Merge pull request #756 from onekey-sec/754-duplicate-entries

fix(fs): duplicate entries handling in FileSystem API.
onekey-sec · Mar 25, 2024 · 695b59f · 695b59f
2 parents 00de4c4 + 1ba3e04
commit 695b59f
Show file tree

Hide file tree

Showing 8 changed files with 22 additions and 1 deletion.
diff --git a/tests/integration/archive/cpio/cpio_binary/__input__/duplicate_entries.cpio b/tests/integration/archive/cpio/cpio_binary/__input__/duplicate_entries.cpio
diff --git a/...ntegration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/blue b/...ntegration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/blue
diff --git a/...ration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/blue.txt b/...ration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/blue.txt
diff --git a/...tion/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/orange.txt b/...tion/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/orange.txt
diff --git a/...egration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/purple b/...egration/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/purple
@@ -0,0 +1 @@
+purple.txt
diff --git a/...tion/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/purple.txt b/...tion/archive/cpio/cpio_binary/__output__/duplicate_entries.cpio_extract/colors/purple.txt
diff --git a/unblob/file_utils.py b/unblob/file_utils.py
@@ -628,7 +628,7 @@ def open(  # noqa: A003
 
     def unlink(self, path):
         """Delete file within extraction path."""
-        logger.debug("unlink file", file_path=path)
+        logger.debug("unlink file", file_path=path, _verbosity=3)
         safe_path = self._get_extraction_path(path, "unlink")
 
         safe_path.unlink(missing_ok=True)
diff --git a/unblob/handlers/archive/cpio.py b/unblob/handlers/archive/cpio.py
@@ -210,6 +210,11 @@ def dump_entries(self, fs: FileSystem):
             if entry.path.name in ("", "."):
                 continue
 
+            # There are cases where CPIO archives have duplicated entries
+            # We then unlink the files to overwrite them and avoid an error.
+            if not stat.S_ISDIR(entry.mode):
+                fs.unlink(entry.path)
+
             if stat.S_ISREG(entry.mode):
                 fs.carve(entry.path, self.file, entry.start_offset, entry.size)
             elif stat.S_ISDIR(entry.mode):