fix: BadZip logic fixed and enhanced

Author: jakub-kocka

repair_wheels.py

@@ -27,6 +27,34 @@
 from _helper_functions import print_color
 
 
+def _wheel_archive_is_valid(path: Path) -> bool:
+    """Return True only if the file is a readable zip with a valid central directory.
+
+    zipfile.is_zipfile() only checks the local header magic; wheels can still be
+    truncated or corrupt such that opening fails with BadZipFile (e.g. bad central
+    directory magic) — which delocate then surfaces.
+    """
+    if not zipfile.is_zipfile(path):
+        return False
+    try:
+        with zipfile.ZipFile(path, "r") as zf:
+            zf.namelist()
+    except zipfile.BadZipFile:
+        return False
+    return True
+
+
+def _stderr_indicates_bad_zip(error_msg: str) -> bool:
+    """True if repair tool output indicates an unreadable/corrupt zip archive."""
+    if not error_msg:
+        return False
+    return (
+        "BadZipFile" in error_msg
+        or "Bad magic number for central directory" in error_msg
+        or "File is not a zip file" in error_msg
+    )
+
+
 def _dedupe_wheel_paths(wheels_dir: Path) -> List[Path]:
     """Collect *.whl under wheels_dir once per inode (rglob can list the same file twice via symlinks)."""
     wheels: List[Path] = []
@@ -209,10 +237,10 @@ def main() -> None:
                 skipped_count += 1
                 continue
 
-        # PEP 427: wheels are zip files; invalid magic usually means truncated/corrupt CI artifact
-        if not zipfile.is_zipfile(wheel):
-            print_color("  -> Deleting file (not a valid zip / wheel archive)", Fore.RED)
-            # missing_ok: duplicate paths or prior partial runs can leave nothing to remove
+        # PEP 427: wheels are zip files; truncated/corrupt CI artifacts may pass is_zipfile
+        # but fail on central directory (delocate: BadZipFile).
+        if not _wheel_archive_is_valid(wheel):
+            print_color("  -> Deleting file (not a valid / readable zip wheel archive)", Fore.RED)
             wheel.unlink(missing_ok=True)
             deleted_count += 1
             continue
@@ -241,6 +269,15 @@ def main() -> None:
         # Check for errors
         error_msg = result.stderr.strip() if result.stderr else ""
 
+        # Corrupt zip / bad central directory (delocate opens the wheel as a zip)
+        if _stderr_indicates_bad_zip(error_msg):
+            print_color("  -> Deleting file (repair tool reported corrupt zip archive)", Fore.RED)
+            for old_wheel in temp_dir.glob("*.whl"):
+                old_wheel.unlink()
+            wheel.unlink(missing_ok=True)
+            deleted_count += 1
+            continue
+
         # Special handling for incorrectly tagged universal2 wheels on macOS
         if (
             current_platform == "Darwin"
@@ -350,8 +387,8 @@ def main() -> None:
                     repaired.rename(wheel)
                     final_path = wheel
                     print_color(f"  -> Repaired successfully: {repaired.name}", Fore.GREEN)
-                if not zipfile.is_zipfile(final_path):
-                    print_color("  -> Deleting repaired output (not a valid zip archive)", Fore.RED)
+                if not _wheel_archive_is_valid(final_path):
+                    print_color("  -> Deleting repaired output (not a valid / readable zip archive)", Fore.RED)
                     final_path.unlink(missing_ok=True)
                     deleted_count += 1
                 else:

test_wheels_install.py

@@ -34,6 +34,22 @@
 WHEELS_DIR = Path("./downloaded_wheels")
 
 
+def wheel_archive_is_readable(path: Path) -> bool:
+    """True if the file is a zip with a readable central directory (a valid wheel container).
+
+    zipfile.is_zipfile() only checks the leading magic; truncated or corrupt wheels can still
+    fail at install with pip's ``Wheel '...' ... is invalid`` (BadZipFile inside pip).
+    """
+    if not zipfile.is_zipfile(path):
+        return False
+    try:
+        with zipfile.ZipFile(path, "r") as zf:
+            zf.namelist()
+    except zipfile.BadZipFile:
+        return False
+    return True
+
+
 def get_python_version_tag() -> str:
     """Get the Python version tag (e.g., '311' for Python 3.11)."""
     return f"{sys.version_info.major}{sys.version_info.minor}"
@@ -148,9 +164,15 @@ def is_compatibility_error(error_message: str) -> bool:
 
 def is_corrupt_wheel_archive_error(error_message: str) -> bool:
     """True if pip failed because the file is not a readable ZIP / wheel archive."""
+    if not error_message:
+        return False
+    # pip.exceptions.InvalidWheel -> "Wheel 'pkg' located at <path> is invalid."
+    if "Wheel '" in error_message and " is invalid." in error_message:
+        return True
     markers = (
         "BadZipFile",
         "Bad magic number for file header",
+        "Bad magic number for central directory",
         "has an invalid wheel",
         "zipfile.BadZipFile",
     )
@@ -216,9 +238,12 @@ def main() -> int:
     print_color("---------- INSTALL WHEELS ----------")
 
     for wheel_path in wheels_to_install:
-        if not zipfile.is_zipfile(wheel_path):
+        if not wheel_archive_is_readable(wheel_path):
             discarded_corrupt += 1
-            discard_corrupt_wheel(wheel_path, "invalid zip — not a valid wheel file (PEP 427)")
+            discard_corrupt_wheel(
+                wheel_path,
+                "unreadable / corrupt zip — not a valid wheel archive (PEP 427)",
+            )
             continue
 
         success, error_message = install_wheel(wheel_path)
@@ -233,8 +258,8 @@ def main() -> int:
             wheel_path.unlink()
             print_color(f"-- {wheel_path.name} (compatibility constraint)", Fore.YELLOW)
         elif is_corrupt_wheel_archive_error(error_message):
-            # Truncated/corrupt artifact or bad repair output; same handling as incompatible:
-            # drop from this test artifact so CI can continue (see module docstring).
+            # Truncated/corrupt artifact or bad repair output; drop from this test artifact
+            # so CI can continue (see module docstring).
             discarded_corrupt += 1
             discard_corrupt_wheel(wheel_path, "invalid / corrupt zip (pip could not read wheel)")
         else: