
@@ -18,13 +22,136 @@ pub use integrity::{IntegrityReport, check as check_integrity};
 /// (or, as a sign-extended negative `i32`, multi-exabyte) allocation, which
 /// would abort the process — an abort that `catch_unpack` below cannot trap.
 /// Real protected binaries are far below this.
-pub(crate) const MAX_IMAGE_SIZE: u64 = 1 << 30; // 1 GiB
+pub(crate) const MAX_IMAGE_SIZE: u64 = senbei_crypto::MAX_IMAGE_SIZE;
+
+#[derive(Clone)]
+pub(crate) struct PanicCapture(Arc<Mutex<Option<PanicDetails>>>);
+
+#[derive(Clone)]
+struct PanicDetails {
+    message: String,
+    file: String,
+    line: u32,
+    column: u32,
+}
+
+thread_local! {
+    static ACTIVE_PANIC_CAPTURE: RefCell<Option<PanicCapture>> = const { RefCell::new(None) };
+}
+
+struct PanicCaptureGuard(Option<PanicCapture>);
+
+impl Drop for PanicCaptureGuard {
+    fn drop(&mut self) {
+        ACTIVE_PANIC_CAPTURE.with(|slot| {
+            slot.replace(self.0.take());
+        });
+    }
+}
+
+impl PanicCapture {
+    fn new() -> Self {
+        Self(Arc::new(Mutex::new(None)))
+    }
+
+    fn record(&self, info: &std::panic::PanicHookInfo<'_>) {
+        let location = info.location();
+        let details = PanicDetails {
+            message: panic_message(info.payload()),
+            file: location
+                .map(|value| value.file().to_owned())
+                .unwrap_or_else(|| "<unknown>".to_owned()),
+            line: location.map_or(0, std::panic::Location::line),
+            column: location.map_or(0, std::panic::Location::column),
+        };
+        let mut captured = self
+            .0
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner);
+        if captured.is_none() {
+            *captured = Some(details);
+        }
+    }
+
+    fn into_error(self, payload: &(dyn std::any::Any + Send)) -> UnpackError {
+        let details = self
+            .0
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner)
+            .clone()
+            .unwrap_or_else(|| PanicDetails {
+                message: panic_message(payload),
+                file: "<unknown>".to_owned(),
+                line: 0,
+                column: 0,
+            });
+        UnpackError::InternalPanic {
+            message: details.message,
+            file: details.file,
+            line: details.line,
+            column: details.column,
+        }
+    }
+
+    fn merge_from(&self, other: &Self) {
+        let details = other
+            .0
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner)
+            .clone();
+        let Some(details) = details else { return };
+        let mut captured = self
+            .0
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner);
+        if captured.is_none() {
+            *captured = Some(details);
+        }
+    }
+}
+
+fn panic_message(payload: &(dyn std::any::Any + Send)) -> String {
+    if let Some(message) = payload.downcast_ref::<&str>() {
+        (*message).to_owned()
+    } else if let Some(message) = payload.downcast_ref::<String>() {
+        message.clone()
+    } else {
+        "non-string panic payload".to_owned()
+    }
+}
+
+fn install_panic_capture_hook() {
+    static INSTALL: std::sync::Once = std::sync::Once::new();
+    INSTALL.call_once(|| {
+        let previous = std::panic::take_hook();
+        std::panic::set_hook(Box::new(move |info| {
+            let capture = ACTIVE_PANIC_CAPTURE
+                .try_with(|slot| slot.borrow().clone())
+                .ok()
+                .flatten();
+            if let Some(capture) = capture {
+                capture.record(info);
+            } else {
+                previous(info);
+            }
+        }));
+    });
+}
+
+pub(crate) fn current_panic_capture() -> Option<PanicCapture> {
+    ACTIVE_PANIC_CAPTURE.with(|slot| slot.borrow().clone())
+}
+
+pub(crate) fn with_panic_capture<R>(capture: Option<PanicCapture>, f: impl FnOnce() -> R) -> R {
+    let previous = ACTIVE_PANIC_CAPTURE.with(|slot| slot.replace(capture));
+    let _guard = PanicCaptureGuard(previous);
+    f()
+}

 /// Run an unpack pipeline, converting any internal panic into a clean
-/// [`UnpackError::Corrupt`] so the public API stays panic-free on any input
-/// (truncated/garbled files chase offsets out of bounds). The default panic
-/// hook is suppressed transiently so a trapped panic does not spill a
-/// backtrace to stderr.
+/// [`UnpackError::InternalPanic`] so the public API stays panic-free on any input
+/// (truncated/garbled files chase offsets out of bounds). The panic location and
+/// payload are captured for diagnostics without printing a backtrace to stderr.
 ///
 /// Note: allocation *failures* abort the process and are NOT caught here; size
 /// requests are bounds-checked against [`MAX_IMAGE_SIZE`] before allocating.
@@ -32,17 +159,15 @@ pub(crate) fn catch_unpack<F>(f: F) -> Result<Vec<u8>, UnpackError>
 where
    F: FnOnce() -> Result<Vec<u8>, UnpackError>,
 {
-    // Hook suppression is skipped on wasm: the prebuilt std cannot unwind
-    // there, so a panic traps immediately — and the suppressed hook would
-    // hide the panic message, leaving a bare `unreachable` with no clue.
-    #[cfg(not(target_arch = "wasm32"))]
-    let prev = std::panic::take_hook();
-    #[cfg(not(target_arch = "wasm32"))]
-    std::panic::set_hook(Box::new(|_| {}));
-    let r = std::panic::catch_unwind(std::panic::AssertUnwindSafe(f));
-    #[cfg(not(target_arch = "wasm32"))]
-    std::panic::set_hook(prev);
-    r.unwrap_or(Err(UnpackError::Corrupt))
+    install_panic_capture_hook();
+    let capture = PanicCapture::new();
+    let r = with_panic_capture(Some(capture.clone()), || {
+        std::panic::catch_unwind(std::panic::AssertUnwindSafe(f))
+    });
+    match r {
+        Ok(result) => result,
+        Err(payload) => Err(capture.into_error(payload.as_ref())),
+    }
 }

 /// Crackproof header magic stored in `keys[1]`/`info[1]`.
@@ -80,11 +205,8 @@ fn key_table(input: &[u8]) -> Option<[u32; 8]> {
    if input.len() < 4128 {
        return None;
    }
-    // Validate PE signature. `checked_add`, not `+`: `usize` is 32-bit on
-    // wasm32, where an `e_lfanew` of 0xFFFF_FFFC..=0xFFFF_FFFF wraps the bound
-    // check, and the slice below then panics with start > end. `detect` runs on
-    // the folder-scan threads and (in the web app) on the main thread outside
-    // the disposable-worker isolation, so it must not panic on any input.
+    // Validate the PE signature with checked arithmetic so a crafted offset
+    // cannot wrap the bounds check on a narrower target.
    let e_lfanew = primitives::get_u32(input, 0x3C);
    let pe_start = e_lfanew as usize;
    if pe_start.checked_add(4).is_none_or(|end| end > input.len()) {
@@ -194,13 +316,76 @@ pub fn unpack_auto_v(input: &[u8], verbose: bool) -> Result<(Kind, Vec<u8>), Unp
                Ok(out) => out,
                Err(dll_err) => match exe::unpack_v(input, verbose) {
                    Ok(out) => out,
-                    // Surface the DLL-pipeline error, not the EXE one: for a
-                    // genuinely corrupt DLL the DLL error is the more relevant
-                    // diagnostic, and the EXE fallback is best-effort.
-                    Err(_) => return Err(dll_err),
+                    Err(exe_err) => {
+                        return Err(UnpackError::PipelineFallbackFailed {
+                            dll: Box::new(dll_err),
+                            exe: Box::new(exe_err),
+                        });
+                    }
                },
            }
        }
    };
    Ok((detected.kind, out))
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn caught_panic_reports_location_and_message() {
+        let error = catch_unpack(|| -> Result<Vec<u8>, UnpackError> {
+            panic!("test panic");
+        })
+        .expect_err("panic must become an error");
+        let UnpackError::InternalPanic {
+            message,
+            file,
+            line,
+            column,
+        } = error
+        else {
+            panic!("unexpected error: {error}");
+        };
+        assert_eq!(message, "test panic");
+        assert!(
+            file.ends_with("senbei-engine/src/windows/mod.rs")
+                || file.ends_with("senbei-engine\\src\\windows\\mod.rs")
+        );
+        assert!(line > 0);
+        assert!(column > 0);
+    }
+
+    #[test]
+    fn worker_panic_keeps_the_worker_source_location() {
+        let error = catch_unpack(|| -> Result<Vec<u8>, UnpackError> {
+            let capture = current_panic_capture();
+            let result = std::thread::spawn(move || {
+                with_panic_capture(capture, || panic!("worker panic"));
+            })
+            .join();
+            if let Err(payload) = result {
+                std::panic::resume_unwind(payload);
+            }
+            Ok(Vec::new())
+        })
+        .expect_err("worker panic must become an error");
+        let UnpackError::InternalPanic {
+            message,
+            file,
+            line,
+            column,
+        } = error
+        else {
+            panic!("unexpected error: {error}");
+        };
+        assert_eq!(message, "worker panic");
+        assert!(
+            file.ends_with("senbei-engine/src/windows/mod.rs")
+                || file.ends_with("senbei-engine\\src\\windows\\mod.rs")
+        );
+        assert!(line > 0);
+        assert!(column > 0);
+    }
+}
@@ -19,20 +19,6 @@
 use std::sync::Mutex;
 use std::sync::atomic::{AtomicBool, Ordering};

-/// Worker-thread cap. `SENBEI_THREADS` overrides it (`1` forces the sequential
-/// path); otherwise the host's available parallelism; otherwise 1.
-pub(crate) fn thread_cap() -> usize {
-    if let Ok(v) = std::env::var("SENBEI_THREADS")
-        && let Ok(n) = v.trim().parse::<usize>()
-        && n >= 1
-    {
-        return n;
-    }
-    std::thread::available_parallelism()
-        .map(|n| n.get())
-        .unwrap_or(1)
-}
-
 /// Run `f(i, span_base, span)` for every block `i`, fanning out across worker
 /// threads when the spans are disjoint and worthwhile, else sequentially.
 ///
@@ -46,7 +32,7 @@ pub(crate) fn thread_cap() -> usize {
 ///
 /// Returns the first `Err` any block produces; re-raises the first block panic
 /// on the calling thread (so the pipeline's existing `catch_unpack` still
-/// converts it to `UnpackError::Corrupt`).
+/// converts it to `UnpackError::InternalPanic`).
 pub(crate) fn parallel_for<E, F>(
    buf: &mut [u8],
    spans: &[(usize, usize)],
@@ -102,7 +88,7 @@ where
        }
    }

-    let cap = thread_cap();
+    let cap = crate::thread_cap();
    let per = min_per_thread.max(1);
    let workers = if cap > 1 && n >= per.saturating_mul(2) {
        cap.min(n / per)
@@ -125,6 +111,7 @@ where
    let stop = AtomicBool::new(false);
    let first_err: Mutex<Option<E>> = Mutex::new(None);
    let first_panic: Mutex<Option<Box<dyn std::any::Any + Send>>> = Mutex::new(None);
+    let panic_capture = super::current_panic_capture();

