Files
kefu/wechat_rpa/review_assistant_contacts.py
T
2026-09-21 10:34:06 +08:00

414 lines
20 KiB
Python

"""Read-only managed-account customer recipients for review reminders.
This module does not initialize/decrypt databases, send messages, or choose an
account/contact by display name. UI refreshes run it on a background worker.
"""
from __future__ import annotations
from collections import OrderedDict
from datetime import datetime
from functools import cmp_to_key
import unicodedata
import math
from pathlib import Path
import re
import sqlite3
import threading
from runtime_paths import application_data_dir
_ACCOUNT = re.compile(r"[0-9]{1,24}\Z")
_PRIVATE = re.compile(r"M:[0-9]{1,24}\Z")
_LOCK = threading.RLock()
_CACHE = OrderedDict()
def recipient_peer(account, conv_id) -> str:
"""Parse an exact direct-chat ID; customer membership is checked separately."""
account, conv_id = str(account or ""), str(conv_id or "")
if not _ACCOUNT.fullmatch(account):
return ""
if _PRIVATE.fullmatch(conv_id):
peer = conv_id[2:]
return peer if peer != account else ""
if not conv_id.startswith("S:"):
return ""
parts = conv_id[2:].split("_")
if len(parts) != 2 or any(not _ACCOUNT.fullmatch(part) for part in parts):
return ""
if parts.count(account) != 1:
return ""
return next(part for part in parts if part != account)
def active_context() -> dict:
try:
from wecom_native_sender import discover, readonly_build_validation_scope
except ImportError:
return {"account_id": "", "account_name": "", "ready": False,
"reason": "当前版本不支持医疗助理审核提醒"}
try:
with readonly_build_validation_scope():
identity = discover()
account = str(identity.get("accountId") or "")
if not _ACCOUNT.fullmatch(account):
raise ValueError("未取得可核验的当前企微账号")
return {"account_id": account, "account_name": account, "ready": True,
"reason": "", "pid": int(identity.get("pid") or 0)}
except Exception as exc:
return {"account_id": "", "account_name": "", "ready": False,
"reason": f"当前企微账号尚未核验:{str(exc)[:240]}"}
def _roots():
data, source = application_data_dir(), Path(__file__).resolve().parent
return [data / "wxwork_reply_cache", data / "archive_auto_backup" / "decrypted",
data / "wxwork_decrypted", source / "wxwork_reply_cache",
source / "archive_auto_backup" / "decrypted", source / "wxwork_decrypted"]
def _modified(path):
# An active SQLite WAL can be newer than the main database file.
stamps = [path.stat().st_mtime_ns]
try:
wal_stat = Path(str(path) + "-wal").stat()
if wal_stat.st_size:
stamps.append(wal_stat.st_mtime_ns)
except OSError:
pass
return max(stamps)
def _databases(account, cache_roots=None):
if not _ACCOUNT.fullmatch(str(account)):
raise ValueError("企微账号格式无效")
found = {}
for root in cache_roots if cache_roots is not None else _roots():
directory = Path(root) / account
# Reject a linked directory belonging to a different account.
if directory.resolve().name != account:
continue
for name in ("message.db", "session.db", "user.db"):
path = directory / name
if not path.is_file() or path.resolve().parent != directory.resolve():
continue
try:
if name not in found or _modified(path) > _modified(found[name]):
found[name] = path
except OSError:
continue
if "user.db" not in found:
raise ValueError("当前企微账号的客户关系库尚不可读,请先在微信消息库中完成读取")
return found
def _revision(databases):
result = []
for name, path in sorted(databases.items()):
for candidate in (path, Path(str(path) + "-wal")):
try:
stat = candidate.stat()
# Opening a read-only WAL database may create an empty sidecar.
# An empty WAL carries no changes and is equivalent to no WAL.
if candidate != path and not stat.st_size:
result.append((name, str(candidate), None))
continue
result.append((name, str(candidate.resolve()), stat.st_size, stat.st_mtime_ns,
stat.st_ctime_ns, stat.st_ino))
except OSError:
result.append((name, str(candidate), None))
return tuple(result)
def _open(path):
connection = sqlite3.connect(path.resolve().as_uri() + "?mode=ro", uri=True, timeout=2)
connection.execute("PRAGMA query_only=ON")
return connection
def _names(path, table, wanted, key="id"):
if path is None:
return []
connection = _open(path)
try:
columns = {str(row[1]) for row in connection.execute(f'PRAGMA table_info("{table}")')}
fields = [field for field in wanted if field in columns]
if key not in fields:
return []
return [dict(zip(fields, row)) for row in connection.execute(f"SELECT {','.join(fields)} FROM {table}")]
finally:
connection.close()
def _text(value):
return str(value or "").strip()
def recipient_display(name, identity, name_known=None):
"""Presentation only; never supplies or changes a recipient address."""
name, identity = _text(name), _text(identity)
known = (name not in {"", "微信联系人", "未命名微信联系人", "客户联系人"}) if not isinstance(name_known, bool) else name_known
known = bool(known and name)
number = identity.split(":", 1)[-1]
return {"nameKnown": known,
"displayName": name if known else (f"未读取昵称 · 尾号 {number[-6:]}" if number else "尚未选择")}
def _sort_contacts(items, order):
# Use the same Chinese collation in both source and packaged Qt clients.
# Normalize leading emoji/punctuation so decorative nicknames do not crowd
# the first page. Unknown names remain after named contacts in name order.
try:
from PySide6.QtCore import QCollator, QLocale, Qt
collator = QCollator(QLocale("zh_CN"))
collator.setNumericMode(True)
collator.setCaseSensitivity(Qt.CaseSensitivity.CaseInsensitive)
compare_names = collator.compare
except ImportError:
compare_names = lambda left, right: (left > right) - (left < right)
def normalized(row):
text = unicodedata.normalize("NFKC", _text(row.get("name"))).casefold()
return "".join(ch for ch in text if ch.isalnum()) or text
def compare(left, right):
# Keep sendable and directory entries together even in the all view.
a, b = not left.get("selectable", True), not right.get("selectable", True)
if a != b:
return (a > b) - (a < b)
if order == "recent" and not a:
x, y = left.get("lastTimestamp", 0), right.get("lastTimestamp", 0)
if x != y:
return (y > x) - (y < x)
x, y = not left.get("nameKnown", True), not right.get("nameKnown", True)
if x != y:
return (x > y) - (x < y)
value = compare_names(normalized(left), normalized(right))
if value:
return value
return (left["id"] > right["id"]) - (left["id"] < right["id"])
return sorted(items, key=cmp_to_key(compare))
def _contact_time(value):
try:
stamp = float(value or 0)
if not math.isfinite(stamp) or stamp < 0:
stamp = 0
if stamp > 10 ** 12:
stamp /= 1000
return stamp, datetime.fromtimestamp(stamp).strftime("%Y-%m-%d %H:%M") if stamp else ""
except (ValueError, TypeError, OSError, OverflowError):
return 0, ""
# Observed customer records on supported client 5.0.11.6018 use these values.
# This is an allowlist for that verified build, not a general status-bit enum.
# Unknown values may be displayed but must never enable a recipient.
_VERIFIED_CUSTOMER_STATUSES = frozenset({2049, 2057})
_SYSTEM_CUSTOMER_NAMES = frozenset({
"企业微信团队", "微信团队", "微信支付", "腾讯客服", "腾讯新闻", "企业码会话管理",
})
def _system_customer(names, conv_id=""):
from reply_session_policy import excluded_session_reason, normalize_session_name
return any(normalize_session_name(name) in _SYSTEM_CUSTOMER_NAMES
or excluded_session_reason(name, conv_id) for name in names if name)
def _integer(value):
# Do not turn unknown encodings or floating values into verified flags.
if isinstance(value, bool):
return None
if isinstance(value, int):
return value
if isinstance(value, str) and re.fullmatch(r"-?[0-9]+", value):
return int(value)
return None
def _customer_metadata(user_db, account):
"""Read customer identity metadata only, in one read-only SQLite snapshot."""
connection = _open(user_db)
try:
connection.execute("BEGIN")
columns = {row[1] for row in connection.execute('PRAGMA table_info("external_user_relation_v3")')}
required = {"user_id", "status", "stranger_type", "add_customer_time"}
if not required.issubset(columns):
raise ValueError("当前账号的‘我的客户’关系表缺失或不完整,请先刷新微信消息库;不会使用私人微信通讯录代替")
def rows(table, wanted, key):
fields = {row[1] for row in connection.execute(f'PRAGMA table_info("{table}")')}
selected = [field for field in wanted if field in fields]
if key not in selected:
return []
return [dict(zip(selected, row)) for row in connection.execute(f"SELECT {','.join(selected)} FROM {table}")]
users = {str(row["id"]): row for row in rows(
"user_table", ("id", "name", "real_name", "account", "corp_id"), "id")}
deleted = {str(row["user_id"]) for row in rows("delete_external_userV1", ("user_id",), "user_id")}
blocked = {str(row["value"]) for row in rows("blacklist_external_userids", ("value",), "value")}
coworkers = {str(row["user_id"]) for row in rows("user_dept_tableV2", ("user_id",), "user_id")}
denied = deleted | blocked | coworkers
self_corp = _integer(users.get(account, {}).get("corp_id"))
customers = {}
for row in rows("external_user_relation_v3",
("user_id", "status", "stranger_type", "add_customer_time", "remarks", "real_remarks"), "user_id"):
uid = str(row["user_id"])
if not _ACCOUNT.fullmatch(uid) or uid == account or uid in denied:
continue
user = users.get(uid, {})
peer_corp = _integer(user.get("corp_id"))
if self_corp not in (None, 0) and peer_corp == self_corp:
continue
# A relation to a stranger, or an un-added profile, is not My Customers.
if _integer(row.get("stranger_type")) != 0 or (_integer(row.get("add_customer_time")) or 0) <= 0:
continue
aliases = [_text(row.get(field)) for field in ("remarks", "real_remarks")]
aliases.extend(_text(user.get(field)) for field in ("name", "real_name"))
if _system_customer(aliases):
continue
verified = _integer(row.get("status")) in _VERIFIED_CUSTOMER_STATUSES
customers[uid] = {"name": next((value for value in aliases if value), "客户联系人"),
"aliases": aliases, "verified": verified,
"reason": "" if verified else "客户关系状态尚未核验,暂不能设为医疗助理;请刷新微信消息库后重试"}
own = users.get(account, {})
account_name = _text(own.get("name")) or _text(own.get("real_name")) or account
return account_name, customers
finally:
connection.close()
def _contacts(account, cache_roots=None, *, fresh=False):
databases = _databases(account, cache_roots)
revision = _revision(databases)
key = (account, revision)
with _LOCK:
if not fresh and key in _CACHE:
_CACHE.move_to_end(key)
return _CACHE[key]
account_name, customers = _customer_metadata(databases["user.db"], account)
sessions = {}
for row in _names(databases.get("session.db"), "conversation_table",
("id", "name", "roomname_remark", "is_blocked")):
sessions[str(row["id"])] = row
conversations = {}
if databases.get("message.db") is not None:
connection = _open(databases["message.db"])
try:
columns = {str(row[1]) for row in connection.execute('PRAGMA table_info("message_table")')}
if "conversation_id" not in columns:
raise ValueError("当前账号消息库缺少会话标识,请刷新微信消息库")
timestamp = "MAX(CAST(send_time AS REAL))" if "send_time" in columns else "0"
# Do not read message bodies, previews, phone numbers or binary blobs.
rows = connection.execute(
f"SELECT conversation_id,{timestamp} FROM message_table "
"WHERE conversation_id LIKE 'M:%' OR conversation_id LIKE 'S:%' GROUP BY conversation_id").fetchall()
finally:
connection.close()
for conv_id, stamp in rows:
conv_id = str(conv_id or "")
peer = recipient_peer(account, conv_id)
if peer not in customers:
continue
session = sessions.get(conv_id, {})
if _integer(session.get("is_blocked")) not in (None, 0):
continue
session_names = [_text(session.get(field)) for field in ("name", "roomname_remark")]
if _system_customer(session_names, conv_id):
continue
conversations.setdefault(peer, []).append((conv_id, stamp))
items = []
for uid, customer in customers.items():
targets = conversations.get(uid, []) if customer["verified"] else []
# An actual cached conversation is required. Never synthesize M:/S:.
if not targets:
targets = [("customer:" + uid, 0)]
for conv_id, stamp in targets:
selectable = recipient_peer(account, conv_id) == uid and customer["verified"]
stamp, last_time = _contact_time(stamp)
reason = "" if selectable else customer["reason"] or "尚无可核验的客户单聊。请在当前托管企业微信中与此客户建立单聊并产生一条消息,再刷新消息库。"
kind = "wecom_customer" if conv_id.startswith("S:") else "wechat_customer" if conv_id.startswith("M:") else "customer"
name = customer["name"]
item = {"id": conv_id, "peer_id": uid, "name": name,
"recipient_kind": kind,
"kind": "企业微信客户" if kind == "wecom_customer" else "微信客户" if kind == "wechat_customer" else "我的客户",
"lastTimestamp": stamp, "lastTime": last_time,
"identityLabel": f"我的客户 · 联系编号 {uid}",
"selectable": selectable, "unavailableReason": reason,
"searchText": " ".join([name, conv_id, uid, *customer["aliases"]])}
item.update(recipient_display(name, uid))
items.append(item)
if _revision(databases) != revision:
raise ValueError("读取期间客户或会话数据已更新,请刷新联系人列表后重试")
items = _sort_contacts(items, "recent")
result = (account_name, items)
_CACHE[key] = result
while len(_CACHE) > 8:
_CACHE.popitem(last=False)
return result
def list_recipients(account: str, query="", limit=50, offset=0, *, cache_roots=None, status="all", order="recent") -> dict:
context = active_context()
if not context["ready"] or str(context["account_id"]) != str(account):
return {**context, "ready": False, "items": [], "total": 0, "has_more": False,
"reason": context["reason"] or "当前企业微信账号已变化,请重新读取账号"}
try:
if status not in {"all", "sendable", "directory"} or order not in {"recent", "name"}:
raise ValueError("联系人筛选或排序方式无效")
name, contacts = _contacts(str(account), cache_roots)
current = active_context()
if not current["ready"] or str(current["account_id"]) != str(account):
raise ValueError("读取期间企业微信账号已变化,请重新读取账号")
needle = str(query or "").strip().casefold()[:100]
matches = [item for item in contacts if not needle or needle in item.get("searchText", f"{item['name']} {item['id']}").casefold()]
all_total = len(matches)
all_selectable = sum(bool(item.get("selectable", True)) for item in matches)
if status != "all":
matches = [item for item in matches if bool(item.get("selectable", True)) == (status == "sendable")]
matches = _sort_contacts(matches, order)
offset, limit = max(0, int(offset)), max(1, min(100, int(limit)))
return {**context, "account_name": name, "items": [dict(row) for row in matches[offset:offset + limit]],
"total": len(matches), "offset": offset, "has_more": offset + limit < len(matches),
"selectable_total": sum(bool(item.get("selectable", True)) for item in matches),
"contact_total": sum(not item.get("selectable", True) for item in matches),
"all_total": all_total, "all_selectable_total": all_selectable,
"all_contact_total": all_total - all_selectable, "status": status, "order": order}
except (ValueError, OSError, sqlite3.Error) as exc:
return {**context, "ready": False, "items": [], "total": 0, "has_more": False, "reason": str(exc)[:300]}
def validate_recipient(account: str, conv_id: str, *, cache_roots=None) -> dict:
if not recipient_peer(account, conv_id):
return {"ok": False, "reason": "请选择当前托管账号‘我的客户’中已有单聊的联系人", "recipient": None}
context = active_context()
if not context["ready"] or context["account_id"] != str(account):
return {"ok": False, "reason": context["reason"] or "当前企业微信账号已变化,请重新选择", "recipient": None}
try:
name, contacts = _contacts(str(account), cache_roots, fresh=True)
recipient = next((dict(item) for item in contacts if item["id"] == conv_id), None)
if recipient is None or not recipient.get("selectable", False):
customer = next((item for item in contacts if item.get("peer_id") == recipient_peer(account, conv_id)), {})
raise ValueError(customer.get("unavailableReason") or "所选联系人不在当前账号有效的‘我的客户’单聊中,请刷新后重新选择")
current = active_context()
if not current["ready"] or current["account_id"] != str(account):
raise ValueError("核对期间企业微信账号已变化,请重新选择")
return {"ok": True, "reason": "", "account_id": str(account), "account_name": name, "recipient": recipient}
except (ValueError, OSError, sqlite3.Error) as exc:
return {"ok": False, "reason": str(exc)[:300], "recipient": None}
def notification_preview(account="当前企微账号", recipient=None) -> str:
return ("【客户待审核提醒】\n"
f"托管企微账号:{account}\n"
"客户:示例客户(实际提醒包含会话标识)\n"
"客户咨询:这里显示本轮待处理的客户消息\n"
"待发送回复:这里显示 AI 生成、尚未发送的回复\n"
"审核原因:这里显示命中的审核原因\n"
"请用同一企业微信账号直接回复该客户,结束本次审核。后续自动回复仍逐条风控。\n"
"——以上仅为内容示例,不会发送消息。")