haha

worker/tasks/boss_recruit.py

@@ -13,7 +13,7 @@ import json
 import random
 import re
 import time
-from typing import Any, Callable, Coroutine, Dict, List
+from typing import Any, Callable, Coroutine, Dict, List, Optional
 
 from common.protocol import TaskType
 from worker.bit_browser import BitBrowserAPI
@@ -170,7 +170,14 @@ class BossRecruitHandler(BaseTaskHandler):
                     self.ensure_not_cancelled(cancel_event)
                     action_state = self._ask_and_exchange_wechat_like_script(tab)
 
-                contacts = self._extract_contacts(messages)
+                panel_texts = self._collect_chat_panel_texts(tab)
+                contacts = self._extract_contacts(messages, extra_texts=panel_texts)
+                contact_written = bool(contacts["wechat"] or contacts["phone"])
+                if has_contact_keyword and not contact_written:
+                    self.logger.warning(
+                        "[%s] 历史消息含联系方式关键词，但未提取到有效联系方式，疑似识别失败",
+                        name,
+                    )
                 collected.append(
                     {
                         "name": name,
@@ -178,6 +185,7 @@ class BossRecruitHandler(BaseTaskHandler):
                         "job_id": friend_job_id,
                         "wechat": contacts["wechat"],
                         "phone": contacts["phone"],
+                        "contact_written": contact_written,
                         "has_contact_keyword": has_contact_keyword,
                         **action_state,
                     }
@@ -189,6 +197,10 @@ class BossRecruitHandler(BaseTaskHandler):
                 err_msg = f"处理第 {i} 个会话出错: {e}"
                 self.logger.error(err_msg)
                 errors.append(err_msg)
+            finally:
+                if i < total:
+                    self.ensure_not_cancelled(cancel_event)
+                    self._sleep_between_sessions()
 
         self.ensure_not_cancelled(cancel_event)
         return {"details": collected, "errors": errors}
@@ -244,6 +256,11 @@ class BossRecruitHandler(BaseTaskHandler):
             return messages
         return []
 
+    @staticmethod
+    def _sleep_between_sessions() -> None:
+        """会话间随机停顿，降低频繁切换带来的风控风险。"""
+        time.sleep(random.uniform(1.8, 4.2))
+
     def _ask_and_exchange_wechat_like_script(self, tab) -> dict:
         state = {
             "asked_wechat": False,
@@ -485,17 +502,40 @@ class BossRecruitHandler(BaseTaskHandler):
 
         return texts
 
+    def _collect_chat_panel_texts(self, tab, max_items: int = 80) -> list[str]:
+        """从当前聊天面板读取可见消息文本，补充接口历史消息。"""
+        texts: list[str] = []
+        try:
+            items = tab.eles("css:.message-item", timeout=1)
+        except Exception:
+            return texts
+
+        if not items:
+            return texts
+
+        for e in items[-max_items:]:
+            try:
+                text = (e.text or "").strip()
+            except Exception:
+                text = ""
+            if text:
+                texts.append(text)
+        return texts
+
     def _has_contact_keyword(self, messages: list) -> bool:
         for text in self._history_texts(messages):
             if any(k in text for k in CONTACT_KEYWORDS):
                 return True
         return False
 
-    def _extract_contacts(self, messages: list) -> dict:
+    def _extract_contacts(self, messages: list, extra_texts: Optional[list[str]] = None) -> dict:
         wechats: list[str] = []
         phones: list[str] = []
+        all_texts = self._history_texts(messages)
+        if extra_texts:
+            all_texts.extend([str(t).strip() for t in extra_texts if str(t).strip()])
 
-        for text in self._history_texts(messages):
+        for text in all_texts:
             wechats.extend(self._extract_wechat(text))
             phones.extend(self._extract_phone(text))
 
@@ -518,7 +558,6 @@ class BossRecruitHandler(BaseTaskHandler):
             r"微信[：:\s]*([a-zA-Z0-9_\-]{6,20})",
             r"wx[：:\s]*([a-zA-Z0-9_\-]{6,20})",
             r"wechat[：:\s]*([a-zA-Z0-9_\-]{6,20})",
-            r"([a-zA-Z][a-zA-Z0-9_\-]{5,19})",
         ]
 
         for pattern in patterns: