feat: remove Asset when there is no references left + bugfixes + more tests

app/api/assets_routes.py

@@ -277,9 +277,13 @@ async def upload_asset(request: web.Request) -> web.Response:
             os.remove(tmp_path)
         msg = str(e)
         if "HASH_MISMATCH" in msg or msg.strip().upper() == "HASH_MISMATCH":
-            return _error_response(400, "HASH_MISMATCH", "Uploaded file hash does not match provided hash.")
+            return _error_response(
+                400,
+                "HASH_MISMATCH",
+                "Uploaded file hash does not match provided hash.",
+            )
         return _error_response(400, "BAD_REQUEST", "Invalid inputs.")
-    except Exception:
+    except Exception as e:
         if tmp_path and os.path.exists(tmp_path):
             os.remove(tmp_path)
         return _error_response(500, "INTERNAL", "Unexpected server error.")
@@ -343,10 +347,14 @@ async def delete_asset(request: web.Request) -> web.Response:
     except Exception:
         return _error_response(400, "INVALID_ID", f"AssetInfo id '{asset_info_id_raw}' is not a valid UUID.")
 
+    delete_content = request.query.get("delete_content")
+    delete_content = True if delete_content is None else delete_content.lower() not in {"0", "false", "no"}
+
     try:
         deleted = await assets_manager.delete_asset_reference(
             asset_info_id=asset_info_id,
             owner_id=UserManager.get_request_user_id(request),
+            delete_content_if_orphan=delete_content,
         )
     except Exception:
         return _error_response(500, "INTERNAL", "Unexpected server error.")

app/assets_manager.py

@@ -27,6 +27,8 @@ from .database.services import (
     create_asset_info_for_existing_asset,
     fetch_asset_info_asset_and_tags,
     get_asset_info_by_id,
+    list_cache_states_by_asset_hash,
+    asset_info_exists_for_hash,
 )
 from .api import schemas_in, schemas_out
 from ._assets_helpers import (
@@ -371,11 +373,40 @@ async def update_asset(
     )
 
 
-async def delete_asset_reference(*, asset_info_id: str, owner_id: str) -> bool:
+async def delete_asset_reference(*, asset_info_id: str, owner_id: str, delete_content_if_orphan: bool = True) -> bool:
+    """Delete single AssetInfo. If this was the last reference to Asset and delete_content_if_orphan=True (default),
+    delete the Asset row as well and remove all cached files recorded for that asset_hash.
+    """
     async with await create_session() as session:
-        r = await delete_asset_info_by_id(session, asset_info_id=asset_info_id, owner_id=owner_id)
+        info_row = await get_asset_info_by_id(session, asset_info_id=asset_info_id)
+        asset_hash = info_row.asset_hash if info_row else None
+        deleted = await delete_asset_info_by_id(session, asset_info_id=asset_info_id, owner_id=owner_id)
+        if not deleted:
+            await session.commit()
+            return False
+
+        if not delete_content_if_orphan or not asset_hash:
+            await session.commit()
+            return True
+
+        still_exists = await asset_info_exists_for_hash(session, asset_hash=asset_hash)
+        if still_exists:
+            await session.commit()
+            return True
+
+        states = await list_cache_states_by_asset_hash(session, asset_hash=asset_hash)
+        file_paths = [s.file_path for s in (states or []) if getattr(s, "file_path", None)]
+
+        asset_row = await get_asset_by_hash(session, asset_hash=asset_hash)
+        if asset_row is not None:
+            await session.delete(asset_row)
+
         await session.commit()
-    return r
+        for p in file_paths:
+            with contextlib.suppress(Exception):
+                if p and os.path.isfile(p):
+                    os.remove(p)
+    return True
 
 
 async def create_asset_from_hash(

app/database/services.py

@@ -36,6 +36,17 @@ async def get_asset_info_by_id(session: AsyncSession, *, asset_info_id: str) ->
     return await session.get(AssetInfo, asset_info_id)
 
 
+async def asset_info_exists_for_hash(session: AsyncSession, *, asset_hash: str) -> bool:
+    return (
+        await session.execute(
+            sa.select(sa.literal(True))
+            .select_from(AssetInfo)
+            .where(AssetInfo.asset_hash == asset_hash)
+            .limit(1)
+        )
+    ).first() is not None
+
+
 async def check_fs_asset_exists_quick(
     session,
     *,
@@ -586,7 +597,7 @@ async def create_asset_info_for_existing_asset(
     tag_origin: str = "manual",
     owner_id: str = "",
 ) -> AssetInfo:
-    """Create a new AssetInfo referencing an existing Asset (no content write)."""
+    """Create a new AssetInfo referencing an existing Asset. If row already exists, return it unchanged."""
     now = utcnow()
     info = AssetInfo(
         owner_id=owner_id,
@@ -597,8 +608,25 @@ async def create_asset_info_for_existing_asset(
         updated_at=now,
         last_access_time=now,
     )
-    session.add(info)
-    await session.flush()  # get info.id
+    try:
+        async with session.begin_nested():
+            session.add(info)
+            await session.flush()  # get info.id
+    except IntegrityError:
+        existing = (
+                await session.execute(
+                    select(AssetInfo)
+                    .where(
+                    AssetInfo.asset_hash == asset_hash,
+                    AssetInfo.name == name,
+                    AssetInfo.owner_id == owner_id,
+                )
+                .limit(1)
+            )
+        ).scalars().first()
+        if not existing:
+            raise RuntimeError("AssetInfo upsert failed to find existing row after conflict.")
+        return existing
 
     # Uncomment next code, and remove code after it, once the hack with "metadata[filename" is not needed anymore
     # if user_metadata is not None: