浅念
c261b2bc9a
- main.py + config.py: CORS origins read from Settings.cors_origins (comma-separated).
Empty falls back to dev origin localhost:5173; never wildcard. Methods locked
to GET/PATCH/POST; allow_credentials=False.
- scraper/media.py: download_url now passes follow_redirects=False to httpx.
OSS URLs return 200 directly; a 3xx is treated as a download failure (caught
by existing except clauses) instead of opening an SSRF vector to internal hosts.
- routers/voice_clips.py: PATCH /voice-clips/{id}/match accepts an optional
If-Match header carrying the client's last-seen updated_at. On mismatch the
row was modified by another request and the server returns 409. Without the
header the behaviour is unchanged (backward compatible). Response now echoes
the post-update updated_at so the client can fold it into the next request.
- routers/organization.py: /api/accounts now UNION ALLs the wechat_account and
whatsapp_account projections at the SQL level and pushes COUNT + pagination
(offset/limit) down to the DB. Stable ordering on (channel, account_id).
Removes the in-memory materialization that broke at scale.
Tests: 395 passed. Added optimistic-lock coverage (If-Match accepted / stale 409 /
absent header backward-compat). Updated existing PATCH tests for the new
updated_at response field.
Co-Authored-By: Claude <noreply@anthropic.com>
82 lines
3.8 KiB
Python
82 lines
3.8 KiB
Python
from functools import lru_cache
|
|
from pathlib import Path
|
|
|
|
from pydantic_settings import BaseSettings, SettingsConfigDict
|
|
|
|
|
|
class Settings(BaseSettings):
|
|
"""Application settings sourced from environment / .env file."""
|
|
|
|
model_config = SettingsConfigDict(
|
|
env_file=".env",
|
|
env_file_encoding="utf-8",
|
|
extra="ignore",
|
|
)
|
|
|
|
yunke_base_url: str = "https://crm2.yunkecn.com"
|
|
yunke_username: str = ""
|
|
yunke_password: str = ""
|
|
db_url: str = "mysql+asyncmy://user:pass@localhost:3306/yunke"
|
|
redis_url: str = "redis://localhost:6379/0"
|
|
media_dir: Path = Path("./media")
|
|
# Fernet key for encrypting stored credentials. Required for real runs;
|
|
# empty default lets tests construct Settings without a key.
|
|
encryption_key: str = ""
|
|
|
|
# Existing-session creds (optional; bypasses programmatic login — e.g. for
|
|
# validation with a live browser session). Read from .env / environment.
|
|
yk_userid: str = ""
|
|
yk_company: str = ""
|
|
yk_cookie: str = ""
|
|
# Shared bearer token for the viewing API. Empty (default) -> auth DISABLED
|
|
# (we trust the network boundary; this is a local/intranet tool). When set,
|
|
# every API request must present this value via ``Authorization: Bearer
|
|
# <token>`` or ``?token=<token>``. See app/api/auth.py.
|
|
api_token: str = ""
|
|
|
|
# Comma-separated list of allowed CORS origins for the viewing API. Empty
|
|
# (default) -> only the dev fallback (``http://localhost:5173``) is allowed;
|
|
# set ``CORS_ORIGINS=https://app.example.com`` in prod to lock cross-origin
|
|
# access down to explicit origins. Never ``*`` (wildcard + credentialed APIs
|
|
# are an open door; we accept no credentials but still restrict origins to
|
|
# avoid reflecting arbitrary attacker-controlled origins in ``Acces-Control-
|
|
# Allow-Origin``).
|
|
cors_origins: str = ""
|
|
|
|
# --- Anti-risk-control / stealth tuning ---------------------------------
|
|
# Tuned 2026-08-11 after triggering 风控 during a heavy session (~500+
|
|
# consecutive requests across groups/WA/voice-clips/SMS in ~1 hour).
|
|
# The previous defaults (1.0s + 0.8s jitter = ~1.0..1.8s cadence) were
|
|
# fast enough to trigger 云客's volume-based risk control when many
|
|
# entities are synced back-to-back. New strategy: slower base cadence +
|
|
# larger jitter (human-like irregularity) + hard per-run cap + longer
|
|
# backoff with more retries (patience over aggression).
|
|
# See app.session.yunke_client for the implementation.
|
|
|
|
# Lower bound on seconds between successive upstream requests.
|
|
# 2.0s base — roughly mimics a human clicking through pages (not a bot
|
|
# hammering at 1 req/s). Combined with jitter this gives ~2.0..4.5s cadence.
|
|
request_min_interval: float = 2.0
|
|
# Random extra 0..jitter seconds. Large jitter (2.5s) ensures the cadence
|
|
# is highly irregular — no periodic signature for 云客 to fingerprint.
|
|
request_jitter: float = 2.5
|
|
# Hard cap: 300 requests per run (~10 minutes at 2s avg). Prevents a
|
|
# runaway sync from hammering 云客 for hours. 0 = unlimited (only for
|
|
# manual debugging). A daily `sync --entity all` will likely need multiple
|
|
# runs (or increase this cap) — by design, to stay under 风控 radar.
|
|
max_requests_per_run: int = 300
|
|
# Exponential backoff: start at 5s, double each retry, cap at 120s.
|
|
# More patient than the old 2s/60s/4 — gives 云客 time to cool down.
|
|
risk_backoff_base: float = 5.0
|
|
risk_backoff_max: float = 120.0
|
|
# 6 retries (was 4) — more chances to recover before giving up.
|
|
# Total worst-case backoff: 5+10+20+40+80+120 = 275s (~4.6 min) before
|
|
# the final attempt; generous recovery window.
|
|
risk_backoff_attempts: int = 6
|
|
|
|
|
|
@lru_cache
|
|
def get_settings() -> Settings:
|
|
"""Return the process-wide Settings singleton."""
|
|
return Settings()
|