refactor: consolidate platform engines into senbei-engine

This commit is contained in:
bfloat16
2026-09-06 19:31:19 +08:00
parent cbfacbc31f
commit d436a200ba
66 changed files with 1148 additions and 1012 deletions
+20 -78
View File
@@ -1,98 +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 and protected
Android (AArch64) shared libraries: a Cargo workspace with a pure, panic-free,
no-I/O PE unpacker core (`senbei-pe/`, built on `senbei-crypto/`), il2cpp
metadata de-obfuscators (`senbei-metadata/` for the Windows structural
variant, `senbei-android-metadata/` for the Android seeded-permutation and
embedded-blob variants), the native-only Android pipeline
(`senbei-android-crypto/`, `senbei-android-engine/`, `senbei-android-elf/`),
filesystem/CLI orchestration (`senbei-io/`, including the Android
single-library/package glue in `senbei-io/src/android.rs`), the `senbei`
binary (`senbei-cli/`), WebAssembly bindings (`senbei-wasm/`, outside the
workspace; builds into `web/pkg/`), and the static browser frontend assets
(`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 `docs/design.md` before changing architecture or pipeline boundaries.
## Commands
```cmd
cargo build --release :: CLI (default member: senbei-cli)
cargo test --release --workspace :: full suite (golden corpus: samples/, git-ignored)
cargo build --release
cargo test --release --workspace
cargo clippy --workspace --all-targets -- -D warnings
cargo fmt --all
cd senbei-wasm && wasm-pack build --target web --release --out-dir ../web/pkg :: 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). The
Android corpus lives in `samples/android/` (one extracted app tree per
subdirectory) and is covered by `tests/android_samples.rs`;
`SENBEI_ANDROID_SAMPLES` overrides that location. 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 PE unpacker core stays pure**: `senbei-pe` and `senbei-crypto` have no
file I/O, no `unsafe`, no panics across the public boundary, no
platform-specific code. Everything `senbei-wasm` compiles must keep building
for `wasm32-unknown-unknown` (`cargo check --target wasm32-unknown-unknown`
at the workspace root covers it — the Android crates do compile to wasm, but
nothing on the wasm path calls them).
- **The Android crates are native-only orchestration-style crates**:
`senbei-android-engine`/`senbei-android-elf` memory-map inputs and write a
module workspace to disk (the restore is a two-phase design consuming that
workspace). Keep them off the web app's code paths; `senbei-io`'s
`android.rs` is the only caller the CLI uses.
- **`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 basic format parsing and address mapping only. `senbei-engine/src/windows/` contains the PE unpacking pipeline; `senbei-engine/src/android/` contains Android extraction and ELF restoration. `senbei-crypto/src/android/` and `senbei-metadata/src/android/` contain Android-specific primitives; Windows metadata code is under `senbei-metadata/src/windows/`. 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. Update `docs/usage.md` when CLI behavior changes.
Generated
+29 -48
View File
@@ -418,53 +418,11 @@ dependencies = [
"syn 3.0.4",
]
[[package]]
name = "senbei-android-crypto"
version = "1.2.0"
dependencies = [
"aes",
"thiserror",
]
[[package]]
name = "senbei-android-elf"
version = "1.2.0"
dependencies = [
"memmap2",
"senbei-android-crypto",
"serde",
"serde_json",
"sha2",
"tempfile",
"thiserror",
]
[[package]]
name = "senbei-android-engine"
version = "1.2.0"
dependencies = [
"goblin",
"memmap2",
"senbei-android-crypto",
"serde",
"serde_json",
"sha2",
"tempfile",
"thiserror",
]
[[package]]
name = "senbei-android-metadata"
version = "1.2.0"
dependencies = [
"serde",
"thiserror",
]
[[package]]
name = "senbei-cli"
version = "1.2.0"
dependencies = [
"senbei-engine",
"senbei-io",
"senbei-metadata",
"sha2",
@@ -475,6 +433,29 @@ dependencies = [
name = "senbei-crypto"
version = "1.2.0"
dependencies = [
"aes",
"thiserror",
]
[[package]]
name = "senbei-elf"
version = "1.2.0"
dependencies = [
"goblin",
"thiserror",
]
[[package]]
name = "senbei-engine"
version = "1.2.0"
dependencies = [
"goblin",
"memmap2",
"senbei-crypto",
"serde",
"serde_json",
"sha2",
"tempfile",
"thiserror",
]
@@ -487,11 +468,8 @@ dependencies = [
"indicatif",
"libc",
"owo-colors",
"senbei-android-elf",
"senbei-android-engine",
"senbei-android-metadata",
"senbei-engine",
"senbei-metadata",
"senbei-pe",
"sha2",
"tempfile",
"walkdir",
@@ -502,12 +480,15 @@ dependencies = [
[[package]]
name = "senbei-metadata"
version = "1.2.0"
dependencies = [
"serde",
"thiserror",
]
[[package]]
name = "senbei-pe"
version = "1.2.0"
dependencies = [
"senbei-crypto",
"thiserror",
]
+4 -8
View File
@@ -1,11 +1,9 @@
[workspace]
members = [
"senbei-android-crypto",
"senbei-android-elf",
"senbei-android-engine",
"senbei-android-metadata",
"senbei-cli",
"senbei-crypto",
"senbei-elf",
"senbei-engine",
"senbei-io",
"senbei-metadata",
"senbei-pe",
@@ -43,11 +41,9 @@ windows = { version = "0.62", features = [
"Win32_System_SystemInformation",
] }
zip = { version = "8", default-features = false, features = ["deflate"] }
senbei-android-crypto = { path = "senbei-android-crypto" }
senbei-android-elf = { path = "senbei-android-elf" }
senbei-android-engine = { path = "senbei-android-engine" }
senbei-android-metadata = { path = "senbei-android-metadata" }
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" }
+33 -66
View File
@@ -1,89 +1,56 @@
# Senbei
A static unpacker for Crackproof-protected 64-bit and 32-bit PE files and
protected Android (AArch64) shared libraries. Point it at a file, an app
package, 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 64-bit and 32-bit PE files and protected Android AArch64 shared libraries. Point it at a file, an app package, or a folder and it writes decrypted copies without launching the protected program.
> _"Crackproof"? It's senbei (煎餅 — rice cracker). Cracks itself._
Senbei reads protected input bytes and replays the unpacking algorithm statically. The command-line tool adds filesystem scanning, progress reporting, and logs; `senbei-wasm` provides the browser binding.
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/).
## Crates
## Legal notice and intended use
The workspace contains eight crates: `senbei-cli`, `senbei-crypto`, `senbei-io`, `senbei-metadata`, `senbei-pe`, `senbei-elf`, `senbei-engine`, and `senbei-wasm`.
**Read this before using Senbei.**
`senbei-pe` and `senbei-elf` contain only basic format parsing and address mapping. Protection-specific code is in `senbei-engine/src/windows/` and `senbei-engine/src/android/`. Platform-specific crypto and metadata code is grouped under `senbei-crypto/src/android/`, `senbei-metadata/src/windows/`, and `senbei-metadata/src/android/`.
- 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.
## Supported Inputs
## What it handles
- Protected Windows `.exe` and `.dll` files, including external `<name>.exe._` and `<name>.dll._` payloads.
- `global-metadata.dat` files with supported method-token layouts.
- Protected Android `.so` files and Android `.apk`, `.apks`, and `.xapk` packages.
| 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. |
| Android `.so` | Protected AArch64 shared library, statically restored (hollowed sections + stripped dynamic tables rebuilt). |
| `.apk` / `.apks` / `.xapk` | App packages; protected entries inside are restored, preserving the package's internal layout. |
Windows scanning probes only `.exe`, `.dll`, and `global-metadata.dat`; companion payloads are consumed through their matching stub and are not counted as skipped files. Android scanning probes only `.so` and `global-metadata.dat`. Android packages are inspected from their ZIP manifests and only matching `.so` and metadata entries are extracted.
Detection is content-based (header key-table at offset 4096, magic `KONN`),
not extension-based — app packages are the one exception, recognised by
extension plus the zip magic because they are containers. Anything
unrecognized is left untouched.
## Quick start
## Quick Start
```cmd
cargo build --release
senbei protected.exe
:: -> unpack\protected.unpack.exe
senbei game.apk
:: -> unpack\game.apk\lib\arm64-v8a\libil2cpp.unpack.so
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.
Outputs are written below an `unpack` directory unless `--out` is supplied. Every restored PE or ELF image passes a structural validation step before it is reported as successful.
## Documentation
## Tests
- [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
```cmd
cargo test --release --workspace
cargo clippy --workspace --all-targets -- -D warnings
cargo fmt --all -- --check
```
The local `test/` corpus can be passed to the CLI for real sample verification. The tracked `samples/` corpus is optional and remains user-managed.
## Web Build
```cmd
cd senbei-wasm
wasm-pack build --target web --release --out-dir ../web/pkg
```
The generated package is written to the ignored `web/pkg/` directory and can be served with any static HTTP server.
## Legal Notice
Use Senbei only for software you own or are authorized to analyze. The project is intended for lawful reverse engineering, security research, preservation, and interoperability.
## License
+31 -215
View File
@@ -1,238 +1,54 @@
# 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.
Senbei is a fully static unpacker. It reads protected bytes, replays the protection algorithm, validates the result, and writes a recovered image without launching or attaching to the protected program.
## Crate layout
## Crate Layout
Senbei is a Cargo workspace split into a pure core and thin shells around it:
The workspace is organized into eight crates. `senbei-cli` is the command-line entry point, `senbei-io` owns filesystem orchestration, `senbei-wasm` provides browser bindings, `senbei-pe` and `senbei-elf` provide basic format parsing, `senbei-crypto` provides shared primitives, `senbei-metadata` restores metadata, and `senbei-engine` owns protection-specific pipelines.
- **`senbei-pe/`** — 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.
- **`senbei-crypto/`** — cryptographic, checksum, compression, and bytecode
primitives the core is built from. Same purity rules as `senbei-pe`.
- **`senbei-metadata/`** — il2cpp `global-metadata.dat` method-token
de-obfuscation (format version 31; other versions are left untouched).
- **`senbei-android-crypto/`** — container primitives of the Android
(AArch64) protection scheme: the word/record ciphers, the GF(2³²)
transform, the AES-augmented segment transform, and the Huffman/LZ decoder.
- **`senbei-android-engine/`** — stage-1/stage-2 extraction: finds the
appended payload section, decrypts the stage-1 header and stage-2 payload,
and walks the recursive record streams to decode every module. Native-only
(memory-maps the input, writes the module set to a workspace directory).
- **`senbei-android-elf/`** — the restore: replays the decoded target-image
and fixup containers onto a hollowed ELF and rebuilds the dynamic-linker
tables (hash tables, symbols, relocations) the protector stripped.
Native-only.
- **`senbei-android-metadata/`** — the Android metadata variants: the seeded
five-round MethodDef-RID permutation restore (v31), seed discovery, and the
embedded-metadata XOR unwrap (`keystream.rs`).
- **`senbei-io/`** — filesystem and orchestration: recursive folder scanning,
per-run log file, progress bar, Explorer-friendly exit pause, the
single-file/folder orchestration in `job.rs` (incl. the wasm-safe in-memory
byte API used by the web frontend), and `android.rs` — the Android
single-library / folder / app-package orchestration.
- **`senbei-cli/`** — the `senbei` binary: argument parsing + dispatch. The
integration test suite (incl. the golden corpus test) lives in
`senbei-cli/tests/`.
Single-platform source stays directly under `src/`. Multi-platform crates keep platform code below `src/windows/` and `src/android/`, with shared code directly below `src/`.
```
senbei-cli/
└── src/main.rs argument parsing + dispatch
senbei-io/src/
├── job.rs single-file + folder orchestration, out-naming,
│ companion splice, stub overlay/TLS restore,
│ pipeline routing (incl. the wasm-safe byte API)
├── android.rs Android single-library / folder / package
│ orchestration, cross-source dedup
├── scan.rs recursive target discovery (PE + metadata + Android)
├── logfile.rs per-run timestamped log
├── ui.rs progress bar + status lines
└── pause.rs Explorer-friendly exit pause
senbei-metadata/src/
└── metadata.rs il2cpp global-metadata.dat de-obfuscation
```text
senbei-cli/src/main.rs
senbei-crypto/src/
├── primitives.rs decrypt_data* steps, key derivation
├── bytecode.rs bytecode VM
├── tables.rs constant tables
└── crc32.rs checksum
senbei-pe/src/engine/ pure, panic-free, no-I/O core
├── mod.rs detection + unpack_auto dispatch
├── error.rs structured error taxonomy
├── integrity.rs static post-unpack sanity check
├── parallel.rs deterministic block-parallel fan-out
├── layout/ layout discovery + validation
│ ├── dd8.rs .text dd8 key-formula + shift selection
│ ├── discovery.rs layout candidate discovery (trial-and-validate)
│ └── image.rs PE image reconstruction helpers
├── exe/
│ ├── pipeline.rs EXE pipeline (PE32+ and PE32 orchestration)
│ └── pipeline/pe32.rs PE32-specific EXE restore
└── dll/
└── pipeline.rs native + managed DLL pipeline
senbei-android-crypto/src/
└── protector.rs container ciphers, GF(2^32), Huffman/LZ decoder
senbei-android-engine/src/
├── stage1.rs payload-section discovery + stage-1 header/payload
├── stream.rs record-stream parsing
├── extract.rs recursive module extraction (writes the workspace)
├── probe.rs protected-library content probe
└── report.rs machine-readable extraction report
senbei-android-elf/src/
├── restore.rs image restore + dynamic-table rebuild
├── layout.rs ELF layout parsing
├── artifact.rs module-workspace index loading
└── hash.rs SysV/GNU hash table rebuild
senbei-android-metadata/src/
├── method_tokens.rs seeded RID permutation restore + seed discovery
├── embedded.rs embedded-metadata blob locate + XOR unwrap
└── keystream.rs recovered keystream table (one observed build)
senbei-crypto/src/android/
senbei-elf/src/
senbei-engine/src/windows/
senbei-engine/src/android/
senbei-io/src/
senbei-io/src/android/
senbei-metadata/src/windows/
senbei-metadata/src/android/
senbei-pe/src/
senbei-wasm/src/
```
## Detection and routing
`senbei-pe` and `senbei-elf` are format crates only. They do not depend on the unpacking engines, filesystem code, or platform protection logic.
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`). The folder scan additionally
classifies Android targets: an ELF64/AArch64 prefix promotes the file to a
full protection probe (`senbei_android_engine::is_protected_libil2cpp`), and a
package extension plus zip magic marks an app package for container
extraction.
## Windows Engine
`unpack_auto` then dispatches:
`senbei-engine/src/windows/` contains PE detection, layout discovery, EXE and DLL restoration, deterministic block parallelism, and structural integrity checks. Candidate layouts are trial-decrypted and validated before an output is accepted.
- `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).
External companion inputs are reconstructed as `stub[..4096]` followed by the matching `._` payload. The stub's export and TLS data is overlaid after unpacking because those regions are not present in the encrypted companion.
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.
## Android Engine
## External-companion inputs
`senbei-engine/src/android/extract/` decrypts the stage-1 header and stage-2 record streams and writes a temporary module workspace. `senbei-engine/src/android/restore/` applies decoded image and fixup containers to the hollowed ELF and rebuilds dynamic-linker tables. Both phases validate bounds and table placement before writing output.
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.
Android protection primitives are in `senbei-crypto/src/android/`. Android metadata restoration is in `senbei-metadata/src/android/` and only rewrites MethodDef token fields. The Windows structural metadata transform is in `senbei-metadata/src/windows/`.
## Pipelines
## Scanning and Packages
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.
Folder scanning uses platform target names to avoid opening bulk assets: Windows candidates are `.exe`, `.dll`, and `global-metadata.dat`; Android candidates are `.so` and `global-metadata.dat`. A Windows `.exe._` or `.dll._` companion is auxiliary input for its sibling stub and is excluded from the skipped count.
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.
APK, APKS, and XAPK files are containers. Senbei reads their ZIP manifests first, follows nested APK entries when necessary, and extracts only `.so` and exact `global-metadata.dat` entries. Extraction streams directly to temporary files, so compressed and decompressed copies are not held in memory together.
## The Android pipeline
## Validation
The Android scheme hollows an ELF64/AArch64 shared object: section bodies are
zeroed in the file and the original bytes move into an encrypted payload
appended as a `SHT_LOUSER` section (invisible to the dynamic loader). Restore
is two-phase:
Every heuristic layout uses trial-and-validate. A candidate that fails structural checks, checksums, or table bounds is rejected and the next candidate is tried. A failed restore is reported as an error rather than emitting a silently damaged binary.
1. **Extract** (`senbei-android-engine`): decrypt the stage-1 parameter block
and stage-2 payload from the payload section, then walk the recursive
record streams — each decoded module may interpret a further nested stream
— into a temporary module workspace with a JSON index.
2. **Restore** (`senbei-android-elf`): decode the target-image container onto
a copy of the hollowed file, apply the compact fixup database (the
relocations stripped from `.rela.dyn`), and rebuild the dynamic-linker
tables the loader needs (SysV/GNU hash, symbol and string tables,
`.rela.dyn`/`.rela.plt`). Validation is structural and total: mismatched
container sizes, descriptor bounds, or a rebuilt table overhanging its
section fail the restore rather than emit a broken image.
The PE integrity check verifies headers, section ranges, entry-point mapping, import names, relocation requirements, and managed metadata signatures. Android restoration validates ELF ranges, decoded container sizes, fixup bounds, and rebuilt dynamic tables.
il2cpp metadata comes in three shapes, all routed through
`job::deobfuscate_metadata_to` / `android::restore_metadata_bytes`:
## WebAssembly
- **structural (Windows `-GMD`)**: sparse method tokens remapped to the
contiguous per-module range, keyless, idempotent (`senbei-metadata`).
- **seeded permutation (Android v31)**: MethodDef RIDs permuted by a keyed
five-round transform; the seed is recovered by intersecting per-image key
residues, and the restore validates every RID — a wrong seed errors and the
structural remap takes over (`senbei-android-metadata`).
- **embedded blob**: no metadata file in the app at all; a slim blob sits in
the library's data section under a per-word XOR layer. After a restore the
blob is located by content (two known plaintext header words against the
embedded keystream) and unwrapped to a standalone `global-metadata.dat`.
Key derivation is untraced — the shipped keystream covers the one observed
build, and other builds simply never match the probe.
Packages (`.apk`/`.apks`/`.xapk`) are containers, not targets: entries are
extracted to a temporary workspace and content-probed like loose files.
Cross-source duplicates (a library loose in the tree *and* inside its
package) are restored once, preferring the loose file, then the `.apk`, then
bundle splits.
## 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.
The browser binding depends on `senbei-engine` through the I/O byte API. Native filesystem and Android package orchestration remain outside the browser workflow. Each browser unpack runs in a disposable worker because WebAssembly cannot recover from a caught panic in the same way as native code.
+34 -99
View File
@@ -2,125 +2,60 @@
## 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`.
The pinned Rust toolchain is defined in `rust-toolchain.toml`. Build the CLI with `cargo build --release`; the binary is written to `target/release/senbei.exe` on Windows.
```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)).
The workspace crates are portable where their APIs are pure. The browser binding is outside the workspace and is checked with `cargo check --manifest-path senbei-wasm/Cargo.toml` or built with `wasm-pack`.
## Testing
```cmd
cargo test --release
cargo test --release --workspace
cargo clippy --workspace --all-targets -- -D warnings
cargo fmt --all -- --check
```
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).
The tracked test suite is safe without protected samples. The optional local `samples/` corpus is user-managed and the ignored `test/` folder can be used for real Windows and Android runs.
> **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.
For an Android package, use one command at a time because a protected `.so` can be hundreds of megabytes. APK, APKS, and XAPK tests read the ZIP manifest first and extract only `.so` and `global-metadata.dat` entries.
## Debugging levers (environment variables)
## 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).
- `SENBEI_ANDROID_SAMPLES` — override the Android corpus location (default
`samples/android/`; see `samples/README.md`). The Android corpus test pins
restored outputs with SHA-256 sidecar files next to each protected input
and documents known restore gaps with empty `<base>.restore-fails` markers.
- `DD8_SHIFT` overrides the PE page-XOR shift; `99` skips that stage.
- `SEL_DIAG` prints PE layout-selector diagnostics.
- `SENBEI_THREADS` caps deterministic block fan-out; `1` forces the sequential reference path.
- `SENBEI_SCAN_ALL` enables the explicit scan-all mode for selected target names.
- `SENBEI_ANDROID_SAMPLES` overrides the Android sample corpus location.
## Conventions
- The `senbei-pe/` core (and its `senbei-crypto/` base) 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).
Format crates stay free of filesystem I/O and protection-specific logic. Windows engine code lives below `senbei-engine/src/windows/`, Android engine code below `senbei-engine/src/android/`, and shared code stays directly under each crate's `src/`.
## Repository layout
Layout heuristics must trial and validate every candidate. A failed validation is an error or a fall-through, never a silently accepted offset.
```
senbei/
├── Cargo.toml workspace root (members: the senbei-* crates)
├── rust-toolchain.toml pinned toolchain + targets
├── senbei-cli/ senbei binary (default member)
│ └── tests/ CLI, detection, golden, and folder tests
├── senbei-pe/ pure unpacker core (see docs/design.md)
├── senbei-crypto/ crypto/compression primitives
├── senbei-metadata/ il2cpp metadata de-obfuscation
├── senbei-io/ filesystem, scanning, CLI orchestration
├── senbei-wasm/ WebAssembly bindings crate (own Cargo.lock,
│ outside the workspace; builds into web/pkg/)
├── samples/ local-only test corpus (git-ignored)
├── web/ static browser frontend assets (+ built pkg/)
├── docs/ usage, design, and development documentation
└── .github/ CI workflows and issue templates
Outputs must remain byte-identical against the available golden corpus. Run the full workspace tests after changing a pipeline or a metadata layout.
Folder scanning uses explicit target names to avoid opening bulk assets. External `.exe._` and `.dll._` files are auxiliary data for their sibling stubs and are not independent scan targets.
## Repository Layout
```text
senbei-cli/ command-line binary and integration tests
senbei-crypto/ shared crypto and Android crypto primitives
senbei-elf/ basic ELF parsing
senbei-engine/ Windows and Android unpacking engines
senbei-io/ filesystem, package, scanning, and CLI orchestration
senbei-metadata/ Windows and Android metadata restoration
senbei-pe/ basic PE parsing
senbei-wasm/ browser bindings and its own lockfile
web/ static browser frontend
samples/ optional local corpus
```
## Web build
See [web/README.md](../web/README.md). In short:
## Web Build
```cmd
cd senbei-wasm
wasm-pack build --target web --release --out-dir ../web/pkg
```
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.
Serve `web/` with a static HTTP server after the build. The browser never uploads input files.
+25 -133
View File
@@ -1,163 +1,55 @@
# Usage
```
senbei <file|folder> [--out DIR] [-v|--verbose] [-q|--quiet]... [--scan-all]
[--no-log] [--no-pause] [-V|--version] [-h|--help]
```text
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
## 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:
The output is written below `<parent>/unpack/` with `.unpack` inserted before the extension. `--out DIR` changes both the output and log directory.
```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.
For `global-metadata.dat`, Senbei writes `global-metadata.unpack.dat` only when method tokens change. Unsupported metadata versions remain untouched and are reported as skipped.
## Android targets
## Android Targets
Senbei also restores Android (AArch64) protected shared libraries and app
packages:
Protected `.so` files are restored from their encrypted payload sections and written as `libil2cpp.unpack.so` or the corresponding input name. APK, APKS, and XAPK files are treated as containers: their manifests are read first, nested APKs are followed when necessary, and only `.so` and exact `global-metadata.dat` entries are extracted.
- **`.so`** — a protected library is hollowed out on disk: its original
sections live in an encrypted payload appended to the file, and senbei
rebuilds the static image from it. Output: `libil2cpp.unpack.so`.
- **`.apk`** — entries are extracted to a temporary workspace and
content-probed like loose files; protected libraries and metadata blobs
inside are restored to `<out>/<apk name>/<entry path>`.
- **`.apks` / `.xapk`** — split-package bundles; each nested `.apk` is opened
and searched the same way, under `<out>/<bundle name>/<split name>/...`.
If a restored library contains embedded metadata, the unwrapped blob is written beside it as `global-metadata.unpack.dat`. Identical loose and package entries are restored once, preferring the loose file.
When a restored il2cpp library carries its metadata embedded in its data
section (no standalone `global-metadata.dat` in the app at all), senbei
unwraps the blob and writes it next to the library as
`global-metadata.unpack.dat`. One observed packaging variant wraps the blob in
a per-word XOR layer whose keys are generated at runtime and stored nowhere;
senbei ships the keystream recovered from the one build known to use it and
content-probes for it — builds with a different keystream are silently
skipped (the library itself is still fully restored).
## Folder Mode
The same content may appear loose in a folder, in its `.apk`, and in a bundle
side by side: identical content is restored once, at the loose file's
destination. A restored library is validated structurally by the restore
itself (the rebuild refuses inconsistent layouts); a protected library that
fails validation counts as an error, not a suspect.
Folder mode walks recursively, skips directories named `unpack`, and mirrors recognized outputs below `<root>/unpack/` or `--out DIR`. Windows candidates are `.exe`, `.dll`, and `global-metadata.dat`; Android candidates are `.so` and `global-metadata.dat`. A matching `.exe._` or `.dll._` payload is consumed by its stub and is excluded from the skipped count.
## Folder mode
The summary has the form `12 unpacked · 3 skipped · 0 errors · 1 suspect · 2 metadata`; the package count is appended when packages were opened. Each file is isolated so one failed target does not stop the folder run.
Senbei walks the directory recursively, skips any subdirectory literally named
`unpack`, and unpacks every file it recognises as protected (by content, not
extension — renamed files and `.bak` backups are still found; packages are the
one exception, recognised by extension plus the zip magic because they are
containers). 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**:
## Integrity Check
```cmd
senbei "C:\Games\MyGame"
:: -> C:\Games\MyGame\unpack\...
:: -> C:\Games\MyGame\unpack\senbei-YYYYMMDD-HHMMSS.log
```
PE outputs are checked for valid headers, section ranges, entry-point mapping, readable import names, relocation requirements, and managed metadata signatures. Android outputs are validated during ELF restoration, including decoded container sizes, fixup bounds, and rebuilt dynamic tables.
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
```
The `packages` count appears (as `· N packages`) only when Android app
packages were processed.
## Integrity check
(PE outputs only — Android restores carry their own structural validation; see
[Android targets](#android-targets).)
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`).
A clean report is not a proof of correctness, but a non-clean report is a reliable broken-output signal. Suspect PE files are still written and counted separately.
## Flags
| Flag | Behavior |
| --- | --- |
| `--out DIR` | Write outputs (and the log, unless `--no-log`) under `DIR`. |
| `-v`, `--verbose` | Print detailed per-stage progress (and the destination path) for each file — `[N/9]` stages for PE targets, container/segment lines for Android libraries. 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. |
| `--out DIR` | Write outputs and logs below `DIR`. |
| `-v`, `--verbose` | Print per-stage progress. |
| `-q`, `--quiet` | Hide progress and per-file lines; repeat to suppress all standard output. |
| `--no-log` | Do not write a run log. |
| `--scan-all` | Probe every selected target-name candidate, including files below the size floor. |
| `--no-pause` | Disable the Explorer-friendly Windows exit prompt. |
| `-V`, `--version` | Print the 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
## Exit Codes
| Code | Meaning |
| --- | --- |
| `0` | Success (single file restored, 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.
| `0` | The requested restore completed without errors. |
| `1` | A target failed, a scan probe was unreadable, or a single-file restore errored. |
| `2` | The command line was invalid. |
+15 -97
View File
@@ -1,105 +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.
## Android corpus (`samples/android/`)
The `android/` subfolder holds Android samples, one **extracted app tree** per
subdirectory (the layout an APK unpacks to: `lib/<abi>/*.so`,
`assets/.../global-metadata.dat`, ...). The test
(`tests/android_samples.rs`) finds protected AArch64 libraries by content and
restores them through the real pipeline. `SENBEI_ANDROID_SAMPLES` overrides
the corpus location.
Sidecar conventions (all next to the protected `.so` input):
| File | Meaning |
| ---- | ------- |
| `<base>.golden.so.sha256` | Expected SHA-256 of the restored library |
| `<base>.golden.metadata.sha256` | Expected SHA-256 of the unwrapped embedded metadata blob (when the library carries one) |
| `<base>.restore-fails` | Empty marker: this input's restore is a known gap and *must* fail (a future fix fails the test, prompting marker removal) |
A missing sidecar is a warning (with the computed digest printed, ready to
promote), never a failure. App packages (`.apk`/`.apks`/`.xapk`) dropped into
a tree are exercised by folder mode as containers.
`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.
-10
View File
@@ -1,10 +0,0 @@
//! Static restoration of the current protected AArch64 `libil2cpp.so`.
mod artifact;
mod error;
mod hash;
mod layout;
mod restore;
pub use error::Error;
pub use restore::{RestoreOptions, RestoreReport, restore_libil2cpp};
-20
View File
@@ -1,20 +0,0 @@
[package]
name = "senbei-android-engine"
version.workspace = true
edition.workspace = true
rust-version.workspace = true
license.workspace = true
description = "Static Stage 1 and Stage 2 extraction for Senbei Android"
[dependencies]
goblin.workspace = true
memmap2.workspace = true
serde.workspace = true
serde_json.workspace = true
sha2.workspace = true
tempfile.workspace = true
thiserror.workspace = true
senbei-android-crypto.workspace = true
[lints]
workspace = true
-14
View File
@@ -1,14 +0,0 @@
[package]
name = "senbei-android-metadata"
version.workspace = true
edition.workspace = true
rust-version.workspace = true
license.workspace = true
description = "IL2CPP metadata restoration for Senbei Android"
[dependencies]
serde.workspace = true
thiserror.workspace = true
[lints]
workspace = true
+1
View File
@@ -13,6 +13,7 @@ path = "src/main.rs"
[dependencies]
senbei-io.workspace = true
senbei-engine.workspace = true
[dev-dependencies]
senbei-io.workspace = true
+1
View File
@@ -6,4 +6,5 @@ license.workspace = true
description = "Cryptographic and compression primitives for Senbei"
[dependencies]
aes.workspace = true
thiserror.workspace = true
@@ -1,4 +1,4 @@
//! Cryptographic and container primitives used by Senbei Android.
//! Android container cryptography and decoding primitives.
mod protector;
@@ -359,7 +359,7 @@ pub struct HuffmanLzDecoder {
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() % 3 != 0 {
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 {
@@ -542,6 +542,7 @@ impl HuffmanLzDecoder {
}
/// Apply the native word transform and optional AES-256-CBC decryption.
#[allow(clippy::chunks_exact_to_as_chunks)]
pub fn transform_segment(
data: &[u8],
seed: u32,
+1
View File
@@ -1,5 +1,6 @@
//! Cryptographic, checksum, compression, and bytecode primitives.
pub mod android;
pub mod bytecode;
pub mod crc32;
pub mod primitives;
@@ -1,13 +1,13 @@
[package]
name = "senbei-android-crypto"
name = "senbei-elf"
version.workspace = true
edition.workspace = true
rust-version.workspace = true
license.workspace = true
description = "Protector container primitives for Senbei Android"
description = "ELF format parsing and structural utilities for Senbei"
[dependencies]
aes.workspace = true
goblin.workspace = true
thiserror.workspace = true
[lints]
+54
View File
@@ -0,0 +1,54 @@
//! Basic ELF format parsing shared by the unpacking engine.
use goblin::elf::{Elf, header::EM_AARCH64, program_header::PT_LOAD};
use thiserror::Error;
#[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,
}
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 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))
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn rejects_non_elf() {
assert!(matches!(parse(b"not elf"), Err(Error::Parse(_))));
}
}
@@ -1,19 +1,20 @@
[package]
name = "senbei-android-elf"
name = "senbei-engine"
version.workspace = true
edition.workspace = true
rust-version.workspace = true
license.workspace = true
description = "AArch64 ELF restoration for Senbei Android"
description = "Platform unpacking engines for Senbei"
[dependencies]
goblin.workspace = true
memmap2.workspace = true
serde.workspace = true
serde_json.workspace = true
sha2.workspace = true
tempfile.workspace = true
thiserror.workspace = true
senbei-android-crypto.workspace = true
senbei-crypto.workspace = true
[lints]
workspace = true
@@ -19,7 +19,7 @@ pub enum Error {
#[error("serialize extraction index: {0}")]
Json(#[from] serde_json::Error),
#[error("embedded Stage 2 decoder configuration: {0}")]
EmbeddedConfig(#[source] senbei_android_crypto::Error),
EmbeddedConfig(#[source] senbei_crypto::android::Error),
#[error(
"depth {depth} stream 0x{stream_id:02X} interpreter 0x{interpreter_id:02X} configuration: {source}"
)]
@@ -28,7 +28,7 @@ pub enum Error {
stream_id: u32,
interpreter_id: u32,
#[source]
source: senbei_android_crypto::Error,
source: senbei_crypto::android::Error,
},
#[error(
"depth {depth} stream 0x{stream_id:02X} record {record_index} command 0x{command_id:02X} {part}: {source}"
@@ -40,7 +40,7 @@ pub enum Error {
command_id: u32,
part: &'static str,
#[source]
source: senbei_android_crypto::Error,
source: senbei_crypto::android::Error,
},
#[error("{0}")]
Invalid(String),
@@ -1,14 +1,12 @@
//! Pure-static Stage 1 decryption and recursive Stage 2 module extraction.
mod error;
mod extract;
mod pipeline;
mod probe;
mod report;
mod stage1;
mod stream;
pub use error::Error;
pub use extract::{ExtractOptions, extract_stage2};
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};
@@ -4,20 +4,20 @@ use std::io::Write;
use std::path::{Path, PathBuf};
use memmap2::MmapOptions;
use senbei_android_crypto::{Module9bConfig, decode_container};
use senbei_crypto::android::{Module9bConfig, decode_container};
use serde_json::to_vec_pretty;
use sha2::{Digest, Sha256};
use tempfile::NamedTempFile;
use crate::error::{Error, Result, invalid};
use crate::report::{
use super::error::{Error, Result, invalid};
use super::report::{
ArtifactReport, DecoderReport, ExtractionReport, ModuleRegistryEntry, RecordReport,
Stage1Report, StreamParent, StreamReport,
};
use crate::stage1::{
use super::stage1::{
DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE, SHT_LOUSER, Stage1Result, inspect,
};
use crate::stream::{DIRECT_FLAG, Record, parse_record_stream};
use super::stream::{DIRECT_FLAG, Record, parse_record_stream};
/// Inputs and output locations for one complete static Stage 2 extraction.
#[derive(Debug, Clone)]
@@ -1,8 +1,8 @@
use std::path::Path;
use senbei_android_crypto::Module9bConfig;
use senbei_crypto::android::Module9bConfig;
use crate::stage1::{self, DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE};
use super::stage1::{self, DEFAULT_CIPHER_CONSTANT, DEFAULT_OUTER_SIZE};
/// Return whether `data` has a supported protected AArch64 IL2CPP layout.
#[must_use]
@@ -2,7 +2,7 @@ use std::path::Path;
use goblin::elf::{Elf, header::EM_AARCH64};
use crate::error::{Error, Result, invalid};
use super::error::{Error, Result, invalid};
pub(crate) const SHT_LOUSER: u32 = 0x8000_0000;
pub const DEFAULT_CIPHER_CONSTANT: u32 = 0xbf20_165d;
@@ -1,6 +1,6 @@
use senbei_android_crypto::gf32_mul_fixed;
use senbei_crypto::android::gf32_mul_fixed;
use crate::error::{Error, Result, invalid};
use super::error::{Error, Result, invalid};
pub(crate) const RECORD_SIZE: usize = 0x5c;
pub(crate) const DIRECT_FLAG: u32 = 2;
+10
View File
@@ -0,0 +1,10 @@
//! Android AArch64 extraction and ELF restoration.
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};
@@ -3,7 +3,7 @@ use std::path::{Path, PathBuf};
use serde_json::Value;
use crate::error::{Error, Result, invalid};
use super::error::{Error, Result, invalid};
const REQUIRED_IDS: [u32; 3] = [0x9b, 0x9d, 0x9e];
@@ -13,7 +13,7 @@ pub enum Error {
#[error("cannot parse module index: {0}")]
Json(#[from] serde_json::Error),
#[error(transparent)]
Crypto(#[from] senbei_android_crypto::Error),
Crypto(#[from] senbei_crypto::android::Error),
#[error("{0}")]
Invalid(String),
}
@@ -1,4 +1,4 @@
use crate::error::{Error, Result, invalid};
use super::error::{Error, Result, invalid};
#[must_use]
pub(crate) fn elf_hash(name: &[u8]) -> u32 {
@@ -1,4 +1,4 @@
use crate::error::{Error, Result, invalid};
use super::error::{Error, Result, invalid};
pub(crate) const SHT_NOBITS: u32 = 8;
pub(crate) const SHT_LOUSER: u32 = 0x8000_0000;
+8
View File
@@ -0,0 +1,8 @@
mod artifact;
mod error;
mod hash;
mod layout;
mod pipeline;
pub use error::Error;
pub use pipeline::{RestoreOptions, RestoreReport, restore_libil2cpp};
@@ -5,17 +5,17 @@ use std::path::{Path, PathBuf};
use std::time::Instant;
use memmap2::{Mmap, MmapMut, MmapOptions};
use senbei_android_crypto::{
use senbei_crypto::android::{
ContainerHeader, HuffmanLzDecoder, Module9bConfig, ProtectedDescriptor, transform_segment,
};
use serde::Serialize;
use sha2::{Digest, Sha256};
use tempfile::NamedTempFile;
use crate::artifact::load_artifacts;
use crate::error::{Error, Result, invalid};
use crate::hash::{build_gnu_hash, build_sysv_hash};
use crate::layout::{
use super::artifact::load_artifacts;
use super::error::{Error, Result, invalid};
use super::hash::{build_gnu_hash, build_sysv_hash};
use super::layout::{
ElfLayout, SHF_ALLOC, SHT_LOUSER, SHT_NOBITS, SectionHeader, align_up, read_i64, read_u32,
read_u64, slice, slice_u64, usize_from_u64,
};
+14
View File
@@ -0,0 +1,14 @@
//! 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 {
windows::thread_cap()
}
@@ -407,7 +407,7 @@ impl<'a> Unpacker<'a> {
// non-critical for false-positive rejection.
if v8 < info6 {
let delta = info6.wrapping_sub(v8);
if delta <= 0x1000 && delta.is_multiple_of(0x200) {
if delta <= 0x1000 && delta % 0x200 == 0 {
anchor = Some(probe);
break;
}
@@ -350,8 +350,8 @@ mod tests {
};
assert_eq!(message, "test panic");
assert!(
file.ends_with("senbei-pe/src/engine/mod.rs")
|| file.ends_with("senbei-pe\\src\\engine\\mod.rs")
file.ends_with("senbei-engine/src/windows/mod.rs")
|| file.ends_with("senbei-engine\\src\\windows\\mod.rs")
);
assert!(line > 0);
assert!(column > 0);
@@ -382,8 +382,8 @@ mod tests {
};
assert_eq!(message, "worker panic");
assert!(
file.ends_with("senbei-pe/src/engine/mod.rs")
|| file.ends_with("senbei-pe\\src\\engine\\mod.rs")
file.ends_with("senbei-engine/src/windows/mod.rs")
|| file.ends_with("senbei-engine\\src\\windows\\mod.rs")
);
assert!(line > 0);
assert!(column > 0);
+1 -4
View File
@@ -10,11 +10,8 @@ anyhow.workspace = true
flate2.workspace = true
indicatif.workspace = true
owo-colors.workspace = true
senbei-android-elf.workspace = true
senbei-android-engine.workspace = true
senbei-android-metadata.workspace = true
senbei-engine.workspace = true
senbei-metadata.workspace = true
senbei-pe.workspace = true
sha2.workspace = true
tempfile.workspace = true
walkdir.workspace = true
@@ -5,24 +5,24 @@
//! 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_android_engine`]) and rebuilds the static image
//! ([`senbei_android_elf`]). Some il2cpp builds additionally embed their
//! ([`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_android_metadata::extract_embedded_metadata`]).
//! ([`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::io::Read;
use std::io::{BufWriter, Write};
use std::path::{Path, PathBuf};
use anyhow::{Context, Result, bail};
use flate2::read::DeflateDecoder;
use senbei_android_elf::{RestoreOptions, restore_libil2cpp};
use senbei_android_engine::{ExtractOptions, extract_stage2, is_protected_libil2cpp};
use senbei_engine::android::{ExtractOptions, extract_stage2, is_protected_libil2cpp};
use senbei_engine::android::{RestoreOptions, restore_libil2cpp};
use sha2::{Digest, Sha256};
use zip::ZipArchive;
@@ -97,7 +97,7 @@ pub fn restore_so_file(input: &Path, dest: &Path, verbose: bool) -> Result<Optio
.context("restore protected library")?;
let restored =
std::fs::read(dest).with_context(|| format!("read restored `{}`", dest.display()))?;
Ok(senbei_android_metadata::extract_embedded_metadata(
Ok(senbei_metadata::android::extract_embedded_metadata(
&restored,
))
}
@@ -122,20 +122,19 @@ pub fn content_identity(data: &[u8]) -> String {
/// 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_android_metadata::discover_method_token_seeds(data)
&& discovery.version == 31
&& discovery.images.iter().any(|image| !image.clean)
if let Ok(discovery) = senbei_metadata::android::discover_method_token_seeds(data)
&& matches!(discovery.version, 31 | 39)
{
let mut seeds = discovery.seed_candidates.clone();
if seeds.is_empty() {
seeds.push(senbei_android_metadata::DEFAULT_METHOD_TOKEN_SEED);
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_android_metadata::restore_method_tokens(data, seed) {
if let Ok((out, report)) = senbei_metadata::android::restore_method_tokens(data, seed) {
return Ok((
out,
senbei_metadata::Report {
@@ -234,7 +233,9 @@ pub fn restore_package(
nested.push((index, name));
}
} else {
direct.push((index, name));
if crate::scan::is_android_entry_name(&name) {
direct.push((index, name));
}
}
}
drop(archive);
@@ -262,7 +263,9 @@ pub fn restore_package(
let Some(entry_name) = entry_name else {
bail!("unsafe entry path in `{}`", nested_label.display());
};
entries.push((nested_index, entry_name));
if crate::scan::is_android_entry_name(&entry_name) {
entries.push((nested_index, entry_name));
}
}
}
drop(nested_archive);
@@ -324,6 +327,7 @@ fn restore_package_entry(
}
if is_so {
drop(data);
return Ok(match restore_so_file(&entry_path, dest, verbose) {
Ok(embedded) => {
let mut outcomes = vec![outcome(EntryKind::So, EntryStatus::Restored)];
@@ -407,27 +411,23 @@ fn extract_entry(
// `:` 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 compressed_size = usize::try_from(entry.compressed_size())
.map_err(|_| anyhow::anyhow!("entry compressed size exceeds usize"))?;
let output_size =
usize::try_from(entry.size()).map_err(|_| anyhow::anyhow!("entry size exceeds usize"))?;
let mut compressed = vec![0_u8; compressed_size];
entry.read_exact(&mut compressed)?;
let mut output = Vec::with_capacity(output_size);
match entry.compression() {
zip::CompressionMethod::Stored => output.extend_from_slice(&compressed),
let output_size = entry.size();
let mut output = BufWriter::new(std::fs::File::create(&destination)?);
let written = match entry.compression() {
zip::CompressionMethod::Stored => std::io::copy(&mut entry, &mut output)?,
zip::CompressionMethod::Deflated => {
DeflateDecoder::new(compressed.as_slice()).read_to_end(&mut output)?;
let mut decoder = DeflateDecoder::new(&mut entry);
std::io::copy(&mut decoder, &mut output)?
}
method => bail!("unsupported compression method {method:?} in entry `{key}`"),
}
if output.len() != output_size {
};
output.flush()?;
if written != output_size {
bail!(
"entry `{key}` decompressed to 0x{:x}, expected 0x{output_size:x}",
output.len()
written
);
}
std::fs::write(&destination, &output)?;
Ok(destination)
}
/// Lowercase hex of a digest output (sha2 0.11's `Array` no longer formats as
+2 -2
View File
@@ -1,4 +1,4 @@
use senbei_pe as unpacker;
use senbei_engine as unpacker;
use std::path::{Path, PathBuf};
/// Crackproof header key table lives at this fixed file offset. For the
@@ -818,7 +818,7 @@ pub fn run_file_v(
// handled entry-by-entry. Anything else falls through to the PE pipeline.
let is_android_so = crate::android::is_elf64_aarch64(&prefix)
&& std::fs::read(input)
.map(|bytes| senbei_android_engine::is_protected_libil2cpp(&bytes))
.map(|bytes| senbei_engine::android::is_protected_libil2cpp(&bytes))
.unwrap_or(false);
let is_android_package = !is_android_so && crate::android::is_app_package(input, &prefix);
+102 -32
View File
@@ -1,4 +1,4 @@
use senbei_pe::detect;
use senbei_engine::detect;
use std::io::Read;
use std::path::{Path, PathBuf};
use walkdir::WalkDir;
@@ -16,7 +16,7 @@ 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 [`senbei_pe::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
@@ -25,14 +25,59 @@ const DETECT_PREFIX: u64 = 8 * 1024;
/// 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 executable allow-list: unknown
/// extensions are still probed. Extensionless files are handled separately by
/// [`denied_name`] because asset stores commonly contain tens of thousands of
/// extensionless chunks; exhaustive probing remains available through
/// `--scan-all`.
const METADATA_FILE_NAME: &str = "global-metadata.dat";
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(METADATA_FILE_NAME))
}
fn is_target_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")
|| ext.eq_ignore_ascii_case("so")
})
}
fn is_android_package_name(path: &Path) -> bool {
path.extension()
.and_then(|ext| ext.to_str())
.is_some_and(|ext| {
ext.eq_ignore_ascii_case("apk")
|| ext.eq_ignore_ascii_case("apks")
|| ext.eq_ignore_ascii_case("xapk")
})
}
/// External Windows payloads are consumed through their sibling `.exe`/`.dll`
/// stub. They are valid input bytes, but are not independent unpack targets.
pub(crate) fn is_windows_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;
};
let stub_path = Path::new(stub_name);
stub_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_android_entry_name(path: &Path) -> bool {
is_metadata_name(path)
|| path
.extension()
.and_then(|ext| ext.to_str())
.is_some_and(|ext| ext.eq_ignore_ascii_case("so"))
}
/// File extensions that are bulk data by construction and can never be a target.
///
/// Set `SENBEI_SCAN_ALL=1` (or pass `--scan-all`) to probe every file regardless.
const DENY_EXT: &[&str] = &[
@@ -89,9 +134,7 @@ const DENY_EXT: &[&str] = &[
"sr",
];
/// Whether `path` can be skipped from its name alone. Extensionless files and
/// files whose extension is on [`DENY_EXT`] are not opened during a default
/// scan. `--scan-all` remains available when exhaustive probing is required.
/// Whether `path` can be skipped from its name alone.
fn denied_name(path: &Path) -> bool {
let Some(ext) = path.extension() else {
return true;
@@ -151,16 +194,14 @@ pub struct ScanResult {
/// 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.
///
/// 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
@@ -224,6 +265,15 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> ScanResult {
if !entry.file_type().is_file() {
continue;
}
if is_windows_companion(entry.path()) {
continue;
}
if !is_metadata_name(entry.path())
&& !is_target_extension(entry.path())
&& !is_android_package_name(entry.path())
{
continue;
}
if !scan_all {
// Name checks come first so extensionless asset chunks never
// trigger even an explicit metadata query.
@@ -247,7 +297,7 @@ pub fn find_targets_opts(root: &Path, scan_all: bool) -> ScanResult {
// `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 = senbei_pe::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);
@@ -314,9 +364,8 @@ 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, then the
/// Android probes. Returns `None` when the file could not be classified at
/// 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.
@@ -337,22 +386,31 @@ pub fn scan_all_env() -> bool {
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() {
return Class::Crackproof;
if is_android_package_name(path) && crate::android::is_app_package(path, &head) {
return Class::AndroidPackage;
}
if senbei_metadata::is_metadata(&head) {
if is_metadata_name(path) && senbei_metadata::is_metadata(&head) {
return Class::Metadata;
}
if crate::android::is_elf64_aarch64(&head)
if path
.extension()
.and_then(|ext| ext.to_str())
.is_some_and(|ext| ext.eq_ignore_ascii_case("exe") || ext.eq_ignore_ascii_case("dll"))
&& detect(&head).is_some()
{
return Class::Crackproof;
}
if path
.extension()
.and_then(|ext| ext.to_str())
.is_some_and(|ext| ext.eq_ignore_ascii_case("so"))
&& crate::android::is_elf64_aarch64(&head)
&& std::fs::read(path)
.map(|bytes| senbei_android_engine::is_protected_libil2cpp(&bytes))
.map(|bytes| senbei_engine::android::is_protected_libil2cpp(&bytes))
.unwrap_or(false)
{
return Class::AndroidSo;
}
if crate::android::is_app_package(path, &head) {
return Class::AndroidPackage;
}
Class::None
}));
r.ok()
@@ -403,7 +461,7 @@ mod tests {
}
#[test]
fn extensionless_targets_require_exhaustive_scan() {
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];
@@ -414,7 +472,7 @@ mod tests {
assert!(filtered.metadata.is_empty());
let exhaustive = find_targets_opts(root, true);
assert_eq!(exhaustive.metadata.len(), 1);
assert!(exhaustive.metadata.is_empty());
}
/// A file below the Crackproof key-table bound is skipped without being
@@ -479,4 +537,16 @@ mod tests {
);
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!(is_windows_companion(&root.join("app.exe._")));
}
}
+1 -1
View File
@@ -1,6 +1,6 @@
use indicatif::{ProgressBar, ProgressStyle};
use owo_colors::OwoColorize;
use senbei_pe::{IntegrityReport, Kind};
use senbei_engine::{IntegrityReport, Kind};
use std::path::Path;
/// Create a progress bar for `n` items. Hidden when `quiet` is true.
+4
View File
@@ -4,3 +4,7 @@ version.workspace = true
edition.workspace = true
license.workspace = true
description = "Unity il2cpp metadata de-obfuscation for Senbei"
[dependencies]
serde.workspace = true
thiserror.workspace = true
@@ -20,7 +20,7 @@
//! they are rewritten to the standard il2cpp metadata magic and version so the
//! output is a well-formed `global-metadata.dat`.
use crate::keystream::{HEADER_KEYS, SEGMENTS};
use super::keystream::{HEADER_KEYS, SEGMENTS};
/// Standard il2cpp metadata sanity magic written over the patched header.
const STANDARD_MAGIC: u32 = 0xfab1_1baf;
@@ -171,6 +171,9 @@ pub fn restore_method_tokens(data: &[u8], seed: u32) -> Result<(Vec<u8>, Report)
return Err(Error::NotMetadata);
}
let version = read_u32(data, 4)?;
if version == 39 {
return restore_v39(data, seed);
}
if version != SUPPORTED_VERSION {
return Err(Error::UnsupportedVersion(version));
}
@@ -368,6 +371,9 @@ pub fn discover_method_token_seeds(data: &[u8]) -> Result<SeedDiscoveryReport> {
return Err(Error::NotMetadata);
}
let version = read_u32(data, 4)?;
if version == 39 {
return discover_v39(data);
}
if version != SUPPORTED_VERSION {
return Ok(SeedDiscoveryReport {
version,
@@ -550,6 +556,458 @@ fn decrypt_rid_with_key(rid: u32, low: u32, high: u32, key: u32) -> u32 {
value + low
}
const V39_METHODS: usize = 5;
const V39_PARAMETERS: usize = 10;
const V39_GENERIC_CONTAINERS: usize = 14;
const V39_INTERFACE_OFFSETS: usize = 18;
const V39_TYPES: usize = 19;
const V39_IMAGES: usize = 20;
#[derive(Debug, Clone, Copy)]
struct V39Layout {
method_offset: usize,
method_count: usize,
method_stride: usize,
method_token_offset: usize,
type_definition_index_width: usize,
type_offset: usize,
type_count: usize,
type_stride: usize,
type_method_start_offset: usize,
type_method_count_offset: usize,
image_offset: usize,
image_count: usize,
image_stride: usize,
}
fn v39_section(data: &[u8], index: usize) -> Result<(usize, usize, usize)> {
let header = 8_usize
.checked_add(
index
.checked_mul(12)
.ok_or_else(|| Error::Malformed("v39 section header offset overflow".to_owned()))?,
)
.ok_or_else(|| Error::Malformed("v39 section header offset overflow".to_owned()))?;
let offset = read_u32(data, header)? as usize;
let size = read_u32(data, header + 4)? as usize;
let count = read_u32(data, header + 8)? as usize;
bytes(data, offset, size)?;
Ok((offset, size, count))
}
fn v39_index_width(count: usize) -> usize {
if count <= u8::MAX as usize {
1
} else if count <= u16::MAX as usize {
2
} else {
4
}
}
fn read_v39_index(data: &[u8], offset: usize, width: usize) -> Result<usize> {
match width {
1 => Ok(bytes(data, offset, 1)?[0] as usize),
2 => Ok(read_u16(data, offset)? as usize),
4 => Ok(read_u32(data, offset)? as usize),
_ => malformed("v39 index has an unsupported width"),
}
}
fn parse_v39(data: &[u8]) -> Result<V39Layout> {
let (method_offset, method_size, method_count) = v39_section(data, V39_METHODS)?;
let (_, parameter_size, parameter_count) = v39_section(data, V39_PARAMETERS)?;
let (_, _, generic_count) = v39_section(data, V39_GENERIC_CONTAINERS)?;
let (_interface_offset, interface_size, interface_count) =
v39_section(data, V39_INTERFACE_OFFSETS)?;
let (type_offset, type_size, type_count) = v39_section(data, V39_TYPES)?;
let (image_offset, image_size, image_count) = v39_section(data, V39_IMAGES)?;
let parameter_index_width = v39_index_width(parameter_count);
let generic_container_index_width = v39_index_width(generic_count);
let type_definition_index_width = v39_index_width(type_count);
let type_index_width = if interface_count == 0 {
4
} else {
let element_size = interface_size
.checked_div(interface_count)
.ok_or_else(|| Error::Malformed("v39 interface-offset size is invalid".to_owned()))?;
element_size
.checked_sub(4)
.filter(|width| matches!(width, 1 | 2 | 4))
.ok_or_else(|| Error::Malformed("v39 type-index width is invalid".to_owned()))?
};
let method_stride = 20_usize
.checked_add(type_definition_index_width)
.and_then(|size| size.checked_add(type_index_width))
.and_then(|size| size.checked_add(parameter_index_width))
.and_then(|size| size.checked_add(generic_container_index_width))
.ok_or_else(|| Error::Malformed("v39 method stride overflow".to_owned()))?;
let type_stride = 68_usize
.checked_add(
type_index_width
.checked_mul(3)
.ok_or_else(|| Error::Malformed("v39 type stride overflow".to_owned()))?,
)
.and_then(|size| size.checked_add(generic_container_index_width))
.ok_or_else(|| Error::Malformed("v39 type stride overflow".to_owned()))?;
let image_stride = 32_usize
.checked_add(
type_definition_index_width
.checked_mul(2)
.ok_or_else(|| Error::Malformed("v39 image stride overflow".to_owned()))?,
)
.ok_or_else(|| Error::Malformed("v39 image stride overflow".to_owned()))?;
if method_count.checked_mul(method_stride) != Some(method_size)
|| type_count.checked_mul(type_stride) != Some(type_size)
|| image_count.checked_mul(image_stride) != Some(image_size)
|| parameter_count == 0 && parameter_size != 0
{
return malformed("v39 table size does not match its compact entry layout");
}
let type_method_start_offset = 16_usize
.checked_add(
type_index_width
.checked_mul(3)
.ok_or_else(|| Error::Malformed("v39 type method offset overflow".to_owned()))?,
)
.and_then(|offset| offset.checked_add(generic_container_index_width))
.ok_or_else(|| Error::Malformed("v39 type method offset overflow".to_owned()))?;
let type_method_count_offset = type_method_start_offset
.checked_add(7 * 4)
.ok_or_else(|| Error::Malformed("v39 type method count offset overflow".to_owned()))?;
let method_token_offset = 4_usize
.checked_add(type_definition_index_width)
.and_then(|offset| offset.checked_add(type_index_width))
.and_then(|offset| offset.checked_add(4))
.and_then(|offset| offset.checked_add(parameter_index_width))
.and_then(|offset| offset.checked_add(generic_container_index_width))
.ok_or_else(|| Error::Malformed("v39 method token offset overflow".to_owned()))?;
if method_token_offset
.checked_add(4)
.is_none_or(|end| end > method_stride)
|| type_method_count_offset
.checked_add(2)
.is_none_or(|end| end > type_stride)
{
return malformed("v39 compact layout fields exceed their records");
}
// Touch the section base so malformed headers fail before any output copy.
Ok(V39Layout {
method_offset,
method_count,
method_stride,
method_token_offset,
type_definition_index_width,
type_offset,
type_count,
type_stride,
type_method_start_offset,
type_method_count_offset,
image_offset,
image_count,
image_stride,
})
}
fn v39_image_methods(data: &[u8], layout: V39Layout, image: usize) -> Result<Vec<usize>> {
let image_base = layout
.image_offset
.checked_add(
image
.checked_mul(layout.image_stride)
.ok_or_else(|| Error::Malformed("v39 image offset overflow".to_owned()))?,
)
.ok_or_else(|| Error::Malformed("v39 image offset overflow".to_owned()))?;
let type_start = read_v39_index(data, image_base + 8, layout.type_definition_index_width)?;
let type_count = read_u32(data, image_base + 8 + layout.type_definition_index_width)? as usize;
let type_end = type_start
.checked_add(type_count)
.ok_or_else(|| Error::Malformed("v39 image type range overflow".to_owned()))?;
if type_end > layout.type_count {
return malformed("v39 image type range exceeds the type table");
}
let mut methods = Vec::new();
for type_index in type_start..type_end {
let type_base = layout
.type_offset
.checked_add(
type_index
.checked_mul(layout.type_stride)
.ok_or_else(|| Error::Malformed("v39 type offset overflow".to_owned()))?,
)
.ok_or_else(|| Error::Malformed("v39 type offset overflow".to_owned()))?;
let method_start = read_u32(data, type_base + layout.type_method_start_offset)?;
let method_count = read_u16(data, type_base + layout.type_method_count_offset)? as usize;
if method_start == u32::MAX || method_count == 0 {
continue;
}
let method_start = method_start as usize;
let method_end = method_start
.checked_add(method_count)
.ok_or_else(|| Error::Malformed("v39 type method range overflow".to_owned()))?;
if method_end > layout.method_count {
return malformed("v39 type method range exceeds the method table");
}
methods.extend(method_start..method_end);
}
Ok(methods)
}
#[allow(clippy::too_many_arguments)]
fn v39_report(
layout: V39Layout,
changed_tokens: usize,
images_with_methods: usize,
visited_methods: usize,
already_correct_before: usize,
correct_after: usize,
transformed_images: usize,
seed: u32,
) -> Report {
Report {
version: 39,
seed: format!("0x{seed:08X}"),
encryption_status: if changed_tokens == 0 {
"clean".to_owned()
} else {
"encrypted".to_owned()
},
images: layout.image_count,
images_with_methods,
types: layout.type_count,
methods: layout.method_count,
visited_methods,
already_correct_before,
correct_after,
changed_tokens,
transformed_images,
}
}
fn restore_v39(data: &[u8], seed: u32) -> Result<(Vec<u8>, Report)> {
let layout = parse_v39(data)?;
let mut owners = vec![u32::MAX; layout.method_count];
let mut output = data.to_vec();
let mut images_with_methods = 0;
let mut visited_methods = 0;
let mut already_correct_before = 0;
let mut correct_after = 0;
let mut changed_tokens = 0;
let mut transformed_images = 0;
for image in 0..layout.image_count {
let methods = v39_image_methods(data, layout, image)?;
if methods.is_empty() {
continue;
}
images_with_methods += 1;
visited_methods += methods.len();
let method_base = *methods.iter().min().ok_or_else(|| {
Error::Malformed("v39 nonempty image lost its method minimum".to_owned())
})?;
let method_last = *methods.iter().max().ok_or_else(|| {
Error::Malformed("v39 nonempty image lost its method maximum".to_owned())
})?;
if method_last - method_base + 1 != methods.len() {
return validation(format!("v39 image {image} method block is not contiguous"));
}
for &method in &methods {
if owners[method] != u32::MAX {
return malformed(format!("v39 method {method} belongs to multiple images"));
}
owners[method] = image as u32;
}
let mut tokens = Vec::with_capacity(methods.len());
let mut clean = true;
for method in methods {
let offset =
layout.method_offset + method * layout.method_stride + layout.method_token_offset;
let token = read_u32(data, offset)?;
if token & 0xff00_0000 != METHOD_TOKEN_TABLE {
return malformed(format!("v39 method {method} has a non-MethodDef token"));
}
let expected = (method - method_base + 1) as u32;
let rid = token & 0x00ff_ffff;
if rid == expected {
already_correct_before += 1;
} else {
clean = false;
}
tokens.push((offset, token, expected));
}
if clean {
correct_after += tokens.len();
continue;
}
transformed_images += 1;
let low = tokens
.iter()
.map(|(_, token, _)| token & 0x00ff_ffff)
.min()
.unwrap();
let high = tokens
.iter()
.map(|(_, token, _)| token & 0x00ff_ffff)
.max()
.unwrap();
if high - low + 1 != tokens.len() as u32 {
return validation(format!(
"v39 image {image} RID interval is not a permutation"
));
}
for (offset, token, expected) in tokens {
let restored = decrypt_rid(token & 0x00ff_ffff, low, high, seed)?;
if restored != expected {
return validation(format!(
"v39 restored RID {restored} != expected {expected}"
));
}
let restored_token = METHOD_TOKEN_TABLE | restored;
if restored_token != token {
output[offset..offset + 4].copy_from_slice(&restored_token.to_le_bytes());
changed_tokens += 1;
}
correct_after += 1;
}
}
if owners.contains(&u32::MAX) {
return malformed("v39 method definitions are not all owned by an image");
}
if visited_methods != layout.method_count || correct_after != layout.method_count {
return validation(format!(
"v39 method coverage mismatch: visited={visited_methods}, correct={correct_after}, total={}",
layout.method_count
));
}
Ok((
output,
v39_report(
layout,
changed_tokens,
images_with_methods,
visited_methods,
already_correct_before,
correct_after,
transformed_images,
seed,
),
))
}
fn discover_v39(data: &[u8]) -> Result<SeedDiscoveryReport> {
let layout = parse_v39(data)?;
let mut reports = Vec::with_capacity(layout.image_count);
for image in 0..layout.image_count {
let methods = v39_image_methods(data, layout, image)?;
if methods.is_empty() {
reports.push(ImageKeyDiscovery {
image,
method_count: 0,
modulus: 0,
clean: true,
seed_residues: Vec::new(),
});
continue;
}
let base = *methods.iter().min().unwrap();
let last = *methods.iter().max().unwrap();
if last - base + 1 != methods.len() {
return validation(format!("v39 image {image} method block is not contiguous"));
}
let values = methods
.iter()
.map(|&method| {
let offset = layout.method_offset
+ method * layout.method_stride
+ layout.method_token_offset;
let token = read_u32(data, offset)?;
if token & 0xff00_0000 != METHOD_TOKEN_TABLE {
return malformed(format!("v39 method {method} has a non-MethodDef token"));
}
Ok((token & 0x00ff_ffff, (method - base + 1) as u32))
})
.collect::<Result<Vec<_>>>()?;
let count = u32::try_from(values.len())
.map_err(|_| Error::Validation("v39 image method count exceeds u32".to_owned()))?;
let clean = values.iter().all(|(rid, expected)| rid == expected);
if clean {
reports.push(ImageKeyDiscovery {
image,
method_count: count,
modulus: count / 2,
clean: true,
seed_residues: Vec::new(),
});
continue;
}
let low = values.iter().map(|(rid, _)| *rid).min().unwrap();
let high = values.iter().map(|(rid, _)| *rid).max().unwrap();
if high - low + 1 != count || count < 2 {
return validation(format!(
"v39 image {image} RID interval is not a permutation"
));
}
let half = count / 2;
let quarter = count / 4;
let mut residues = Vec::new();
for residue in 0..half {
let key = quarter + residue;
if values
.iter()
.all(|(rid, expected)| decrypt_rid_with_key(*rid, low, high, key) == *expected)
{
residues.push(residue);
}
}
reports.push(ImageKeyDiscovery {
image,
method_count: count,
modulus: half,
clean: false,
seed_residues: residues,
});
}
let constraints = reports
.iter()
.filter(|image| !image.clean)
.collect::<Vec<_>>();
let mut candidates = Vec::new();
if let Some(anchor) = constraints.iter().max_by_key(|image| image.modulus) {
// Returning every 32-bit seed is not representable for tiny synthetic
// images (for example, a two-entry image has billions of candidates).
// The caller still tries the known default seed and validates it fully.
if anchor.modulus < 1024 {
return Ok(SeedDiscoveryReport {
version: 39,
images: reports,
seed_candidates: candidates,
});
}
for &residue in &anchor.seed_residues {
let modulus = u64::from(anchor.modulus);
let mut candidate = u64::from(residue);
while candidate <= u64::from(u32::MAX) {
if constraints.iter().all(|image| {
image.modulus != 0
&& image
.seed_residues
.iter()
.any(|value| candidate % u64::from(image.modulus) == u64::from(*value))
}) {
candidates.push(candidate as u32);
}
candidate = candidate.saturating_add(modulus);
}
}
}
candidates.sort_unstable();
candidates.dedup();
Ok(SeedDiscoveryReport {
version: 39,
images: reports,
seed_candidates: candidates,
})
}
#[cfg(test)]
mod tests {
use super::*;
@@ -656,4 +1114,66 @@ mod tests {
Err(Error::Validation(_))
));
}
#[test]
fn restores_compact_v39_method_tokens_without_touching_other_fields() {
let method_stride = 25;
let type_stride = 75;
let image_stride = 34;
let method_offset = 0x400;
let type_offset = 0x600;
let image_offset = 0x700;
let method_count = 7_u32;
let mut data = vec![0_u8; image_offset + image_stride];
put_u32(&mut data, 0, MAGIC);
put_u32(&mut data, 4, 39);
let section = |data: &mut [u8], index: usize, offset: usize, size: usize, count: usize| {
let header = 8 + index * 12;
put_u32(data, header, offset as u32);
put_u32(data, header + 4, size as u32);
put_u32(data, header + 8, count as u32);
};
section(
&mut data,
V39_METHODS,
method_offset,
method_stride * method_count as usize,
method_count as usize,
);
section(&mut data, V39_PARAMETERS, 0x300, 1, 1);
section(&mut data, V39_GENERIC_CONTAINERS, 0x320, 1, 1);
section(&mut data, V39_INTERFACE_OFFSETS, 0x340, 6, 1);
section(&mut data, V39_TYPES, type_offset, type_stride, 1);
section(&mut data, V39_IMAGES, image_offset, image_stride, 1);
data.resize(image_offset + image_stride, 0);
// Compact v39 type definition: firstMethod at offset 23, methodCount at 51.
put_u32(&mut data, type_offset + 23, 0);
put_u16(&mut data, type_offset + 51, method_count as u16);
// Compact v39 image definition: firstTypeIndex (one byte) and typeCount.
data[image_offset + 8] = 0;
put_u32(&mut data, image_offset + 9, 1);
for expected in 1..=method_count {
let token = METHOD_TOKEN_TABLE
| encrypted_rid(expected, method_count, DEFAULT_METHOD_TOKEN_SEED);
put_u32(
&mut data,
method_offset + (expected as usize - 1) * method_stride + 13,
token,
);
}
let (out, report) =
restore_method_tokens(&data, DEFAULT_METHOD_TOKEN_SEED).expect("v39 restore");
assert_eq!(report.version, 39);
assert_eq!(report.changed_tokens, method_count as usize);
for expected in 1..=method_count {
let offset = method_offset + (expected as usize - 1) * method_stride + 13;
assert_eq!(
read_u32(&out, offset).unwrap(),
METHOD_TOKEN_TABLE | expected
);
}
let discovery = discover_method_token_seeds(&data).expect("v39 discovery");
assert_eq!(discovery.version, 39);
assert!(discovery.seed_candidates.is_empty());
}
}
+4 -3
View File
@@ -1,5 +1,6 @@
//! Unity il2cpp metadata de-obfuscation.
//! Unity il2cpp metadata restoration.
mod metadata;
pub mod android;
pub mod windows;
pub use metadata::*;
pub use windows::*;
+5
View File
@@ -0,0 +1,5 @@
//! Windows metadata restoration.
mod metadata;
pub use metadata::*;
+1 -2
View File
@@ -3,8 +3,7 @@ name = "senbei-pe"
version.workspace = true
edition.workspace = true
license.workspace = true
description = "PE detection, unpacking, and validation for Senbei"
description = "PE format parsing and address mapping for Senbei"
[dependencies]
senbei-crypto.workspace = true
thiserror.workspace = true
+137 -3
View File
@@ -1,5 +1,139 @@
//! PE detection, unpacking, and structural validation.
//! Basic PE format parsing and address mapping.
mod engine;
use thiserror::Error;
pub use engine::*;
#[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()
}
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)
}
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))
}
+6 -41
View File
@@ -429,7 +429,7 @@ dependencies = [
]
[[package]]
name = "senbei-android-crypto"
name = "senbei-crypto"
version = "1.2.0"
dependencies = [
"aes",
@@ -437,25 +437,12 @@ dependencies = [
]
[[package]]
name = "senbei-android-elf"
version = "1.2.0"
dependencies = [
"memmap2",
"senbei-android-crypto",
"serde",
"serde_json",
"sha2",
"tempfile",
"thiserror",
]
[[package]]
name = "senbei-android-engine"
name = "senbei-engine"
version = "1.2.0"
dependencies = [
"goblin",
"memmap2",
"senbei-android-crypto",
"senbei-crypto",
"serde",
"serde_json",
"sha2",
@@ -463,21 +450,6 @@ dependencies = [
"thiserror",
]
[[package]]
name = "senbei-android-metadata"
version = "1.2.0"
dependencies = [
"serde",
"thiserror",
]
[[package]]
name = "senbei-crypto"
version = "1.2.0"
dependencies = [
"thiserror",
]
[[package]]
name = "senbei-io"
version = "1.2.0"
@@ -487,11 +459,8 @@ dependencies = [
"indicatif",
"libc",
"owo-colors",
"senbei-android-elf",
"senbei-android-engine",
"senbei-android-metadata",
"senbei-engine",
"senbei-metadata",
"senbei-pe",
"sha2",
"tempfile",
"walkdir",
@@ -502,12 +471,8 @@ dependencies = [
[[package]]
name = "senbei-metadata"
version = "1.2.0"
[[package]]
name = "senbei-pe"
version = "1.2.0"
dependencies = [
"senbei-crypto",
"serde",
"thiserror",
]
@@ -516,9 +481,9 @@ name = "senbei-wasm"
version = "1.2.0"
dependencies = [
"console_error_panic_hook",
"senbei-engine",
"senbei-io",
"senbei-metadata",
"senbei-pe",
"wasm-bindgen",
]
+1 -1
View File
@@ -11,7 +11,7 @@ crate-type = ["cdylib"]
[dependencies]
senbei-io = { path = "../senbei-io" }
senbei-metadata = { path = "../senbei-metadata" }
senbei-pe = { path = "../senbei-pe" }
senbei-engine = { path = "../senbei-engine" }
wasm-bindgen = "0.2"
console_error_panic_hook = "0.1"
+6 -6
View File
@@ -101,12 +101,12 @@ impl MetadataResult {
}
}
fn kind_str(kind: senbei_pe::Kind) -> &'static str {
fn kind_str(kind: senbei_engine::Kind) -> &'static str {
match kind {
senbei_pe::Kind::NativeExe => "native-exe",
senbei_pe::Kind::ManagedExe => "managed-exe",
senbei_pe::Kind::NativeDll => "native-dll",
senbei_pe::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",
}
}
@@ -120,7 +120,7 @@ pub fn detect(input: &[u8]) -> Option<String> {
if senbei_metadata::is_metadata(input) {
return Some("metadata".to_string());
}
senbei_pe::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.
+10 -61
View File
@@ -1,76 +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 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
```
senbei-wasm/ the senbei-wasm cdylib crate (own Cargo.lock, outside the
workspace; depends on the senbei-pe/-io/-metadata crates)
└── src/lib.rs #[wasm_bindgen] bindings: detect / unpack_file /
unpack_file_force_exe / deobfuscate_metadata
web/
├── 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; build from senbei-wasm/)
```
`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.