Sanitize vendored crate checksums for PPA builds

Author: inureyes

.github/workflows/launchpad_ppa.yml

@@ -137,10 +137,12 @@ jobs:
         rm -rf vendor .cargo
         mkdir -p .cargo
         cargo vendor --locked vendor > .cargo/config.toml
+        python3 debian/sanitize-vendor.py
         grep -q 'replace-with = "vendored-sources"' .cargo/config.toml
         grep -q '\[source.vendored-sources\]' .cargo/config.toml
         grep -q '^directory = "vendor"$' .cargo/config.toml
         test -d vendor
+        ! find vendor -name '*.orig' -print | grep -q .
 
     - name: Prepare package version
       if: steps.check_distro.outputs.should_run == 'true'

debian/sanitize-vendor.py

@@ -0,0 +1,37 @@
+#!/usr/bin/env python3
+
+import json
+from pathlib import Path
+
+
+def main() -> int:
+    vendor_root = Path("vendor")
+    if not vendor_root.is_dir():
+        raise SystemExit("vendor directory not found")
+
+    checksum_files = list(vendor_root.rglob(".cargo-checksum.json"))
+    for checksum_file in checksum_files:
+        crate_dir = checksum_file.parent
+        data = json.loads(checksum_file.read_text())
+        files = data.get("files", {})
+
+        removed = []
+        for rel_path in list(files.keys()):
+            if rel_path.endswith(".orig"):
+                removed.append(rel_path)
+                files.pop(rel_path, None)
+                target = crate_dir / rel_path
+                if target.exists():
+                    target.unlink()
+
+        if removed:
+            checksum_file.write_text(
+                json.dumps(data, separators=(",", ":"), sort_keys=True) + "\n"
+            )
+            print(f"sanitized {checksum_file} ({len(removed)} .orig entries)")
+
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())