fix(repair): dedupe wheel paths and tolerate missing files when deleting bad zips

Author: jakub-kocka

repair_wheels.py

@@ -17,13 +17,34 @@
 import zipfile
 
 from pathlib import Path
+from typing import Set
+from typing import Tuple
 from typing import Union
 
 from colorama import Fore
 
 from _helper_functions import print_color
 
 
+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] = []
+    seen: Set[Tuple[int, int]] = set()
+    for p in sorted(wheels_dir.rglob("*.whl")):
+        try:
+            if not p.is_file():
+                continue
+            st = p.stat()
+            key = (st.st_dev, st.st_ino)
+        except OSError:
+            continue
+        if key in seen:
+            continue
+        seen.add(key)
+        wheels.append(p)
+    return wheels
+
+
 def get_platform() -> str:
     return platform.system()
 
@@ -91,7 +112,7 @@ def fix_universal2_wheel_name(wheel_path: Path, error_msg: str) -> Union[Path, s
     if "'arm64,x86_64'" in error_msg or "'x86_64,arm64'" in error_msg:
         # Missing BOTH architectures - wheel is corrupted, delete it
         print_color("  -> Deleting corrupted wheel (missing native binaries for all architectures)", Fore.RED)
-        wheel_path.unlink()
+        wheel_path.unlink(missing_ok=True)
         return "delete"
     elif "'x86_64'" in error_msg:
         # Missing x86_64, so it only has arm64
@@ -138,8 +159,8 @@ def main() -> None:
     temp_dir: Path = Path("./temp_repair")
     temp_dir.mkdir(exist_ok=True)
 
-    # Find all wheel files
-    wheels: list[Path] = list(wheels_dir.rglob("*.whl"))
+    # Find all wheel files (dedupe: same inode can appear twice via symlinks / layout quirks)
+    wheels: list[Path] = _dedupe_wheel_paths(wheels_dir)
 
     if not wheels:
         print_color(f"No wheels found in {wheels_dir} - nothing to repair", Fore.YELLOW)
@@ -190,7 +211,8 @@ def main() -> None:
         # 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)
-            wheel.unlink()
+            # missing_ok: duplicate paths or prior partial runs can leave nothing to remove
+            wheel.unlink(missing_ok=True)
             deleted_count += 1
             continue
 
@@ -255,7 +277,7 @@ def main() -> None:
             and "This does not look like a platform wheel, no ELF executable" in error_msg
         ):
             print_color("  -> Deleting corrupted wheel", Fore.RED)
-            wheel.unlink()
+            wheel.unlink(missing_ok=True)
             deleted_count += 1
             continue
 
@@ -302,19 +324,19 @@ def main() -> None:
             if repaired:
                 # A repaired wheel was created successfully
                 if repaired.name != wheel.name:
-                    wheel.unlink()  # Remove original
+                    wheel.unlink(missing_ok=True)  # Remove original
                     final_path = wheel.parent / repaired.name
                     repaired.rename(final_path)
                     print_color(f"  -> Replaced with repaired wheel: {repaired.name}", Fore.GREEN)
                 else:
                     # Name unchanged
-                    wheel.unlink()
+                    wheel.unlink(missing_ok=True)
                     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)
-                    final_path.unlink()
+                    final_path.unlink(missing_ok=True)
                     deleted_count += 1
                 else:
                     repaired_count += 1