backend/app/services/cleanup_service.py

@@ -675,23 +675,29 @@ def cancel_audit_run(session: Session, *, audit_id: int) -> None:
 _ARCHIVE_EXT_FOR_FORMAT = {"zip": ".zip", "rar": ".rar", "7z": ".7z"}
 
 
-def _reextract_archive_to_post(importer, archive_path: Path, post, source_row, artist) -> list[int]:
+def _reextract_archive_to_post(
+    importer, archive_path: Path, post, source_row, artist, images_root: Path,
+) -> list[int]:
     """Extract one stored archive and link its members to `post`.
 
     The stored attachment has no adjacent sidecar (it lives in the sha-addressed
-    attachment store), so reconstruct the post's sidecar from the DB and re-run
+    attachment store). Stage a copy + a reconstructed sidecar UNDER the artist's
-    the normal import path: `attach_in_place` extracts the members and
+    library dir (`images_root/<slug>/<platform>/<post>/`) — the importer
-    `_apply_sidecar` → `find_or_create_post` re-attaches them to the SAME Post
+    re-derives the artist from the path AND copies members relative to it, so the
-    (matched by source_id + external_post_id). Returns the new member image ids.
+    members land in the real library and resolve to the right artist — then re-run
+    `attach_in_place`: the archive extracts and `find_or_create_post` re-attaches
+    the members to the SAME Post (source_id + external_post_id). Removes only the
+    staged archive + sidecar afterward; the imported member files stay. Returns
+    the new member image ids.
     """
     import json
     import shutil
-    import tempfile
 
     from .archive_extractor import detect_archive_format
 
     fmt = detect_archive_format(archive_path)
     ext = _ARCHIVE_EXT_FOR_FORMAT.get(fmt or "", ".zip")
+    platform = source_row.platform if source_row is not None else "imported"
     sidecar = {
         "category": source_row.platform if source_row is not None
         else (post.raw_metadata or {}).get("category"),
@@ -701,12 +707,20 @@ def _reextract_archive_to_post(importer, archive_path: Path, post, source_row, a
         "published_at": post.post_date.isoformat() if post.post_date else None,
         "url": post.post_url,
     }
-    with tempfile.TemporaryDirectory(prefix="fc_reextract_") as td:
+    work = images_root / artist.slug / platform / str(post.external_post_id)
-        tmp = Path(td) / f"archive{ext}"  # clean ext → is_archive + find_sidecar
+    work.mkdir(parents=True, exist_ok=True)
-        shutil.copy2(archive_path, tmp)
+    staged = work / f"archive{ext}"  # clean ext → is_archive + find_sidecar
-        tmp.with_suffix(".json").write_text(json.dumps(sidecar))
+    sidecar_path = staged.with_suffix(".json")
-        res = importer.attach_in_place(tmp, artist=artist, source=source_row)
+    try:
+        shutil.copy2(archive_path, staged)
+        sidecar_path.write_text(json.dumps(sidecar))
+        res = importer.attach_in_place(staged, artist=artist, source=source_row)
         return list(res.member_image_ids or [])
+    finally:
+        # Drop only the staged archive + sidecar; the extracted member files
+        # were copied into the library alongside them and must stay.
+        staged.unlink(missing_ok=True)
+        sidecar_path.unlink(missing_ok=True)
 
 
 def reextract_archive_attachments(session: Session, *, images_root: Path) -> dict:
@@ -727,7 +741,8 @@ def reextract_archive_attachments(session: Session, *, images_root: Path) -> dic
 
     summary = {
         "scanned": 0, "archives": 0, "members_imported": 0,
-        "posts_touched": 0, "skipped_no_post": 0, "errors": 0,
+        "posts_touched": 0, "skipped_no_post": 0, "skipped_no_artist": 0,
+        "errors": 0,
     }
     settings = ImportSettings.load_sync(session)
     importer = Importer(
@@ -756,9 +771,19 @@ def reextract_archive_attachments(session: Session, *, images_root: Path) -> dic
             summary["skipped_no_post"] += 1
             continue
         artist = session.get(Artist, att.artist_id) if att.artist_id else None
+        if artist is None and post.artist_id:
+            artist = session.get(Artist, post.artist_id)
+        if artist is None or not artist.slug:
+            # The importer re-derives the artist from the staged path, so we need
+            # a real artist+slug to anchor under. (Shouldn't happen for
+            # subscription posts; skip rather than orphan the members.)
+            summary["skipped_no_artist"] += 1
+            continue
         source_row = session.get(Source, post.source_id) if post.source_id else None
         try:
-            ids = _reextract_archive_to_post(importer, stored, post, source_row, artist)
+            ids = _reextract_archive_to_post(
+                importer, stored, post, source_row, artist, images_root,
+            )
             session.commit()
         except Exception as exc:  # one bad archive must not strand the rest
             session.rollback()