收口855 provider运行时并同步适配路线图

This commit is contained in:
liuwei
2026-05-07 10:01:57 +08:00
parent 712dda5b41
commit fe8b7171fd
4 changed files with 524 additions and 266 deletions

View File

@@ -5,8 +5,10 @@ from wechat_ipad.providers.legacy_855.friends import FriendMixin
from wechat_ipad.providers.legacy_855.group import ChatroomMixin
from wechat_ipad.providers.legacy_855.login import LoginMixin
from wechat_ipad.providers.legacy_855.message import MessageMixin
from wechat_ipad.providers.legacy_855.runtime import Legacy855RuntimeMixin
from wechat_ipad.providers.legacy_855.tools import ToolMixin
from wechat_ipad.providers.legacy_855.user import UserMixin
from wechat_ipad.providers.legacy_855.base import WechatAPIClientBase
class Legacy855WechatClient(
@@ -17,6 +19,7 @@ class Legacy855WechatClient(
ChatroomMixin,
UserMixin,
ToolMixin,
Legacy855RuntimeMixin,
WechatProviderBase,
):
"""855/859 风格 server 的独立 Provider。
@@ -30,6 +33,19 @@ class Legacy855WechatClient(
provider_name = "legacy_855"
server_type = "legacy_855"
def __init__(self, ip: str, port: int, **kwargs):
"""初始化 855 provider。
说明:
1. 旧 `wechat_ipad/client` 的多继承结构没有显式构造入口,迁移后这里补一个稳定初始化点;
2. 基础连接信息仍写入 `WechatAPIClientBase`,消息发送队列继续沿用 `MessageMixin` 的实现;
3. 运行时状态由 `Legacy855RuntimeMixin` 单独初始化,便于后续 864 provider 走自己的模型。
"""
del kwargs
WechatAPIClientBase.__init__(self, ip, port)
MessageMixin.__init__(self)
self._init_runtime_state()
async def send_at_message(self, wxid: str, content: str, at: list[str]) -> tuple[int, int, int]:
"""发送 @ 消息,兼容现有插件调用方式。"""
if not self.wxid:
@@ -47,4 +63,3 @@ class Legacy855WechatClient(
output = content
return await self.send_text_message(wxid, output, at)

View File

@@ -0,0 +1,386 @@
import asyncio
import time
from typing import Any, Awaitable, Callable
import toml
AsyncCallback = Callable[..., Awaitable[None]]
class Legacy855RuntimeMixin:
"""855/859 风格 server 的运行时编排实现。
设计说明:
1. 855 的差异不只是接口路径,而是“客户端自己负责保活和拉消息”的运行模型;
2. 这里把登录、心跳、历史消息消化、实时轮询、掉线恢复集中收口到 provider 内部;
3. `Robot` 只再关心“登录成功后做什么”“收到消息后怎么处理”,避免后续接 864 时继续改主链路。
"""
def _init_runtime_state(self) -> None:
"""初始化运行时状态字段。
这里不用 `__init__` 参与多继承链,而是由 provider 显式调用:
1. 现有 mixin 组合里已经有消息队列初始化逻辑;
2. 显式初始化更容易看清哪些状态只属于 855 runtime
3. 也能避免后续再因为 MRO 调用顺序引入隐蔽问题。
"""
self._runtime_running = False
self._runtime_recovery_lock = asyncio.Lock()
def stop_runtime(self) -> None:
"""请求停止当前 provider 的运行时主循环。"""
self._runtime_running = False
def is_runtime_running(self) -> bool:
"""返回当前 provider 运行时是否处于运行态。"""
return bool(getattr(self, "_runtime_running", False))
async def run_runtime(
self,
*,
ipad_config: dict,
config_path: str,
logger,
on_login_ready: AsyncCallback,
on_history_message: AsyncCallback,
on_message: AsyncCallback,
on_idle_payload: AsyncCallback | None = None,
on_logout: AsyncCallback | None = None,
on_runtime_state_change: AsyncCallback | None = None,
) -> None:
"""启动 855 provider 的完整运行时。
参数说明:
1. `ipad_config` 与 `config_path` 由上层传入provider 只负责更新和落盘登录态;
2. `on_*` 回调保持尽量少,只暴露业务层真正需要接手的几个时机;
3. 这样既避免 `Robot` 再写协议细节,也不额外引入复杂的事件总线或状态机层。
"""
wxid = str(ipad_config.get("wxid", "") or "").strip()
device_name = str(ipad_config.get("device_name", "") or "").strip()
device_id = str(ipad_config.get("device_id", "") or "").strip()
if not device_name:
device_name = self.create_device_name()
if not device_id:
device_id = self.create_device_id()
await self._ensure_login(
wxid=wxid,
device_name=device_name,
device_id=device_id,
ipad_config=ipad_config,
config_path=config_path,
logger=logger,
)
# 登录后的项目初始化若失败,应直接中断启动:
# 1. 这里会初始化联系人缓存、插件注入、消息存储等关键依赖;
# 2. 如果仅记录异常继续运行,后续消息循环只会在半初始化状态下引发更多连锁问题;
# 3. 因此这里刻意不吞异常,让启动期问题尽早暴露。
await on_login_ready(self.get_login_identity(device_name=device_name, device_id=device_id))
logger.info(f"wechat_ipad登录设备信息: device_name: {device_name} device_id: {device_id}")
logger.info("wechat_ipad登录成功")
logger.info("处理堆积消息中")
await self._drain_startup_history(on_history_message=on_history_message, logger=logger)
logger.info("处理堆积消息完毕")
await self._set_runtime_running(True, on_runtime_state_change=on_runtime_state_change, logger=logger)
heartbeat_task = asyncio.create_task(
self._heartbeat_loop(
heartbeat_func=self.heartbeat,
interval_seconds=60,
logger=logger,
loop_name="wechat_ipad心跳",
on_logout=on_logout,
on_runtime_state_change=on_runtime_state_change,
)
)
heartbeat_long_task = asyncio.create_task(
self._heartbeat_loop(
heartbeat_func=self.heartbeat_long,
interval_seconds=120,
logger=logger,
loop_name="wechat_ipad长连接心跳",
on_logout=on_logout,
on_runtime_state_change=on_runtime_state_change,
)
)
try:
logger.info("开始处理wechat_ipad消息")
while self.is_runtime_running():
try:
data_temp = await self.sync_message()
except Exception as e:
logger.error(f"获取新消息失败 {e}")
recovered = await self._try_recover_from_logout(
reason=e,
logger=logger,
on_logout=on_logout,
on_runtime_state_change=on_runtime_state_change,
)
if not recovered:
if not self.is_runtime_running():
break
await asyncio.sleep(5)
continue
data = data_temp.get("AddMsgs")
if data:
for message in data:
await self._safe_callback(
on_message,
message,
logger=logger,
callback_name="on_message",
)
else:
# 对于 855 而言,`Ret=0 + KeyBuf` 只是正常空轮询返回,不需要额外刷屏日志。
if on_idle_payload:
await self._safe_callback(
on_idle_payload,
data_temp,
logger=logger,
callback_name="on_idle_payload",
)
await asyncio.sleep(2)
finally:
heartbeat_task.cancel()
heartbeat_long_task.cancel()
await asyncio.gather(heartbeat_task, heartbeat_long_task, return_exceptions=True)
await self._set_runtime_running(False, on_runtime_state_change=on_runtime_state_change, logger=logger)
async def _ensure_login(
self,
*,
wxid: str,
device_name: str,
device_id: str,
ipad_config: dict,
config_path: str,
logger,
) -> None:
"""保证当前 provider 已完成登录,并把登录结果写回配置。
这里沿用现有 855 的行为:
1. 优先复用缓存唤醒;
2. 唤醒失败或无缓存时回退到二维码登录;
3. 登录成功后继续把 wxid / device 信息写回 `config.toml`,保持现有部署习惯不变。
"""
if await self.is_logged_in(wxid):
self.wxid = wxid
profile = await self.get_profile()
self.nickname = profile.get("NickName", {}).get("string", "")
self.alias = profile.get("Alias", "")
self.phone = profile.get("BindMobile", {}).get("string", "")
self.signature = profile.get("Signature", "")
logger.info(
f"wechat_ipad登录账号信息: wxid: {self.wxid} 昵称: {self.nickname} 微信号: {self.alias} 手机号: {self.phone}"
)
return
while not await self.is_logged_in(wxid):
uuid = ""
url = ""
try:
if await self.get_cached_info(wxid):
uuid = await self.awaken_login(wxid)
logger.info(f"获取到登录uuid: {uuid}")
else:
uuid, url = await self.get_qr_code(device_id=device_id, device_name=device_name, print_qr=True)
logger.info(f"获取到登录uuid: {uuid}")
logger.info(f"获取到登录二维码: {url}")
except Exception as e:
logger.error(f"登录过程出错: {e}")
uuid, url = await self.get_qr_code(device_id=device_id, device_name=device_name, print_qr=True)
logger.info(f"获取到登录uuid: {uuid}")
logger.info(f"获取到登录二维码: {url}")
while True:
logger.info(f"uuid: {uuid}, url: {url}")
stat, data = await self.check_login_uuid(uuid, device_id=device_id)
if stat:
break
logger.info(f"等待登录中,过期倒计时:{data}")
await asyncio.sleep(5)
self._apply_login_result(data=data, logger=logger)
ipad_config["wxid"] = self.wxid
ipad_config["device_name"] = device_name
ipad_config["device_id"] = device_id
ipad_config["login_time"] = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
with open(config_path, "w", encoding="utf-8") as f:
toml.dump(ipad_config, f)
break
def _apply_login_result(self, *, data: dict, logger) -> None:
"""把登录接口返回的用户信息统一写回当前 provider。"""
acct_section = data.get("acctSectResp", {}) or {}
self.wxid = acct_section.get("userName", "")
self.nickname = acct_section.get("nickName", "")
self.alias = acct_section.get("alias", "")
self.phone = acct_section.get("bindMobile", "")
self.signature = data.get("Signature", "")
logger.info(
f"wechat_ipad登录账号信息: wxid: {self.wxid} 昵称: {self.nickname} 微信号: {self.alias} 手机号: {self.phone}"
)
async def _drain_startup_history(self, *, on_history_message: AsyncCallback, logger) -> None:
"""在实时主循环前先消化堆积消息。
这里保持旧逻辑的退出条件:
1. 连续多次轮询不到 `AddMsgs` 才认为历史堆积已经处理完;
2. 每批历史消息仍交给上层回调决定如何归档;
3. provider 只负责拉取与调度,不把历史消息也混入实时业务处理。
"""
empty_rounds = 0
while True:
data = await self.sync_message()
add_msgs = data.get("AddMsgs")
if not add_msgs:
if empty_rounds > 2:
break
empty_rounds += 1
continue
logger.debug(f"接受到 {len(add_msgs)} 条历史消息,开始仅落库归档")
for raw_message in add_msgs:
await self._safe_callback(
on_history_message,
raw_message,
logger=logger,
callback_name="on_history_message",
)
await asyncio.sleep(1)
async def _heartbeat_loop(
self,
*,
heartbeat_func: Callable[[], Awaitable[bool]],
interval_seconds: int,
logger,
loop_name: str,
on_logout: AsyncCallback | None,
on_runtime_state_change: AsyncCallback | None,
) -> None:
"""统一承接心跳与长心跳循环,减少 855 provider 内部重复代码。"""
logger.info(f"开启{loop_name}")
while self.is_runtime_running():
try:
success = await heartbeat_func()
if success:
logger.debug(f"{loop_name}进行中")
else:
logger.warning(f"{loop_name}失败")
except Exception as e:
logger.error(f"{loop_name}: {e}")
recovered = await self._try_recover_from_logout(
reason=e,
logger=logger,
on_logout=on_logout,
on_runtime_state_change=on_runtime_state_change,
)
if not recovered and not self.is_runtime_running():
break
await asyncio.sleep(interval_seconds)
async def _try_recover_from_logout(
self,
*,
reason: Exception | str,
logger,
on_logout: AsyncCallback | None,
on_runtime_state_change: AsyncCallback | None,
) -> bool:
"""处理 855 provider 的掉线恢复逻辑。
关键点:
1. 855 的掉线恢复是 provider 运行模型的一部分,因此也应该收口在 provider 内部;
2. 这里用锁把恢复流程串行化,避免心跳线程与消息轮询线程同时触发二次登录;
3. 上层只接收一个“检测到掉线”的通知,用于发告警或记录运维日志。
"""
if not self._is_logout_reason(reason):
return False
async with self._runtime_recovery_lock:
# 进入锁后再判断一次,避免并发恢复时第二个协程重复执行二次登录。
if not self.is_runtime_running():
return False
await self._safe_callback(
on_logout,
str(reason),
logger=logger,
callback_name="on_logout",
)
try:
logger.info("定时进行二次登录动作")
resp = await self.twice_auto_auth()
if resp:
logger.info("定时二次登录成功!")
return True
logger.error("定时二次登录失败!")
except Exception as e:
logger.error(f"login_twice_auto_auth error: {e}")
await self._set_runtime_running(False, on_runtime_state_change=on_runtime_state_change, logger=logger)
return False
async def _set_runtime_running(
self,
running: bool,
*,
on_runtime_state_change: AsyncCallback | None,
logger,
) -> None:
"""同步 provider 运行态,并通知上层镜像状态。"""
self._runtime_running = running
if on_runtime_state_change:
await self._safe_callback(
on_runtime_state_change,
running,
logger=logger,
callback_name="on_runtime_state_change",
)
async def _safe_callback(
self,
callback: AsyncCallback | None,
*args: Any,
logger,
callback_name: str,
) -> None:
"""统一保护上层回调,避免单个业务异常直接打断 provider 主循环。"""
if callback is None:
return
try:
await callback(*args)
except Exception as e:
logger.exception(f"执行回调失败: {callback_name}, error: {e}")
@staticmethod
def _is_logout_reason(reason: Exception | str) -> bool:
"""判断当前异常是否属于 855 provider 约定的掉线场景。"""
return "用户可能退出" in str(reason)
def get_login_identity(self, *, device_name: str = "", device_id: str = "") -> dict[str, Any]:
"""返回当前登录身份的轻量归一化结构。
第一阶段先继续使用 dict
1. 便于 `Robot` 直接消费,不额外引入 dataclass
2. 后续如果 864 也需要对齐结构,可以在 provider 内继续增量补字段;
3. 这里同时把 device 信息带上,方便上层统一打印和展示。
"""
return {
"wxid": self.wxid,
"nickname": self.nickname,
"alias": self.alias,
"phone": self.phone,
"signature": getattr(self, "signature", ""),
"device_name": device_name,
"device_id": device_id,
}