feat(deep-scan): trigger accepts 'deep'; scan_directory(mode) + enqueue backfill_phash

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-19 15:09:19 -04:00
parent ae67e67299
commit 9535533385
3 changed files with 45 additions and 7 deletions
+3 -4
View File
@@ -15,14 +15,13 @@ import_admin_bp = Blueprint("import_admin", __name__, url_prefix="/api/import")
async def trigger_scan():
body = await request.get_json(silent=True) or {}
mode = body.get("mode", "quick")
if mode != "quick":
# FC-2d will accept 'deep'. For now, reject anything else.
return jsonify({"error": f"mode {mode!r} not supported in FC-2a; use 'quick'"}), 400
if mode not in ("quick", "deep"):
return jsonify({"error": f"mode {mode!r} not supported; use 'quick' or 'deep'"}), 400
# Enqueue the scan task in Celery. The task creates the batch itself.
from ..tasks.scan import scan_directory
async_result = scan_directory.delay(triggered_by="manual")
async_result = scan_directory.delay(triggered_by="manual", mode=mode)
return jsonify({"celery_task_id": async_result.id, "mode": mode}), 202
+12 -3
View File
@@ -34,8 +34,12 @@ def _sync_session_factory():
@celery.task(name="backend.app.tasks.scan.scan_directory", bind=True)
def scan_directory(self, triggered_by: str = "manual") -> int:
"""Walks the import root and creates ImportTasks. Returns the batch id."""
def scan_directory(self, triggered_by: str = "manual",
mode: str = "quick") -> int:
"""Walks the import root and creates ImportTasks. `mode` is 'quick'
or 'deep'; deep also enqueues a library-wide phash backfill and (via
Importer.deep) re-derives on already-imported files. Returns the
batch id."""
SessionLocal = _sync_session_factory()
with SessionLocal() as session:
settings = session.execute(
@@ -46,7 +50,7 @@ def scan_directory(self, triggered_by: str = "manual") -> int:
batch = ImportBatch(
triggered_by=triggered_by,
source_path=str(import_root),
scan_mode="quick", # FC-2a is quick only; FC-2d adds 'deep'
scan_mode=mode,
status="running",
)
session.add(batch)
@@ -84,4 +88,9 @@ def scan_directory(self, triggered_by: str = "manual") -> int:
import_media_file.delay(task.id)
session.commit()
if mode == "deep":
from .maintenance import backfill_phash
backfill_phash.delay()
return batch_id
+30
View File
@@ -90,3 +90,33 @@ async def test_clear_completed(client, db):
resp = await client.post("/api/import/clear-completed", json={"age_days": 7})
body = await resp.get_json()
assert body["deleted"] == 1
@pytest.mark.asyncio
async def test_trigger_accepts_deep(client, monkeypatch):
# Stub the task dispatch — assert the API accepts 'deep' and forwards
# mode, without running the real (eager) scan_directory.
from backend.app.tasks import scan as scan_mod
seen = {}
class _Res:
id = "celery-deep"
def _fake_delay(*, triggered_by, mode):
seen["triggered_by"] = triggered_by
seen["mode"] = mode
return _Res()
monkeypatch.setattr(scan_mod.scan_directory, "delay", _fake_delay)
resp = await client.post("/api/import/trigger", json={"mode": "deep"})
assert resp.status_code == 202
body = await resp.get_json()
assert body["mode"] == "deep"
assert seen == {"triggered_by": "manual", "mode": "deep"}
@pytest.mark.asyncio
async def test_trigger_still_rejects_unknown_mode(client):
resp = await client.post("/api/import/trigger", json={"mode": "wat"})
assert resp.status_code == 400