    std::thread::scope(|scope| {
        for _ in 0..workers {
@@ -133,6 +120,7 @@ where
            let first_err = &first_err;
            let first_panic = &first_panic;
            let f = &f;
+            let panic_capture = panic_capture.clone();
            scope.spawn(move || {
                loop {
                    if stop.load(Ordering::Relaxed) {
@@ -141,9 +129,15 @@ where
                    let next = iter.lock().unwrap().next();
                    let Some((i, piece)) = next else { break };
                    let span = piece.unwrap();
-                    let r = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
-                        f(i, spans[i].0, span)
-                    }));
+                    // Keep details local until this panic wins `first_panic`;
+                    // otherwise simultaneous workers could pair one worker's
+                    // location with another worker's propagated payload.
+                    let block_capture = panic_capture.as_ref().map(|_| super::PanicCapture::new());
+                    let r = super::with_panic_capture(block_capture.clone(), || {
+                        std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
+                            f(i, spans[i].0, span)
+                        }))
+                    });
                    match r {
                        Ok(Ok(())) => {}
                        Ok(Err(e)) => {
@@ -157,6 +151,11 @@ where
                        Err(panic) => {
                            let mut slot = first_panic.lock().unwrap();
                            if slot.is_none() {
+                                if let (Some(parent), Some(block)) =
+                                    (&panic_capture, &block_capture)
+                                {
+                                    parent.merge_from(block);
+                                }
                                *slot = Some(panic);
                            }
                            stop.store(true, Ordering::Relaxed);
@@ -0,0 +1,30 @@
+[package]
+name = "senbei-io"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+description = "Filesystem, scanning, logging, and CLI orchestration for Senbei"
+
+[dependencies]
+anyhow.workspace = true
+senbei-crypto.workspace = true
+indicatif.workspace = true
+memmap2.workspace = true
+owo-colors.workspace = true
+senbei-engine.workspace = true
+senbei-elf.workspace = true
+senbei-pe.workspace = true
+senbei-metadata.workspace = true
+sha2.workspace = true
+tempfile.workspace = true
+walkdir.workspace = true
+zip.workspace = true
+
+[target.'cfg(windows)'.dependencies]
+windows.workspace = true
+
+[target.'cfg(all(not(windows), not(target_arch = "wasm32")))'.dependencies]
+libc.workspace = true
+
+[dev-dependencies]
+tempfile.workspace = true
@@ -0,0 +1,462 @@
+//! Android target orchestration: protected AArch64 shared libraries (`.so`),
+//! app packages (`.apk` / `.apks` / `.xapk`), and the Android variant of the
+//! il2cpp method-token obfuscation.
+//!
+//! The protection scheme hollows out an ELF64/AArch64 shared object and moves
+//! the original bytes into an encrypted payload appended as a `SHT_LOUSER`
+//! section; restoration extracts the stage-2 module set
+//! ([`senbei_engine::android`]) and rebuilds the static image
+//! ([`senbei_engine::android`]). Some il2cpp builds additionally embed their
+//! metadata blob — XOR-wrapped, with no standalone `global-metadata.dat` in
+//! the assets — inside the library's data section; after a successful restore
+//! the blob is located by content and unwrapped
+//! ([`senbei_metadata::android::extract_embedded_metadata`]).
+//!
+//! All functions in this module are native filesystem orchestration; the web
+//! app (wasm) never touches them.
+
+use std::collections::HashSet;
+use std::fs::File;
+use std::io::{BufWriter, Read, Seek, Write};
+use std::path::{Path, PathBuf};
+
+use anyhow::{Context, Result, bail};
+use memmap2::{Mmap, MmapOptions};
+use senbei_crypto::hex_digest;
+use senbei_engine::android::{ExtractOptions, extract_stage2, is_protected_libil2cpp};
+use senbei_engine::android::{RestoreOptions, restore_libil2cpp};
+use sha2::{Digest, Sha256};
+use zip::ZipArchive;
+
+pub use crate::METADATA_FILE_NAME;
+
+/// Package extensions recognised as Android app packages. Packages are
+/// *containers*: membership is decided by extension plus the ZIP magic, while
+/// only `.so` and `global-metadata.dat` entries are read.
+const PACKAGE_EXTENSIONS: [&str; 3] = ["apk", "apks", "xapk"];
+
+pub(crate) fn is_package_name(path: &Path) -> bool {
+    path.extension()
+        .and_then(|value| value.to_str())
+        .is_some_and(|value| {
+            PACKAGE_EXTENSIONS
+                .iter()
+                .any(|ext| value.eq_ignore_ascii_case(ext))
+        })
+}
+
+pub(crate) fn is_so_name(path: &Path) -> bool {
+    path.extension()
+        .and_then(|value| value.to_str())
+        .is_some_and(|value| value.eq_ignore_ascii_case("so"))
+}
+
+pub(crate) fn is_android_entry_name(path: &Path) -> bool {
+    path.file_name()
+        .and_then(|name| name.to_str())
+        .is_some_and(|name| name.eq_ignore_ascii_case(METADATA_FILE_NAME))
+        || is_so_name(path)
+}
+
+/// Whether `prefix` (the first bytes of a file) is an ELF64/AArch64 image.
+/// Only those can be protected Android libraries, so the folder scan uses this
+/// cheap check to decide when the full-file protection probe is worth its
+/// read.
+pub fn is_elf64_aarch64(prefix: &[u8]) -> bool {
+    senbei_elf::is_aarch64_prefix(prefix)
+}
+
+/// Whether `path` is an Android app package: a recognised package extension
+/// and the local-file-header zip magic in `prefix`.
+pub fn is_app_package(path: &Path, prefix: &[u8]) -> bool {
+    is_package_name(path) && prefix.starts_with(b"PK\x03\x04")
+}
+
+/// Probe a file on disk: true when it is a protected AArch64 library.
+/// Reads the whole file (the payload section is found through the
+/// section-header table at the end); call only after [`is_elf64_aarch64`]
+/// has matched a prefix.
+pub fn is_protected_so_file(path: &Path) -> bool {
+    let Ok(file) = File::open(path) else {
+        return false;
+    };
+    let Ok(bytes) = map_read_only(&file, path) else {
+        return false;
+    };
+    is_elf64_aarch64(&bytes) && is_protected_libil2cpp(&bytes)
+}
+
+pub fn file_content_identity(path: &Path) -> std::io::Result<String> {
+    let file = File::open(path)?;
+    // SAFETY: the file remains open for the mapping lifetime and the mapping
+    // is read-only.
+    let bytes = unsafe { MmapOptions::new().map(&file)? };
+    Ok(content_identity(&bytes))
+}
+
+/// Restore one protected `.so` to `dest`.
+///
+/// The stage-2 module set is extracted into a temporary workspace (it is an
+/// implementation detail of the two-phase restore, not user-facing output).
+/// Returns the unwrapped embedded metadata blob when the restored image
+/// carries one (see the module docs); the caller decides where to write it.
+pub fn restore_so_file(input: &Path, dest: &Path, verbose: bool) -> Result<Option<Vec<u8>>> {
+    let temporary = tempfile::tempdir().context("create stage-2 workspace")?;
+    let stage2_dir = temporary.path().join("stage2");
+    extract_stage2(&ExtractOptions::with_defaults(
+        input.to_path_buf(),
+        stage2_dir.clone(),
+    ))
+    .context("extract stage-1/stage-2 payload")?;
+    restore_libil2cpp(&RestoreOptions {
+        input: input.to_path_buf(),
+        output: dest.to_path_buf(),
+        index: stage2_dir.join("index.json"),
+        dump_auxiliary: None,
+        outer_only: false,
+        preserve_entrypoint: false,
+        verbose,
+    })
+    .context("restore protected library")?;
+    let restored =
+        std::fs::read(dest).with_context(|| format!("read restored `{}`", dest.display()))?;
+    Ok(senbei_metadata::android::extract_embedded_metadata(
+        &restored,
+    ))
+}
+
+/// Content identity for cross-source deduplication: the same library may
+/// appear loose in a tree, in its `.apk`, and again in an `.apks`/`.xapk`
+/// bundle — restore it once, at the highest-priority source's destination.
+pub fn content_identity(data: &[u8]) -> String {
+    let mut digest = Sha256::new();
+    digest.update(data);
+    hex_digest(&digest.finalize())
+}
+
+/// Restore an il2cpp metadata blob (Android seeded permutation first, then the
+/// structural remap used by the Windows builds).
+///
+/// The Android variant obfuscates MethodDef RIDs with a keyed five-round
+/// permutation; the correct seed is recovered by intersecting per-image key
+/// residues, and the restore *validates* every restored RID against its
+/// canonical per-module index — so an unusable seed fails loudly and the
+/// caller falls through to the structural remap, which targets the same
+/// canonical form. Both paths are no-ops (`remapped == 0`) on an
+/// already-clean blob.
+pub fn restore_metadata_bytes(data: &[u8]) -> anyhow::Result<(Vec<u8>, senbei_metadata::Report)> {
+    if let Ok(discovery) = senbei_metadata::android::discover_method_token_seeds(data)
+        && matches!(discovery.version, 29 | 31 | 39)
+    {
+        let mut seeds = discovery.seed_candidates.clone();
+        if seeds.is_empty() {
+            seeds.push(senbei_metadata::android::DEFAULT_METHOD_TOKEN_SEED);
+        }
+        // Trial-and-validate: a wrong seed fails the restore's full-coverage
+        // RID check, so ambiguous candidates cost one extra pass each and a
+        // build with an unseeded permutation falls through to the structural
+        // remap rather than producing a silently wrong file.
+        for seed in seeds {
+            if let Ok((out, report)) = senbei_metadata::android::restore_method_tokens(data, seed) {
+                return Ok((
+                    out,
+                    senbei_metadata::Report {
+                        version: report.version,
+                        methods: report.methods,
+                        remapped: report.changed_tokens,
+                        modules: report.images_with_methods,
+                    },
+                ));
+            }
+        }
+    }
+    let (out, report) = senbei_metadata::deobfuscate(data).map_err(anyhow::Error::new)?;
+    Ok((out, report))
+}
+
+/// What happened to one archive entry (or one loose Android target).
+#[derive(Debug)]
+pub struct EntryOutcome {
+    /// Human-readable source label, e.g. `base.apk::lib/arm64-v8a/libil2cpp.so`.
+    pub label: String,
+    /// Where the restored bytes were written (meaningless unless `status` is
+    /// `Restored`).
+    pub dest: PathBuf,
+    pub kind: EntryKind,
+    pub status: EntryStatus,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum EntryKind {
+    /// A protected shared library, restored.
+    So,
+    /// An il2cpp metadata blob, de-obfuscated (`remapped` tokens changed).
+    Metadata { remapped: usize },
+    /// A metadata blob unwrapped from a restored library's data section.
+    EmbeddedMetadata,
+}
+
+#[derive(Debug)]
+pub enum EntryStatus {
+    Restored,
+    /// Byte-identical content was already restored from a higher-priority
+    /// source; no output written.
+    Duplicate,
+    /// Content-probed but not a target (unprotected library).
+    NotTarget,
+    /// A metadata blob whose tokens were already canonical; no copy written.
+    Unchanged,
+    /// Recognised as a target but the restore failed.
+    Failed(anyhow::Error),
+}
+
+/// Restore every protected library and metadata blob inside one app package.
+///
+/// `rel` is the package's path relative to the scanned root (or its bare file
+/// name in single-file mode); outputs mirror the package's internal layout
+/// under `out_root/rel/`, with [`crate::job::out_name`] renaming. `seen`
+/// carries content identities already restored from higher-priority sources
+/// (loose files first, then `.apk`, then bundles) across the whole run.
+pub fn restore_package(
+    package: &Path,
+    rel: &Path,
+    out_root: &Path,
+    seen: &mut HashSet<String>,
+    verbose: bool,
+) -> Result<Vec<EntryOutcome>> {
+    let bundle = package
+        .extension()
+        .and_then(|value| value.to_str())
+        .is_some_and(|value| {
+            value.eq_ignore_ascii_case("apks") || value.eq_ignore_ascii_case("xapk")
+        });
+    let mut archive = open_package(package)?;
+    let temporary = tempfile::tempdir().context("create package workspace")?;
+    let mut outcomes = Vec::new();
+
+    let mut direct = Vec::new();
+    let mut nested = Vec::new();
+    for index in 0..archive.len() {
+        let (name, is_dir) = {
+            let entry = archive.by_index(index)?;
+            (entry.enclosed_name(), entry.is_dir())
+        };
+        if is_dir {
+            continue;
+        }
+        let Some(name) = name else {
+            bail!("unsafe entry path in package `{}`", package.display());
+        };
+        if bundle {
+            if name
+                .extension()
+                .and_then(|value| value.to_str())
+                .is_some_and(|value| value.eq_ignore_ascii_case("apk"))
+            {
+                nested.push((index, name));
+            }
+        } else {
+            if is_android_entry_name(&name) {
+                direct.push((index, name));
+            }
+        }
+    }
+    for (index, name) in direct {
+        let label = format!("{}::{}", rel.display(), name.display());
+        let dest = out_root.join(rel).join(crate::job::out_name(&name));
+        let mut entry_outcomes = restore_package_entry(
+            &mut archive,
+            index,
+            &label,
+            &dest,
+            &temporary,
+            seen,
+            verbose,
+        )
+        .with_context(|| format!("extract `{label}`"))?;
+        outcomes.append(&mut entry_outcomes);
+    }
+    for (index, name) in nested {
+        let nested_label = rel.join(&name);
+        let nested_path = extract_entry(&mut archive, index, &temporary, &nested_label)
+            .with_context(|| format!("extract `{}`", nested_label.display()))?;
+        let mut nested_archive = open_package(&nested_path)?;
+        let mut entries = Vec::new();
+        for nested_index in 0..nested_archive.len() {
+            let (entry_name, is_dir) = {
+                let entry = nested_archive.by_index(nested_index)?;
+                (entry.enclosed_name(), entry.is_dir())
+            };
+            if !is_dir {
+                let Some(entry_name) = entry_name else {
+                    bail!("unsafe entry path in `{}`", nested_label.display());
+                };
+                if is_android_entry_name(&entry_name) {
+                    entries.push((nested_index, entry_name));
+                }
+            }
+        }
+        // Keep the nested package's stem in the output layout so two splits
+        // carrying same-named entries cannot collide.
+        let base = rel.join(name.with_extension(""));
+        for (nested_index, entry_name) in entries {
+            let label = format!("{}::{}", nested_label.display(), entry_name.display());
+            let dest = out_root.join(&base).join(crate::job::out_name(&entry_name));
+            let mut entry_outcomes = restore_package_entry(
+                &mut nested_archive,
+                nested_index,
+                &label,
+                &dest,
+                &temporary,
+                seen,
+                verbose,
+            )
+            .with_context(|| format!("extract `{label}`"))?;
+            outcomes.append(&mut entry_outcomes);
+        }
+    }
+    Ok(outcomes)
+}
+
+/// Probe one extracted package entry and restore it when it is a target.
+/// Returns one outcome per produced/consumed artifact: the entry itself, plus
+/// an `EmbeddedMetadata` outcome when the restored library carried a blob.
+fn restore_package_entry<R: Read + Seek>(
+    archive: &mut ZipArchive<R>,
+    index: usize,
+    label: &str,
+    dest: &Path,
+    temporary: &tempfile::TempDir,
+    seen: &mut HashSet<String>,
+    verbose: bool,
+) -> Result<Vec<EntryOutcome>> {
+    let entry_path = extract_entry(archive, index, temporary, Path::new(label))?;
+    let entry_file =
+        File::open(&entry_path).with_context(|| format!("open extracted `{label}`"))?;
+    let entry_data = map_read_only(&entry_file, &entry_path)?;
+
+    let is_so = is_elf64_aarch64(&entry_data) && is_protected_libil2cpp(&entry_data);
+    let is_meta = !is_so && senbei_metadata::is_metadata(&entry_data);
+    let outcome = |kind, status| EntryOutcome {
+        label: label.to_owned(),
+        dest: dest.to_path_buf(),
+        kind,
+        status,
+    };
+    if !is_so && !is_meta {
+        return Ok(vec![outcome(EntryKind::So, EntryStatus::NotTarget)]);
+    }
+    if !seen.insert(content_identity(&entry_data)) {
+        let kind = if is_so {
+            EntryKind::So
+        } else {
+            EntryKind::Metadata { remapped: 0 }
+        };
+        return Ok(vec![outcome(kind, EntryStatus::Duplicate)]);
+    }
+
+    if is_so {
+        drop(entry_data);
+        drop(entry_file);
+        return Ok(match restore_so_file(&entry_path, dest, verbose) {
+            Ok(embedded) => {
+                let mut outcomes = vec![outcome(EntryKind::So, EntryStatus::Restored)];
+                if let Some(blob) = embedded {
+                    let meta_dest = embedded_metadata_dest(dest);
+                    let status = match write_metadata_blob(&meta_dest, &blob) {
+                        Ok(()) => EntryStatus::Restored,
+                        Err(error) => EntryStatus::Failed(error),
+                    };
+                    outcomes.push(EntryOutcome {
+                        label: format!("{label} (embedded metadata)"),
+                        dest: meta_dest,
+                        kind: EntryKind::EmbeddedMetadata,
+                        status,
+                    });
+                }
+                outcomes
+            }
+            Err(error) => vec![outcome(EntryKind::So, EntryStatus::Failed(error))],
+        });
+    }
+
+    // Metadata entry: write only when the restore actually changed tokens —
+    // a clean blob needs no copy (same contract as loose metadata files).
+    let kind_and_status = match restore_metadata_bytes(&entry_data) {
+        Ok((out, report)) if report.remapped > 0 => {
+            let kind = EntryKind::Metadata {
+                remapped: report.remapped,
+            };
+            match write_metadata_blob(dest, &out) {
+                Ok(()) => (kind, EntryStatus::Restored),
+                Err(error) => (kind, EntryStatus::Failed(error)),
+            }
+        }
+        Ok(_) => (EntryKind::Metadata { remapped: 0 }, EntryStatus::Unchanged),
+        Err(error) => (
+            EntryKind::Metadata { remapped: 0 },
+            EntryStatus::Failed(error),
+        ),
+    };
+    Ok(vec![outcome(kind_and_status.0, kind_and_status.1)])
+}
+
+/// Output path for a metadata blob unwrapped from a restored library: next to
+/// the library, under the standard file name (with the usual `.unpack` infix).
+pub fn embedded_metadata_dest(restored_so: &Path) -> PathBuf {
+    let dir = restored_so.parent().unwrap_or_else(|| Path::new("."));
+    dir.join(crate::job::out_name(Path::new(METADATA_FILE_NAME)))
+}
+
+/// Write a metadata blob, creating the parent directory. The restore writes
+/// its own output atomically; metadata blobs use the shared orchestration
+/// atomic writer to keep the same mid-write failure semantics.
+fn write_metadata_blob(dest: &Path, data: &[u8]) -> Result<()> {
+    if let Some(parent) = dest.parent() {
+        std::fs::create_dir_all(parent)
+            .with_context(|| format!("create `{}`", parent.display()))?;
+    }
+    crate::atomic::write_atomic(dest, data)
+        .map_err(anyhow::Error::from)
+        .context("write metadata output")
+}
+
+fn open_package(path: &Path) -> Result<ZipArchive<std::fs::File>> {
+    let file = std::fs::File::open(path).with_context(|| format!("open `{}`", path.display()))?;
+    ZipArchive::new(file).with_context(|| format!("read package `{}`", path.display()))
+}
+
+/// Stream one package entry to a temporary, seekable file. The Android engine
+/// needs random access to ELF section tables, while the ZIP reader itself is
+/// consumed directly without creating an in-memory compressed or decompressed
+/// copy.
+fn extract_entry<R: Read + Seek>(
+    archive: &mut ZipArchive<R>,
+    index: usize,
+    temporary: &tempfile::TempDir,
+    label: &Path,
+) -> Result<PathBuf> {
+    let mut entry = archive.by_index(index)?;
+    let key = format!("{}-{index:08x}", label.display());
+    // `:` appears in `package::entry` labels and is invalid in Windows file
+    // names; sanitize every path-ish separator.
+    let destination = temporary.path().join(key.replace(['\\', '/', ':'], "_"));
+    let output_size = entry.size();
+    let mut output = BufWriter::new(std::fs::File::create(&destination)?);
+    let written = std::io::copy(&mut entry, &mut output)?;
+    output.flush()?;
+    if written != output_size {
+        bail!(
+            "entry `{key}` decompressed to 0x{:x}, expected 0x{output_size:x}",
+            written
+        );
+    }
+    Ok(destination)
+}
+
+fn map_read_only(file: &File, path: &Path) -> Result<Mmap> {
+    // SAFETY: the file descriptor remains open for the returned mapping's
+    // lifetime, and this mapping is read-only.
+    unsafe { MmapOptions::new().map(file) }
+        .with_context(|| format!("map extracted `{}`", path.display()))
+}
@@ -0,0 +1,16 @@
+//! Shared atomic filesystem writes for native orchestration.
+
+use std::path::{Path, PathBuf};
+
+/// Write `bytes` through a sibling temporary file and replace `dest` only after
+/// the complete write succeeds.
+pub(crate) fn write_atomic(dest: &Path, bytes: &[u8]) -> std::io::Result<()> {
+    let mut temporary_name = dest.as_os_str().to_os_string();
+    temporary_name.push(".senbei-tmp");
+    let temporary = PathBuf::from(temporary_name);
+    let result = std::fs::write(&temporary, bytes).and_then(|()| std::fs::rename(&temporary, dest));
+    if result.is_err() {
+        let _ = std::fs::remove_file(&temporary);
+    }
+    result
+}
@@ -0,0 +1,850 @@
+use std::path::{Path, PathBuf};
+
+use crate::atomic::write_atomic;
+pub use crate::windows::{
+    UnpackedImage, unpack_bytes, unpack_bytes_force_exe, unpack_one, unpack_one_v,
+};
+
+/// Summary of a folder-mode run.
+#[derive(Default)]
+pub struct Summary {
+    pub unpacked: usize,
+    pub skipped: usize,
+    pub errors: usize,
+    /// Files that unpacked without error but failed the static integrity check
+    /// — likely to crash at runtime (e.g. 0xC0000005). Counted in addition to
+    /// `unpacked` (a suspect file is still written).
+    pub suspect: usize,
+    /// il2cpp `global-metadata.dat` files de-obfuscated (method tokens remapped),
+    /// including blobs unwrapped from restored Android libraries.
+    pub metadata: usize,
+    /// Android app packages (`.apk`/`.apks`/`.xapk`) opened and searched.
+    pub packages: usize,
+    /// Wall-clock duration of the folder run in milliseconds.
+    pub duration_ms: u128,
+}
+
+impl Summary {
+    /// The summary line shared by CLI output and the log file.
+    pub fn line(&self) -> String {
+        let mut line = format!(
+            "{} unpacked · {} skipped · {} errors · {} suspect · {} metadata",
+            self.unpacked, self.skipped, self.errors, self.suspect, self.metadata
+        );
+        if self.packages > 0 {
+            line.push_str(&format!(" · {} packages", self.packages));
+        }
+        line
+    }
+}
+
+/// Default output root for a folder unpack: `<root>/unpack`.
+pub fn default_out_root_for_folder(root: &Path) -> PathBuf {
+    root.join("unpack")
+}
+
+/// Default output root for a single-file unpack: `<parent>/unpack` (or `./unpack`
+/// when the input has no parent directory).
+pub fn default_out_root_for_file(input: &Path) -> PathBuf {
+    let parent = input
+        .parent()
+        .filter(|p| !p.as_os_str().is_empty())
+        .unwrap_or_else(|| Path::new("."));
+    parent.join("unpack")
+}
+
+/// Unpack all Crackproof-protected files under `root`, writing results into
+/// a mirrored subtree under `out_dir` (or `root/unpack` if None).
+///
+/// Each file is processed independently: a panic or error in one file is
+/// isolated and counted as an error; the loop continues.
+pub fn run_folder(root: &Path, out_dir: Option<&Path>, quiet: bool) -> anyhow::Result<Summary> {
+    run_folder_v(root, out_dir, if quiet { 1 } else { 0 }, false, false)
+}
+
+/// Like [`run_folder`], but prints detailed `[N/9]` step progress (and a final
+/// `Write to <dest>` line) for each EXE when `verbose` is true.
+///
+/// `quiet` is a level: `>= 1` suppresses per-file UI lines and the progress bar.
+/// When `no_log` is true, no `senbei-*.log` is created under the out root.
+pub fn run_folder_v(
+    root: &Path,
+    out_dir: Option<&Path>,
+    quiet: u8,
+    verbose: bool,
+    no_log: bool,
+) -> anyhow::Result<Summary> {
+    run_folder_opts(
+        root,
+        out_dir,
+        quiet,
+        verbose,
+        no_log,
+        crate::scan::scan_all_env(),
+    )
+}
+
+/// Like [`run_folder_v`], but with the scan pre-filter explicitly controlled.
+///
+/// When `scan_all` is true selected target names below the minimum size are
+/// also opened and content-probed. Other filenames are never opened.
+pub fn run_folder_opts(
+    root: &Path,
+    out_dir: Option<&Path>,
+    quiet: u8,
+    verbose: bool,
+    no_log: bool,
+    scan_all: bool,
+) -> anyhow::Result<Summary> {
+    let t0 = std::time::Instant::now();
+    let out_root = out_dir
+        .map(Path::to_path_buf)
+        .unwrap_or_else(|| default_out_root_for_folder(root));
+    std::fs::create_dir_all(&out_root)?;
+    let log = if no_log {
+        None
+    } else {
+        let log = crate::logfile::Log::create(&out_root)?;
+        log.step(&format!("Senbei {}", env!("CARGO_PKG_VERSION")));
+        log.step(&format!(
+            "started {}",
+            crate::logfile::local_stamp_display()
+        ));
+        log.step(&format!("input {}", root.display()));
+        log.step(&format!("out {}", out_root.display()));
+        Some(log)
+    };
+    // Single merged directory walk: returns Crackproof unpack candidates, il2cpp
+    // metadata blobs, and Android targets from one traversal (see
+    // [`crate::scan::find_targets_opts`]). Files the free directory metadata
+    // already rules out are never opened — on asset-heavy trees the per-file
+    // open+read latency, not the traversal, is the whole cost.
+    let scan = crate::scan::find_targets_opts(root, scan_all);
+    let candidates = scan.crackproof.as_slice();
+    let metas = scan.metadata.as_slice();
+    let scan_stats = &scan.stats;
+    // Files the scan could not classify are potential missed targets, not
+    // clean skips: an unreadable directory or a locked il2cpp game assembly must
+    // fail the run (exit 1) rather than report "0 errors" over a partial scan.
+    let scan_failed = scan_stats.walk_errors + scan_stats.probe_errors;
+    if scan_failed > 0 && quiet == 0 {
+        eprintln!(
+            "warning: {} file(s) could not be read during the scan and may be missed targets",
+            scan_failed
+        );
+    }
+    if let Some(log) = &log {
+        if scan_stats.walk_errors > 0 {
+            log.step(&format!(
+                "scan: {} directory entry(s) unreadable",
+                scan_stats.walk_errors
+            ));
+        }
+        if scan_stats.probe_errors > 0 {
+            log.step(&format!(
+                "scan: {} file(s) failed content probe (unreadable or detector panic)",
+                scan_stats.probe_errors
+            ));
+        }
+    }
+    let suppress_file_lines = quiet >= 1;
+    // Quiet wins over verbose: step progress only when quiet == 0 (spec: verbose
+    // lines only when quiet == 0; quiet ≥ 2 must stay fully silent even with -v).
+    let verbose_steps = verbose && quiet == 0;
+    // Verbose mode prints multi-line `[N/9]` step output per file straight to
+    // stdout; an active progress bar would be clobbered by it, so hide the bar
+    // (its per-file ok/err lines still print) when verbose is on.
+    let android_targets = scan.android_so.len() + scan.android_packages.len();
+    let bar = crate::ui::progress(
+        (candidates.len() + android_targets) as u64,
+        quiet >= 1 || verbose,
+    );
+    let mut s = Summary {
+        skipped: scan_stats.skipped,
+        errors: scan_failed,
+        ..Summary::default()
+    };
+
+    // Silence the default panic hook's stderr spew during per-file processing.
+    let default_hook = std::panic::take_hook();
+    std::panic::set_hook(Box::new(|_| {})); // suppress "thread panicked" messages
+
+    for input in candidates {
+        let rel = rel_in_tree(root, input);
+        let dest = out_root.join(out_name(&rel));
+
+        // Wrap in catch_unwind so a single bad file never aborts the folder run.
+        let input_owned = input.clone();
+        let dest_owned = dest.clone();
+        let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
+            unpack_one_v(&input_owned, &dest_owned, verbose_steps)
+        }));
+
+        match result {
+            Ok(Ok((kind, report))) => {
+                s.unpacked += 1;
+                crate::ui::ok(&bar, suppress_file_lines, &rel, kind, &dest);
+                if let Some(log) = &log {
+                    log.step(&format!("OK {rel:?} -> {dest:?} ({kind:?})"));
+                }
+                if !report.ok() {
+                    s.suspect += 1;
+                    crate::ui::suspect(&bar, suppress_file_lines, &rel, &report);
+                    if let Some(log) = &log {
+                        log.step(&format!("SUSPECT {rel:?}: {}", report.issues.join("; ")));
+                    }
+                }
+            }
+            Ok(Err(e)) => {
+                s.errors += 1;
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {rel:?}: {e:#}"));
+                }
+            }
+            Err(panic) => {
+                s.errors += 1;
+                let e = anyhow::anyhow!("unexpected panic: {}", panic_payload(&panic));
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "ERR {rel:?}: panic during unpack: {}",
+                        panic_payload(&panic)
+                    ));
+                }
+            }
+        }
+        bar.inc(1);
+    }
+
+    // Android pass: protected AArch64 libraries and app packages. Loose `.so`
+    // files restore first so the cross-source dedup keeps them over a copy
+    // inside a package (loose beats `.apk` beats `.apks`/`.xapk` bundle).
+    let mut android_seen = std::collections::HashSet::new();
+    // Hashing a protected library costs a full read, so only pay it when a
+    // duplicate source can actually exist in this run.
+    let android_dedup = scan.android_so.len() > 1 || !scan.android_packages.is_empty();
+    for input in &scan.android_so {
+        let rel = rel_in_tree(root, input);
+        let dest = out_root.join(out_name(&rel));
+        // Unreadable here is fine: the restore reports the same error.
+        if android_dedup
+            && let Ok(identity) = crate::android::file_content_identity(input)
+            && !android_seen.insert(identity)
+        {
+            s.skipped += 1;
+            if let Some(log) = &log {
+                log.step(&format!("SKIP {rel:?}: duplicate of an earlier target"));
+            }
+            bar.inc(1);
+            continue;
+        }
+        let input_owned = input.clone();
+        let dest_owned = dest.clone();
+        let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
+            crate::android::restore_so_file(&input_owned, &dest_owned, verbose_steps)
+        }));
+        match result {
+            Ok(Ok(embedded)) => {
+                s.unpacked += 1;
+                crate::ui::ok_label(
+                    &bar,
+                    suppress_file_lines,
+                    &rel.display().to_string(),
+                    "So",
+                    &dest,
+                );
+                if let Some(log) = &log {
+                    log.step(&format!("OK {rel:?} -> {dest:?} (Android SO)"));
+                }
+                match write_embedded_metadata(embedded, &dest) {
+                    Ok(Some(meta_dest)) => {
+                        s.metadata += 1;
+                        crate::ui::ok_label(
+                            &bar,
+                            suppress_file_lines,
+                            &format!("{} (embedded metadata)", rel.display()),
+                            "metadata",
+                            &meta_dest,
+                        );
+                        if let Some(log) = &log {
+                            log.step(&format!("META {rel:?} (embedded) -> {meta_dest:?}"));
+                        }
+                    }
+                    Ok(None) => {}
+                    Err(e) => {
+                        s.errors += 1;
+                        crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                        if let Some(log) = &log {
+                            log.step(&format!("ERR {rel:?}: embedded metadata: {e:#}"));
+                        }
+                    }
+                }
+            }
+            Ok(Err(e)) => {
+                s.errors += 1;
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {rel:?}: {e:#}"));
+                }
+            }
+            Err(panic) => {
+                s.errors += 1;
+                let e = anyhow::anyhow!("unexpected panic: {}", panic_payload(&panic));
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "ERR {rel:?}: panic during restore: {}",
+                        panic_payload(&panic)
+                    ));
+                }
+            }
+        }
+        bar.inc(1);
+    }
+    for package in &scan.android_packages {
+        let rel = rel_in_tree(root, package);
+        s.packages += 1;
+        let package_owned = package.clone();
+        let rel_owned = rel.clone().into_owned();
+        let out_root_owned = out_root.clone();
+        let mut seen_taken = std::mem::take(&mut android_seen);
+        let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
+            let outcomes = crate::android::restore_package(
+                &package_owned,
+                &rel_owned,
+                &out_root_owned,
+                &mut seen_taken,
+                verbose_steps,
+            );
+            (outcomes, seen_taken)
+        }));
+        match result {
+            Ok((Ok(outcomes), seen_back)) => {
+                android_seen = seen_back;
+                apply_package_outcomes(outcomes, &mut s, &bar, suppress_file_lines, &log);
+            }
+            Ok((Err(e), seen_back)) => {
+                android_seen = seen_back;
+                s.errors += 1;
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {rel:?}: {e:#}"));
+                }
+            }
+            Err(panic) => {
+                // The dedup set may be in an unknown state after a panic; a
+                // re-scan costs a duplicate restore at worst, never corruption.
+                let e = anyhow::anyhow!("unexpected panic: {}", panic_payload(&panic));
+                s.errors += 1;
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "ERR {rel:?}: panic during package restore: {}",
+                        panic_payload(&panic)
+                    ));
+                }
+            }
+        }
+        bar.inc(1);
+    }
+
+    // il2cpp metadata pass. Crackproof's `-GMD` option obfuscates the method
+    // tokens in `global-metadata.dat`; de-obfuscate any we find so the unpacked
+    // il2cpp game assembly resolves methods instead of indexing its per-module
+    // tables out of bounds (see [`senbei_metadata`]). This is additive to the
+    // Crackproof module unpack above — the metadata blob is not itself a
+    // Crackproof file.
+    for meta in metas.iter() {
+        let rel = rel_in_tree(root, meta);
+        let dest = out_root.join(out_name(&rel));
+        let meta_owned = meta.clone();
+        let dest_owned = dest.clone();
+        let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
+            deobfuscate_metadata_to(&meta_owned, &dest_owned, verbose_steps)
+        }));
+        match result {
+            Ok(Ok(report)) if report.remapped > 0 => {
+                s.metadata += 1;
+                crate::ui::metadata(&bar, suppress_file_lines, &rel, report.remapped, &dest);
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "META {rel:?} -> {dest:?}: v{} remapped {} method tokens",
+                        report.version, report.remapped
+                    ));
+                }
+            }
+            // Recognised metadata that needed no change (not -GMD-obfuscated):
+            // leave it untouched and don't write a redundant copy.
+            Ok(Ok(report)) => {
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "META {rel:?}: v{} already de-obfuscated",
+                        report.version
+                    ));
+                }
+            }
+            Ok(Err(e)) => {
+                // A metadata version we don't handle is NOT a run failure: the
+                // game is simply not -GMD-obfuscated in a layout we know, the
+                // file is left untouched, and the PE unpacks around it may be
+                // fully successful. Count it as skipped (with a visible note),
+                // matching the "anything that doesn't match is left untouched"
+                // contract. Genuine corruption (Malformed) stays an error —
+                // silently exiting 0 would let a failed de-obfuscation pass CI
+                // while the il2cpp game assembly still crashes.
+                if let Some(v) = unsupported_version(&e) {
+                    s.skipped += 1;
+                    if !suppress_file_lines {
+                        eprintln!(
+                            "- {}  unsupported metadata version {v}, left untouched",
+                            rel.display()
+                        );
+                    }
+                    if let Some(log) = &log {
+                        log.step(&format!(
+                            "META SKIP {rel:?}: unsupported metadata version {v}"
+                        ));
+                    }
+                } else {
+                    s.errors += 1;
+                    crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                    if let Some(log) = &log {
+                        log.step(&format!("META ERR {rel:?}: {e:#}"));
+                    }
+                }
+            }
+            Err(panic) => {
+                s.errors += 1;
+                let e = anyhow::anyhow!(
+                    "unexpected panic during de-obfuscation: {}",
+                    panic_payload(&panic)
+                );
+                crate::ui::err(&bar, suppress_file_lines, &rel, &e);
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "META ERR {rel:?}: panic during de-obfuscation: {}",
+                        panic_payload(&panic)
+                    ));
+                }
+            }
+        }
+    }
+
+    // Restore the original panic hook.
+    std::panic::set_hook(default_hook);
+
+    bar.finish_and_clear();
+    s.duration_ms = t0.elapsed().as_millis();
+    if let Some(log) = &log {
+        log.step(&format!("done in {} ms", s.duration_ms));
+        log.step(&format!("summary: {}", s.line()));
+    }
+    Ok(s)
+}
+
+/// Single-file (PE or metadata) with the same log/header/footer/timing as folder mode.
+///
+/// Always returns `Ok(Summary)` for per-file unpack outcomes (including failures,
+/// which set `errors: 1`) so callers always receive `duration_ms`. Fatal `Err`
+/// only when the out dir / log cannot be created.
+pub fn run_file_v(
+    input: &Path,
+    out_dir: Option<&Path>,
+    quiet: u8,
+    verbose: bool,
+    no_log: bool,
+) -> anyhow::Result<Summary> {
+    let t0 = std::time::Instant::now();
+    let out_root = out_dir
+        .map(Path::to_path_buf)
+        .unwrap_or_else(|| default_out_root_for_file(input));
+    std::fs::create_dir_all(&out_root)?;
+
+    let log = if no_log {
+        None
+    } else {
+        let log = crate::logfile::Log::create(&out_root)?;
+        log.step(&format!("Senbei {}", env!("CARGO_PKG_VERSION")));
+        log.step(&format!(
+            "started {}",
+            crate::logfile::local_stamp_display()
+        ));
+        log.step(&format!("input {}", input.display()));
+        log.step(&format!("out {}", out_root.display()));
+        Some(log)
+    };
+
+    let name = out_name(Path::new(input.file_name().unwrap_or_default()));
+    let dest = out_root.join(name);
+    let mut s = Summary::default();
+
+    let prefix = {
+        use std::io::Read;
+        let mut buf = vec![0u8; 8 * 1024];
+        match std::fs::File::open(input).and_then(|mut f| f.read(&mut buf).map(|n| (buf, n))) {
+            Ok((buf, n)) => {
+                let mut b = buf;
+                b.truncate(n);
+                b
+            }
+            Err(_) => Vec::new(),
+        }
+    };
+    let is_meta = senbei_metadata::is_metadata(&prefix);
+    // Android single-file targets are routed by content: a protected AArch64
+    // library probe needs the whole file (its payload section is found through
+    // the section-header table at the end), while a package is a container
+    // handled entry-by-entry. Anything else falls through to the PE pipeline.
+    let is_android_so =
+        crate::android::is_elf64_aarch64(&prefix) && crate::android::is_protected_so_file(input);
+    let is_android_package = !is_android_so && crate::android::is_app_package(input, &prefix);
+
+    if is_meta {
+        match deobfuscate_metadata_to(input, &dest, verbose && quiet == 0) {
+            Ok(report) if report.remapped > 0 => {
+                s.metadata = 1;
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "META {:?} -> {:?}: v{} remapped {} method tokens",
+                        input, dest, report.version, report.remapped
+                    ));
+                }
+                if quiet == 0 {
+                    println!(
+                        "✓ metadata v{} -> {:?} ({} method tokens remapped)",
+                        report.version, dest, report.remapped
+                    );
+                }
+            }
+            Ok(report) => {
+                if let Some(log) = &log {
+                    log.step(&format!(
+                        "META {:?}: v{} already de-obfuscated",
+                        input, report.version
+                    ));
+                }
+                if quiet == 0 {
+                    println!(
+                        "metadata v{}: already de-obfuscated, nothing to do",
+                        report.version
+                    );
+                }
+            }
+            Err(e) => {
+                s.errors = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("META ERR {:?}: {e:#}", input));
+                }
+                // Level 1 quiet: banner/summary/duration only (match folder mode).
+                if quiet == 0 {
+                    eprintln!("error: {e:#}");
+                }
+            }
+        }
+    } else if is_android_so {
+        match crate::android::restore_so_file(input, &dest, verbose && quiet == 0) {
+            Ok(embedded) => {
+                s.unpacked = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("OK {:?} -> {:?} (Android SO)", input, dest));
+                }
+                if quiet == 0 {
+                    println!("✓ So  {}  ->  {}", input.display(), dest.display());
+                }
+                match write_embedded_metadata(embedded, &dest) {
+                    Ok(Some(meta_dest)) => {
+                        s.metadata += 1;
+                        if let Some(log) = &log {
+                            log.step(&format!("META {:?} (embedded) -> {:?}", input, meta_dest));
+                        }
+                        if quiet == 0 {
+                            println!(
+                                "✓ metadata  {} (embedded)  ->  {}",
+                                input.display(),
+                                meta_dest.display()
+                            );
+                        }
+                    }
+                    Ok(None) => {}
+                    Err(e) => {
+                        s.errors += 1;
+                        if let Some(log) = &log {
+                            log.step(&format!("ERR {:?}: embedded metadata: {e:#}", input));
+                        }
+                        if quiet == 0 {
+                            eprintln!("error: embedded metadata: {e:#}");
+                        }
+                    }
+                }
+            }
+            Err(e) => {
+                s.errors = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {:?}: {e:#}", input));
+                }
+                if quiet == 0 {
+                    eprintln!("error: {e:#}");
+                }
+            }
+        }
+    } else if is_android_package {
+        s.packages = 1;
+        let rel = PathBuf::from(input.file_name().unwrap_or_default());
+        let mut seen = std::collections::HashSet::new();
+        match crate::android::restore_package(
+            input,
+            &rel,
+            &out_root,
+            &mut seen,
+            verbose && quiet == 0,
+        ) {
+            Ok(outcomes) => {
+                let bar = crate::ui::progress(0, true);
+                apply_package_outcomes(outcomes, &mut s, &bar, quiet >= 1, &log);
+            }
+            Err(e) => {
+                s.errors = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {:?}: {e:#}", input));
+                }
+                if quiet == 0 {
+                    eprintln!("error: {e:#}");
+                }
+            }
+        }
+    } else {
+        match unpack_one_v(input, &dest, verbose && quiet == 0) {
+            Ok((kind, report)) => {
+                s.unpacked = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("OK {:?} -> {:?} ({kind:?})", input, dest));
+                }
+                if quiet == 0 {
+                    println!("✓ {:?} -> {:?}", kind, dest);
+                }
+                if !report.ok() {
+                    s.suspect = 1;
+                    if let Some(log) = &log {
+                        log.step(&format!(
+                            "SUSPECT {:?}: {}",
+                            input,
+                            report.issues.join("; ")
+                        ));
+                    }
+                    if quiet == 0 {
+                        eprintln!(
+                            "! integrity check failed (likely to crash at runtime): {}",
+                            report.issues.join("; ")
+                        );
+                    }
+                }
+            }
+            Err(e) => {
+                s.errors = 1;
+                if let Some(log) = &log {
+                    log.step(&format!("ERR {:?}: {e:#}", input));
+                }
+                if quiet == 0 {
+                    eprintln!("error: {e:#}");
+                }
+            }
+        }
+    }
+
+    s.duration_ms = t0.elapsed().as_millis();
+    if let Some(log) = &log {
+        log.step(&format!("done in {} ms", s.duration_ms));
+        log.step(&format!("summary: {}", s.line()));
+    }
+    Ok(s)
+}
+
+/// Path of `p` relative to `root`, for mirroring into the output tree.
+///
+/// Falls back to just the file name when `p` is not under `root` (e.g. a
+/// `\\?\`-prefixed root against plain candidate paths): `Path::join` with an
+/// *absolute* path replaces the output root outright, which would write the
+/// output back over the source tree instead of under `--out`.
+fn rel_in_tree<'a>(root: &Path, p: &'a Path) -> std::borrow::Cow<'a, Path> {
+    match p.strip_prefix(root) {
+        Ok(rel) => std::borrow::Cow::Borrowed(rel),
+        Err(_) => std::borrow::Cow::Owned(PathBuf::from(p.file_name().unwrap_or_default())),
+    }
+}
+
+/// Insert `.unpack` before the last dot in the **file name**, preserving any
+/// parent directories. If the file name has no dot, append `.unpack`.
+///
+/// The dot search is scoped to the file-name component only: a relative path
+/// like `v1.2/launcher` (dotted directory, extension-less file) must become
+/// `v1.2/launcher.unpack`, not `v1.unpack.2/launcher`.
+pub fn out_name(input: &Path) -> PathBuf {
+    let file = input
+        .file_name()
+        .map(|s| s.to_string_lossy().into_owned())
+        .unwrap_or_default();
+    let renamed = match file.rfind('.') {
+        Some(i) => format!("{}.unpack{}", &file[..i], &file[i..]),
+        None => format!("{file}.unpack"),
+    };
+    match input.parent() {
+        Some(parent) if !parent.as_os_str().is_empty() => parent.join(renamed),
+        _ => PathBuf::from(renamed),
+    }
+}
+
+/// Extract a printable message from a caught panic payload.
+fn panic_payload(panic: &(dyn std::any::Any + Send)) -> String {
+    if let Some(s) = panic.downcast_ref::<&str>() {
+        (*s).to_string()
+    } else if let Some(s) = panic.downcast_ref::<String>() {
+        s.clone()
+    } else {
+        "<non-string payload>".to_string()
+    }
+}
+
+/// If `e`'s chain contains [`senbei_metadata::Error::UnsupportedVersion`],
+/// return the version. Used to apply the folder-mode "leave untouched, don't
+/// fail the run" policy to metadata versions this build can't de-obfuscate.
+fn unsupported_version(e: &anyhow::Error) -> Option<u32> {
+    for cause in e.chain() {
+        if let Some(senbei_metadata::Error::UnsupportedVersion(v)) =
+            cause.downcast_ref::<senbei_metadata::Error>()
+        {
+            return Some(*v);
+        }
+    }
+    None
+}
+
+/// De-obfuscate an il2cpp `global-metadata.dat` to `dest`.
+///
+/// Crackproof's `-GMD` option scrambles each `Il2CppMethodDefinition`'s token
+/// into a sparse, original-metadata-style value; il2cpp expects the contiguous
+/// per-module index it indexes its codegen tables with, so a statically-unpacked
+/// il2cpp game assembly reads garbage and crashes during init. This rewrites
+/// the tokens back to their canonical form (see [`senbei_metadata::deobfuscate`]).
+///
+/// The output is written only when something actually changed
+/// (`report.remapped > 0`); an already-clean metadata is left untouched and no
+/// redundant copy is produced. Returns the [`metadata::Report`] either way so
+/// the caller can report what happened.
+pub fn deobfuscate_metadata_to(
+    input: &Path,
+    dest: &Path,
+    verbose: bool,
+) -> anyhow::Result<senbei_metadata::Report> {
+    let data = std::fs::read(input)?;
+    // The Android seeded-permutation variant is tried first (it validates
+    // every restored RID); the structural remap is the fallback and the
+    // Windows path. The [`senbei_metadata::Error`] is preserved in the chain
+    // (rather than stringified) so the folder driver can apply its
+    // unsupported-version policy.
+    let (out, report) = crate::android::restore_metadata_bytes(&data)
+        .map_err(|e| e.context(format!("{input:?}")))?;
+    if report.remapped > 0 {
+        if let Some(parent) = dest.parent() {
+            std::fs::create_dir_all(parent)?;
+        }
+        write_atomic(dest, &out)?;
+        if verbose {
+            println!("Write to {}", dest.display());
+        }
+    }
+    Ok(report)
+}
+
+/// Write an embedded metadata blob (unwrapped from a restored Android
+/// library) next to the restored library. Returns the destination when a
+/// blob was written.
+fn write_embedded_metadata(
+    embedded: Option<Vec<u8>>,
+    so_dest: &Path,
+) -> anyhow::Result<Option<PathBuf>> {
+    let Some(blob) = embedded else {
+        return Ok(None);
+    };
+    let dest = crate::android::embedded_metadata_dest(so_dest);
+    if let Some(parent) = dest.parent() {
+        std::fs::create_dir_all(parent)?;
+    }
+    write_atomic(&dest, &blob)?;
+    Ok(Some(dest))
+}
+
+/// Fold one package's per-entry outcomes into the run summary, UI, and log.
+fn apply_package_outcomes(
+    outcomes: Vec<crate::android::EntryOutcome>,
+    s: &mut Summary,
+    bar: &indicatif::ProgressBar,
+    quiet: bool,
+    log: &Option<crate::logfile::Log>,
+) {
+    use crate::android::{EntryKind, EntryStatus};
+    for outcome in outcomes {
+        match outcome.status {
+            EntryStatus::Restored => {
+                match outcome.kind {
+                    EntryKind::So => {
+                        s.unpacked += 1;
+                        crate::ui::ok_label(bar, quiet, &outcome.label, "So", &outcome.dest);
+                    }
+                    EntryKind::Metadata { remapped } => {
+                        s.metadata += 1;
+                        crate::ui::metadata(
+                            bar,
+                            quiet,
+                            Path::new(&outcome.label),
+                            remapped,
+                            &outcome.dest,
+                        );
+                    }
+                    EntryKind::EmbeddedMetadata => {
+                        s.metadata += 1;
+                        crate::ui::ok_label(bar, quiet, &outcome.label, "metadata", &outcome.dest);
+                    }
+                }
+                if let Some(log) = log {
+                    log.step(&format!("OK {} -> {:?}", outcome.label, outcome.dest));
+                }
+            }
+            EntryStatus::Duplicate | EntryStatus::NotTarget | EntryStatus::Unchanged => {
+                s.skipped += 1;
+                if let Some(log) = log {
+                    log.step(&format!("SKIP {} ({:?})", outcome.label, outcome.kind));
+                }
+            }
+            EntryStatus::Failed(e) => {
+                s.errors += 1;
+                crate::ui::err(bar, quiet, Path::new(&outcome.label), &e);
+                if let Some(log) = log {
+                    log.step(&format!("ERR {}: {e:#}", outcome.label));
+                }
+            }
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    /// Review regression: when the candidate path is not under `root` (e.g. a
+    /// `\\?\`-prefixed root against plain walk paths), the output name must
+    /// fall back to the bare file name — joining the absolute path would
+    /// replace the output root and write back over the source tree.
+    #[test]
+    fn rel_in_tree_falls_back_to_file_name_outside_root() {
+        let root = Path::new(r"D:\out-of-tree-root");
+        let abs = Path::new(r"C:\game\bin\app.exe");
+        let rel = rel_in_tree(root, abs);
+        assert_eq!(rel.as_ref(), Path::new("app.exe"));
+
+        // And the normal case still preserves the tree structure.
+        let under = Path::new(r"D:\out-of-tree-root\bin\app.exe");
+        let rel = rel_in_tree(root, under);
+        assert_eq!(rel.as_ref(), Path::new(r"bin\app.exe"));
+    }
+}
@@ -0,0 +1,13 @@
+//! Filesystem and command-line orchestration.
+
+/// File name of an IL2CPP metadata blob shared by both platform scanners.
+pub const METADATA_FILE_NAME: &str = "global-metadata.dat";
+
+pub mod android;
+mod atomic;
+pub mod job;
+pub mod logfile;
+pub mod pause;
+pub mod scan;
+pub mod ui;
+pub mod windows;
@@ -1,4 +1,4 @@
-use crate::unpacker::detect;
+use senbei_engine::detect;
 use std::io::Read;
 use std::path::{Path, PathBuf};
 use walkdir::WalkDir;
@@ -16,94 +16,19 @@ const DETECT_PREFIX: u64 = 8 * 1024;
 /// Smallest file that can possibly be a target, so anything shorter is skipped
 /// without ever being opened.
 ///
-/// A Crackproof module needs ≥ 4128 bytes for [`crate::unpacker::detect`]'s key
+/// A Crackproof module needs ≥ 4128 bytes for [`senbei_engine::detect`]'s key
 /// table (it reads the dword at 4124), so the bound is exact for the unpack
 /// path. An il2cpp `global-metadata.dat` only needs 4 bytes to match its magic,
 /// but its header alone runs to offset 0xB0 and the images/types/methods tables
 /// it indexes make every real one megabytes long — a sub-4 KiB "metadata" could
-/// only ever fail [`crate::metadata::deobfuscate`] with `Malformed`, so nothing
+/// only ever fail [`senbei_metadata::deobfuscate`] with `Malformed`, so nothing
 /// processable is lost.
 const MIN_SIZE: u64 = 4128;

