fix(parse): prevent Zip Slip path traversal in _extract_zip (CWE-22)

[ze-mu-zhou]

Description

CodeRepositoryParser._extract_zip() calls zipfile.extractall() without validating member paths. An attacker can craft a malicious zip containing entries like ../../evil.txt, /etc/passwd, C:\evil.txt, ..\..\evil.txt, \server\share\evil.txt, or symlink entries to write files outside the target directory during extraction (Zip Slip, CWE-22).

This PR replaces extractall() with per-member extraction and implements 5-layer defense-in-depth:

1. Skip directory entries (info.is_dir() + stat.S_ISDIR) — directories are created implicitly when extracting their children
2. Skip symlink entries (stat.S_ISLNK) — prevents symlink-based escapes to external paths
3. Pre-extraction rejection of raw filename — raise ValueError if raw path contains .. components, is absolute, or has a drive letter
4. Post-normalization verification — normalize path using the same algorithm as zipfile._extract_member, then verify with Path.is_relative_to()
5. Post-extraction verification — verify the actual path written to disk, clean up and raise ValueError if it escapes

Related Issue

No existing issue. Vulnerability discovered via code audit.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactoring (no functional changes)
• Performance improvement
• Test update

Changes Made

• Replace zip_ref.extractall(target_dir) with per-member safe extraction and 5-layer path validation (code.py +50/-2)
• Add import stat and from pathlib import PurePosixPath
• Add tests/misc/test_extract_zip.py (11 regression tests)
• Update tests/README.md with missing entries for the misc/ test directory

Testing

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have tested this on the following platforms:
  • Linux
  • macOS
  • Windows

11 test cases covering the following scenarios:

Category	Test Case	Expected Behavior
Normal extraction	Multi-file zip, stem name return	No regression, files extracted correctly
../ traversal	../../evil.txt	raise ValueError, no file escape
Absolute path	/etc/passwd	raise ValueError, no file escape
Nested traversal	foo/../../evil.txt	raise ValueError, no file escape
Windows drive path	C:\evil.txt	raise ValueError, no file escape
Backslash traversal	..\..\evil.txt	raise ValueError, no file escape
UNC path	\server\share\evil.txt	raise ValueError, no file escape
Symlink entry	external_attr set to S_IFLNK	Silently skipped, not extracted
.. normalization	./.. (contains .., normalizes to empty)	raise ValueError (fail-fast, .. rejected before normalization)
Directory entry	mydir/	Silently skipped, child files extracted normally

All attack vectors verified as successfully blocked on Windows + Python 3.13.

Checklist

• My code follows the project's coding style
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Additional Notes

Behavioral impact: No change for legitimate zip files. extractall() is replaced with equivalent per-member extract() calls (internally, extractall is just a loop over extract). The only new behavior is raising ValueError on malicious entries.

Performance impact: Negligible. All added checks are in-memory string operations (split, replace, is_relative_to) with no additional I/O.

Fail-fast design decision: The implementation raises ValueError immediately upon detecting a malicious entry, rather than skipping it and continuing. A zip containing malicious entries is considered untrustworthy as a whole — continuing to extract remaining members carries risk. Fail-fast ensures the caller is notified and can handle the exception.

Python compatibility: The project requires Python 3.9+, and zipfile's built-in ../ sanitization was only added in Python 3.12+. On Python 3.9–3.11, extractall() performs no path traversal protection at all, making this fix necessary. Even on Python 3.12+, this fix provides independent defense covering attack vectors that zipfile does not handle (symlink entries, absolute paths, drive letter paths).

[MaojiaSheng]

good job