This commit is contained in:
HuangHai
2026-01-16 20:00:41 +08:00
parent 8c2e8bcb0a
commit bd4c2c80e3
5 changed files with 30 additions and 20 deletions

View File

@@ -218,7 +218,7 @@ class TeLaiDianCrawler(BaseCrawler):
logger.info(f"详情页快照已保存: {detail_check_path}")
# 简单验证:如果标题包含 "我的卡券"、"优惠券"、"新人福利" 等,说明点错了
is_wrong_page = await self.check_wrong_page(d, detail_check_path)
is_wrong_page = await self.check_wrong_page(d, detail_check_path, context="detail")
if is_wrong_page:
logger.warning(f"检测到进入了错误页面,尝试返回列表...")
# 尝试点击左上角的返回箭头,如果没有,则执行系统 back
@@ -268,15 +268,24 @@ class TeLaiDianCrawler(BaseCrawler):
self.pending_price_tasks.clear()
logger.info("[收尾] 后台价格任务已全部完成")
async def check_wrong_page(self, d, image_path):
async def check_wrong_page(self, d, image_path, context="list"):
"""
检查是否误触进入了错误的页面(如:我的卡券、活动页等)
"""
data = await self.read_image_kit.check_wrong_page_vlm(image_path)
is_detail = data.get("is_detail_page", True)
if not is_detail:
logger.warning(f"⚠️ 确认进入错误页面: {data.get('page_type')} ({data.get('reason')})")
return not is_detail
page_type = data.get("page_type")
reason = data.get("reason")
if is_detail:
return False
if context == "detail" and page_type == "other":
logger.warning(f"⚠️ 详情页首轮检测结果为 other,将视为加载异常页面,继续尝试解析。原因: {reason}")
return False
logger.warning(f"⚠️ 确认进入错误页面: {page_type} ({reason})")
return True
async def crawl_detail_logic(self, d, station_info):
"""