feat: audit media catalog integrity

rf4_research/media_assets.py

@@ -120,6 +120,40 @@ def review_asset(
     return item
 
 
+def audit_media_catalog(root: Path) -> dict:
+    manifest = json.loads((root / "manifest.json").read_text(encoding="utf-8"))
+    statuses: dict[str, int] = {}
+    issues: list[str] = []
+    referenced: set[str] = set()
+    for item in manifest.get("assets", []):
+        status = str(item.get("status", "unknown"))
+        statuses[status] = statuses.get(status, 0) + 1
+        local_path = item.get("local_path")
+        if status in {"stored", "approved"}:
+            if not isinstance(local_path, str):
+                issues.append(f"{item['asset_url']}: {status} asset has no local_path")
+                continue
+            referenced.add(local_path)
+            target = root / local_path
+            if not target.is_file():
+                issues.append(f"{item['asset_url']}: local file is missing")
+                continue
+            body = target.read_bytes()
+            if hashlib.sha256(body).hexdigest() != item.get("sha256"):
+                issues.append(f"{item['asset_url']}: SHA-256 mismatch")
+            try:
+                width, height, mime = inspect_image(body)
+                if (width, height, mime) != (item.get("width"), item.get("height"), item.get("content_type")):
+                    issues.append(f"{item['asset_url']}: image metadata mismatch")
+            except ValueError as exc:
+                issues.append(f"{item['asset_url']}: {exc}")
+        if status == "approved" and (not item.get("entity_key") or item.get("entity_type") not in {"fish", "waterbody", "tackle", "reference"}):
+            issues.append(f"{item['asset_url']}: approved asset has no valid canonical mapping")
+    files_root = root / "files"
+    orphaned = sorted(str(path.relative_to(root)) for path in files_root.rglob("*") if path.is_file() and str(path.relative_to(root)) not in referenced) if files_root.exists() else []
+    return {"total": sum(statuses.values()), "statuses": statuses, "issues": issues, "orphaned_files": orphaned}
+
+
 def inspect_image(body: bytes) -> tuple[int, int, str]:
     try:
         with Image.open(io.BytesIO(body)) as image:

rf4_research/media_cli.py

@@ -8,7 +8,7 @@ import urllib.error
 import urllib.request
 
 from .community_cli import USER_AGENT, _StrictRedirectHandler, _validate_url_before_io, check_and_reserve, fetch_html, fetch_site_key
-from .media_assets import extract_media_candidates, merge_manifest, reclassify_manifest, review_asset, store_asset
+from .media_assets import audit_media_catalog, extract_media_candidates, merge_manifest, reclassify_manifest, review_asset, store_asset
 
 
 DEFAULT_ROOT = Path("data/media")
@@ -57,10 +57,15 @@ def main(argv: list[str] | None = None) -> int:
     parser.add_argument("--entity-type", choices=("fish", "waterbody", "tackle", "reference"))
     parser.add_argument("--entity-key")
     parser.add_argument("--note")
+    parser.add_argument("--audit", action="store_true", help="Verify manifest metadata, hashes and local files without network access")
     parser.add_argument("--root", type=Path, default=DEFAULT_ROOT)
     parser.add_argument("--state-file", type=Path, default=Path(".cache/community-fetch-state.json"))
     args = parser.parse_args(argv)
     try:
+        if args.audit:
+            report = audit_media_catalog(args.root)
+            print(json.dumps(report, ensure_ascii=False, indent=2))
+            return 1 if report["issues"] or report["orphaned_files"] else 0
         if args.review_url:
             if not args.decision:
                 parser.error("--decision is required with --review-url")