-/// File extensions that are bulk data by construction and can never be a PE
-/// image or an il2cpp metadata blob.
-///
-/// This is deliberately a **deny**-list, not an allow-list: the default is to
-/// probe, so anything unrecognised is still opened. Targets are recognised by
-/// content, not extension, and can carry arbitrary names — there is no closed
-/// set of target extensions an allow-list of `exe`/`dll` could enumerate.
-/// Only extensions that are bulk asset or text formats by construction appear
-/// here.
-///
-/// Set `SENBEI_SCAN_ALL=1` (or pass `--scan-all`) to probe every file regardless.
-const DENY_EXT: &[&str] = &[
-    // Unity and other engine asset containers
-    "ab",
-    "bundle",
-    "unity3d",
-    "manifest",
-    "resource",
-    "ress",
-    "assets",
-    "sharedassets",
-    // audio / video / image / font
-    "acb",
-    "awb",
-    "usm",
-    "wav",
-    "ogg",
-    "mp3",
-    "mp4",
-    "avi",
-    "png",
-    "jpg",
-    "jpeg",
-    "bmp",
-    "gif",
-    "tga",
-    "dds",
-    "svg",
-    "ttf",
-    "otf",
-    // text, markup, config, logs
-    "xml",
-    "json",
-    "txt",
-    "csv",
-    "md",
-    "toml",
-    "ini",
-    "yml",
-    "yaml",
-    "log",
-    "html",
-    "htm",
-    "css",
-    "aspx",
-    "browser",
-    "config",
-    "sig",
-    "map",
-    "pdb",
-    // rhythm-game chart/score data
-    "ma2",
-    "sr",
-];
-
-/// Whether `path`'s extension is on [`DENY_EXT`]. Extensionless files are never
-/// denied (they could be anything).
-fn denied_ext(path: &Path) -> bool {
-    let Some(ext) = path.extension() else {
-        return false;
-    };
-    let Some(ext) = ext.to_str() else {
-        return false;
-    };
-    // Extensions are ASCII in practice; compare case-insensitively without
-    // allocating for the overwhelmingly common non-match.
-    DENY_EXT
-        .iter()
-        .any(|d| d.len() == ext.len() && d.eq_ignore_ascii_case(ext))
+fn is_metadata_name(path: &Path) -> bool {
+    path.file_name()
+        .and_then(|name| name.to_str())
+        .is_some_and(|name| name.eq_ignore_ascii_case(crate::METADATA_FILE_NAME))
 }

 /// Content classification of a single file.
@@ -115,6 +40,25 @@ enum Class {
    Crackproof,
    /// An il2cpp `global-metadata.dat` (de-obfuscation target).
    Metadata,
+    /// A protected AArch64 shared library (Android restore target).
+    AndroidSo,
+    /// An Android app package (`.apk`/`.apks`/`.xapk`) — a container whose
+    /// entries are content-probed individually during the Android pass.
+    AndroidPackage,
+}
+
+/// Everything one [`find_targets_opts`] walk found, plus non-target tallies.
+#[derive(Default)]
+pub struct ScanResult {
+    /// Crackproof-protected PE files.
+    pub crackproof: Vec<PathBuf>,
+    /// il2cpp `global-metadata.dat` blobs.
+    pub metadata: Vec<PathBuf>,
+    /// Protected AArch64 shared libraries.
+    pub android_so: Vec<PathBuf>,
+    /// Android app packages (containers restored entry-by-entry).
+    pub android_packages: Vec<PathBuf>,
+    pub stats: ScanStats,
 }

 /// Walk `root` recursively (skipping any directory literally named `"unpack"`)
@@ -132,21 +76,19 @@ enum Class {
 /// per-file I/O latency, not bandwidth (that tree lives on a user-mode virtual
 /// disk that tops out near 1,300 IOPS). Thread count barely moves it either.
 ///
-/// So the only lever is **probing fewer files**, which is what [`MIN_SIZE`] and
-/// [`DENY_EXT`] do — both decided from the free directory metadata, before any
-/// file is opened. On that tree they cut 46,446 probes to 1,814 and the scan
-/// from ~40 s to ~2 s while still finding every target.
+/// So the only lever is **probing fewer files**, which is what the target-name
+/// filter and [`MIN_SIZE`] do — both decided before any file is opened.
 ///
-/// The surviving probes (open + short read + magic test) are fanned out across
-/// worker threads. Directory traversal itself stays serial (one cheap `readdir`
-/// pass, no file opens) because it feeds the parallel probe.
+/// The selected probes (open + short read + magic test) are fanned out across
+/// worker threads. Directory traversal itself stays serial because it only
+/// collects names and sizes before the parallel probe.
 ///
-/// Thread count follows [`crate::unpacker::parallel::thread_cap`] (honoring
+/// Thread count follows [`senbei_engine::thread_cap`] (honoring
 /// `SENBEI_THREADS`, `1` = fully sequential). Output order is independent of
 /// thread count: each worker owns a disjoint contiguous slice of the path list
 /// and writes the matching disjoint slice of the class list, so results are
 /// deterministic.
-pub fn find_targets(root: &Path) -> (Vec<PathBuf>, Vec<PathBuf>, ScanStats) {
+pub fn find_targets(root: &Path) -> ScanResult {
    find_targets_opts(root, scan_all_env())
 }

@@ -167,9 +109,8 @@ pub struct ScanStats {
 }

 /// [`find_targets`], but with the pre-filter explicitly controlled. When
-/// `scan_all` is true every regular file is probed, restoring the exhaustive
-/// (and on asset-heavy trees, far slower) behavior.
-pub fn find_targets_opts(root: &Path, scan_all: bool) -> (Vec<PathBuf>, Vec<PathBuf>, ScanStats) {
+/// `scan_all` is true selected target names below [`MIN_SIZE`] are also probed.
+pub fn find_targets_opts(root: &Path, scan_all: bool) -> ScanResult {
    // Phase 1: serial traversal collecting regular-file paths only. No file is
    // opened here; `readdir` is fast relative to the content probe that follows,
    // and `entry.metadata()` is served from the directory entry on Windows, so
@@ -193,7 +134,7 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> (Vec<PathBuf>, Vec<Path
        // Skip reparse-point directories (junctions, symlink-dirs): they point
        // outside the scanned tree — walking one would silently unpack an
        // entire foreign tree (e.g. a `samples` junction into the golden corpus).
-        !is_reparse_point(e)
+        !crate::windows::is_reparse_point(e)
    }) {
        let entry = match entry {
            Ok(e) => e,
@@ -205,13 +146,23 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> (Vec<PathBuf>, Vec<Path
        if !entry.file_type().is_file() {
            continue;
        }
+        if crate::windows::is_companion(entry.path()) {
+            continue;
+        }
+        if !is_metadata_name(entry.path())
+            && !crate::windows::is_pe_extension(entry.path())
+            && !crate::android::is_so_name(entry.path())
+            && !crate::android::is_package_name(entry.path())
+        {
+            continue;
+        }
        if !scan_all {
            // Skip on directory metadata alone — never open these.
            let too_small = entry
                .metadata()
                .map(|m| m.len() < MIN_SIZE)
                .unwrap_or(false);
-            if too_small || denied_ext(entry.path()) {
+            if too_small {
                continue;
            }
        }
@@ -223,7 +174,7 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> (Vec<PathBuf>, Vec<Path
    // `Some(Class::None)` means "probed, matched neither detector".
    let n = paths.len();
    let mut class: Vec<Option<Class>> = vec![Some(Class::None); n];
-    let workers = crate::unpacker::parallel::thread_cap().clamp(1, n.max(1));
+    let workers = senbei_engine::thread_cap().clamp(1, n.max(1));
    if workers <= 1 {
        for (p, c) in paths.iter().zip(class.iter_mut()) {
            *c = classify(p);
@@ -241,44 +192,28 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> (Vec<PathBuf>, Vec<Path
        });
    }

-    let mut candidates = Vec::new();
-    let mut metadata = Vec::new();
+    let mut result = ScanResult {
+        stats,
+        ..ScanResult::default()
+    };
    for (p, c) in paths.into_iter().zip(class) {
        match c {
-            Some(Class::Crackproof) => candidates.push(p),
-            Some(Class::Metadata) => metadata.push(p),
-            Some(Class::None) => stats.skipped += 1,
+            Some(Class::Crackproof) => result.crackproof.push(p),
+            Some(Class::Metadata) => result.metadata.push(p),
+            Some(Class::AndroidSo) => result.android_so.push(p),
+            Some(Class::AndroidPackage) => result.android_packages.push(p),
+            Some(Class::None) => result.stats.skipped += 1,
            // Unreadable / panicking probe: NOT skipped — the scan could not
            // classify it, so it may be a target we failed to unpack.
-            None => stats.probe_errors += 1,
+            None => result.stats.probe_errors += 1,
        }
    }
-    (candidates, metadata, stats)
+    result
 }

-/// True if a walked directory entry is a reparse point (junction or symlink).
-///
-/// `DirEntry::file_type` only flags true symlinks; NTFS junctions report as
-/// ordinary directories, so without this check the walker descends into them.
-/// Off-Windows there are no junctions — symlink dirs are already excluded
-/// because `follow_links` is off (their `file_type().is_dir()` is false).
-#[cfg(windows)]
-fn is_reparse_point(e: &walkdir::DirEntry) -> bool {
-    use std::os::windows::fs::MetadataExt;
-    const FILE_ATTRIBUTE_REPARSE_POINT: u32 = 0x400;
-    e.metadata()
-        .map(|m| m.file_attributes() & FILE_ATTRIBUTE_REPARSE_POINT != 0)
-        .unwrap_or(false)
-}
-
-#[cfg(not(windows))]
-fn is_reparse_point(_e: &walkdir::DirEntry) -> bool {
-    false
-}
-
-/// Whether the scan pre-filter is disabled via `SENBEI_SCAN_ALL`. Any value
-/// other than `0`/empty turns exhaustive scanning on. The `--scan-all` flag is
-/// ORed with this.
+/// Whether the size pre-filter is disabled via `SENBEI_SCAN_ALL`. Any value
+/// other than `0`/empty enables probing small selected target names. It never
+/// expands the platform filename boundary.
 pub fn scan_all_env() -> bool {
    match std::env::var("SENBEI_SCAN_ALL") {
        Ok(v) => !matches!(v.trim(), "" | "0"),
@@ -286,11 +221,11 @@ pub fn scan_all_env() -> bool {
    }
 }

-/// Classify one file by content. Reads a short prefix once and tests the
-/// Crackproof detector first, then the il2cpp metadata magic. Returns `None`
-/// when the file could not be classified at all — an I/O error opening it
-/// (locked, permissions) or a panic inside a detector — so the caller counts
-/// it as a probe error rather than a clean "not a target" skip.
+/// Classify one named candidate by content. Reads a short prefix once and tests
+/// the detector for that platform. Returns `None` when the file could not be classified at
+/// all — an I/O error opening it (locked, permissions) or a panic inside a
+/// detector — so the caller counts it as a probe error rather than a clean
+/// "not a target" skip.
 ///
 /// The detector is wrapped in `catch_unwind` because a panic in a scan worker
 /// thread would otherwise abort the whole folder run (a scoped-thread panic
@@ -299,16 +234,30 @@ pub fn scan_all_env() -> bool {
 ///
 /// A Crackproof PE never matches the metadata magic (it is a PE, not a
 /// metadata blob) and vice versa, so the order is immaterial.
+///
+/// The Android library probe needs more than the prefix: the protection
+/// payload lives in a section found via the section-header table at the *end*
+/// of the file, so an ELF64/AArch64 prefix triggers a full-file read. Only
+/// selected `.so` images pay for it.
 fn classify(path: &Path) -> Option<Class> {
    let head = read_prefix(path, DETECT_PREFIX)?;
    let r = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
-        if detect(&head).is_some() {
-            Class::Crackproof
-        } else if crate::metadata::is_metadata(&head) {
-            Class::Metadata
-        } else {
-            Class::None
+        if crate::android::is_package_name(path) && crate::android::is_app_package(path, &head) {
+            return Class::AndroidPackage;
        }
+        if is_metadata_name(path) && senbei_metadata::is_metadata(&head) {
+            return Class::Metadata;
+        }
+        if crate::windows::is_pe_extension(path) && detect(&head).is_some() {
+            return Class::Crackproof;
+        }
+        if crate::android::is_so_name(path)
+            && crate::android::is_elf64_aarch64(&head)
+            && crate::android::is_protected_so_file(path)
+        {
+            return Class::AndroidSo;
+        }
+        Class::None
    }));
    r.ok()
 }
@@ -327,35 +276,56 @@ mod tests {
    use super::*;

    #[test]
-    fn denies_bulk_asset_extensions_case_insensitively() {
-        for p in ["a.ab", "a.XML", "a.Acb", "a.ma2", "a.manifest", "a.PNG"] {
-            assert!(denied_ext(Path::new(p)), "{p} should be denied");
+    fn candidate_names_are_platform_specific() {
+        for p in [
+            "daemon.exe",
+            "GameLib.DLL",
+            "libil2cpp.so",
+            "global-metadata.dat",
+        ] {
+            assert!(
+                is_metadata_name(Path::new(p))
+                    || crate::windows::is_pe_extension(Path::new(p))
+                    || crate::android::is_so_name(Path::new(p)),
+                "{p} should be a candidate"
+            );
        }
-    }
-
-    #[test]
-    fn never_denies_what_a_target_can_be_named() {
-        // Targets are recognised by content, not name — a protected module
-        // can carry any extension, or none — so names like these must always
-        // be probed. An allow-list would have skipped them.
        for p in [
            "app.exe.bak",
            "managed.dll.bak",
-            "daemon.exe",
-            "GameLib.dll",
-            "global-metadata.dat",
-            "noextension",
-            "a.so",
-            "a.bin",
+            "libil2cpp.so.bak",
+            "global-metadata.bin",
+            "asset",
+            "a.ab",
        ] {
-            assert!(!denied_ext(Path::new(p)), "{p} must still be probed");
+            assert!(
+                !is_metadata_name(Path::new(p))
+                    && !crate::windows::is_pe_extension(Path::new(p))
+                    && !crate::android::is_so_name(Path::new(p)),
+                "{p} must not be a candidate"
+            );
        }
    }

-    /// A file below the Crackproof key-table bound is skipped without being
-    /// opened, but a large non-asset file is still probed.
    #[test]
-    fn prefilter_skips_small_and_denied_files_only() {
+    fn extensionless_targets_are_not_candidates() {
+        let td = tempfile::tempdir().unwrap();
+        let root = td.path();
+        let mut blob = vec![0u8; MIN_SIZE as usize + 1];
+        blob[..4].copy_from_slice(&0xFAB1_1BAFu32.to_le_bytes());
+        std::fs::write(root.join("metadata"), &blob).unwrap();
+
+        let filtered = find_targets_opts(root, false);
+        assert!(filtered.metadata.is_empty());
+
+        let exhaustive = find_targets_opts(root, true);
+        assert!(exhaustive.metadata.is_empty());
+    }
+
+    /// A selected file below the Crackproof key-table bound is skipped without
+    /// being opened, while `scan_all` probes it.
+    #[test]
+    fn prefilter_skips_small_selected_files_only() {
        let td = tempfile::tempdir().unwrap();
        let root = td.path();
        std::fs::write(root.join("tiny.dll"), vec![0u8; 100]).unwrap();
@@ -363,15 +333,15 @@ mod tests {
        std::fs::write(root.join("plain.dll"), vec![0u8; 100_000]).unwrap();

        // None of them are Crackproof, so both modes find nothing; the point is
-        // that the filtered walk does not panic and honors `scan_all`.
-        let (c, m, _) = find_targets_opts(root, false);
-        assert!(c.is_empty() && m.is_empty());
-        let (c, m, _) = find_targets_opts(root, true);
-        assert!(c.is_empty() && m.is_empty());
+        // that only the selected names are considered and `scan_all` controls
+        // the size floor.
+        let scan = find_targets_opts(root, false);
+        assert!(scan.crackproof.is_empty() && scan.metadata.is_empty());
+        let scan = find_targets_opts(root, true);
+        assert!(scan.crackproof.is_empty() && scan.metadata.is_empty());
    }

-    /// An il2cpp metadata blob is found by the filtered scan: `.dat` is not on
-    /// the deny-list and a real one is far above `MIN_SIZE`.
+    /// An exact `global-metadata.dat` name is found by the filtered scan.
    #[test]
    fn finds_metadata_through_the_prefilter() {
        let td = tempfile::tempdir().unwrap();
@@ -382,9 +352,9 @@ mod tests {
        // Same magic but too small to be processable — skipped by the size floor.
        std::fs::write(root.join("stub.dat"), &blob[..64]).unwrap();

-        let (_, m, _) = find_targets_opts(root, false);
-        assert_eq!(m.len(), 1);
-        assert!(m[0].ends_with("global-metadata.dat"));
+        let scan = find_targets_opts(root, false);
+        assert_eq!(scan.metadata.len(), 1);
+        assert!(scan.metadata[0].ends_with("global-metadata.dat"));
    }

    /// Review regression: a previous output tree is pruned case-insensitively
@@ -403,12 +373,41 @@ mod tests {
        // A big non-target file at the root: probed, then skipped.
        std::fs::write(root.join("plain.dll"), vec![0u8; 100_000]).unwrap();

-        let (c, m, stats) = find_targets_opts(root, false);
+        let scan = find_targets_opts(root, false);
        assert!(
-            c.is_empty() && m.is_empty(),
+            scan.crackproof.is_empty() && scan.metadata.is_empty(),
            "old output tree must be pruned"
        );
-        assert_eq!(stats.skipped, 1, "the probed non-target counts as skipped");
-        assert_eq!(stats.walk_errors, 0);
+        assert_eq!(
+            scan.stats.skipped, 1,
+            "the probed non-target counts as skipped"
+        );
+        assert_eq!(scan.stats.walk_errors, 0);
+    }
+
+    #[test]
+    fn companion_payload_is_not_counted_as_skipped() {
+        let td = tempfile::tempdir().unwrap();
+        let root = td.path();
+        std::fs::write(root.join("app.exe"), vec![0u8; MIN_SIZE as usize]).unwrap();
+        std::fs::write(root.join("app.exe._"), vec![0u8; MIN_SIZE as usize]).unwrap();
+
+        let scan = find_targets_opts(root, false);
+        assert_eq!(scan.stats.skipped, 1, "only the stub was probed");
+        assert!(crate::windows::is_companion(&root.join("app.exe._")));
+    }
+
+    #[test]
+    fn scan_all_keeps_the_platform_name_boundary() {
+        let td = tempfile::tempdir().unwrap();
+        let root = td.path();
+        let mut metadata = vec![0_u8; MIN_SIZE as usize];
+        metadata[..4].copy_from_slice(&0xFAB1_1BAFu32.to_le_bytes());
+        std::fs::write(root.join("renamed.bin"), &metadata).unwrap();
+        std::fs::write(root.join("global-metadata.dat"), &metadata).unwrap();
+
+        let scan = find_targets_opts(root, true);
+        assert_eq!(scan.metadata.len(), 1);
+        assert!(scan.metadata[0].ends_with("global-metadata.dat"));
    }
 }
@@ -1,6 +1,6 @@
-use crate::unpacker::{IntegrityReport, Kind};
 use indicatif::{ProgressBar, ProgressStyle};
 use owo_colors::OwoColorize;
+use senbei_engine::{IntegrityReport, Kind};
 use std::path::Path;

 /// Create a progress bar for `n` items. Hidden when `quiet` is true.
@@ -19,16 +19,22 @@ pub fn progress(n: u64, quiet: bool) -> ProgressBar {

 /// Print a green success line, suspending the progress bar.
 pub fn ok(bar: &ProgressBar, quiet: bool, rel: &Path, kind: Kind, dest: &Path) {
+    ok_label(
+        bar,
+        quiet,
+        &rel.display().to_string(),
+        &format!("{kind:?}"),
+        dest,
+    );
+}
+
+/// Print a green success line with a free-form kind label (Android targets),
+/// suspending the progress bar.
+pub fn ok_label(bar: &ProgressBar, quiet: bool, rel: &str, label: &str, dest: &Path) {
    if quiet {
        return;
    }
-    let msg = format!(
-        "{} {:?}  {}  ->  {}",
-        "✓".green(),
-        kind,
-        rel.display(),
-        dest.display()
-    );
+    let msg = format!("{} {}  {}  ->  {}", "✓".green(), label, rel, dest.display());
    bar.suspend(|| println!("{msg}"));
 }

