feat: 企微菜单新增 AI 分析开关 + 白名单控制

- 新增 ai_whitelist 配置项，仅白名单用户可操作 AI 开关
- 新增 app/services/ai_state.py 运行时开关模块
- 系统管理菜单新增「AI 分析」按钮 (toggle_ai)
- 监控配置、系统状态显示 AI 状态和 AI 标记统计
- pipeline 改用运行时开关 is_ai_enabled()

.env.example

@@ -32,6 +32,8 @@ LOGHIVE_API_KEY=
 
 # AI 分析 (DeepSeek)
 AI_ENABLED=false
+# AI 管理白名单（企微用户ID，逗号分隔），留空表示所有人可操作
+AI_WHITELIST=
 AI_API_KEY=sk-your-deepseek-api-key
 AI_BASE_URL=https://api.deepseek.com/v1
 AI_MODEL=deepseek-chat

app/config.py

@@ -41,6 +41,7 @@ class Settings(BaseSettings):
 
     # AI 分析 (DeepSeek)
     ai_enabled: bool = False
+    ai_whitelist: str = ""
     ai_api_key: str = ""
     ai_base_url: str = "https://api.deepseek.com/v1"
     ai_model: str = "deepseek-chat"

app/crawler/parsers.py

@@ -151,8 +151,21 @@ def parse_dahuagov_detail_pubdate(html: str) -> datetime | None:
 
 async def extract_page_content(url: str, timeout: int = 30) -> str | None:
     """抓取详情页并用 BeautifulSoup 提取正文纯文本"""
+    import urllib.parse
+
     import httpx
 
+    # ---- 站点专用处理 ----
+
+    # ① 政采云 SPA (HTTP API, 无需渲染)
+    if "zfcg.gxzf.gov.cn" in url and "articleId=" in url:
+        return await _extract_zcy_content(url, timeout)
+
+    # ② 大化县政府网 (静态HTML)
+    if "www.gxdh.gov.cn" in url or "gxdh.gov.cn" in url:
+        return await _extract_dahuagov_content(url, timeout)
+
+    # ---- 通用 HTML 提取 ----
     try:
         headers = {
             "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
@@ -172,15 +185,16 @@ async def extract_page_content(url: str, timeout: int = 30) -> str | None:
                 "div.Custom_UnionStyle", "div.pages_content", "div#content",
                 "div.main-content", "article", ".article", ".detail-content",
                 ".text-content", ".news-content", ".detail-article",
+                "div.article-con", ".trs_editor_view", ".TRS_UEDITOR",
+                ".trs_paper_default", ".article-content",
             ]
             for selector in selectors:
                 container = soup.select_one(selector)
                 if container:
-                    # 移除脚本和样式
                     for tag in container.find_all(["script", "style"]):
                         tag.decompose()
                     text = container.get_text(separator="\n", strip=True)
-                    if len(text) > 50:  # 至少50字才算有效正文
+                    if len(text) > 50:
                         return text
 
             # 兜底：取 body 内所有文本
@@ -189,9 +203,8 @@ async def extract_page_content(url: str, timeout: int = 30) -> str | None:
                 for tag in body.find_all(["script", "style", "nav", "footer", "header"]):
                     tag.decompose()
                 text = body.get_text(separator="\n", strip=True)
-                # 移除过长的空白行
                 lines = [l.strip() for l in text.split("\n") if l.strip()]
-                text = "\n".join(lines[:200])  # 最多取前200行
+                text = "\n".join(lines[:200])
                 if len(text) > 50:
                     return text
 
@@ -200,6 +213,109 @@ async def extract_page_content(url: str, timeout: int = 30) -> str | None:
         return None
 
 
+async def _extract_zcy_content(url: str, timeout: int = 30) -> str | None:
+    """从政采云 SPA 隐藏 API 提取公告正文"""
+    import urllib.parse
+
+    import httpx
+    from bs4 import BeautifulSoup
+
+    params = urllib.parse.parse_qs(urllib.parse.urlparse(url).query)
+    article_id = params.get("articleId", [None])[0]
+    parent_id = params.get("parentId", [None])[0]
+    if not article_id:
+        return None
+
+    api_url = "https://zfcg.gxzf.gov.cn/portal/detail"
+    if parent_id:
+        api_url += f"?articleId={article_id}&parentId={parent_id}"
+    else:
+        api_url += f"?articleId={article_id}"
+
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
+        "Accept": "application/json, text/plain, */*",
+        "Referer": url,
+    }
+
+    try:
+        async with httpx.AsyncClient(timeout=timeout) as client:
+            resp = await client.get(api_url, headers=headers)
+            if resp.status_code != 200:
+                return None
+
+            data = resp.json()
+            if not data.get("success"):
+                return None
+
+            content_html = data.get("result", {}).get("data", {}).get("content", "")
+            if not content_html:
+                return None
+
+            soup = BeautifulSoup(content_html, "html.parser")
+            for tag in soup.find_all(["script", "style"]):
+                tag.decompose()
+            text = soup.get_text(separator="\n", strip=True)
+
+            # 清理过短行和多余空白
+            lines = [l.strip() for l in text.split("\n") if len(l.strip()) > 5]
+            text = "\n".join(lines[:300])
+            return text if len(text) > 50 else None
+    except Exception:
+        return None
+
+
+async def _extract_dahuagov_content(url: str, timeout: int = 30) -> str | None:
+    """从大化县政府网详情页提取正文"""
+    import httpx
+    from bs4 import BeautifulSoup
+
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
+        "Accept": "text/html,application/xhtml+xml",
+        "Accept-Language": "zh-CN,zh;q=0.9",
+    }
+
+    try:
+        async with httpx.AsyncClient(timeout=timeout, follow_redirects=True) as client:
+            resp = await client.get(url, headers=headers)
+            if resp.status_code != 200:
+                return None
+
+            soup = BeautifulSoup(resp.text, "html.parser")
+
+            # 大化县政府网正文容器
+            selectors = [
+                "div.article-con", ".trs_editor_view", ".TRS_UEDITOR",
+                ".trs_paper_default", "div.content", "div.TRS_Editor",
+                "div.article-content", "div.Custom_UnionStyle",
+                "div#content", "div.main-content", "article", ".detail-content",
+            ]
+            for selector in selectors:
+                container = soup.select_one(selector)
+                if container:
+                    for tag in container.find_all(["script", "style"]):
+                        tag.decompose()
+                    text = container.get_text(separator="\n", strip=True)
+                    if len(text) > 50:
+                        lines = [l.strip() for l in text.split("\n") if l.strip()]
+                        return "\n".join(lines[:300])
+
+            # 兜底
+            body = soup.find("body")
+            if body:
+                for tag in body.find_all(["script", "style", "nav", "footer", "header"]):
+                    tag.decompose()
+                text = body.get_text(separator="\n", strip=True)
+                lines = [l.strip() for l in text.split("\n") if l.strip()]
+                text = "\n".join(lines[:200])
+                if len(text) > 50:
+                    return text
+            return None
+    except Exception:
+        return None
+
+
 def _generate_hash(ann: dict[str, Any]) -> str:
     content = (
         f"{ann['title']}|{ann['publish_date'].strftime('%Y-%m-%d')}"

app/services/ai_state.py

@@ -0,0 +1,51 @@
+"""AI 运行时状态 — 支持企微菜单动态开关"""
+
+import logging
+
+from app.config import settings
+
+logger = logging.getLogger(__name__)
+
+# 运行时覆盖值，None 表示使用 settings.ai_enabled
+_runtime_override: bool | None = None
+
+
+def is_ai_enabled() -> bool:
+    """获取 AI 分析当前是否启用（考虑运行时覆盖）"""
+    if _runtime_override is not None:
+        return _runtime_override
+    return settings.ai_enabled
+
+
+def set_ai_enabled(enabled: bool) -> bool:
+    """设置 AI 分析运行时开关，返回是否真的发生了变化"""
+    global _runtime_override
+    current = is_ai_enabled()
+    if enabled == current:
+        return False
+    _runtime_override = enabled
+    status = "启用" if enabled else "禁用"
+    logger.info("AI 分析已通过企微菜单%s", status)
+    return True
+
+
+def get_ai_status_text() -> str:
+    """获取 AI 状态文本"""
+    return "已启用" if is_ai_enabled() else "已禁用"
+
+
+def get_whitelist() -> list[str]:
+    """获取 AI 白名单用户列表"""
+    raw = settings.ai_whitelist
+    if not raw:
+        return []
+    return [u.strip() for u in raw.split(",") if u.strip()]
+
+
+def is_whitelisted(user_id: str) -> bool:
+    """检查用户是否在 AI 白名单中"""
+    whitelist = get_whitelist()
+    if not whitelist:
+        # 白名单为空则所有人都可以操作
+        return True
+    return user_id in whitelist

app/services/pipeline.py

@@ -3,6 +3,7 @@ from typing import Any
 
 from app.config import settings
 from app.crawler.base import PipelineConfig, PipelineResult
+from app.services.ai_state import is_ai_enabled
 from app.services.filter_service import dedup_by_hash
 
 logger = logging.getLogger(__name__)
@@ -45,8 +46,8 @@ class PostCrawlPipeline:
         if to_notify:
             to_notify = await self._exclude_sent(to_notify)
 
-        # 5.5 AI 分析（可选）
-        if settings.ai_enabled and to_notify:
+        # 5.5 AI 分析（可选，支持运行时开关）
+        if is_ai_enabled() and to_notify:
             await self._ai_analyze(to_notify)
 
         # 6. Notify

app/wechat/handler.py

@@ -87,6 +87,8 @@ class WeChatMessageHandler:
             return await self._handle_pause_scheduler(from_user)
         elif event_key == "resume_scheduler":
             return await self._handle_resume_scheduler(from_user)
+        elif event_key == "toggle_ai":
+            return await self._handle_toggle_ai(from_user)
         return None
 
     async def handle_text(self, content: str, from_user: str) -> str | None:
@@ -170,15 +172,22 @@ class WeChatMessageHandler:
 
     async def _handle_monitor_config(self, from_user: str) -> str | None:
         import json
+        from app.services.ai_state import get_ai_status_text
         try:
             keywords = settings.crawler_keywords
             sources = json.loads(settings.announcement_sources)
             source_names = "、".join(v["name"] for v in sources.values())
             text = (
+                f"📋 监控配置\n"
+                f"---\n"
                 f"监控关键词: {', '.join(keywords)}\n"
                 f"爬取页数: {settings.crawler_max_pages} 页\n"
                 f"定时规则: {settings.scheduler_cron}\n"
-                f"公告来源: {source_names}"
+                f"公告来源: {source_names}\n"
+                f"---\n"
+                f"🤖 AI 分析: {get_ai_status_text()}\n"
+                f"AI 模型: {settings.ai_model}\n"
+                f"重点标记: {settings.ai_analysis_title}"
             )
             await self.client.send_text(text, from_user)
         except Exception as e:
@@ -189,6 +198,7 @@ class WeChatMessageHandler:
         from app.api.deps import get_db
         from sqlalchemy import func, select
         from app.models.announcement import Announcement
+        from app.services.ai_state import get_ai_status_text
 
         try:
             async for db in get_db():
@@ -212,13 +222,27 @@ class WeChatMessageHandler:
                 )
                 unsent = unsent_result.scalar() or 0
 
+                # AI 标记统计
+                ai_relevant_result = await db.execute(
+                    select(func.count()).where(
+                        Announcement.ai_relevant == True,  # noqa: E712
+                    ).select_from(Announcement)
+                )
+                ai_relevant = ai_relevant_result.scalar() or 0
+
                 scheduler_status = "已启用" if settings.scheduler_enabled else "已禁用"
                 text = (
+                    f"📊 系统状态\n"
+                    f"---\n"
                     f"累计公告: {total} 条\n"
                     f"今日新增: {today} 条\n"
                     f"待推送: {unsent} 条\n"
+                    f"---\n"
                     f"定时任务: {scheduler_status}\n"
-                    f"定时规则: {settings.scheduler_cron}"
+                    f"定时规则: {settings.scheduler_cron}\n"
+                    f"---\n"
+                    f"🤖 AI 分析: {get_ai_status_text()}\n"
+                    f"AI 标记项目: {ai_relevant} 条"
                 )
                 await self.client.send_text(text, from_user)
         except Exception as e:
@@ -304,6 +328,40 @@ class WeChatMessageHandler:
             logger.error(f"暂停定时任务失败: {e}")
             await self.client.send_text(f"操作失败: {e}", from_user)
 
+    async def _handle_toggle_ai(self, from_user: str) -> str | None:
+        from app.services.ai_state import (
+            get_ai_status_text,
+            get_whitelist,
+            is_ai_enabled,
+            is_whitelisted,
+            set_ai_enabled,
+        )
+
+        # 白名单校验
+        whitelist = get_whitelist()
+        if whitelist and not is_whitelisted(from_user):
+            await self.client.send_text(
+                f"⚠️ 你没有权限操作 AI 分析开关\n"
+                f"当前仅以下用户可操作：\n{', '.join(whitelist)}",
+                from_user,
+            )
+            return None
+
+        current = is_ai_enabled()
+        changed = set_ai_enabled(not current)
+        if not changed:
+            await self.client.send_text(
+                f"AI 分析当前已是「{get_ai_status_text()}」状态，无需切换",
+                from_user,
+            )
+            return None
+
+        await self.client.send_text(
+            f"✅ AI 分析已切换为「{get_ai_status_text()}」\n"
+            f"下次爬取触发时生效",
+            from_user,
+        )
+
     async def _handle_resume_scheduler(self, from_user: str) -> str | None:
         try:
             from app.scheduler.jobs import scheduler

app/wechat/menu.py

@@ -46,6 +46,11 @@ MENU = {
                     "type": "click",
                     "key": "sync_holidays",
                 },
+                {
+                    "name": "AI 分析",
+                    "type": "click",
+                    "key": "toggle_ai",
+                },
             ],
         },
     ]