fix(android): deduplicate only successfully restored inputs

senbei-io/src/android/mod.rs

@@ -15,6 +15,9 @@
 //! All functions in this module are native filesystem orchestration; the web
 //! app (wasm) never touches them.
 
+#[cfg(test)]
+mod tests;
+
 use std::collections::HashSet;
 use std::fs::File;
 use std::io::{BufWriter, Read, Seek, Write};
@@ -480,7 +483,8 @@ fn restore_package_entry<R: Read + Seek>(
     if !is_so && !is_meta {
         return Ok(vec![outcome(EntryKind::So, EntryStatus::NotTarget)]);
     }
-    if !seen.insert(content_identity(&entry_data)) {
+    let identity = content_identity(&entry_data);
+    if seen.contains(&identity) {
         let kind = if is_so {
             EntryKind::So
         } else {
@@ -512,6 +516,12 @@ fn restore_package_entry<R: Read + Seek>(
                             status,
                         });
                     }
+                    if outcomes
+                        .iter()
+                        .all(|outcome| !matches!(outcome.status, EntryStatus::Failed(_)))
+                    {
+                        seen.insert(identity);
+                    }
                     outcomes
                 }
                 Err(error) => vec![outcome(EntryKind::So, EntryStatus::Failed(error))],
@@ -540,6 +550,12 @@ fn restore_package_entry<R: Read + Seek>(
             EntryStatus::Failed(error),
         ),
     };
+    if matches!(
+        kind_and_status.1,
+        EntryStatus::Restored | EntryStatus::Unchanged
+    ) {
+        seen.insert(identity);
+    }
     Ok(vec![outcome(kind_and_status.0, kind_and_status.1)])
 }
 

senbei-io/src/android/tests.rs

@@ -0,0 +1,118 @@
+use super::*;
+use zip::write::SimpleFileOptions;
+
+fn metadata_fixture() -> Vec<u8> {
+    let mut data = vec![0; 0x1a4];
+    let mut put = |offset: usize, value: u32| {
+        data[offset..offset + 4].copy_from_slice(&value.to_le_bytes());
+    };
+    put(0, 0xfab1_1baf);
+    put(4, 31);
+    put(0x30, 0x180);
+    put(0x34, 0x24);
+    put(0xa0, 0x128);
+    put(0xa4, 0x58);
+    put(0xa8, 0x100);
+    put(0xac, 0x28);
+    put(0x10c, 1);
+    put(0x14c, 0);
+    put(0x168, 1);
+    put(0x198, 0x0600_1234);
+    data
+}
+
+fn write_package(path: &Path, entries: &[(&str, &[u8])]) {
+    let mut archive = zip::ZipWriter::new(File::create(path).unwrap());
+    for (name, data) in entries {
+        archive
+            .start_file(*name, SimpleFileOptions::default())
+            .unwrap();
+        archive.write_all(data).unwrap();
+    }
+    archive.finish().unwrap();
+}
+
+#[test]
+fn failed_metadata_output_does_not_poison_deduplication() {
+    let temp = tempfile::tempdir().unwrap();
+    let package = temp.path().join("app.apk");
+    let out = temp.path().join("out");
+    let data = metadata_fixture();
+    write_package(
+        &package,
+        &[
+            ("a/global-metadata.dat", &data),
+            ("b/global-metadata.dat", &data),
+            ("c/global-metadata.dat", &data),
+        ],
+    );
+    std::fs::create_dir_all(out.join("app.apk")).unwrap();
+    std::fs::write(out.join("app.apk/a"), b"blocks the output directory").unwrap();
+    let mut seen = HashSet::new();
+    let outcomes = restore_package(&package, Path::new("app.apk"), &out, &mut seen, false).unwrap();
+    assert_eq!(outcomes.len(), 3);
+    assert!(matches!(outcomes[0].status, EntryStatus::Failed(_)));
+    assert!(matches!(outcomes[1].status, EntryStatus::Restored));
+    assert!(matches!(outcomes[2].status, EntryStatus::Duplicate));
+    assert_eq!(seen.len(), 1);
+    let expected = restore_metadata_bytes(&data).unwrap().0;
+    assert_eq!(std::fs::read(&outcomes[1].dest).unwrap(), expected);
+    assert!(!outcomes[2].dest.exists());
+}
+
+#[test]
+fn repeated_failed_metadata_is_retried() {
+    let temp = tempfile::tempdir().unwrap();
+    let package = temp.path().join("app.apk");
+    let malformed = [0xaf, 0x1b, 0xb1, 0xfa, 31, 0, 0, 0];
+    write_package(
+        &package,
+        &[
+            ("a/global-metadata.dat", &malformed),
+            ("b/global-metadata.dat", &malformed),
+        ],
+    );
+    let mut seen = HashSet::new();
+    let outcomes = restore_package(
+        &package,
+        Path::new("app.apk"),
+        &temp.path().join("out"),
+        &mut seen,
+        false,
+    )
+    .unwrap();
+    assert_eq!(outcomes.len(), 2);
+    assert!(
+        outcomes
+            .iter()
+            .all(|outcome| matches!(outcome.status, EntryStatus::Failed(_)))
+    );
+    assert!(seen.is_empty());
+}
+
+#[test]
+fn unchanged_metadata_is_a_successful_deduplication_source() {
+    let temp = tempfile::tempdir().unwrap();
+    let package = temp.path().join("app.apk");
+    let data = restore_metadata_bytes(&metadata_fixture()).unwrap().0;
+    write_package(
+        &package,
+        &[
+            ("a/global-metadata.dat", &data),
+            ("b/global-metadata.dat", &data),
+        ],
+    );
+    let mut seen = HashSet::new();
+    let outcomes = restore_package(
+        &package,
+        Path::new("app.apk"),
+        &temp.path().join("out"),
+        &mut seen,
+        false,
+    )
+    .unwrap();
+    assert!(matches!(outcomes[0].status, EntryStatus::Unchanged));
+    assert!(matches!(outcomes[1].status, EntryStatus::Duplicate));
+    assert_eq!(seen.len(), 1);
+    assert!(outcomes.iter().all(|outcome| !outcome.dest.exists()));
+}

senbei-io/src/job.rs

@@ -228,10 +228,14 @@ pub fn run_folder_opts(
     for input in &scan.android_so {
         let rel = rel_in_tree(root, input);
         let dest = out_root.join(out_name(&rel));
+        // Failed restores must not hide an identical input encountered later.
         // Unreadable here is fine: the restore reports the same error.
-        if android_dedup
+        let identity = android_dedup
-            && let Ok(identity) = crate::android::file_content_identity(input)
+            .then(|| crate::android::file_content_identity(input).ok())
-            && !android_seen.insert(identity)
+            .flatten();
+        if identity
+            .as_ref()
+            .is_some_and(|id| android_seen.contains(id))
         {
             s.skipped += 1;
             if let Some(log) = &log {
@@ -263,7 +267,13 @@ pub fn run_folder_opts(
                 if let Some(log) = &log {
                     log.step(&format!("OK {rel:?} -> {dest:?} (Android SO)"));
                 }
-                match write_embedded_metadata(embedded, &dest) {
+                let embedded_result = write_embedded_metadata(embedded, &dest);
+                if embedded_result.is_ok()
+                    && let Some(identity) = identity
+                {
+                    android_seen.insert(identity);
+                }
+                match embedded_result {
                     Ok(Some(meta_dest)) => {
                         s.metadata += 1;
                         crate::ui::ok_label(