Fix review findings and remove Uncategorized from sidebar

Backend: cache Google access tokens (drop dead access_token_expires_at,
migration 0007), handle MeTube cleared/canceled events by URL, return
email from /auth/status only when authenticated, move Google base URLs
into settings, run container as non-root.

Frontend: include local feed filters in the query key, remove dead
Saved page and unused assets, drop stale CategoryNav props and classes,
send Content-Type only with a body, remove Uncategorized from sidebar.
This commit is contained in:
vrubelroman 2026-09-17 17:22:22 +00:00
parent 6c704cac97
commit fde9a439df
25 changed files with 420 additions and 174 deletions

View file

@ -117,6 +117,26 @@ def _find_job_by_payload(db: Session, payload: dict) -> DownloadJob | None:
return None
def _find_job_by_key(db: Session, key: str) -> DownloadJob | None:
"""Match a bare MeTube store key (its 'canceled'/'cleared' events carry
just one key, JSON-string-encoded) to our latest job for it. The key is
the URL the download was enqueued under; matching the metube job id first
keeps us tolerant of both shapes."""
job = (
db.query(DownloadJob)
.filter(DownloadJob.metube_job_id == key)
.order_by(DownloadJob.id.desc())
.first()
)
if job is not None:
return job
video = db.query(Video).filter(Video.youtube_url == key).one_or_none()
if video is not None:
return get_latest_job(db, video.id)
return None
async def handle_metube_event(db: Session, event_name: str, raw_payload) -> None:
try:
payload = json.loads(raw_payload) if isinstance(raw_payload, str) else raw_payload
@ -124,21 +144,34 @@ async def handle_metube_event(db: Session, event_name: str, raw_payload) -> None
logger.warning("Could not parse MeTube event payload for %s: %r", event_name, raw_payload)
return
if event_name in ("canceled", "cleared"):
if event_name == "canceled":
if not isinstance(payload, str):
return
job = (
db.query(DownloadJob)
.filter(DownloadJob.metube_job_id == payload)
.order_by(DownloadJob.id.desc())
.first()
)
if job is not None and event_name == "canceled" and job.status in ACTIVE_STATUSES:
job = _find_job_by_key(db, payload)
if job is not None and job.status in ACTIVE_STATUSES:
job.status = "failed"
job.error_message = "Отменено в MeTube"
db.commit()
return
if event_name == "cleared":
# MeTube emits 'cleared' ONLY when a *done* entry is deleted (trash
# via /delete?where=done, or CLEAR_COMPLETED_AFTER): the payload is
# that entry's key (its URL) as a JSON string, one event per item.
# Clearing the queue emits 'canceled' instead. The removed entry is
# usually a download MeTube had before we existed (AGENTS.md rule 10)
# or one we already completed, so an unmatched/empty payload must not
# touch anything: failing all active jobs here would wrongly mark
# downloads MeTube is still running.
if not isinstance(payload, str) or not payload:
return
job = _find_job_by_key(db, payload)
if job is not None and job.status in ACTIVE_STATUSES:
job.status = "failed"
job.error_message = "Очищено в MeTube"
db.commit()
return
if not isinstance(payload, dict):
return

View file

@ -1,6 +1,7 @@
import logging
import os
from datetime import datetime, timezone
import threading
from datetime import datetime, timedelta, timezone
# Google frequently echoes back a scope string that's a superset/reordering
# of what we requested (e.g. we ask for "youtube", Google's token response
@ -34,10 +35,15 @@ SCOPES = [
"https://www.googleapis.com/auth/userinfo.profile",
]
AUTH_URI = "https://accounts.google.com/o/oauth2/auth"
TOKEN_URI = "https://oauth2.googleapis.com/token"
USERINFO_URI = "https://www.googleapis.com/oauth2/v3/userinfo"
REVOKE_URI = "https://oauth2.googleapis.com/revoke"
# Refresh proactively: never hand out a token that could die mid-request.
_ACCESS_TOKEN_REFRESH_BUFFER_SECONDS = 60
# Module-level access-token cache. Google access tokens live ~1h; previously
# every get_credentials() call (sync, unsubscribe, ...) paid for a full token
# exchange with Google. The lock keeps concurrent callers sharing one refresh
# instead of racing each other to the token endpoint.
_credentials_lock = threading.Lock()
_cached_credentials: Credentials | None = None
class OAuthNotConnected(Exception):
@ -49,8 +55,8 @@ def _client_config() -> dict:
"web": {
"client_id": settings.google_client_id,
"client_secret": settings.google_client_secret,
"auth_uri": AUTH_URI,
"token_uri": TOKEN_URI,
"auth_uri": settings.google_auth_uri,
"token_uri": settings.google_token_uri,
"redirect_uris": [settings.google_redirect_uri],
}
}
@ -83,7 +89,7 @@ def exchange_code(code: str, state: str) -> Credentials:
def fetch_userinfo(access_token: str) -> dict:
response = httpx.get(
USERINFO_URI,
settings.google_userinfo_uri,
headers={"Authorization": f"Bearer {access_token}"},
timeout=settings.metube_request_timeout_seconds,
)
@ -93,16 +99,14 @@ def fetch_userinfo(access_token: str) -> dict:
def revoke_token(token: str) -> None:
try:
httpx.post(REVOKE_URI, params={"token": token}, timeout=10)
httpx.post(settings.google_revoke_uri, params={"token": token}, timeout=10)
except Exception:
logger.warning("Failed to revoke Google token", exc_info=True)
def store_credentials(db: Session, google_email: str, credentials: Credentials) -> None:
global _cached_credentials
encrypted = encrypt_token(credentials.refresh_token)
expires_at = credentials.expiry
if expires_at is not None and expires_at.tzinfo is None:
expires_at = expires_at.replace(tzinfo=timezone.utc)
row = db.get(OAuthCredentials, SINGLETON_ID)
if row is None:
@ -111,15 +115,22 @@ def store_credentials(db: Session, google_email: str, credentials: Credentials)
else:
row.google_email = google_email
row.encrypted_refresh_token = encrypted
row.access_token_expires_at = expires_at
db.commit()
# The exchanged credentials carry a fresh access token -- reuse them so
# the immediately following syncs don't pay for a second Google request.
with _credentials_lock:
_cached_credentials = credentials
def clear_credentials(db: Session) -> None:
global _cached_credentials
row = db.get(OAuthCredentials, SINGLETON_ID)
if row is not None:
db.delete(row)
db.commit()
with _credentials_lock:
_cached_credentials = None
def is_connected(db: Session) -> bool:
@ -131,26 +142,48 @@ def get_connected_email(db: Session) -> str | None:
return row.google_email if row else None
def _token_missing_or_expiring(credentials: Credentials) -> bool:
if not credentials.token:
return True
expiry = credentials.expiry
if expiry is None:
# Unknown expiry -- be conservative and refresh.
return True
if expiry.tzinfo is None:
# google-auth reports expiry as a naive UTC datetime.
expiry = expiry.replace(tzinfo=timezone.utc)
return expiry <= datetime.now(timezone.utc) + timedelta(seconds=_ACCESS_TOKEN_REFRESH_BUFFER_SECONDS)
def get_credentials(db: Session) -> Credentials:
global _cached_credentials
row = db.get(OAuthCredentials, SINGLETON_ID)
if row is None:
raise OAuthNotConnected("Google account is not connected")
refresh_token = decrypt_token(row.encrypted_refresh_token)
credentials = Credentials(
token=None,
refresh_token=refresh_token,
token_uri=TOKEN_URI,
client_id=settings.google_client_id,
client_secret=settings.google_client_secret,
scopes=SCOPES,
)
credentials.refresh(GoogleAuthRequest())
expires_at = credentials.expiry
if expires_at is not None and expires_at.tzinfo is None:
expires_at = expires_at.replace(tzinfo=timezone.utc)
row.access_token_expires_at = expires_at
db.commit()
with _credentials_lock:
credentials = _cached_credentials
if credentials is None or credentials.refresh_token != refresh_token:
# No cached token yet, or the account was reconnected with a new
# refresh token. The refresh below does the initial exchange.
credentials = Credentials(
token=None,
refresh_token=refresh_token,
token_uri=settings.google_token_uri,
client_id=settings.google_client_id,
client_secret=settings.google_client_secret,
scopes=SCOPES,
)
_cached_credentials = credentials
return credentials
if _token_missing_or_expiring(credentials):
try:
credentials.refresh(GoogleAuthRequest())
except Exception:
# Don't keep a broken cached object behind.
_cached_credentials = None
raise
return credentials

View file

@ -10,7 +10,10 @@ Verified against the real MeTube source (alexta69/metube, app/main.py + app/ytdl
JSON-*string*-encoded (json.JSONEncoder().encode(...)), not a raw object — must
json.loads() the payload. Relevant keys: id, title, url, status, msg, percent
(float 0-100 or None), filename (already relative to DOWNLOAD_DIR), error.
'canceled'/'cleared' carry just an id (also JSON-string-encoded).
'canceled'/'cleared' carry the download's URL key (JSON-string-encoded),
one event per item; 'cleared' fires only when a done entry is deleted
(trash via /delete?where=done or CLEAR_COMPLETED_AFTER) — clearing the
queue emits 'canceled'.
- MeTube status vocabulary: pending/preparing/scheduled/downloading/postprocessing/
finished/error — mapped to our own vocabulary in sync with download_jobs.
- GET /history returns {"queue": [...], "pending": [...], "done": [...]} of the

View file

@ -195,7 +195,7 @@ def sync_videos(db: Session) -> dict:
continue
duration_seconds = parse_iso8601_duration(item["duration_iso8601"])
youtube_url = f"https://www.youtube.com/watch?v={item['youtube_video_id']}"
youtube_url = settings.youtube_watch_url_template.format(video_id=item["youtube_video_id"])
video = existing.get(item["youtube_video_id"])
if video is None:

View file

@ -7,7 +7,6 @@ from app.config import settings
logger = logging.getLogger(__name__)
API_BASE = "https://www.googleapis.com/youtube/v3"
BATCH_SIZE = 50
@ -68,7 +67,7 @@ def fetch_subscriptions(credentials: Credentials) -> list[dict]:
if page_token:
params["pageToken"] = page_token
response = client.get(f"{API_BASE}/subscriptions", params=params, headers=_headers(credentials))
response = client.get(f"{settings.youtube_api_base_url}/subscriptions", params=params, headers=_headers(credentials))
_raise_for_status(response)
data = response.json()
@ -109,7 +108,7 @@ def fetch_playlist_video_ids(credentials: Credentials, playlist_id: str, max_res
"playlistId": playlist_id,
"maxResults": min(max_results, 50),
}
response = client.get(f"{API_BASE}/playlistItems", params=params, headers=_headers(credentials))
response = client.get(f"{settings.youtube_api_base_url}/playlistItems", params=params, headers=_headers(credentials))
if response.status_code == 404:
return []
_raise_for_status(response)
@ -134,7 +133,7 @@ def fetch_videos_details(credentials: Credentials, video_ids: list[str]) -> list
"id": ",".join(batch),
"maxResults": BATCH_SIZE,
}
response = client.get(f"{API_BASE}/videos", params=params, headers=_headers(credentials))
response = client.get(f"{settings.youtube_api_base_url}/videos", params=params, headers=_headers(credentials))
_raise_for_status(response)
data = response.json()
@ -162,7 +161,7 @@ def fetch_videos_details(credentials: Credentials, video_ids: list[str]) -> list
def unsubscribe(credentials: Credentials, youtube_subscription_id: str) -> None:
with httpx.Client(timeout=settings.metube_request_timeout_seconds) as client:
response = client.delete(
f"{API_BASE}/subscriptions",
f"{settings.youtube_api_base_url}/subscriptions",
params={"id": youtube_subscription_id},
headers=_headers(credentials),
)
@ -183,7 +182,7 @@ def fetch_uploads_playlists(credentials: Credentials, channel_ids: list[str]) ->
"id": ",".join(batch),
"maxResults": BATCH_SIZE,
}
response = client.get(f"{API_BASE}/channels", params=params, headers=_headers(credentials))
response = client.get(f"{settings.youtube_api_base_url}/channels", params=params, headers=_headers(credentials))
_raise_for_status(response)
data = response.json()