247 lines
7.9 KiB
Python
247 lines
7.9 KiB
Python
"""
|
||
微盛群消息一轮：按「询价机器人」席位拉新话，写入 inbox。
|
||
|
||
本文件职责：换票查询、归一、入 inbox、推进测服 last_seq。
|
||
Worker 独立线程轮询，不占 HTTP 槽、不堵回调。入队后由 inbox 线程分发。
|
||
禁止：堵回调；禁止推现网官方 SDK 游标；禁止把存档席当发言人。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import logging
|
||
import time
|
||
from typing import Any, Callable, Optional, Sequence
|
||
|
||
from agent.channel.queue import get_message_store
|
||
from agent.channel.wshoto.client import WshotoError, WshotoOpenClient, wshoto_ready
|
||
from agent.channel.wshoto.parse import archive_record_to_inbound
|
||
from agent.channel.wshoto.rooms import (
|
||
list_watched_rooms,
|
||
load_last_seq,
|
||
mark_room_hot,
|
||
save_last_seq,
|
||
)
|
||
from agent.config import Settings, get_settings
|
||
from agent.redis_coord.client import RedisClient
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
# 新群还没有游标时，只回看最近几分钟，避免每轮翻三小时。
|
||
_NEW_ROOM_LOOKBACK_SEC = 180
|
||
|
||
|
||
def archive_seat_userids(settings: Optional[Settings] = None) -> list[str]:
|
||
"""
|
||
会话存档账号 userid。
|
||
|
||
测服群里的「询价机器人」就是这一席；默认读 WECOM_ARCHIVE_SEAT_USER_IDS（AI001）。
|
||
"""
|
||
cfg = settings or get_settings()
|
||
raw = str(getattr(cfg, "wecom_archive_seat_user_ids", "") or "").strip()
|
||
ids: list[str] = []
|
||
for part in raw.replace(";", ",").split(","):
|
||
uid = part.strip()
|
||
if uid and uid not in ids:
|
||
ids.append(uid)
|
||
return ids
|
||
|
||
|
||
def _record_seq(record: dict[str, Any]) -> int:
|
||
try:
|
||
return int(record.get("seq") or 0)
|
||
except (TypeError, ValueError):
|
||
return 0
|
||
|
||
|
||
def ingest_records(
|
||
records: Sequence[dict[str, Any]],
|
||
*,
|
||
seat_userids: Sequence[str],
|
||
min_seq: int = 0,
|
||
store: Any = None,
|
||
) -> tuple[int, int]:
|
||
"""
|
||
把一页管家记录写入 inbox。
|
||
|
||
返回：(入队成功数, 跳过数)。无 sender / 存档席 / 重复 message_id 都算跳过。
|
||
"""
|
||
book = store if store is not None else get_message_store()
|
||
enqueued = 0
|
||
skipped = 0
|
||
for rec in records:
|
||
if _record_seq(rec) <= min_seq:
|
||
skipped += 1
|
||
continue
|
||
msg = archive_record_to_inbound(rec, seat_userids=seat_userids)
|
||
if msg is None:
|
||
skipped += 1
|
||
continue
|
||
ok, reason = book.enqueue_inbound(msg)
|
||
if ok:
|
||
enqueued += 1
|
||
mark_room_hot(msg.chat_id)
|
||
logger.info(
|
||
"存档入队 sender=%s chat=%s type=%s reason=%s",
|
||
msg.sender_id,
|
||
msg.chat_id,
|
||
msg.msg_type,
|
||
reason,
|
||
)
|
||
else:
|
||
skipped += 1
|
||
return enqueued, skipped
|
||
|
||
|
||
def _needs_target(exc: WshotoError) -> bool:
|
||
text = (exc.msg or "").lower()
|
||
return "target_id" in text or "targetid" in text or "房间" in (exc.msg or "")
|
||
|
||
|
||
def poll_once(
|
||
*,
|
||
client: Optional[WshotoOpenClient] = None,
|
||
settings: Optional[Settings] = None,
|
||
redis: Optional[RedisClient] = None,
|
||
store: Any = None,
|
||
now: Optional[Callable[[], float]] = None,
|
||
) -> dict[str, Any]:
|
||
"""
|
||
拉一页并存入 inbox。
|
||
|
||
已有 last_seq：按 seq 增量查。没有：只看最近若干小时，避免从 0 翻全公司历史。
|
||
已登记协同群按房间号拉（不要带存档席 userid，应用群否则给空页）。
|
||
还没有群时才按席位试拉。
|
||
"""
|
||
cfg = settings or get_settings()
|
||
if not wshoto_ready(cfg):
|
||
return {"ok": False, "reason": "not_configured"}
|
||
seats = archive_seat_userids(cfg)
|
||
if not seats:
|
||
logger.warning("微盛轮询：未配置 WECOM_ARCHIVE_SEAT_USER_IDS，只能靠显示名丢掉询价机器人自己的话")
|
||
api = client or WshotoOpenClient(cfg, redis=redis)
|
||
lookback = max(60, int(getattr(cfg, "wshoto_bootstrap_lookback_sec", 10800) or 10800))
|
||
page_size = int(getattr(cfg, "wshoto_page_size", 50) or 50)
|
||
clock = now or time.time
|
||
rooms = list_watched_rooms(redis=redis)
|
||
if rooms:
|
||
# 应用群必须带房间号、不要带存档席 userid，否则管家返回空页。
|
||
records, mode, last = _poll_by_rooms(
|
||
api,
|
||
redis=redis,
|
||
page_size=page_size,
|
||
lookback=lookback,
|
||
clock=clock,
|
||
)
|
||
else:
|
||
seat = seats[0] if seats else ""
|
||
last = load_last_seq(f"seat:{seat or 'default'}", redis=redis)
|
||
time_start = 0
|
||
time_end = 0
|
||
if last <= 0:
|
||
time_end = int(clock())
|
||
time_start = time_end - lookback
|
||
try:
|
||
records = api.list_group_messages(
|
||
userid=seat,
|
||
last_seq=last,
|
||
page_size=page_size,
|
||
msg_time_start=time_start,
|
||
msg_time_end=time_end,
|
||
)
|
||
mode = "seat"
|
||
except WshotoError as exc:
|
||
if not _needs_target(exc):
|
||
logger.warning("微盛按席拉取失败 code=%s msg=%s", exc.code, exc.msg)
|
||
return {"ok": False, "reason": "list_failed", "code": exc.code}
|
||
records, mode, last = _poll_by_rooms(
|
||
api,
|
||
redis=redis,
|
||
page_size=page_size,
|
||
lookback=lookback,
|
||
clock=clock,
|
||
)
|
||
# 按房间拉时各群自己的 last_seq 已经过滤过，这里不要再用席游标误丢。
|
||
ingest_min = 0 if mode == "rooms" else last
|
||
enqueued, skipped = ingest_records(
|
||
records,
|
||
seat_userids=seats,
|
||
min_seq=ingest_min,
|
||
store=store,
|
||
)
|
||
max_seq = last
|
||
for rec in records:
|
||
seq = _record_seq(rec)
|
||
if seq > max_seq:
|
||
max_seq = seq
|
||
if max_seq > last and not rooms:
|
||
save_last_seq(f"seat:{(seats[0] if seats else 'default')}", max_seq, redis=redis)
|
||
logger.info(
|
||
"微盛轮询 mode=%s fetched=%s enqueued=%s skipped=%s last_seq=%s→%s",
|
||
mode,
|
||
len(records),
|
||
enqueued,
|
||
skipped,
|
||
last,
|
||
max_seq,
|
||
)
|
||
return {
|
||
"ok": True,
|
||
"mode": mode,
|
||
"fetched": len(records),
|
||
"enqueued": enqueued,
|
||
"skipped": skipped,
|
||
"last_seq": max_seq,
|
||
}
|
||
|
||
|
||
def _poll_by_rooms(
|
||
api: WshotoOpenClient,
|
||
*,
|
||
redis: Optional[RedisClient],
|
||
page_size: int,
|
||
lookback: int,
|
||
clock: Callable[[], float],
|
||
) -> tuple[list[dict[str, Any]], str, int]:
|
||
"""只拉已登记协同群。应用群不要带存档席 userid，否则管家给空页。"""
|
||
rooms = list_watched_rooms(redis=redis)
|
||
if not rooms:
|
||
logger.warning("微盛按房间拉取：还没有登记过的协同群")
|
||
return [], "rooms", 0
|
||
rooms = sorted(
|
||
rooms,
|
||
key=lambda r: (0 if load_last_seq(f"room:{r}", redis=redis) <= 0 else 1, r),
|
||
)
|
||
merged: list[dict[str, Any]] = []
|
||
floor = 0
|
||
for room in rooms:
|
||
scope = f"room:{room}"
|
||
last = load_last_seq(scope, redis=redis)
|
||
time_start = 0
|
||
time_end = 0
|
||
if last <= 0:
|
||
time_end = int(clock())
|
||
time_start = time_end - min(lookback, _NEW_ROOM_LOOKBACK_SEC)
|
||
try:
|
||
rows = api.list_group_messages(
|
||
userid="",
|
||
target_id=room,
|
||
last_seq=last,
|
||
page_size=page_size,
|
||
msg_time_start=time_start,
|
||
msg_time_end=time_end,
|
||
)
|
||
except WshotoError as exc:
|
||
logger.warning("微盛房间拉取失败 chat=%s code=%s msg=%s", room, exc.code, exc.msg)
|
||
continue
|
||
max_seq = last
|
||
for rec in rows:
|
||
seq = _record_seq(rec)
|
||
if seq > max_seq:
|
||
max_seq = seq
|
||
if max_seq > last:
|
||
save_last_seq(scope, max_seq, redis=redis)
|
||
if last > floor:
|
||
floor = last
|
||
merged.extend(rows)
|
||
return merged, "rooms", floor
|