@@ -0,0 +1,677 @@
+//! Windows filesystem adapter for PE companion payloads and byte APIs.
+
+use senbei_engine as unpacker;
+use std::path::Path;
+
+use crate::atomic::write_atomic;
+
+pub(crate) fn is_pe_extension(path: &Path) -> bool {
+    path.extension()
+        .and_then(|ext| ext.to_str())
+        .is_some_and(|ext| ext.eq_ignore_ascii_case("exe") || ext.eq_ignore_ascii_case("dll"))
+}
+
+pub(crate) fn is_companion(path: &Path) -> bool {
+    let Some(name) = path.file_name().and_then(|name| name.to_str()) else {
+        return false;
+    };
+    let Some(stub_name) = name.strip_suffix("._") else {
+        return false;
+    };
+    is_pe_extension(Path::new(stub_name))
+}
+
+/// Return whether a directory entry is an NTFS reparse point. The scanner keeps
+/// this host-specific check in the Windows adapter while the traversal itself
+/// remains platform-neutral.
+#[cfg(windows)]
+pub(crate) fn is_reparse_point(entry: &walkdir::DirEntry) -> bool {
+    use std::os::windows::fs::MetadataExt;
+
+    const FILE_ATTRIBUTE_REPARSE_POINT: u32 = 0x400;
+    entry
+        .metadata()
+        .map(|metadata| metadata.file_attributes() & FILE_ATTRIBUTE_REPARSE_POINT != 0)
+        .unwrap_or(false)
+}
+
+#[cfg(not(windows))]
+pub(crate) fn is_reparse_point(_entry: &walkdir::DirEntry) -> bool {
+    false
+}
+
+/// Crackproof header key table lives at this fixed file offset. For the
+/// external-companion layout, the companion payload aligns to the stub here.
+const HEADER_OFF: usize = 4096;
+
+/// Build the unpacker input for `input`, transparently handling the
+/// **external-companion** layout used by some il2cpp games.
+///
+/// In that layout a protected module is split into a thin on-disk loader stub
+/// (`Foo.dll`, whose code sections are stripped to one page) plus an encrypted
+/// `Foo.dll._` companion holding the real payload. The companion is byte-for-byte
+/// the stub's payload region starting at the Crackproof header (offset 4096), so
+/// `stub[..4096] ++ companion` reconstructs the ordinary embedded-payload file
+/// the existing pipelines already unpack. The runtime loader does exactly this:
+/// it maps `Foo.dll._` and feeds it through the standard Crackproof unpack.
+///
+/// The splice fires only when a sibling `<input>._` exists *and* its first 32
+/// bytes equal the stub's header at offset 4096 — a precise signal that the
+/// companion is this stub's payload. Otherwise the file is returned untouched,
+/// so normal (embedded-payload) inputs are unaffected.
+pub(crate) fn read_unpacker_input(input: &Path) -> std::io::Result<UnpackerInput> {
+    let stub = std::fs::read(input)?;
+
+    // Companion path: append "._" to the full file name (Foo.dll -> Foo.dll._).
+    let companion = match input.file_name() {
+        Some(name) => {
+            let mut n = name.to_os_string();
+            n.push("._");
+            input.with_file_name(n)
+        }
+        None => {
+            return Ok(UnpackerInput {
+                bytes: stub,
+                stub: None,
+            });
+        }
+    };
+    if !companion.is_file() {
+        return Ok(UnpackerInput {
+            bytes: stub,
+            stub: None,
+        });
+    }
+    let comp = std::fs::read(&companion)?;
+    match splice_companion(&stub, &comp) {
+        // A splice fired: keep the stub so its plaintext export table can be
+        // overlaid onto the unpacked image (the companion does not carry it).
+        Some(spliced) => Ok(UnpackerInput {
+            bytes: spliced,
+            stub: Some(stub),
+        }),
+        None => Ok(UnpackerInput {
+            bytes: stub,
+            stub: None,
+        }),
+    }
+}
+
+/// The bytes fed to the unpacker, plus the original loader stub when the input
+/// was reconstructed from an external companion. The stub is retained because
+/// the crackproof loader rebuilds the PE export table at runtime from data kept
+/// in the stub — that table is *not* present in the encrypted companion, so the
+/// unpacked image needs it overlaid from the stub afterwards
+/// (see [`overlay_exports_from_stub`]).
+pub(crate) struct UnpackerInput {
+    pub(crate) bytes: Vec<u8>,
+    pub(crate) stub: Option<Vec<u8>>,
+}
+
+/// Overlay the PE export table from the loader `stub` onto the unpacked image
+/// `out`, for the external-companion layout.
+///
+/// In that layout the encrypted companion carries the real `.text`/`il2cpp`
+/// payload but **not** a usable export directory: the crackproof loader rebuilds
+/// exports at runtime from the plaintext copy retained in the stub's `.rdata`.
+/// Statically, the spliced input therefore decrypts to a garbage export
+/// directory (`NumberOfFunctions` etc. are ciphertext), which makes downstream
+/// tools (IL2CppDumper, IDA) choke when they parse it. The fix does what the
+/// loader does: copy the export-directory region byte-for-byte from the stub to
+/// the same RVA in the unpacked image.
+///
+/// No-op (leaves `out` untouched) if there is no export directory, or if the
+/// region cannot be mapped in either image — so a malformed stub can never
+/// corrupt an otherwise-good unpack.
+pub(crate) fn overlay_exports_from_stub(out: &mut [u8], stub: &[u8]) {
+    let (export_rva, export_size) = match pe_export_dir(out) {
+        Some(v) if v.1 != 0 => v,
+        _ => return,
+    };
+    let dst = match rva_to_file_off(out, export_rva) {
+        Some(o) => o,
+        None => return,
+    };
+    let src = match rva_to_file_off(stub, export_rva) {
+        Some(o) => o,
+        None => return,
+    };
+    let n = export_size as usize;
+    if dst + n <= out.len() && src + n <= stub.len() {
+        out[dst..dst + n].copy_from_slice(&stub[src..src + n]);
+    }
+}
+
+/// Restore the CLR regions retained by an external-companion loader stub.
+/// Method bodies come from the unpacked payload and must not be overlaid.
+fn restore_managed_from_stub(out: &mut [u8], stub: &[u8]) -> Result<(), unpacker::UnpackError> {
+    let failure =
+        |region, source| unpacker::UnpackError::ManagedStubRestoreFailed { region, source };
+    let source_headers = senbei_pe::parse(stub).map_err(|e| failure("PE headers", e))?;
+    let (clr_rva, clr_size) = senbei_pe::data_directory(stub, source_headers, 14)
+        .map_err(|e| failure("CLR directory", e))?;
+    if clr_rva == 0 && clr_size == 0 {
+        return Ok(());
+    }
+    if clr_rva == 0 || clr_size < 0x48 {
+        return Err(failure("CLR directory", senbei_pe::Error::Invalid));
+    }
+    let destination_headers = senbei_pe::parse(out).map_err(|e| failure("output PE headers", e))?;
+    senbei_pe::data_directory(out, destination_headers, 14)
+        .map_err(|e| failure("output CLR directory", e))?;
+    let cor = senbei_pe::rva_range(stub, source_headers, clr_rva, 0x48)
+        .map_err(|e| failure("COR20 header", e))?;
+    if read_u32(stub, cor.start) != Some(0x48) {
+        return Err(failure("COR20 header", senbei_pe::Error::Invalid));
+    }
+    let range_pair = |rva, size, region| {
+        let source = senbei_pe::rva_range(stub, source_headers, rva, size)
+            .map_err(|e| failure(region, e))?;
+        let destination = senbei_pe::rva_range(out, destination_headers, rva, size)
+            .map_err(|e| failure(region, e))?;
+        Ok::<_, unpacker::UnpackError>((source, destination))
+    };
+    let mut copies = vec![range_pair(clr_rva, 0x48, "COR20 header")?];
+    for (field, region) in [
+        (0x08, "metadata"),
+        (0x18, "resources"),
+        (0x20, "strong-name signature"),
+        (0x28, "code-manager table"),
+        (0x30, "vtable fixups"),
+        (0x38, "export address jumps"),
+        (0x40, "managed native header"),
+    ] {
+        let rva = read_u32(stub, cor.start + field)
+            .ok_or_else(|| failure(region, senbei_pe::Error::OutOfBounds))?;
+        let size = read_u32(stub, cor.start + field + 4)
+            .ok_or_else(|| failure(region, senbei_pe::Error::OutOfBounds))?;
+        if field != 0x08 && rva == 0 && size == 0 {
+            continue;
+        }
+        if rva == 0 || size == 0 {
+            return Err(failure(region, senbei_pe::Error::Invalid));
+        }
+        let (source, destination) = range_pair(rva, size, region)?;
+        if field == 0x08 && !stub[source.clone()].starts_with(b"BSJB") {
+            return Err(failure(region, senbei_pe::Error::Invalid));
+        }
+        if field == 0x30 {
+            if !size.is_multiple_of(8) {
+                return Err(failure(region, senbei_pe::Error::Invalid));
+            }
+            for fixup in stub[source.clone()].as_chunks::<8>().0 {
+                let slots_rva =
+                    u32::from_le_bytes(fixup[..4].try_into().expect("eight-byte fixup"));
+                let count = u16::from_le_bytes([fixup[4], fixup[5]]) as u32;
+                let flags = u16::from_le_bytes([fixup[6], fixup[7]]);
+                let width = match flags & 3 {
+                    1 => 4,
+                    2 => 8,
+                    _ => return Err(failure(region, senbei_pe::Error::Invalid)),
+                };
+                if count != 0 {
+                    copies.push(range_pair(slots_rva, count * width, "vtable slots")?);
+                }
+            }
+        }
+        copies.push((source, destination));
+    }
+    // Validate all referenced ranges before changing the output.
+    for (source, destination) in copies {
+        out[destination].copy_from_slice(&stub[source]);
+    }
+    let directory = destination_headers.pe_offset
+        + 24
+        + if destination_headers.is_pe32_plus {
+            112
+        } else {
+            96
+        }
+        + 14 * 8;
+    out[directory..directory + 4].copy_from_slice(&clr_rva.to_le_bytes());
+    out[directory + 4..directory + 8].copy_from_slice(&clr_size.to_le_bytes());
+    Ok(())
+}
+
+/// Restore the TLS directory from the loader `stub` onto the unpacked image
+/// `out`, for the external-companion layout.
+///
+/// Crackproof strips the whole `IMAGE_TLS_DIRECTORY` from the encrypted payload
+/// — the data-directory entry, the directory struct, the raw-data template, and
+/// the base relocations for the struct's four 64-bit pointer fields — and
+/// re-installs TLS itself from data kept in the stub when it loads the module.
+/// A statically-unpacked DLL is loaded by the ordinary Windows loader instead,
+/// which needs a valid TLS directory or it never allocates a TLS slot for the
+/// module nor writes `_tls_index`. The module's C++ `thread_local` accesses then
+/// read a garbage TLS slot — observed as a `0xC0000005` deep in IL2CPP type
+/// resolution (a TypeDef token used as a raw `s_TypeInfoTable` index).
+///
+/// The stub retains the full plaintext `.rdata` (only `.text`/`il2cpp` are
+/// stripped to one page), so the directory struct and its raw-data template are
+/// copied back byte-for-byte at their RVAs, the data-directory entry is taken
+/// from the stub header (the unpacked image's was overwritten with the zeroed
+/// saved-header blob), and four DIR64 relocations are appended to `.reloc`.
+///
+/// No-op if the stub declares no TLS directory or if any required region cannot
+/// be mapped/relocated — so it can never corrupt an otherwise-good unpack.
+pub(crate) fn restore_tls_from_stub(out: &mut [u8], stub: &[u8]) {
+    let pe = match read_u32(out, 0x3C) {
+        Some(v) => v as usize,
+        None => return,
+    };
+    if out.get(pe..pe + 4) != Some(&b"PE\0\0"[..]) {
+        return;
+    }
+    // This restore is PE32+-only: it copies a 40-byte IMAGE_TLS_DIRECTORY64,
+    // converts fields with a 64-bit image base, and appends DIR64 relocs. A
+    // PE32 module needs the 24-byte struct / DIR32 handling (the unpacker core
+    // does that itself — see `restore_pe32_tls_from_stub`), so bail rather than
+    // read the data directories at the wrong (PE32+) offset and write garbage.
+    if read_u16(out, pe + 24) != Some(0x20B) {
+        return;
+    }
+    // TLS is data-directory index 9 (PE32+ directories at optional header +112).
+    let tls_dd = match pe.checked_add(24 + 112 + 9 * 8) {
+        Some(v) => v,
+        None => return,
+    };
+    // The genuine entry survives in the stub header; the unpacked image's copy
+    // was clobbered by the (zeroed-TLS) saved-header blob.
+    let (tls_rva, tls_size) = match (read_u32(stub, tls_dd), read_u32(stub, tls_dd + 4)) {
+        (Some(r), Some(s)) if r != 0 && s != 0 => (r, s),
+        _ => return, // module has no TLS — nothing to restore
+    };
+    // Image base (PE32+, optional header +24) converts the struct's absolute VAs
+    // back to RVAs for the raw-data template overlay.
+    let image_base = match read_u64(out, pe + 24 + 24) {
+        Some(v) => v,
+        None => return,
+    };
+
+    // 1) Overlay the IMAGE_TLS_DIRECTORY struct from the stub at its RVA.
+    let dst = match rva_to_file_off(out, tls_rva) {
+        Some(o) => o,
+        None => return,
+    };
+    let src = match rva_to_file_off(stub, tls_rva) {
+        Some(o) => o,
+        None => return,
+    };
+    let n = tls_size as usize;
+    if dst.checked_add(n).is_none_or(|e| e > out.len())
+        || src.checked_add(n).is_none_or(|e| e > stub.len())
+    {
+        return;
+    }
+    out[dst..dst + n].copy_from_slice(&stub[src..src + n]);
+
+    // 2) Restore the data-directory entry so the loader processes TLS at all.
+    write_u32_at(out, tls_dd, tls_rva);
+    write_u32_at(out, tls_dd + 4, tls_size);
+
+    // 3) Overlay the raw-data template [StartAddressOfRawData, EndAddressOfRawData).
+    if let (Some(start_va), Some(end_va)) = (read_u64(out, dst), read_u64(out, dst + 8))
+        && end_va > start_va
+        && start_va >= image_base
+    {
+        let tpl_rva = (start_va - image_base) as u32;
+        let tpl_len = (end_va - start_va) as usize;
+        if let (Some(td), Some(ts)) = (
+            rva_to_file_off(out, tpl_rva),
+            rva_to_file_off(stub, tpl_rva),
+        ) && td.checked_add(tpl_len).is_some_and(|e| e <= out.len())
+            && ts.checked_add(tpl_len).is_some_and(|e| e <= stub.len())
+        {
+            out[td..td + tpl_len].copy_from_slice(&stub[ts..ts + tpl_len]);
+        }
+    }
+
+    // 4) Append DIR64 relocations for the struct's four 64-bit pointer fields
+    //    (Start/End/Index/CallBacks at +0/+8/+0x10/+0x18). Without them the
+    //    loader would leave preferred-base VAs in a rebased image.
+    add_tls_relocs(out, pe, tls_rva);
+}
+
+/// Append a single base-relocation block covering the four 64-bit pointer fields
+/// of the TLS directory struct at `tls_rva`. The block is written immediately
+/// after the existing relocation table (which must be free space and in bounds)
+/// and the BaseReloc directory size is grown to include it. No-op if the table
+/// is absent, the fields straddle a relocation page, or the slot is not free.
+fn add_tls_relocs(out: &mut [u8], pe: usize, tls_rva: u32) {
+    let reloc_dd = pe + 24 + 112 + 5 * 8; // BaseReloc = directory index 5
+    let (reloc_rva, reloc_size) = match (read_u32(out, reloc_dd), read_u32(out, reloc_dd + 4)) {
+        (Some(r), Some(s)) if r != 0 => (r, s),
+        _ => return,
+    };
+    // All four fields (last at +0x18) must share one 0x1000 relocation page.
+    let page = tls_rva & !0xFFF;
+    if (tls_rva.wrapping_add(0x18)) & !0xFFF != page {
+        return;
+    }
+    const BLOCK: usize = 8 + 4 * 2; // header + four DIR64 entries
+    let at = match rva_to_file_off(out, reloc_rva.wrapping_add(reloc_size)) {
+        Some(o) => o,
+        None => return,
+    };
+    if at.checked_add(BLOCK).is_none_or(|e| e > out.len()) {
+        return;
+    }
+    if out[at..at + BLOCK].iter().any(|&b| b != 0) {
+        return; // refuse to clobber existing data
+    }
+    write_u32_at(out, at, page);
+    write_u32_at(out, at + 4, BLOCK as u32);
+    for (i, off) in [0u32, 8, 0x10, 0x18].iter().enumerate() {
+        let entry = (10u16 << 12) | (((tls_rva.wrapping_add(*off)) & 0xFFF) as u16);
+        let p = at + 8 + i * 2;
+        out[p..p + 2].copy_from_slice(&entry.to_le_bytes());
+    }
+    write_u32_at(out, reloc_dd + 4, reloc_size.wrapping_add(BLOCK as u32));
+}
+
+/// Read the Export data-directory (RVA, size) from a PE image, or `None` if the
+/// headers are too short/invalid to parse.
+fn pe_export_dir(buf: &[u8]) -> Option<(u32, u32)> {
+    let headers = senbei_pe::parse(buf).ok()?;
+    senbei_pe::data_directory(buf, headers, 0).ok()
+}
+
+/// Map an RVA to a file offset using the PE section table. Returns `None` if no
+/// section contains the RVA or the headers cannot be parsed.
+fn rva_to_file_off(buf: &[u8], rva: u32) -> Option<usize> {
+    let headers = senbei_pe::parse(buf).ok()?;
+    senbei_pe::rva_to_offset(buf, headers, rva).ok()
+}
+
+fn read_u32(buf: &[u8], off: usize) -> Option<u32> {
+    let b = buf.get(off..off + 4)?;
+    Some(u32::from_le_bytes([b[0], b[1], b[2], b[3]]))
+}
+
+fn read_u16(buf: &[u8], off: usize) -> Option<u16> {
+    let b = buf.get(off..off + 2)?;
+    Some(u16::from_le_bytes([b[0], b[1]]))
+}
+
+fn read_u64(buf: &[u8], off: usize) -> Option<u64> {
+    let b = buf.get(off..off + 8)?;
+    Some(u64::from_le_bytes([
+        b[0], b[1], b[2], b[3], b[4], b[5], b[6], b[7],
+    ]))
+}
+
+/// Write a little-endian `u32` at `off`, silently doing nothing if out of bounds.
+fn write_u32_at(buf: &mut [u8], off: usize, val: u32) {
+    if let Some(slot) = buf.get_mut(off..off + 4) {
+        slot.copy_from_slice(&val.to_le_bytes());
+    }
+}
+
+/// Splice a stub and its external-companion payload into the embedded-payload
+/// form the pipelines expect, or `None` if `comp` is not this stub's payload.
+///
+/// The companion is byte-for-byte the stub's payload region from the Crackproof
+/// header (offset 4096) onward, so the result is `stub[..4096] ++ comp`. The
+/// splice fires only when the first 32 bytes of `comp` equal the stub's header
+/// at offset 4096 — a 32-byte match on the key-table/magic region that confirms
+/// the pairing and leaves ordinary (non-companion) inputs untouched.
+pub(crate) fn splice_companion(stub: &[u8], comp: &[u8]) -> Option<Vec<u8>> {
+    let hdr_end = HEADER_OFF + 32;
+    if stub.len() >= hdr_end && comp.len() >= 32 && stub[HEADER_OFF..hdr_end] == comp[..32] {
+        let mut spliced = Vec::with_capacity(HEADER_OFF + comp.len());
+        spliced.extend_from_slice(&stub[..HEADER_OFF]);
+        spliced.extend_from_slice(comp);
+        return Some(spliced);
+    }
+    None
+}
+
+/// Detect `bytes` and run the right pipeline. Spliced external companions use
+/// the EXE pipeline directly because that layout is definitionally EXE-style.
+///
+/// Routing spliced inputs straight to the EXE pipeline is safe: the
+/// companion layout is definitionally the EXE-style shell (the runtime
+/// loader maps the companion and runs the standard shell unpack), so the DLL
+/// pipeline probe can never be right for it. Output bytes are identical to the
+/// DLL-first + EXE-fallback route for every input that route handles.
+pub(crate) fn unpack_spliced_or_auto(
+    bytes: &[u8],
+    spliced: bool,
+    force_exe: bool,
+    verbose: bool,
+) -> Result<(unpacker::Kind, Vec<u8>), unpacker::UnpackError> {
+    if spliced || force_exe {
+        let detected = unpacker::detect(bytes).ok_or(unpacker::UnpackError::NotCrackproof)?;
+        let out = unpacker::unpack_exe_v(bytes, verbose)?;
+        return Ok((detected.kind, out));
+    }
+    unpacker::unpack_auto_v(bytes, verbose)
+}
+
+/// Unpack a single file to `dest`. Returns the Kind and integrity report on success.
+pub fn unpack_one(
+    input: &Path,
+    dest: &Path,
+) -> anyhow::Result<(unpacker::Kind, unpacker::IntegrityReport)> {
+    unpack_one_v(input, dest, false)
+}
+
+/// Outcome of a byte-level unpack ([`unpack_bytes`]): the image, its detected
+/// kind, and its integrity report. No file I/O is involved.
+pub struct UnpackedImage {
+    pub kind: unpacker::Kind,
+    pub bytes: Vec<u8>,
+    pub integrity: unpacker::IntegrityReport,
+    /// True when the input was reconstructed from an external companion (the
+    /// `._` layout), i.e. the export/TLS overlays ran.
+    pub companion: bool,
+}
+
+/// Unpack in-memory `input` bytes, optionally paired with an external
+/// companion payload `companion` (the `<input>._` file's contents).
+///
+/// This is the in-memory counterpart of [`unpack_one_v`]: splice a matching
+/// companion, unpack, overlay the export table and TLS directory from the stub,
+/// then run the static integrity check.
+pub fn unpack_bytes(
+    input: &[u8],
+    companion: Option<&[u8]>,
+) -> Result<UnpackedImage, unpacker::UnpackError> {
+    unpack_bytes_impl(input, companion, false)
+}
+
+/// Like [`unpack_bytes`], but forces the EXE pipeline (no DLL-pipeline
+/// probe). This is the web app's recovery path: the DLL-first probe relies
+/// on `catch_unwind` to reject EXE-shell-layout DLLs, and panics cannot be
+/// caught on wasm — the probe traps the whole call. The web app runs each
+/// unpack in a disposable Web Worker and retries trapped DLLs with this
+/// entry point, reproducing the CLI's dll-first/exe-fallback routing.
+pub fn unpack_bytes_force_exe(
+    input: &[u8],
+    companion: Option<&[u8]>,
+) -> Result<UnpackedImage, unpacker::UnpackError> {
+    unpack_bytes_impl(input, companion, true)
+}
+
+fn unpack_bytes_impl(
+    input: &[u8],
+    companion: Option<&[u8]>,
+    force_exe: bool,
+) -> Result<UnpackedImage, unpacker::UnpackError> {
+    let spliced = companion.and_then(|c| splice_companion(input, c));
+    let bytes: &[u8] = spliced.as_deref().unwrap_or(input);
+    let (kind, mut out) = unpack_spliced_or_auto(bytes, spliced.is_some(), force_exe, false)?;
+    if spliced.is_some() {
+        overlay_exports_from_stub(&mut out, input);
+        restore_tls_from_stub(&mut out, input);
+        restore_managed_from_stub(&mut out, input)?;
+    }
+    let integrity = unpacker::check_integrity(&out);
+    Ok(UnpackedImage {
+        kind,
+        bytes: out,
+        integrity,
+        companion: spliced.is_some(),
+    })
+}
+
+/// Like [`unpack_one`], but prints detailed `[N/9]` step progress (and a final
+/// `Write to <dest>` line) to stdout when `verbose` is true.
+pub fn unpack_one_v(
+    input: &Path,
+    dest: &Path,
+    verbose: bool,
+) -> anyhow::Result<(unpacker::Kind, unpacker::IntegrityReport)> {
+    let UnpackerInput { bytes, stub } = read_unpacker_input(input)?;
+    let (kind, mut out) = unpack_spliced_or_auto(&bytes, stub.is_some(), false, verbose)?;
+    // External-companion layout: restore the export table from the stub, which
+    // the encrypted companion does not carry (the loader rebuilds it at runtime).
+    if let Some(stub) = stub {
+        overlay_exports_from_stub(&mut out, &stub);
+        // ...and the TLS directory, which Crackproof strips from the payload and
+        // re-installs at runtime; the ordinary loader needs it or thread_local
+        // access crashes (see [`restore_tls_from_stub`]).
+        restore_tls_from_stub(&mut out, &stub);
+        restore_managed_from_stub(&mut out, &stub)?;
+    }
+    let report = unpacker::check_integrity(&out);
+    if let Some(parent) = dest.parent() {
+        std::fs::create_dir_all(parent)?;
+    }
+    write_atomic(dest, &out)?;
+    if verbose {
+        println!("Write to {}", dest.display());
+    }
+    Ok((kind, report))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    const HEADER_OFF: usize = 4096;
+
+    fn stub_with_header(header: &[u8; 32], extra: usize) -> Vec<u8> {
+        let mut stub = vec![0_u8; HEADER_OFF];
+        stub.extend_from_slice(header);
+        stub.extend_from_slice(&vec![0xAA_u8; extra]);
+        stub
+    }
+
+    #[test]
+    fn splices_when_header_matches() {
+        let header = [7_u8; 32];
+        let stub = stub_with_header(&header, 16);
+        let mut companion = header.to_vec();
+        companion.extend_from_slice(&[0x42_u8; 1000]);
+
+        let output = splice_companion(&stub, &companion).expect("should splice");
+        assert_eq!(output.len(), HEADER_OFF + companion.len());
+        assert_eq!(&output[..HEADER_OFF], &stub[..HEADER_OFF]);
+        assert_eq!(&output[HEADER_OFF..], &companion[..]);
+    }
+
+    #[test]
+    fn no_splice_when_header_differs() {
+        let stub = stub_with_header(&[7_u8; 32], 16);
+        let mut companion = vec![9_u8; 32];
+        companion.extend_from_slice(&[0x42_u8; 1000]);
+        assert!(splice_companion(&stub, &companion).is_none());
+    }
+
+    #[test]
+    fn no_splice_when_too_short() {
+        let short_stub = vec![0_u8; HEADER_OFF + 8];
+        let companion = vec![0_u8; 64];
+        assert!(splice_companion(&short_stub, &companion).is_none());
+
+        let stub = stub_with_header(&[1_u8; 32], 0);
+        let short_companion = vec![1_u8; 16];
+        assert!(splice_companion(&stub, &short_companion).is_none());
+    }
+
+    fn managed_fixture(is_pe32_plus: bool, raw: usize) -> Vec<u8> {
+        let mut data = vec![0; raw + 0x600];
+        data[..2].copy_from_slice(b"MZ");
+        data[0x80..0x84].copy_from_slice(b"PE\0\0");
+        let optional_size = if is_pe32_plus { 0xf0u16 } else { 0xe0 };
+        let section = 0x98 + optional_size as usize;
+        let dirs = 0x98 + if is_pe32_plus { 112 } else { 96 };
+        for (offset, value) in [
+            (0x86, 1u16),
+            (0x94, optional_size),
+            (0x98, if is_pe32_plus { 0x20b } else { 0x10b }),
+            (raw + 0x204, 2),
+            (raw + 0x206, if is_pe32_plus { 2 } else { 1 }),
+        ] {
+            data[offset..offset + 2].copy_from_slice(&value.to_le_bytes());
+        }
+        for (offset, value) in [
+            (0x3c, 0x80u32),
+            (0xd0, 0x3000),
+            (0xd4, 0x400),
+            (dirs + 14 * 8, 0x2010),
+            (dirs + 14 * 8 + 4, 0x48),
+            (section + 8, 0x600),
+            (section + 12, 0x2000),
+            (section + 16, 0x600),
+            (section + 20, raw as u32),
+            (raw + 0x10, 0x48),
+            (raw + 0x18, 0x2100),
+            (raw + 0x1c, 0x20),
+            (raw + 0x28, 0x2180),
+            (raw + 0x2c, 8),
+            (raw + 0x40, 0x2200),
+            (raw + 0x44, 8),
+            (raw + 0x200, 0x2280),
+            (raw + 0x280, 0x0600_0001),
+        ] {
+            data[offset..offset + 4].copy_from_slice(&value.to_le_bytes());
+        }
+        data[raw + 0x100..raw + 0x104].copy_from_slice(b"BSJB");
+        data[raw + 0x180..raw + 0x188].copy_from_slice(b"resource");
+        data
+    }
+
+    #[test]
+    fn managed_companion_restores_rva_mapped_regions_without_overwriting_il() {
+        for is_pe32_plus in [false, true] {
+            let stub = managed_fixture(is_pe32_plus, 0x600);
+            let mut out = managed_fixture(is_pe32_plus, 0x400);
+            out[0x400..].fill(0xcc);
+            restore_managed_from_stub(&mut out, &stub).unwrap();
+            for (offset, size) in [
+                (0x10, 0x48),
+                (0x100, 0x20),
+                (0x180, 8),
+                (0x200, 8),
+                (0x280, if is_pe32_plus { 16 } else { 8 }),
+            ] {
+                assert_eq!(
+                    &out[0x400 + offset..0x400 + offset + size],
+                    &stub[0x600 + offset..0x600 + offset + size]
+                );
+            }
+            assert!(out[0x700..0x740].iter().all(|&b| b == 0xcc));
+        }
+    }
+
+    #[test]
+    fn managed_companion_rejects_invalid_metadata_and_unbacked_vtable_slots() {
+        for broken_metadata in [true, false] {
+            let mut stub = managed_fixture(false, 0x600);
+            if broken_metadata {
+                stub[0x700..0x704].fill(0);
+            } else {
+                stub[0x800..0x804].copy_from_slice(&0x2600u32.to_le_bytes());
+            }
+            let mut out = managed_fixture(false, 0x400);
+            let before = out.clone();
+            assert!(matches!(
+                restore_managed_from_stub(&mut out, &stub),
+                Err(unpacker::UnpackError::ManagedStubRestoreFailed { .. })
+            ));
+            assert_eq!(out, before);
+        }
+    }
+}
@@ -0,0 +1,10 @@
+[package]
+name = "senbei-metadata"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+description = "Unity il2cpp metadata de-obfuscation for Senbei"
+
+[dependencies]
+serde.workspace = true
+thiserror.workspace = true
@@ -0,0 +1,137 @@
+//! Extraction of the embedded-metadata packaging variant.
+//!
+//! Some protected il2cpp builds ship no `global-metadata.dat` in the app's
+//! assets at all. Instead a slim metadata blob (an older header format with
+//! custom record layouts) is embedded in the protected library's data section
+//! and wrapped in a per-word XOR layer: a 0x100-byte header whose 64 words each
+//! carry their own key, followed by exactly 256 segments with one u32 key each
+//! at irregular boundaries. At runtime the protector's il2cpp-side modules
+//! regenerate the keys and unwrap the blob in place; the keys are stored
+//! nowhere in the image.
+//!
+//! For the one observed build using this variant the full keystream was
+//! recovered from a ciphertext/plaintext pair and is embedded in
+//! [`crate::keystream`]. Extraction is therefore content-gated: the wrapped
+//! header's first plaintext words are known constants, so a restored image that
+//! does not contain them (every other build) is skipped cheaply and nothing is
+//! written.
+//!
+//! The unwrapped blob stores its patched sanity/version fields byte-swapped;
+//! they are rewritten to the standard il2cpp metadata magic and version so the
+//! output is a well-formed `global-metadata.dat`.
+
+use super::keystream::{HEADER_KEYS, SEGMENTS};
+
+/// Standard il2cpp metadata sanity magic written over the patched header.
+const STANDARD_MAGIC: u32 = 0xfab1_1baf;
+/// Standard header version matching the blob's record layout.
+const STANDARD_VERSION: u32 = 24;
+
+/// Plaintext of the first two wrapped header words (the byte-swapped patched
+/// sanity/version pair). Also the probe pattern: a restored image contains the
+/// embedded blob iff `word[0] ^ HEADER_KEYS[0]` and `word[1] ^ HEADER_KEYS[1]`
+/// equal these constants at some 4-aligned offset.
+const PROBE_WORDS: [u32; 2] = [0x9732_ca38, 0xbac4_374f];
+
+/// Size of the wrapped blob: the last segment's end offset.
+pub fn embedded_metadata_size() -> usize {
+    SEGMENTS[SEGMENTS.len() - 1].0 as usize
+}
+
+/// Locate and unwrap the embedded metadata blob in a restored library image.
+///
+/// Returns a standalone, well-formed `global-metadata.dat`, or `None` when the
+/// image carries no blob wrapped with the known keystream.
+pub fn extract_embedded_metadata(image: &[u8]) -> Option<Vec<u8>> {
+    let total = embedded_metadata_size();
+    let offset = find_wrapped_header(image)?;
+    let blob = image.get(offset..offset.checked_add(total)?)?;
+
+    let mut out = blob.to_vec();
+    for (i, &key) in HEADER_KEYS.iter().enumerate() {
+        xor_word(&mut out, 4 * i, key);
+    }
+    let mut pos = 0x100_usize;
+    for &(end, key) in &SEGMENTS {
+        let end = end as usize;
+        let mut o = pos;
+        while o + 4 <= end {
+            xor_word(&mut out, o, key);
+            o += 4;
+        }
+        pos = end;
+    }
+    out[0..4].copy_from_slice(&STANDARD_MAGIC.to_le_bytes());
+    out[4..8].copy_from_slice(&STANDARD_VERSION.to_le_bytes());
+    Some(out)
+}
+
+/// Scan `image` for the wrapped header probe pattern (4-aligned).
+fn find_wrapped_header(image: &[u8]) -> Option<usize> {
+    let mut off = 0;
+    while off + 8 <= image.len() {
+        let word = u32::from_le_bytes(image[off..off + 4].try_into().ok()?);
+        if word ^ HEADER_KEYS[0] == PROBE_WORDS[0] {
+            let next = u32::from_le_bytes(image[off + 4..off + 8].try_into().ok()?);
+            if next ^ HEADER_KEYS[1] == PROBE_WORDS[1] {
+                return Some(off);
+            }
+        }
+        off += 4;
+    }
+    None
+}
+
+fn xor_word(data: &mut [u8], offset: usize, key: u32) {
+    let word = u32::from_le_bytes(data[offset..offset + 4].try_into().expect("word in bounds"));
+    data[offset..offset + 4].copy_from_slice(&(word ^ key).to_le_bytes());
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    /// Wrap a synthetic blob with the keystream, then unwrap it back.
+    #[test]
+    fn roundtrip_wrapped_blob() {
+        let total = embedded_metadata_size();
+        let mut image = vec![0_u8; total + 0x40];
+        // Plaintext blob: standard probe words, then a ramp.
+        image[0..4].copy_from_slice(&PROBE_WORDS[0].to_le_bytes());
+        image[4..8].copy_from_slice(&PROBE_WORDS[1].to_le_bytes());
+        for o in (8..total).step_by(4) {
+            let v = (o as u32).wrapping_mul(0x9e37_79b1);
+            image[o..o + 4].copy_from_slice(&v.to_le_bytes());
+        }
+        // Wrap with the keystream.
+        for (i, &key) in HEADER_KEYS.iter().enumerate() {
+            xor_word(&mut image, 4 * i, key);
+        }
+        let mut pos = 0x100_usize;
+        for &(end, key) in &SEGMENTS {
+            let mut o = pos;
+            while o + 4 <= end as usize {
+                xor_word(&mut image, o, key);
+                o += 4;
+            }
+            pos = end as usize;
+        }
+
+        let out = extract_embedded_metadata(&image).expect("blob found");
+        assert_eq!(out.len(), total);
+        // Header rewritten to the standard magic/version…
+        assert_eq!(&out[0..4], &STANDARD_MAGIC.to_le_bytes());
+        assert_eq!(&out[4..8], &STANDARD_VERSION.to_le_bytes());
+        // …and the body round-trips.
+        for o in (8..total).step_by(4) {
+            let v = (o as u32).wrapping_mul(0x9e37_79b1);
+            assert_eq!(&out[o..o + 4], &v.to_le_bytes(), "word at {o:#x}");
+        }
+    }
+
+    #[test]
+    fn no_blob_in_plain_data() {
+        let image = vec![0xAB_u8; 0x1000];
+        assert!(extract_embedded_metadata(&image).is_none());
+    }
+}
@@ -0,0 +1,274 @@
+/// Per-word XOR keystream for the embedded-metadata packaging variant,
+/// recovered from a ciphertext/plaintext pair of one observed build.
+/// Key derivation for future builds is untraced; other builds simply do
+/// not match the header probe and are left untouched.
+pub(crate) const HEADER_KEYS: [u32; 64] = [
+    0x39184c70, 0xd901afd4, 0x19b98815, 0x132906ed, 0x663e8ace, 0x299b1952, 0xe5404ab8, 0xd93b331c,
+    0xb67d3761, 0x42da9259, 0xc29c7a59, 0x17cb841c, 0xd0bcb9c6, 0x21db779b, 0x43874deb, 0x89bf697b,
+    0x0b7f97b4, 0xbe1c59f7, 0xc653ad92, 0x8cdf4336, 0x5e0b6b68, 0x1bd4d668, 0x7250ed61, 0x31a36491,
+    0xaf144dcd, 0xc1e387d0, 0x9d6df5b7, 0x78514f32, 0xc2648cbf, 0x3b8272a5, 0xd2053679, 0x4b18af77,
+    0x71b9ebdd, 0x0094daaa, 0xf3adfed8, 0xc0d082bc, 0xae5e523c, 0xa8dec0be, 0x090a7784, 0x2c0483d6,
+    0x95f0e8f7, 0x234de6d4, 0xa7464527, 0x3b1c531d, 0xc2b31d82, 0xe1c60be0, 0x3d65a0c2, 0x2ea7d77a,
+    0x4ababadb, 0xce484b16, 0x59ab3f99, 0x10a9a463, 0x70e2f78a, 0x0ed71c9c, 0xf8996b2e, 0xff637928,
+    0xf413313d, 0x77c57bf9, 0xdab41dba, 0x0cd2ccbc, 0x3b2fbde3, 0x0b19b14d, 0xd2645dbc, 0x318113d4,
+];
+
+/// (segment end offset, segment key) pairs; offsets relative to blob start.
+pub(crate) const SEGMENTS: [(u32, u32); 256] = [
+    (0x3915c, 0xbb5dda1a),
+    (0x736b4, 0x906dbe0f),
+    (0xaedc4, 0x1e4ca8bd),
+    (0xc506c, 0xe603cb21),
+    (0xdfe34, 0x93bec702),
+    (0xe8f10, 0x9a9f429f),
+    (0x139088, 0x125a8b3f),
+    (0x20bd30, 0x69a6395f),
+    (0x20c548, 0xca807b9a),
+    (0x20c774, 0x8b8880c4),
+    (0x300518, 0x48087852),
+    (0x36c07c, 0x32aa7b5b),
+    (0x448b7c, 0x2e668589),
+    (0x4592b0, 0x292e07d9),
+    (0x45d374, 0x83b0a0ef),
+    (0x474520, 0x8983245d),
+    (0x47a1d4, 0xefb941b7),
+    (0x4bdc74, 0x7c3b3458),
+    (0x4c34c8, 0xee0a87b3),
+    (0x4f1068, 0xf6a2069f),
+    (0x51601c, 0x2e83612b),
+    (0x549d40, 0xb413a58f),
+    (0x56a714, 0x95596da3),
+    (0x573c98, 0x68513e8d),
+    (0x59d058, 0xc4ff5f9a),
+    (0x5c4b34, 0x249ed022),
+    (0x5f19bc, 0xc27272d3),
+    (0x5f47c8, 0xd73aa37b),
+    (0x627df4, 0x002334ba),
+    (0x648f54, 0x868bb6c9),
+    (0x6718a4, 0x17ff0ef4),
+    (0x6a88b8, 0x22cfbc5f),
+    (0x742dbc, 0x152072dd),
+    (0x75603c, 0xbd31be45),
+    (0x783238, 0x45911d6a),
+    (0x7b94d8, 0x6f281add),
+    (0x800060, 0xcf58c8d0),
+    (0x819b50, 0xb40f0276),
+    (0x82dea4, 0x1a1a8402),
+    (0x880210, 0xf2c0824a),
+    (0x8e4f08, 0x86c9ba90),
+    (0x8ea544, 0x0e928544),
+    (0x931454, 0xc3fa017b),
+    (0x94eb70, 0x1dbe612a),
+    (0x95993c, 0x902498fe),
+    (0x98d2b0, 0xb7760451),
+    (0x992034, 0x711cddfc),
+    (0x9e14a0, 0x8bd95e64),
+    (0xa1d2d0, 0xbfdce920),
+    (0xa21ed8, 0x90cf0372),
+    (0xa4d2b8, 0x91e88c9c),
+    (0xa76f8c, 0x9c721e61),
+    (0xac5ba4, 0xbda16e3e),
+    (0xaf070c, 0xe02b6799),
+    (0xaf3a78, 0x32953b4e),
+    (0xb32510, 0x47ea48db),
+    (0xb46550, 0x1443e512),
+    (0xb54998, 0x9e123a75),
+    (0xb5e24c, 0xe11a8efd),
+    (0xb625cc, 0x3facfbf4),
+    (0xb66ec8, 0x76c0c452),
+    (0xb67ad0, 0x4de4ed6c),
+    (0xb7ba5c, 0xe622d97a),
+    (0xb85a90, 0x6f564f8b),
+    (0xbff8b8, 0x3e25d671),
+    (0xc03e50, 0x3563fc2b),
+    (0xc6e958, 0xda8bc3b0),
+    (0xc87f7c, 0x5a9d2269),
+    (0xcb36a4, 0x0ab420cc),
+    (0xcbe4d0, 0x9bbb091e),
+    (0xccd7dc, 0x9e4fd577),
+    (0xd078c0, 0x4b655ae1),
+    (0xd275dc, 0x5ca2a2f4),
+    (0xd2c840, 0xdb437f0d),
+    (0xd3296c, 0x66487f75),
+    (0xd7cbe8, 0xf5427945),
+    (0xd8e0a0, 0x9a65bdb6),
+    (0xda2ed4, 0x46dea4b3),
+    (0xda6f1c, 0xb9916a02),
+    (0xdee9ac, 0x18800a5c),
+    (0xe3673c, 0x4afab3cd),
+    (0xe65420, 0x52e80204),
+    (0xe861f8, 0x639a02d7),
+    (0xeb61c0, 0x21077eba),
+    (0xed51dc, 0x17be91d8),
+    (0xf048a4, 0xd30cc8cb),
+    (0xf3b274, 0xdfb43f3f),
+    (0xf76ac8, 0x63a8b363),
+    (0xf84b64, 0x16508a16),
+    (0xf8bb2c, 0x22ce110d),
+    (0xfb3390, 0xf09a4eb2),
+    (0xff07bc, 0xd2bb0e2c),
+    (0x102d32c, 0xb424012c),
+    (0x10795b0, 0x07338bb9),
+    (0x108d65c, 0x5d68f86e),
+    (0x10ce528, 0x1826c952),
+    (0x10d3528, 0xa7473860),
+    (0x10dca58, 0x92435967),
+    (0x1115c78, 0x061200f4),
+    (0x1171098, 0x94f538a1),
+    (0x117ebf0, 0xd8731d88),
+    (0x1186638, 0x4381b3f9),
+    (0x118afdc, 0xf25ff376),
+    (0x11ee3b4, 0x29605488),
+    (0x11f182c, 0x04367932),
+    (0x11f41dc, 0xaeaccadd),
+    (0x11ff0f4, 0x7c4d358e),
+    (0x120caac, 0xacbc8412),
+    (0x12437cc, 0x3e0ac7e9),
+    (0x124edf8, 0x06f523fd),
+    (0x1263ba0, 0x0a1b9763),
+    (0x12943f0, 0x24a86ba4),
+    (0x12d9230, 0x0cd82e2e),
+    (0x12fd9b4, 0xf3903fb9),
+    (0x135a198, 0x2887f4a3),
+    (0x1366180, 0x9f0d7ca5),
+    (0x13680a4, 0x61e9a459),
+    (0x13a1b44, 0xe61623a4),
+    (0x13a860c, 0xdc44c798),
+    (0x13c024c, 0xc90f7be6),
+    (0x1475c00, 0xc2f338b3),
+    (0x1480aa8, 0xb7b0609e),
+    (0x14fb82c, 0x748e3939),
+    (0x1511184, 0x98426fcf),
+    (0x153c144, 0x1a452d5d),
+    (0x1547838, 0x7dd360e9),
+    (0x15565d4, 0x1d8f093b),
+    (0x156e298, 0x102a1524),
+    (0x159df70, 0xe42613f7),
+    (0x15a13d0, 0xafc5fdc6),
+    (0x15e7f24, 0x84fcd342),
+    (0x15f0878, 0x55038958),
+    (0x1614210, 0xe0602ae4),
+    (0x1631b3c, 0xce2765f6),
+    (0x164eb70, 0xf772dac5),
+    (0x1688b68, 0x5f1a72c9),
+    (0x16d5f8c, 0x7c77747d),
+    (0x16e76dc, 0xac0e16fb),
+    (0x1726374, 0x4a1e7fd7),
+    (0x173455c, 0x870856b4),
+    (0x17697d8, 0xbb2f0a5c),
+    (0x176ed60, 0xc937b386),
+    (0x1784fa8, 0x5e676ab2),
+    (0x17ae3a0, 0xdbf662a1),
+    (0x1866c7c, 0x4e3f1a7d),
+    (0x186b844, 0xe30fce60),
+    (0x18b507c, 0xdfc73c88),
+    (0x18c2f64, 0xb7ee08e0),
+    (0x18c8010, 0xd1471a25),
+    (0x18de290, 0x292e6310),
+    (0x19140d0, 0x9f346f05),
+    (0x192c590, 0xf1eb61bf),
+    (0x194fca0, 0x8888b1df),
+    (0x1959d34, 0x92b89d15),
+    (0x196c0c4, 0x5e152de5),
+    (0x19a5710, 0x866e7bfa),
+    (0x19abfd4, 0x3084ae26),
+    (0x19b1550, 0x0581836f),
+    (0x19b7214, 0xeefc34eb),
+    (0x19c523c, 0xc980335d),
+    (0x19dc4c0, 0x019084e6),
+    (0x19dfb8c, 0xdb1a21a7),
+    (0x19fbf3c, 0xec84cc17),
+    (0x1a29b18, 0xcb31da7d),
+    (0x1a4c670, 0xc5fe570e),
+    (0x1a97024, 0xbbd80964),
+    (0x1ac33bc, 0xe186586d),
+    (0x1acd124, 0x1e413252),
+    (0x1ad9bac, 0x48fc4c75),
+    (0x1b1b728, 0x8071d7a5),
+    (0x1b31d78, 0x9d958013),
+    (0x1badb24, 0x2f236951),
+    (0x1bccc00, 0x7023c620),
+    (0x1bdab2c, 0x88b1e4b8),
+    (0x1c000dc, 0x9e43291a),
+    (0x1c9f0cc, 0x27a7d592),
+    (0x1cd1328, 0x9c0bcc88),
+    (0x1cd79a0, 0x63e0ed75),
+    (0x1d0e484, 0xf51a0d3d),
+    (0x1d17b10, 0xbfd2a7ac),
+    (0x1d930c4, 0xf6b9e877),
+    (0x1db115c, 0xf3eb7e37),
+    (0x1df16b4, 0x682326ff),
+    (0x1e389c0, 0xea11f566),
+    (0x1eb7e48, 0x3dc5fa76),
+    (0x1ec38fc, 0x296ffc1d),
+    (0x1ee87a0, 0x1b9f7fd4),
+    (0x1f19f88, 0x78972e8f),
+    (0x1f33a0c, 0x390c2deb),
+    (0x1f4e0fc, 0xe05e8c6b),
+    (0x1f5a718, 0x367432ae),
+    (0x1f61dcc, 0x7063e58a),
+    (0x1f85878, 0x21c00cea),
+    (0x1fc043c, 0x2676aaaa),
+    (0x1ffdb94, 0xc270eb02),
+    (0x202a618, 0x3a98aed2),
+    (0x2037b34, 0x115d5afc),
+    (0x203d92c, 0x11bced76),
+    (0x203da14, 0xf2628105),
+    (0x2066014, 0x97f32700),
+    (0x208a908, 0xa68e2f71),
+    (0x20ab8ac, 0x1daa2a78),
+    (0x20ba504, 0x73919ef6),
+    (0x20e71e0, 0x0b3fd1d3),
+    (0x2102278, 0x6c123def),
+    (0x21166dc, 0xee354161),
+    (0x2126478, 0x299493f4),
+    (0x2137090, 0x05ae2007),
+    (0x2148270, 0x34b52663),
+    (0x21482ac, 0xe381b5b6),
+    (0x21813b8, 0x94244de1),
+    (0x21a41e8, 0x02c38df5),
+    (0x21a8c4c, 0xf72700dd),
+    (0x21abbac, 0x34c2e7b5),
+    (0x21bcb24, 0x442739ad),
+    (0x21cbe84, 0x6e40d22c),
+    (0x21e2798, 0xdbf774d0),
+    (0x21f892c, 0xe90f1e0c),
+    (0x222beec, 0xa27f27f3),
+    (0x22394f4, 0x7f999a4f),
+    (0x22437ec, 0xf12d28f8),
+    (0x22480b0, 0xf58f3a7d),
+    (0x2261a0c, 0x89b28301),
+    (0x22a76c8, 0x1fe501e2),
+    (0x22b2018, 0xf079db5f),
+    (0x22cc610, 0xaf7d17b7),
+    (0x22cd4f8, 0x71c010cb),
+    (0x22d016c, 0x4a8daed0),
+    (0x22e1c04, 0xe1201aca),
+    (0x22f9994, 0xf3f0e4ee),
+    (0x2384f5c, 0x6b8a5eb1),
+    (0x23d5ecc, 0x5298a9c4),
+    (0x23e15b0, 0xc7bf0afb),
+    (0x23e248c, 0x2d67fecf),
+    (0x2407898, 0x4eef422a),
+    (0x241695c, 0x33ba9ce8),
+    (0x243e8b0, 0x833c1d2c),
+    (0x2460b64, 0x819c96ee),
+    (0x247caec, 0x0ebccbd6),
+    (0x24832b4, 0xf789d4b6),
+    (0x24938b8, 0x9f63baeb),
+    (0x24a7c64, 0x3384e552),
+    (0x24bce94, 0x7bfec208),
+    (0x24bd8f4, 0x9b5260cc),
+    (0x24ce8ec, 0xaf854888),
+    (0x24e741c, 0xda82f062),
+    (0x254401c, 0xbb1a5d5a),
+    (0x25a6a64, 0x24d202d3),
+    (0x2617878, 0x6ac71e5f),
+    (0x2617df0, 0x0e76bd90),
+    (0x268e9b8, 0x874d931c),
+    (0x26a8848, 0xefefd680),
+    (0x26e4f10, 0xfc2799f7),
+    (0x26e93d8, 0x1930ad55),
+    (0x26eea84, 0xfa2f742f),
+    (0x2701f30, 0xed9b92c4),
+];
@@ -0,0 +1,11 @@
+//! Static IL2CPP metadata restoration interfaces.
+
+mod embedded;
+mod keystream;
+mod method_tokens;
+
+pub use embedded::{embedded_metadata_size, extract_embedded_metadata};
+pub use method_tokens::{
+    DEFAULT_METHOD_TOKEN_SEED, Error, ImageKeyDiscovery, Report, SeedDiscoveryReport,
+    discover_method_token_seeds, restore_method_tokens,
+};
@@ -0,0 +1,12 @@
+//! Shared IL2CPP metadata header primitives.
+
+/// IL2CPP global-metadata sanity magic.
+pub(crate) const MAGIC: u32 = 0xFAB1_1BAF;
+
+/// Cheap check used by both platform scanners before opening a full metadata
+/// file.
+#[must_use]
+pub fn is_metadata(data: &[u8]) -> bool {
+    data.get(0..4)
+        .is_some_and(|bytes| u32::from_le_bytes([bytes[0], bytes[1], bytes[2], bytes[3]]) == MAGIC)
+}
@@ -0,0 +1,9 @@
+//! Unity il2cpp metadata restoration.
+
+pub mod android;
+mod common;
+mod structural;
+pub mod windows;
+
+pub use common::is_metadata;
+pub use structural::*;
@@ -27,8 +27,7 @@
 //! metadata (its tokens already equal `local_index + 1`), so it is safe to run on
 //! any il2cpp game — `remapped == 0` then reports that nothing changed.

-/// il2cpp `global-metadata.dat` sanity magic (`Il2CppGlobalMetadataHeader.sanity`).
-const MAGIC: u32 = 0xFAB1_1BAF;
+use crate::common::MAGIC;

 /// Metadata format version this de-obfuscator understands. The struct strides
 /// and header field offsets below are specific to it; other versions are left
@@ -0,0 +1,3 @@
+//! Compatibility namespace for the shared structural metadata transform.
+
+pub use crate::structural::*;
@@ -0,0 +1,9 @@
+[package]
+name = "senbei-pe"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+description = "PE format parsing and address mapping for Senbei"
+
+[dependencies]
+thiserror.workspace = true
@@ -0,0 +1,233 @@
+//! Basic PE format parsing and address mapping.
+
+use thiserror::Error;
+
+#[derive(Debug, Error, Clone, PartialEq, Eq)]
+pub enum Error {
+    #[error("input is not a PE image")]
+    Invalid,
+    #[error("PE range is outside the input")]
+    OutOfBounds,
+}
+
+pub type Result<T> = std::result::Result<T, Error>;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct Section {
+    pub virtual_address: u32,
+    pub virtual_size: u32,
+    pub raw_offset: u32,
+    pub raw_size: u32,
+    pub characteristics: u32,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct Headers {
+    pub pe_offset: usize,
+    pub is_pe32_plus: bool,
+    pub image_base: u64,
+    pub size_of_image: u32,
+    pub entry_rva: u32,
+    pub sections_offset: usize,
+    pub sections: u16,
+}
+
+pub fn parse(data: &[u8]) -> Result<Headers> {
+    if data.get(0..2) != Some(b"MZ") {
+        return Err(Error::Invalid);
+    }
+    let pe_offset = read_u32(data, 0x3c)? as usize;
+    if data.get(pe_offset..pe_offset + 4) != Some(b"PE\0\0") {
+        return Err(Error::Invalid);
+    }
+    let sections = read_u16(data, pe_offset + 6)?;
+    let optional_size = read_u16(data, pe_offset + 20)? as usize;
+    let optional = pe_offset.checked_add(24).ok_or(Error::OutOfBounds)?;
+    let magic = read_u16(data, optional)?;
+    let is_pe32_plus = magic == 0x20b;
+    if !is_pe32_plus && magic != 0x10b {
+        return Err(Error::Invalid);
+    }
+    let entry_rva = read_u32(data, optional + 16)?;
+    let image_base = if is_pe32_plus {
+        read_u64(data, optional + 24)?
+    } else {
+        read_u32(data, optional + 28)? as u64
+    };
+    let size_of_image = read_u32(data, optional + 56)?;
+    let sections_offset = optional
+        .checked_add(optional_size)
+        .ok_or(Error::OutOfBounds)?;
+    let table_size = usize::from(sections)
+        .checked_mul(40)
+        .ok_or(Error::OutOfBounds)?;
+    data.get(sections_offset..sections_offset + table_size)
+        .ok_or(Error::OutOfBounds)?;
+    Ok(Headers {
+        pe_offset,
+        is_pe32_plus,
+        image_base,
+        size_of_image,
+        entry_rva,
+        sections_offset,
+        sections,
+    })
+}
+
+pub fn sections(data: &[u8], headers: Headers) -> Result<Vec<Section>> {
+    (0..headers.sections)
+        .map(|index| {
+            let offset = headers
+                .sections_offset
+                .checked_add(usize::from(index) * 40)
+                .ok_or(Error::OutOfBounds)?;
+            Ok(Section {
+                virtual_size: read_u32(data, offset + 8)?,
+                virtual_address: read_u32(data, offset + 12)?,
+                raw_size: read_u32(data, offset + 16)?,
+                raw_offset: read_u32(data, offset + 20)?,
+                characteristics: read_u32(data, offset + 36)?,
+            })
+        })
+        .collect()
+}
+
+/// Read one PE data-directory entry as `(RVA, size)`.
+pub fn data_directory(data: &[u8], headers: Headers, index: u16) -> Result<(u32, u32)> {
+    let directory_base = headers
+        .pe_offset
+        .checked_add(24)
+        .and_then(|offset| offset.checked_add(if headers.is_pe32_plus { 112 } else { 96 }))
+        .ok_or(Error::OutOfBounds)?;
+    let offset = directory_base
+        .checked_add(
+            usize::from(index)
+                .checked_mul(8)
+                .ok_or(Error::OutOfBounds)?,
+        )
+        .ok_or(Error::OutOfBounds)?;
+    Ok((read_u32(data, offset)?, read_u32(data, offset + 4)?))
+}
+
+/// Return the COFF characteristics bit field.
+pub fn characteristics(data: &[u8], headers: Headers) -> Result<u16> {
+    read_u16(
+        data,
+        headers
+            .pe_offset
+            .checked_add(22)
+            .ok_or(Error::OutOfBounds)?,
+    )
+}
+
+pub fn rva_to_offset(data: &[u8], headers: Headers, rva: u32) -> Result<usize> {
+    if rva < headers.sections_offset as u32 {
+        return Ok(rva as usize);
+    }
+    for section in sections(data, headers)? {
+        let span = section.virtual_size.max(section.raw_size);
+        if rva >= section.virtual_address && rva < section.virtual_address.saturating_add(span) {
+            let offset = section
+                .raw_offset
+                .checked_add(rva - section.virtual_address)
+                .ok_or(Error::OutOfBounds)? as usize;
+            if offset < data.len() {
+                return Ok(offset);
+            }
+        }
+    }
+    Err(Error::OutOfBounds)
+}
+
+/// Map a complete RVA range backed by file bytes in the headers or one section.
+/// Unlike a virtual mapping, this rejects a section's zero-filled tail.
+pub fn rva_range(
+    data: &[u8],
+    headers: Headers,
+    rva: u32,
+    size: u32,
+) -> Result<std::ops::Range<usize>> {
+    let header_size = read_u32(data, headers.pe_offset + 24 + 60)?;
+    let offset = if rva < header_size && size <= header_size - rva {
+        rva
+    } else {
+        sections(data, headers)?
+            .into_iter()
+            .find_map(|section| {
+                let delta = rva.checked_sub(section.virtual_address)?;
+                if delta >= section.raw_size || size > section.raw_size - delta {
+                    return None;
+                }
+                section.raw_offset.checked_add(delta)
+            })
+            .ok_or(Error::OutOfBounds)?
+    } as usize;
+    let end = offset
+        .checked_add(size as usize)
+        .ok_or(Error::OutOfBounds)?;
+    data.get(offset..end).ok_or(Error::OutOfBounds)?;
+    Ok(offset..end)
+}
+
+fn read_u16(data: &[u8], offset: usize) -> Result<u16> {
+    let bytes: [u8; 2] = data
+        .get(offset..offset + 2)
+        .ok_or(Error::OutOfBounds)?
+        .try_into()
+        .map_err(|_| Error::OutOfBounds)?;
+    Ok(u16::from_le_bytes(bytes))
+}
+
+fn read_u32(data: &[u8], offset: usize) -> Result<u32> {
+    let bytes: [u8; 4] = data
+        .get(offset..offset + 4)
+        .ok_or(Error::OutOfBounds)?
+        .try_into()
+        .map_err(|_| Error::OutOfBounds)?;
+    Ok(u32::from_le_bytes(bytes))
+}
+
+fn read_u64(data: &[u8], offset: usize) -> Result<u64> {
+    let bytes: [u8; 8] = data
+        .get(offset..offset + 8)
+        .ok_or(Error::OutOfBounds)?
+        .try_into()
+        .map_err(|_| Error::OutOfBounds)?;
+    Ok(u64::from_le_bytes(bytes))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn rva_ranges_require_file_backing_for_every_byte() {
+        let mut data = [0u8; 0x400];
+        let headers = Headers {
+            pe_offset: 0x40,
+            is_pe32_plus: false,
+            image_base: 0,
+            size_of_image: 0x2000,
+            entry_rva: 0x1000,
+            sections_offset: 0x100,
+            sections: 1,
+        };
+        for (offset, value) in [
+            (0x94, 0x200u32),
+            (0x108, 0x100),
+            (0x10c, 0x1000),
+            (0x110, 0x80),
+            (0x114, 0x200),
+        ] {
+            data[offset..offset + 4].copy_from_slice(&value.to_le_bytes());
+        }
+        assert_eq!(rva_range(&data, headers, 0x1000, 0x80), Ok(0x200..0x280));
+        assert_eq!(rva_range(&data, headers, 0x100, 0x100), Ok(0x100..0x200));
+        for (rva, size) in [(0x1070, 0x20), (0x1080, 1), (0x1f0, 0x20), (u32::MAX, 4)] {
+            assert_eq!(
+                rva_range(&data, headers, rva, size),
+                Err(Error::OutOfBounds)
+            );
+        }
+    }
+}
@@ -0,0 +1,871 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+version = 4
+
+[[package]]
+name = "aes"
+version = "0.9.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "35f0f96ce78e38c3dc6d8948aa8163d06385be74000f3c7a95bf1eef35d3ea32"
+dependencies = [
+ "cipher",
+ "cpubits",
+ "cpufeatures",
+]
+
+[[package]]
+name = "anyhow"
+version = "1.0.104"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "330a5ed07fa54e4702c9d6c4174f74427fc0ef6e214bbd677ae50a5099946470"
+
+[[package]]
+name = "bitflags"
+version = "2.13.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b588b76d00fde79687d7646a9b5bdf3cc0f655e0bbd080335a95d7e96f3587da"
+
+[[package]]
+name = "block-buffer"
+version = "0.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d2f6c7dbe95a6ed67ad9f18e57daf93a2f034c524b99fd2b76d18fdfeb6660aa"
+dependencies = [
+ "hybrid-array",
+]
+
+[[package]]
+name = "bumpalo"
+version = "3.20.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "72f5acc6cb2ba439de613abc23857ec3d78374d8ed5ac84e9d11336e87da8649"
+
+[[package]]
+name = "cfg-if"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
+
+[[package]]
+name = "cipher"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e8cf2a2c93cd704877c0858356ed03480ff301ee950b43f1cbe4573b088bfa6c"
+dependencies = [
+ "crypto-common",
+ "inout",
+]
+
+[[package]]
+name = "console"
+version = "0.16.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4fe5f465a4f6fee88fad41b85d990f84c835335e85b5d9e6e63e0d06d28cba7c"
+dependencies = [
+ "encode_unicode",
+ "libc",
+ "unicode-width",
+ "windows-sys",
+]
+
+[[package]]
+name = "console_error_panic_hook"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a06aeb73f470f66dcdbf7223caeebb85984942f22f1adb2a088cf9668146bbbc"
+dependencies = [
+ "cfg-if",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "const-oid"
+version = "0.10.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a6ef517f0926dd24a1582492c791b6a4818a4d94e789a334894aa15b0d12f55c"
+
+[[package]]
+name = "cpubits"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "15b85f9c39137c3a891689859392b1bd49812121d0d61c9caf00d46ed5ce06ae"
+
+[[package]]
+name = "cpufeatures"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5ca28b0ae3115b884660db4118d803791fd6756b6e88f39c0f3f7859060d7566"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "crc32fast"
+version = "1.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8498c871161e1742aaa9d52551b2d6ebdd4c3d45a3be423e3728f33b955be550"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "crypto-common"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ce6e4c961d6cd6c9a86db418387425e8bdeaf05b3c8bc1411e6dca4c252f1453"
+dependencies = [
+ "hybrid-array",
+]
+
+[[package]]
+name = "digest"
+version = "0.11.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f1dd6dbb5841937940781866fa1281a1ff7bd3bf827091440879f9994983d5c2"
+dependencies = [
+ "block-buffer",
+ "const-oid",
+ "crypto-common",
+]
+
+[[package]]
+name = "encode_unicode"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34aa73646ffb006b8f5147f3dc182bd4bcb190227ce861fc4a4844bf8e3cb2c0"
+
+[[package]]
+name = "equivalent"
+version = "1.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f"
+
+[[package]]
+name = "errno"
+version = "0.3.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
+dependencies = [
+ "libc",
+ "windows-sys",
+]
+
+[[package]]
+name = "fastrand"
+version = "2.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "da7c62ceae207dd37ea5b845da6a0696c799f85e97da1ab5b7910be3c1c80223"
+
+[[package]]
+name = "flate2"
+version = "1.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6e634e2e0ebac1ee034020da1ca582e17ffe4e0f5e985823721e168928136dcb"
+dependencies = [
+ "zlib-rs",
+]
+
+[[package]]
+name = "futures-core"
+version = "0.3.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "92d699e522242e69e3003b94ecc1f960f3a5e015aa7c5d7486e65ad01dd94f5e"
+
+[[package]]
+name = "futures-task"
+version = "0.3.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cd417de3d1d015fc3bfd2b1ea46dfc7bab72ef86f1cc7cc9c78e728b34a6d1fd"
+
+[[package]]
+name = "futures-util"
+version = "0.3.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0d50a92467f8ba5dd6e3ee5d4bd04d73ab2e4e1c44474a0674821dfce14b79bc"
+dependencies = [
+ "futures-core",
+ "futures-task",
+ "pin-project-lite",
+ "slab",
+]
+
+[[package]]
+name = "getrandom"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "300e883d756b2e4ec94e02791f39b04b522276138852cfc41d9fb7e904106099"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "r-efi",
+]
+
+[[package]]
+name = "goblin"
+version = "0.10.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "17582616a7718cca54cec18e534a76c7c4aec11a8b9a85695712f262fd15a4c8"
+dependencies = [
+ "log",
+ "plain",
+ "scroll",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.17.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ed5909b6e89a2db4456e54cd5f673791d7eca6732202bbf2a9cc504fe2f9b84a"
+
+[[package]]
+name = "hybrid-array"
+version = "0.4.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "707114b52a152fa7bdb290cd7cd5912d9467273b6d74e21b8d81aca1f8533f6b"
+dependencies = [
+ "typenum",
+]
+
+[[package]]
+name = "indexmap"
+version = "2.14.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "07aa2048142242915a31d35844fb311e0e53fcca590c3a0a40dcf1b841fa09eb"
+dependencies = [
+ "equivalent",
+ "hashbrown",
+]
+
+[[package]]
+name = "indicatif"
+version = "0.18.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9433806cd6b4ec1aba79c021c7e4c58fb4c3b9977c085062e611ac929998fb0c"
+dependencies = [
+ "console",
+ "portable-atomic",
+ "unicode-width",
+ "unit-prefix",
+ "web-time",
+]
+
+[[package]]
+name = "inout"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4250ce6452e92010fdf7268ccc5d14faa80bb12fc741938534c58f16804e03c7"
+dependencies = [
+ "hybrid-array",
+]
+
+[[package]]
+name = "itoa"
+version = "1.0.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8f42a60cbdf9a97f5d2305f08a87dc4e09308d1276d28c869c684d7777685682"
+
+[[package]]
+name = "js-sys"
+version = "0.3.104"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0e0c1080212aad755ea003d18543e8768dd432c48819efd73a7bf1e39b7a5a3a"
+dependencies = [
+ "cfg-if",
+ "futures-util",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "libc"
+version = "0.2.189"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3eaf3ede3fee6db1a4c2ee091bf8a8b4dccdc6d17f656fb07896ee72867612f2"
+
+[[package]]
+name = "linux-raw-sys"
+version = "0.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "32a66949e030da00e8c7d4434b251670a91556f4144941d37452769c25d58a53"
+
+[[package]]
+name = "log"
+version = "0.4.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f9f8bd3e56ce4dfc153cf470fffbfa98c7620958b312ca5c3a4b8d5181fd13c6"
+
+[[package]]
+name = "memchr"
+version = "2.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf8baf1c55e62ffcace7a9f06f4bd9cd3f0c4beb022d3b367256b91b87513d98"
+
+[[package]]
+name = "memmap2"
+version = "0.9.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d1219ed1b7f229ee7104d281dd01d6802fe28bb6e95d292942c4daacdeb798c0"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "once_cell"
+version = "1.21.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"
+
+[[package]]
+name = "owo-colors"
+version = "4.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "13c45bb4a6ae1280ec0803b1ef9d3455eb50f01efbbe1447ab020f1d54fba9d8"
+
+[[package]]
+name = "pin-project-lite"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"
+
+[[package]]
+name = "plain"
+version = "0.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b4596b6d070b27117e987119b4dac604f3c58cfb0b191112e24771b2faeac1a6"
+
+[[package]]
+name = "portable-atomic"
+version = "1.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "05c8b63e8d9609db387f0324918f81d68fe27748f084ef092fb35954d0539a85"
+
+[[package]]
+name = "proc-macro2"
+version = "1.0.107"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "985e7ec9bb745e6ce6535b544d84d6cd6f7ad8bd711c398938ae983b91a766d9"
+dependencies = [
+ "unicode-ident",
+]
+
+[[package]]
+name = "quote"
+version = "1.0.47"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fbf4db142a473a8d80c26bbf18454ed458bf8d26c8219c331daecfdbd079001"
+dependencies = [
+ "proc-macro2",
+]
+
+[[package]]
+name = "r-efi"
+version = "6.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f8dcc9c7d52a811697d2151c701e0d08956f92b0e24136cf4cf27b57a6a0d9bf"
+
+[[package]]
+name = "rustix"
+version = "1.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6fe4565b9518b83ef4f91bb47ce29620ca828bd32cb7e408f0062e9930ba190"
+dependencies = [
+ "bitflags",
+ "errno",
+ "libc",
+ "linux-raw-sys",
+ "windows-sys",
+]
+
+[[package]]
+name = "rustversion"
+version = "1.0.23"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf54715a573b99ac80df0bc206da022bcd442c974952c7b9720069370852e21f"
+
+[[package]]
+name = "same-file"
+version = "1.0.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502"
+dependencies = [
+ "winapi-util",
+]
+
+[[package]]
+name = "scroll"
+version = "0.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c1257cd4248b4132760d6524d6dda4e053bc648c9070b960929bf50cfb1e7add"
+dependencies = [
+ "scroll_derive",
+]
+
+[[package]]
+name = "scroll_derive"
+version = "0.13.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e1a36a382ed65dbcc0ab47fd5e9a94112417ccd34560a392ef3b7b0f0ec39148"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 3.0.4",
+]
+
+[[package]]
+name = "senbei-crypto"
+version = "1.3.1"
+dependencies = [
+ "aes",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-elf"
+version = "1.3.1"
+dependencies = [
+ "goblin",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-engine"
+version = "1.3.1"
+dependencies = [
+ "memmap2",
+ "senbei-crypto",
+ "senbei-elf",
+ "senbei-pe",
+ "serde",
+ "serde_json",
+ "sha2",
+ "tempfile",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-io"
+version = "1.3.1"
+dependencies = [
+ "anyhow",
+ "indicatif",
+ "libc",
+ "memmap2",
+ "owo-colors",
+ "senbei-crypto",
+ "senbei-elf",
+ "senbei-engine",
+ "senbei-metadata",
+ "senbei-pe",
+ "sha2",
+ "tempfile",
+ "walkdir",
+ "windows",
+ "zip",
+]
+
+[[package]]
+name = "senbei-metadata"
+version = "1.3.1"
+dependencies = [
+ "serde",
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-pe"
+version = "1.3.1"
+dependencies = [
+ "thiserror",
+]
+
+[[package]]
+name = "senbei-wasm"
+version = "1.3.1"
+dependencies = [
+ "console_error_panic_hook",
+ "senbei-engine",
+ "senbei-io",
+ "senbei-metadata",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "serde"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4148590afebada386688f18773da617792bf2ef03ffc1e4cbd2b1d45b023e0ba"
+dependencies = [
+ "serde_core",
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_core"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "67dca2c9c51e58a4791a4b1ed58308b39c64224d349a935ab5039aa360942a48"
+dependencies = [
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_derive"
+version = "1.0.229"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e7a5d71263a5a7d47b41f6b3f06ba276f10cc18b0931f1799f710578e2309348"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 3.0.4",
+]
+
+[[package]]
+name = "serde_json"
+version = "1.0.151"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c841b55ecdae098c80dcae9cf767f6f8a0c2cdb3416bbef72181df4d0fe73f14"
+dependencies = [
+ "itoa",
+ "memchr",
+ "serde",
+ "serde_core",
+ "zmij",
+]
+
+[[package]]
+name = "sha2"
+version = "0.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "446ba717509524cb3f22f17ecc096f10f4822d76ab5c0b9822c5f9c284e825f4"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "digest",
+]
+
+[[package]]
+name = "simd-adler32"
+version = "0.3.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3a219298ac11a56ea9a6d2120044824d6f01aeb034955e7af7bc16858527deea"
+
+[[package]]
+name = "slab"
+version = "0.4.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0c790de23124f9ab44544d7ac05d60440adc586479ce501c1d6d7da3cd8c9cf5"
+
+[[package]]
+name = "syn"
+version = "2.0.119"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "872831b642d1a07999a962a351ed35b955ea2cfc8f3862091e2a240a84f17297"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
+]
+
+[[package]]
+name = "syn"
+version = "3.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6275cddf4610d1775e6d1fe9469b2e77d0f39fd98fb7450901b821e0c53649f"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
+]
+
+[[package]]
+name = "tempfile"
+version = "3.27.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "32497e9a4c7b38532efcdebeef879707aa9f794296a4f0244f6f69e9bc8574bd"
+dependencies = [
+ "fastrand",
+ "getrandom",
+ "once_cell",
+ "rustix",
+ "windows-sys",
+]
+
+[[package]]
+name = "thiserror"
+version = "2.0.20"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ec86235f5fcc2a73650310756d2ac5b138a5780bbbdfae3eeccec992c435ba4f"
+dependencies = [
+ "thiserror-impl",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "2.0.20"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bc04cd3e1236dd4a98afca4569f2deb3f120e5422a4023be2cb683f8486292af"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 3.0.4",
+]
+
+[[package]]
+name = "typed-path"
+version = "0.12.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e28f89b80c87b8fb0cf04ab448d5dd0dd0ade2f8891bae878de66a75a28600e"
+
+[[package]]
+name = "typenum"
+version = "1.20.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6f5e870be6c3b371b77fe0ee0bafb859fa4964b4404c27de1d380043c4dda20"
+
+[[package]]
+name = "unicode-ident"
+version = "1.0.24"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6e4313cd5fcd3dad5cafa179702e2b244f760991f45397d14d4ebf38247da75"
+
+[[package]]
+name = "unicode-width"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b4ac048d71ede7ee76d585517add45da530660ef4390e49b098733c6e897f254"
+
+[[package]]
+name = "unit-prefix"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "81e544489bf3d8ef66c953931f56617f423cd4b5494be343d9b9d3dda037b9a3"
+
+[[package]]
+name = "walkdir"
+version = "2.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "29790946404f91d9c5d06f9874efddea1dc06c5efe94541a7d6863108e3a5e4b"
+dependencies = [
+ "same-file",
+ "winapi-util",
+]
+
+[[package]]
+name = "wasm-bindgen"
+version = "0.2.127"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1b70935747edd64d89de3efa29d73789b806c15798f8e7dca4d8ac356b50ce70"
+dependencies = [
+ "cfg-if",
+ "once_cell",
+ "rustversion",
+ "wasm-bindgen-macro",
+ "wasm-bindgen-shared",
+]
+
+[[package]]
+name = "wasm-bindgen-macro"
+version = "0.2.127"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "77775f8f3f7217702089053b94958f8f54061a3f663417df76e19cbdcca29bc1"
+dependencies = [
+ "quote",
+ "wasm-bindgen-macro-support",
+]
+
+[[package]]
+name = "wasm-bindgen-macro-support"
+version = "0.2.127"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e11d33f857dc2fb11b8bc75aee111aa9cbeb12cd9f25efd3d4c2a3dd4e235284"
+dependencies = [
+ "bumpalo",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.119",
+ "wasm-bindgen-shared",
+]
+
+[[package]]
+name = "wasm-bindgen-shared"
+version = "0.2.127"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7ef64dbcc55df09c7e5a46182d181c2cfa3e925f3da937ea764728b4bbb9dcbf"
+dependencies = [
+ "unicode-ident",
+]
+
+[[package]]
+name = "web-time"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5a6580f308b1fad9207618087a65c04e7a10bc77e02c8e84e9b00dd4b12fa0bb"
+dependencies = [
+ "js-sys",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "winapi-util"
+version = "0.1.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
+dependencies = [
+ "windows-sys",
+]
+
+[[package]]
+name = "windows"
+version = "0.62.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "527fadee13e0c05939a6a05d5bd6eec6cd2e3dbd648b9f8e447c6518133d8580"
+dependencies = [
+ "windows-collections",
+ "windows-core",
+ "windows-future",
+ "windows-numerics",
+]
+
+[[package]]
+name = "windows-collections"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "23b2d95af1a8a14a3c7367e1ed4fc9c20e0a26e79551b1454d72583c97cc6610"
+dependencies = [
+ "windows-core",
+]
+
+[[package]]
+name = "windows-core"
+version = "0.62.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b8e83a14d34d0623b51dce9581199302a221863196a1dde71a7663a4c2be9deb"
+dependencies = [
+ "windows-implement",
+ "windows-interface",
+ "windows-link",
+ "windows-result",
+ "windows-strings",
+]
+
+[[package]]
+name = "windows-future"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e1d6f90251fe18a279739e78025bd6ddc52a7e22f921070ccdc67dde84c605cb"
+dependencies = [
+ "windows-core",
+ "windows-link",
+ "windows-threading",
+]
+
+[[package]]
+name = "windows-implement"
+version = "0.60.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "053e2e040ab57b9dc951b72c264860db7eb3b0200ba345b4e4c3b14f67855ddf"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.119",
+]
+
+[[package]]
+name = "windows-interface"
+version = "0.59.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3f316c4a2570ba26bbec722032c4099d8c8bc095efccdc15688708623367e358"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.119",
+]
+
+[[package]]
+name = "windows-link"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f0805222e57f7521d6a62e36fa9163bc891acd422f971defe97d64e70d0a4fe5"
+
+[[package]]
+name = "windows-numerics"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6e2e40844ac143cdb44aead537bbf727de9b044e107a0f1220392177d15b0f26"
+dependencies = [
+ "windows-core",
+ "windows-link",
+]
+
+[[package]]
+name = "windows-result"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7781fa89eaf60850ac3d2da7af8e5242a5ea78d1a11c49bf2910bb5a73853eb5"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-strings"
+version = "0.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7837d08f69c77cf6b07689544538e017c1bfcf57e34b4c0ff58e6c2cd3b37091"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-sys"
+version = "0.61.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ae137229bcbd6cdf0f7b80a31df61766145077ddf49416a728b02cb3921ff3fc"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-threading"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3949bd5b99cafdf1c7ca86b43ca564028dfe27d66958f2470940f73d86d75b37"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "zip"
+version = "8.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2d04a6b5381502aa6087c94c669499eb1602eb9c5e8198e534de571f7154809b"
+dependencies = [
+ "crc32fast",
+ "flate2",
+ "indexmap",
+ "memchr",
+ "typed-path",
+ "zopfli",
+]
+
+[[package]]
+name = "zlib-rs"
+version = "0.6.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34b31d188d9d685a4f9c7b46d6e36631b07058d2cfe190267adce54dc230bf12"
+
+[[package]]
+name = "zmij"
+version = "1.0.23"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "29666d0abbfad1e3dc4dcf6144730dd3a3ab225bbbdac83319345b1b44ccfc1b"
+
+[[package]]
+name = "zopfli"
+version = "0.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f05cd8797d63865425ff89b5c4a48804f35ba0ce8d125800027ad6017d2b5249"
+dependencies = [
+ "bumpalo",
+ "crc32fast",
+ "log",
+ "simd-adler32",
+]
@@ -0,0 +1,22 @@
+[package]
+name = "senbei-wasm"
+version = "1.3.1"
+edition = "2024"
+rust-version = "1.98.1"
+description = "WebAssembly bindings for senbei (browser frontend assets live in web/)"
+license = "AGPL-3.0-only"
+
+[lib]
+crate-type = ["cdylib"]
+
+[dependencies]
+senbei-io = { path = "../senbei-io" }
+senbei-metadata = { path = "../senbei-metadata" }
+senbei-engine = { path = "../senbei-engine" }
+wasm-bindgen = "0.2"
+console_error_panic_hook = "0.1"
+
+[profile.release]
+opt-level = "z"
+lto = true
+codegen-units = 1
@@ -0,0 +1,662 @@
+                    GNU AFFERO GENERAL PUBLIC LICENSE
+                       Version 3, 19 November 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU Affero General Public License is a free, copyleft license for
+software and other kinds of works, specifically designed to ensure
+cooperation with the community in the case of network server software.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+our General Public Licenses are intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  Developers that use our General Public Licenses protect your rights
+with two steps: (1) assert copyright on the software, and (2) offer
+you this License which gives you legal permission to copy, distribute
+and/or modify the software.
+
+  A secondary benefit of defending all users' freedom is that
+improvements made in alternate versions of the program, if they
+receive widespread use, become available for other developers to
+incorporate.  Many developers of free software are heartened and
+encouraged by the resulting cooperation.  However, in the case of
+software used on network servers, this result may fail to come about.
+The GNU General Public License permits making a modified version and
+letting the public access it on a server without ever releasing its
+source code to the public.
+
+  The GNU Affero General Public License is designed specifically to
+ensure that, in such cases, the modified source code becomes available
+to the community.  It requires the operator of a network server to
+provide the source code of the modified version running there to the
+users of that server.  Therefore, public use of a modified version, on
+a publicly accessible server, gives the public access to the source
+code of the modified version.
+
+  An older license, called the Affero General Public License and
+published by Affero, was designed to accomplish similar goals.  This is
+a different license, not a version of the Affero GPL, but Affero has
+released a new version of the Affero GPL which permits relicensing under
+this license.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU Affero General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Remote Network Interaction; Use with the GNU General Public License.
+
+  Notwithstanding any other provision of this License, if you modify the
+Program, your modified version must prominently offer all users
+interacting with it remotely through a computer network (if your version
+supports such interaction) an opportunity to receive the Corresponding
+Source of your version by providing access to the Corresponding Source
+from a network server at no charge, through some standard or customary
+means of facilitating copying of software.  This Corresponding Source
+shall include the Corresponding Source for any work covered by version 3
+of the GNU General Public License that is incorporated pursuant to the
+following paragraph.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the work with which it is combined will remain governed by version
+3 of the GNU General Public License.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU Affero General Public License from time to time.  Such new versions
+will be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU Affero General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU Affero General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU Affero General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU Affero General Public License as published
+    by the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU Affero General Public License for more details.
+
+    You should have received a copy of the GNU Affero General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If your software can interact with users remotely through a computer
+network, you should also make sure that it provides a way for users to
+get its source.  For example, if your program is a web application, its
+interface could display a "Source" link that leads users to an archive
+of the code.  There are many ways you could offer source, and different
+solutions will be better for different programs; see section 13 for the
+specific requirements.
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU AGPL, see
+<https://www.gnu.org/licenses/>.
+
@@ -101,12 +101,12 @@ impl MetadataResult {
    }
 }

-fn kind_str(kind: senbei::unpacker::Kind) -> &'static str {
+fn kind_str(kind: senbei_engine::Kind) -> &'static str {
    match kind {
-        senbei::unpacker::Kind::NativeExe => "native-exe",
-        senbei::unpacker::Kind::ManagedExe => "managed-exe",
-        senbei::unpacker::Kind::NativeDll => "native-dll",
-        senbei::unpacker::Kind::ManagedDll => "managed-dll",
+        senbei_engine::Kind::NativeExe => "native-exe",
+        senbei_engine::Kind::ManagedExe => "managed-exe",
+        senbei_engine::Kind::NativeDll => "native-dll",
+        senbei_engine::Kind::ManagedDll => "managed-dll",
    }
 }

@@ -117,10 +117,10 @@ fn kind_str(kind: senbei::unpacker::Kind) -> &'static str {
 /// anything unrecognized.
 #[wasm_bindgen]
 pub fn detect(input: &[u8]) -> Option<String> {
-    if senbei::metadata::is_metadata(input) {
+    if senbei_metadata::is_metadata(input) {
        return Some("metadata".to_string());
    }
-    senbei::unpacker::detect(input).map(|d| kind_str(d.kind).to_string())
+    senbei_engine::detect(input).map(|d| kind_str(d.kind).to_string())
 }

 /// Unpack a protected module.
@@ -134,7 +134,7 @@ pub fn unpack_file(
    input: &[u8],
    companion: Option<Vec<u8>>,
 ) -> Result<UnpackResult, JsError> {
-    let r = senbei::job::unpack_bytes(input, companion.as_deref())
+    let r = senbei_io::job::unpack_bytes(input, companion.as_deref())
        .map_err(|e| JsError::new(&e.to_string()))?;
    Ok(UnpackResult {
        kind: kind_str(r.kind).to_string(),
@@ -153,7 +153,7 @@ pub fn unpack_file(
 #[wasm_bindgen]
 pub fn deobfuscate_metadata(data: &[u8]) -> Result<MetadataResult, JsError> {
    let (bytes, report) =
-        senbei::metadata::deobfuscate(data).map_err(|e| JsError::new(&e.to_string()))?;
+        senbei_metadata::deobfuscate(data).map_err(|e| JsError::new(&e.to_string()))?;
    Ok(MetadataResult {
        bytes,
        version: report.version,
@@ -164,14 +164,14 @@ pub fn deobfuscate_metadata(data: &[u8]) -> Result<MetadataResult, JsError> {
 }

 /// Unpack a protected module, forcing the EXE pipeline (no DLL-pipeline
-/// probe). See [`senbei::job::unpack_bytes_force_exe`] for why the web app
+/// probe). See [`senbei_io::job::unpack_bytes_force_exe`] for why the web app
 /// needs this recovery path.
 #[wasm_bindgen]
 pub fn unpack_file_force_exe(
    input: &[u8],
    companion: Option<Vec<u8>>,
 ) -> Result<UnpackResult, JsError> {
-    let r = senbei::job::unpack_bytes_force_exe(input, companion.as_deref())
+    let r = senbei_io::job::unpack_bytes_force_exe(input, companion.as_deref())
        .map_err(|e| JsError::new(&e.to_string()))?;
    Ok(UnpackResult {
        kind: kind_str(r.kind).to_string(),
@@ -1,7 +0,0 @@
-pub mod job;
-pub mod logfile;
-pub mod metadata;
-pub mod pause;
-pub mod scan;
-pub mod ui;
-pub mod unpacker;
@@ -1,10 +0,0 @@
-//! Shared test fixtures.
-#![allow(dead_code)]
-
-use std::path::PathBuf;
-
-/// Path to `senbei/samples` — the user-managed corpus dropped in by hand.
-/// Git-ignored except its README; tests here run against whatever is present.
-pub fn samples_dir() -> PathBuf {
-    PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("samples")
-}
@@ -1,441 +0,0 @@
-# This file is automatically @generated by Cargo.
-# It is not intended for manual editing.
-version = 4
-
-[[package]]
-name = "anyhow"
-version = "1.0.104"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "330a5ed07fa54e4702c9d6c4174f74427fc0ef6e214bbd677ae50a5099946470"
-
-[[package]]
-name = "bumpalo"
-version = "3.20.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "72f5acc6cb2ba439de613abc23857ec3d78374d8ed5ac84e9d11336e87da8649"
-
-[[package]]
-name = "cfg-if"
-version = "1.0.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
-
-[[package]]
-name = "console"
-version = "0.16.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4fe5f465a4f6fee88fad41b85d990f84c835335e85b5d9e6e63e0d06d28cba7c"
-dependencies = [
- "encode_unicode",
- "libc",
- "unicode-width",
- "windows-sys",
-]
-
-[[package]]
-name = "console_error_panic_hook"
-version = "0.1.7"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a06aeb73f470f66dcdbf7223caeebb85984942f22f1adb2a088cf9668146bbbc"
-dependencies = [
- "cfg-if",
- "wasm-bindgen",
-]
-
-[[package]]
-name = "encode_unicode"
-version = "1.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "34aa73646ffb006b8f5147f3dc182bd4bcb190227ce861fc4a4844bf8e3cb2c0"
-
-[[package]]
-name = "futures-core"
-version = "0.3.33"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2cd50c473c80f6d7c3670a752354b8e569b1a7cbfdc0419ec88e5edad85e0dc7"
-
-[[package]]
-name = "futures-task"
-version = "0.3.33"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b231ed28831efb4a61a08580c4bc233ec56bc009f4cd8f52da2c3cb97df0c109"
-
-[[package]]
-name = "futures-util"
-version = "0.3.33"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a77a90a256fce34da66415271e30f94ee91c57b04b8a2c042d9cf3220179deaa"
-dependencies = [
- "futures-core",
- "futures-task",
- "pin-project-lite",
- "slab",
-]
-
-[[package]]
-name = "indicatif"
-version = "0.18.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9433806cd6b4ec1aba79c021c7e4c58fb4c3b9977c085062e611ac929998fb0c"
-dependencies = [
- "console",
- "portable-atomic",
- "unicode-width",
- "unit-prefix",
- "web-time",
-]
-
-[[package]]
-name = "js-sys"
-version = "0.3.103"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53b44bfcdb3f8d5837a46dae1ca9660a837176eee74a28b229bc626816589102"
-dependencies = [
- "cfg-if",
- "futures-util",
- "wasm-bindgen",
-]
-
-[[package]]
-name = "libc"
-version = "0.2.189"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3eaf3ede3fee6db1a4c2ee091bf8a8b4dccdc6d17f656fb07896ee72867612f2"
-
-[[package]]
-name = "once_cell"
-version = "1.21.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"
-
-[[package]]
-name = "owo-colors"
-version = "4.3.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d211803b9b6b570f68772237e415a029d5a50c65d382910b879fb19d3271f94d"
-
-[[package]]
-name = "pin-project-lite"
-version = "0.2.17"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"
-
-[[package]]
-name = "portable-atomic"
-version = "1.14.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d20d5497ef88037a52ff98267d066e7f11fcc5e99bbfbd58a42336193aacec3"
-
-[[package]]
-name = "proc-macro2"
-version = "1.0.107"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "985e7ec9bb745e6ce6535b544d84d6cd6f7ad8bd711c398938ae983b91a766d9"
-dependencies = [
- "unicode-ident",
-]
-
-[[package]]
-name = "quote"
-version = "1.0.47"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1fbf4db142a473a8d80c26bbf18454ed458bf8d26c8219c331daecfdbd079001"
-dependencies = [
- "proc-macro2",
-]
-
-[[package]]
-name = "rustversion"
-version = "1.0.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cf54715a573b99ac80df0bc206da022bcd442c974952c7b9720069370852e21f"
-
-[[package]]
-name = "same-file"
-version = "1.0.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502"
-dependencies = [
- "winapi-util",
-]
-
-[[package]]
-name = "senbei"
-version = "1.0.0"
-dependencies = [
- "anyhow",
- "indicatif",
- "libc",
- "owo-colors",
- "thiserror",
- "walkdir",
- "windows",
-]
-
-[[package]]
-name = "senbei-web"
-version = "1.0.0"
-dependencies = [
- "console_error_panic_hook",
- "senbei",
- "wasm-bindgen",
-]
-
-[[package]]
-name = "slab"
-version = "0.4.12"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0c790de23124f9ab44544d7ac05d60440adc586479ce501c1d6d7da3cd8c9cf5"
-
-[[package]]
-name = "syn"
-version = "2.0.119"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "872831b642d1a07999a962a351ed35b955ea2cfc8f3862091e2a240a84f17297"
-dependencies = [
- "proc-macro2",
- "quote",
- "unicode-ident",
-]
-
-[[package]]
-name = "syn"
-version = "3.0.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53e9bae58849f64dfa4f5d5ae372c8341f7305f82a3868709269343628b659a3"
-dependencies = [
- "proc-macro2",
- "quote",
- "unicode-ident",
-]
-
-[[package]]
-name = "thiserror"
-version = "2.0.19"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09a43598840e33d5b0331f38c5e30d13bb11c11210a4b58f0d9b18a5a5eefcd9"
-dependencies = [
- "thiserror-impl",
-]
-
-[[package]]
-name = "thiserror-impl"
-version = "2.0.19"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "43cbfe0cf76104d42a574802844187e84a305e531ed54455f11fbde0f10541cd"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 3.0.3",
-]
-
-[[package]]
-name = "unicode-ident"
-version = "1.0.24"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e6e4313cd5fcd3dad5cafa179702e2b244f760991f45397d14d4ebf38247da75"
-
-[[package]]
-name = "unicode-width"
-version = "0.2.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b4ac048d71ede7ee76d585517add45da530660ef4390e49b098733c6e897f254"
-
-[[package]]
-name = "unit-prefix"
-version = "0.5.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "81e544489bf3d8ef66c953931f56617f423cd4b5494be343d9b9d3dda037b9a3"
-
-[[package]]
-name = "walkdir"
-version = "2.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "29790946404f91d9c5d06f9874efddea1dc06c5efe94541a7d6863108e3a5e4b"
-dependencies = [
- "same-file",
- "winapi-util",
-]
-
-[[package]]
-name = "wasm-bindgen"
-version = "0.2.126"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4b067c0c11094aef6b7a801c1e34a26affafdf3d051dba08456b868789aaf9a4"
-dependencies = [
- "cfg-if",
- "once_cell",
- "rustversion",
- "wasm-bindgen-macro",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-macro"
-version = "0.2.126"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "167ce5e579f6bcf889c4f7175a8a5a585de84e8ff93976ce393efa5f2837aab1"
-dependencies = [
- "quote",
- "wasm-bindgen-macro-support",
-]
-
-[[package]]
-name = "wasm-bindgen-macro-support"
-version = "0.2.126"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f3997c7839262f4ef12cf90b818d6340c18e80f263f1a94bf157d0ec4420380e"
-dependencies = [
- "bumpalo",
- "proc-macro2",
- "quote",
- "syn 2.0.119",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-shared"
-version = "0.2.126"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dc1b4cb0cc549fcf58d7dfc081778139b3d283a081644e833e84682ad71cea24"
-dependencies = [
- "unicode-ident",
-]
-
-[[package]]
-name = "web-time"
-version = "1.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5a6580f308b1fad9207618087a65c04e7a10bc77e02c8e84e9b00dd4b12fa0bb"
-dependencies = [
- "js-sys",
- "wasm-bindgen",
-]
-
-[[package]]
-name = "winapi-util"
-version = "0.1.11"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
-dependencies = [
- "windows-sys",
-]
-
-[[package]]
-name = "windows"
-version = "0.62.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "527fadee13e0c05939a6a05d5bd6eec6cd2e3dbd648b9f8e447c6518133d8580"
-dependencies = [
- "windows-collections",
- "windows-core",
- "windows-future",
- "windows-numerics",
-]
-
-[[package]]
-name = "windows-collections"
-version = "0.3.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "23b2d95af1a8a14a3c7367e1ed4fc9c20e0a26e79551b1454d72583c97cc6610"
-dependencies = [
- "windows-core",
-]
-
-[[package]]
-name = "windows-core"
-version = "0.62.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b8e83a14d34d0623b51dce9581199302a221863196a1dde71a7663a4c2be9deb"
-dependencies = [
- "windows-implement",
- "windows-interface",
- "windows-link",
- "windows-result",
- "windows-strings",
-]
-
-[[package]]
-name = "windows-future"
-version = "0.3.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e1d6f90251fe18a279739e78025bd6ddc52a7e22f921070ccdc67dde84c605cb"
-dependencies = [
- "windows-core",
- "windows-link",
- "windows-threading",
-]
-
-[[package]]
-name = "windows-implement"
-version = "0.60.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "053e2e040ab57b9dc951b72c264860db7eb3b0200ba345b4e4c3b14f67855ddf"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.119",
-]
-
-[[package]]
-name = "windows-interface"
-version = "0.59.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3f316c4a2570ba26bbec722032c4099d8c8bc095efccdc15688708623367e358"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.119",
-]
-
-[[package]]
-name = "windows-link"
-version = "0.2.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f0805222e57f7521d6a62e36fa9163bc891acd422f971defe97d64e70d0a4fe5"
-
-[[package]]
-name = "windows-numerics"
-version = "0.3.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6e2e40844ac143cdb44aead537bbf727de9b044e107a0f1220392177d15b0f26"
-dependencies = [
- "windows-core",
- "windows-link",
-]
-
-[[package]]
-name = "windows-result"
-version = "0.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7781fa89eaf60850ac3d2da7af8e5242a5ea78d1a11c49bf2910bb5a73853eb5"
-dependencies = [
- "windows-link",
-]
-
-[[package]]
-name = "windows-strings"
-version = "0.5.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7837d08f69c77cf6b07689544538e017c1bfcf57e34b4c0ff58e6c2cd3b37091"
-dependencies = [
- "windows-link",
-]
-
-[[package]]
-name = "windows-sys"
-version = "0.61.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ae137229bcbd6cdf0f7b80a31df61766145077ddf49416a728b02cb3921ff3fc"
-dependencies = [
- "windows-link",
-]
-
-[[package]]
-name = "windows-threading"
-version = "0.2.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3949bd5b99cafdf1c7ca86b43ca564028dfe27d66958f2470940f73d86d75b37"
-dependencies = [
- "windows-link",
-]
@@ -1,19 +0,0 @@
-[package]
-name = "senbei-web"
-version = "1.0.1"
-edition = "2024"
-description = "WebAssembly browser frontend for senbei"
-license = "AGPL-3.0-only"
-
-[lib]
-crate-type = ["cdylib"]
-
-[dependencies]
-senbei = { path = ".." }
-wasm-bindgen = "0.2"
-console_error_panic_hook = "0.1"
-
-[profile.release]
-opt-level = "z"
-lto = true
-codegen-units = 1
@@ -1,75 +1,25 @@
-# Senbei web
+# Senbei Web

-Senbei running in the browser: the unpacker core compiled to WebAssembly,
-wrapped in a small static page. Everything is client-side — files are read
-into the page, unpacked locally, and offered back as downloads. Nothing is
-uploaded; there is no server component.
+Senbei runs in the browser through the `senbei-wasm` crate. Files are read locally, unpacked in a worker, and offered back as downloads; no server receives input bytes.

 ## Features

- A legal notice is shown as a blocking dialog on page open; the tool is
-  unusable until it is acknowledged.
- Dropped files land in a file list, not unpacked immediately: review the
-  batch, remove mistakes, then press **Unpack**. A module and its `._`
-  companion can be dropped in any order (or in separate drops) — companions
-  auto-pair by name (`Foo.dll._` → `Foo.dll`) and show as a badge on the
-  module's row; removing a module removes its companion too.
- Rows show state at a glance: black while staged, an animated blue bar
-  while unpacking, green on success (with a download button) and red on
-  failure.
- Drop one or more protected `.exe` / `.dll` modules → get `<name>.unpack.*`
-  downloads.
- Drop an il2cpp `global-metadata.dat` → de-obfuscated
-  `global-metadata.unpack.dat` (only when tokens actually change).
- Each output passes the same static integrity check as the CLI; suspect
-  outputs are flagged with the specific defects found.
+- Protected `.exe` and `.dll` files produce `<name>.unpack.*` downloads.
+- External `.exe._` and `.dll._` companions are paired by filename.
+- `global-metadata.dat` produces `global-metadata.unpack.dat` when tokens change.
+- Each output receives the same static integrity check as the CLI.

-## Architecture notes
+Every unpack uses a disposable Web Worker so a WebAssembly trap cannot freeze the page. A trapped DLL can be retried through the forced-EXE path, matching native routing.

- Every unpack runs in a **disposable Web Worker** (fresh wasm instance per
-  file): the UI stays responsive on 100 MB+ modules, and a wasm trap is
-  isolated to that worker.
- **Why workers matter for correctness:** the DLL-first routing probe relies
-  on `catch_unwind` to reject EXE-shell-layout DLLs, and panics cannot be
-  caught in WebAssembly — the probe traps the whole call. When a DLL unpack
-  traps, the app retries once in a new worker with the forced-EXE pipeline
-  (`unpack_file_force_exe`), reproducing the CLI's dll-first/exe-fallback
-  outcome. Spliced companion inputs skip the probe entirely (they are always
-  EXE-shell layout), exactly like the CLI.
- Rust panic messages are forwarded to the browser console
-  (`console_error_panic_hook`) — check devtools when reporting an issue.
-
-## Building
-
-Requires a Rust toolchain (`rust-toolchain.toml` in the repo root pins one,
-including the `wasm32-unknown-unknown` target) and
-[wasm-pack](https://rustwasm.github.io/wasm-pack/installer/).
+## Build

 ```cmd
-cd web
-wasm-pack build --target web --release
+cd senbei-wasm
+wasm-pack build --target web --release --out-dir ../web/pkg
 ```

-This produces `web/pkg/` (git-ignored). Then serve the `web/` directory with
-any static file server and open `index.html`:
-
-```cmd
-python -m http.server -d web 8000
-:: -> http://localhost:8000
-```
-
-(Opening `index.html` via `file://` won't work — ES modules require HTTP.)
+Serve `web/` with a static HTTP server, for example `python -m http.server -d web 8000`. Opening `index.html` with `file://` does not work because browser modules require HTTP.

 ## Layout

-```
-web/
-├── Cargo.toml     senbei-web cdylib crate (depends on the senbei lib)
-├── src/lib.rs     #[wasm_bindgen] bindings: detect / unpack_file /
-│                  unpack_file_force_exe / deobfuscate_metadata
-├── index.html     the page
-├── app.js         dropzone, file list, worker orchestration, downloads
-├── worker.js      one-shot unpack worker (fresh wasm instance per file)
-├── style.css
-└── pkg/           wasm-pack output (git-ignored)
-```
+`senbei-wasm/src/lib.rs` contains the bindings. `web/app.js` manages the dropzone and downloads, `web/worker.js` runs one unpack job per worker, and `web/pkg/` contains ignored wasm-pack output.
@@ -1,4 +1,4 @@
-import init, { detect, deobfuscate_metadata } from './pkg/senbei_web.js';
+import init, { detect, deobfuscate_metadata } from './pkg/senbei_wasm.js';

 const dropzone = document.getElementById('dropzone');
 const picker = document.getElementById('picker');
@@ -50,6 +50,22 @@ const KIND_LABEL = {
  'native-dll': 'protected native DLL',
  'managed-dll': 'protected managed DLL',
  metadata: 'il2cpp metadata',
+  'android-package':
+    'Android app package — the web build cannot unpack these yet; use the senbei CLI',
+  'android-so':
+    'Android AArch64 library — the web build cannot unpack these yet; use the senbei CLI',
+};
+
+// Android targets are recognized by extension so the row can explain the
+// situation instead of reporting a protected file as unrecognized: the
+// Android pipeline is filesystem orchestration (senbei-io) with no wasm
+// build, so these files need the CLI. The pseudo-kinds are labels only —
+// they never reach the worker.
+const ANDROID_KIND = {
+  apk: 'android-package',
+  apks: 'android-package',
+  xapk: 'android-package',
+  so: 'android-so',
 };

 const COMPANION_SVG =
@@ -91,8 +107,13 @@ async function stageFiles(list) {
    // the PE header fields); read a small slice, not the whole file.
    const head = new Uint8Array(await file.slice(0, 65536).arrayBuffer());
    // Companions are ciphertext fragments; detect() only makes sense on the
-    // base module, so skip it for `._` files.
-    const kind = file.name.endsWith('._') ? undefined : detect(head);
+    // base module, so skip it for `._` files. Android targets short-circuit
+    // detect() as well: an ELF/zip never classifies as a protected PE, and
+    // the row must carry the android pseudo-kind for its status line.
+    const ext = file.name.slice(file.name.lastIndexOf('.') + 1).toLowerCase();
+    const kind = file.name.endsWith('._')
+      ? undefined
+      : (ANDROID_KIND[ext] ?? detect(head));
    const old = files.get(file.name);
    files.set(file.name, {
      file,
@@ -284,7 +305,10 @@ function render() {

  const unpackable = [...files].some(
    ([name, e]) =>
-      !name.endsWith('._') && e.kind !== undefined && e.state === 'staged',
+      !name.endsWith('._') &&
+      e.kind !== undefined &&
+      !e.kind.startsWith('android-') &&
+      e.state === 'staged',
  );
  unpackBtn.disabled = !unpackable;
  actions.hidden = files.size === 0;
@@ -364,6 +388,10 @@ unpackBtn.addEventListener('click', async () => {
        continue;
      }

+      // Android rows are informational only (no wasm pipeline); their staged
+      // status line already says to use the CLI.
+      if (entry.kind.startsWith('android-')) continue;
+
      entry.state = 'working';
      render();
      try {
@@ -56,7 +56,9 @@
    <p class="hint">
      Protected <code>.exe</code> / <code>.dll</code> modules, optional
      <code>._</code> companions, or an il2cpp
-      <code>global-metadata.dat</code>.
+      <code>global-metadata.dat</code>. Android packages
+      (<code>.apk</code> / <code>.apks</code> / <code>.xapk</code>) and
+      <code>.so</code> libraries are recognized but need the CLI.
    </p>
    <input type="file" id="picker" multiple hidden>
  </div>
@@ -333,6 +333,26 @@ footer {
 }
 footer a { color: var(--dim); }

+/* --- narrow screens --- */
+
+/* On a narrow viewport the status column (flex: 1, right-aligned) is
+   squeezed by a long file name into a one-word-per-line vertical strip.
+   Stack the row instead: name + icons on the first line, the status on its
+   own full-width line below, left-aligned. */
+@media (max-width: 560px) {
+  main { padding: 1.5rem 0.9rem 2.5rem; }
+
+  #dropzone { padding: 1.75rem 1rem; }
+
+  .file .row { flex-wrap: wrap; }
+  .file .name { flex: 1; min-width: 0; }
+  .file .status {
+    order: 6; /* after the download/remove icons */
+    flex-basis: 100%;
+    text-align: left;
+  }
+}
+
 /* --- animations --- */

@keyframes fadeIn {
@@ -10,7 +10,7 @@
 //    instance is never reused). That reproduces the CLI's
 //    dll-first/exe-fallback routing without a catchable panic.

-import init, { unpack_file, unpack_file_force_exe } from './pkg/senbei_web.js';
+import init, { unpack_file, unpack_file_force_exe } from './pkg/senbei_wasm.js';

 let ready = null;
Author	SHA1	Message	Date
Momoko-Ayase	ba380a5774	docs: point README at GitBook and restore the legal-notice heading [skip ci] Product docs already live on crackproof-research; keep a minimal README (usage, legal notice, license) so CI can still extract LEGAL-NOTICE.md.	2026-09-15 23:27:24 +08:00
Momoko-Ayase	d6224e639c	chore: bump version to 1.3.1	2026-09-10 20:58:44 +08:00
Momoko-Ayase	9b8024b132	fix(windows): re-arm neutralized TLS field relocations for PE32 DLLs The packer demotes the four base-relocation entries covering the TLS directory's VA fields to IMAGE_REL_BASED_ABSOLUTE padding, because its own loader fixes TLS up by hand. Restored verbatim, a DLL mapped off its preferred base keeps stale VAs in its TLS directory and the OS loader faults in LdrpAllocateTlsEntry while writing the TLS slot index through the unrelocated AddressOfIndex (observed as a 0xc0000005 startup failure). Walk the restored BaseReloc blocks and promote those entries back to IMAGE_REL_BASED_HIGHLOW, gated on is_dll so EXE output stays byte-identical.	2026-09-10 20:42:39 +08:00
Momoko-Ayase	512a627066	web: recognize Android targets and point users at the CLI Dropping an .apk/.apks/.xapk package or a .so library reported "not recognized — will be skipped", which reads as "not protected" when the real gap is that the Android pipeline (filesystem orchestration in senbei-io) has no wasm build. These files now get an explicit android pseudo-kind at staging time: the row explains that the CLI handles them, and they neither enable the Unpack button nor error out mid-run.	2026-09-10 09:32:58 +08:00
Momoko-Ayase	a230c37281	web: stack file-row status below the name on narrow screens On phone-width viewports the right-aligned status column shared a flex line with a long file name and wrapped one word per line. Below 560px the row now wraps: name and action icons keep the first line and the status takes a full-width line underneath.	2026-09-10 09:29:23 +08:00
Momoko-Ayase	4f59e25652	ci: declare rustfmt/clippy components in rust-toolchain.toml The 1.98.1 pin installs without rustfmt and clippy, so the fmt and clippy jobs fail at the shim before running. Listing the components in the toolchain file makes rustup auto-install them on every host, including the CI runners that invoke cargo with no setup action.	2026-09-10 09:28:50 +08:00
bfloat16	f862633512	chore: bump version to 1.3.0	2026-09-07 22:59:25 +08:00
bfloat16	ed2731f8e0	fix(windows): restore managed companion DLLs	2026-09-07 21:43:54 +08:00
bfloat16	53ef36c837	build: require Rust 1.98.1	2026-09-07 21:43:15 +08:00
bfloat16	6250ca4e98	refactor: align platform crate boundaries	2026-09-07 19:29:54 +08:00
bfloat16	aa1bcaa2eb	fix(android): support compact ELF dynamic table layouts	2026-09-07 16:31:38 +08:00
bfloat16	2d92360d87	fix(elf): validate section names from ELF string table	2026-09-07 13:23:16 +08:00
bfloat16	4d73406ab1	fix(metadata): support Android v29 method layouts	2026-09-07 00:43:34 +08:00
bfloat16	776d246065	fix(scan): stream Android package targets	2026-09-06 22:25:41 +08:00
bfloat16	d436a200ba	refactor: consolidate platform engines into senbei-engine	2026-09-06 19:31:19 +08:00
Momoko-Ayase	cbfacbc31f	Upgrade dependencies to latest stable zip 0.6.6 -> 8.6.0 (the 0.6 line is unmaintained), aes 0.8 -> 0.9 (BlockCipherDecrypt trait replaces BlockDecrypt), sha2 0.10 -> 0.11 (Array no longer formats as hex; local hex_digest helpers). Outputs are byte-identical across the upgrade: full golden corpus and Android corpus sidecars all pass.	2026-09-02 03:19:06 +08:00
Momoko-Ayase	d3dd1a8ff8	Merge Android (AArch64) shared-library restoration, bump to 1.2.0 Adds the Android protection-scheme pipeline: hollowed ELF64/AArch64 libraries are restored statically (stage-1/stage-2 module extraction, container decode, dynamic-linker table rebuild), with app-package (.apk/.apks/.xapk) container handling, cross-source content dedup, and il2cpp metadata support for the Android variants (seeded RID permutation; embedded XOR-wrapped blob extraction). The single senbei CLI now routes single .so files, packages, and folders by content; outputs follow the existing .unpack-infix naming under <root>/unpack or --out. PE behavior is unchanged (35/35 goldens).	2026-09-02 03:09:21 +08:00
Momoko-Ayase	0c3f29f93a	Split web/ into senbei-wasm crate + static assets The Rust bindings move from web/src to a top-level senbei-wasm crate (still outside the workspace, own Cargo.lock), matching the other senbei-* crates. web/ keeps only the static frontend; wasm-pack emits the JS/wasm package into web/pkg/ via --out-dir. JS glue renamed senbei_web -> senbei_wasm with the crate.	2026-08-30 22:44:15 +08:00
Momoko-Ayase	f80749ffd6	Merge bfloat16-senbei workspace restructure, bump to 1.1.0 Adopts the fork's workspace split (senbei-cli / senbei-crypto / senbei-io / senbei-metadata / senbei-pe), its structured error taxonomy, entry-transform and layout validation, PE32 dd8 key-formula selection with a skip floor, the CRT entry-stub dd8 oracle, and the extensionless-file scan skip. Kept from senbei on top of the restructure: - ManagedExe detection/routing and the CLR (COR20 + BSJB) metadata restore in the EXE pipeline. - The RET+int3 padding fingerprint as the primary dd8 padding signal, ahead of the mutated-position 0xCC fallback. - docs/, .github/, samples/, tests/ (moved to senbei-cli/tests), and the web/ wasm frontend (rewired to the split crates), all of which the fork had dropped. - The fork's README compatibility matrix is not taken: it names real games, which the public-repo hygiene rules forbid. - The wasm32 localtime fallback in logfile and unpack_bytes_force_exe (the web app's trap-recovery entry point), both lost in the restructure. Golden corpus: 35/35 byte-identical. clippy -D warnings clean; wasm32 check clean for the full workspace.	2026-08-30 22:31:21 +08:00
bfloat16	35076848b6	chore: ignore local test corpus	2026-08-16 03:49:11 +08:00
bfloat16	18886272e5	docs: note embedded metadata compatibility	2026-08-16 03:43:12 +08:00
bfloat16	b534de872d	refactor: move library implementations into modules	2026-08-16 03:38:42 +08:00
bfloat16	a9aaf95e01	docs: add Android compatibility matrix	2026-08-16 03:31:25 +08:00
bfloat16	7f827d6400	feat: add folder-based Android unpack workflow	2026-08-16 02:38:07 +08:00
bfloat16	131ced6db5	feat: add static stage extraction and metadata detection	2026-08-16 01:47:25 +08:00
bfloat16	b1d3699df3	feat: add static Android il2cpp restoration	2026-08-16 00:34:08 +08:00
bfloat16	9433b4dcca	Initial commit	2026-08-15 22:20:03 +08:00
bfloat16	763bdbb21f	docs(readme): organize compatibility matrix by game	2026-08-13 10:49:44 +08:00
bfloat16	c5982a64b9	fix(unpacker): classify checksum range failures	2026-08-13 10:49:07 +08:00
bfloat16	55a31a2371	doc(none): add README.md	2026-08-12 23:57:00 +08:00
bfloat16	caadbd5325	fix(unpacker): validate executable entry transforms	2026-08-11 19:24:45 +08:00
bfloat16	ab1a14c9d1	perf(scan): skip extensionless files by default	2026-08-11 19:24:32 +08:00
bfloat16	e9ead4dc5f	refactor: init	2026-08-11 14:19:56 +08:00
bfloat16	a89900a812	fix(unpacker): refine layout validation and diagnostics	2026-08-11 11:42:12 +08:00
bfloat16	67178d34af	feat: Optimize error reporting	2026-08-11 10:51:37 +08:00