'commit'

Apps/YeLiTe/Crawler.py

@@ -115,25 +115,40 @@ class YiLaiTeCrawler(BaseCrawler):
                 if await self.redis_kit.get_data(redis_key):
                     continue
                 
+                # 点击进入前截图，用于对比是否成功进入二级页
+                before_click_path = take_screenshot(d, f"before_{clean_station_name(name)}")
+                before_md5 = get_image_content_md5(before_click_path, top_ratio=SAFE_EXCLUDE_RATIO, bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO)
+
                 # 点击进入
                 logger.info(f">>> 发现新场站: {name} 坐标: {point}")
                 d.click(point[0], point[1])
                 await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
                 
                 # 分析详情页 (采用异步后台模式)
-                detail_shot = take_screenshot(d, f"detail_{clean_station_name(name)}_{int(time.time())}.jpg")
-                
-                # 启动后台任务处理详情页
-                task = asyncio.create_task(self.analyze_detail_background(name, detail_shot))
-                background_tasks.append(task)
-                
-                processed_count += 1
-                new_stations_in_page += 1
-                await self.redis_kit.set_data(redis_key, "1", expire=86400*7)
-                
-                # 返回
-                d.press("back")
-                await asyncio.sleep(WAIT_BACK_TO_LIST)
+                detail_shot = take_screenshot(d, f"detail_{clean_station_name(name)}_{int(time.time())}")
+                after_md5 = get_image_content_md5(detail_shot, top_ratio=SAFE_EXCLUDE_RATIO, bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO)
+
+                # 清理临时对比图
+                if os.path.exists(before_click_path): os.remove(before_click_path)
+
+                if before_md5 != after_md5:
+                    logger.info(f"成功进入详情页: {name}")
+                    # 启动后台任务处理详情页
+                    task = asyncio.create_task(self.analyze_detail_background(name, detail_shot))
+                    background_tasks.append(task)
+                    
+                    processed_count += 1
+                    new_stations_in_page += 1
+                    await self.redis_kit.set_data(redis_key, "1", expire=86400*7)
+                    
+                    # 返回
+                    d.press("back")
+                    await asyncio.sleep(WAIT_BACK_TO_LIST)
+                else:
+                    logger.warning(f"点击场站 {name} 后页面似乎未跳转，跳过返回操作")
+                    if os.path.exists(detail_shot): os.remove(detail_shot)
+                    # 即使没进去，也记录一下，避免短时间内重复尝试
+                    await self.redis_kit.set_data(redis_key, "1", expire=3600)
             
             if new_stations_in_page == 0 and stations:
                 no_new_data_count += 1