'commit'

Apps/TeLaiDian/Crawler.py

@@ -248,17 +248,80 @@ class TeLaiDianCrawler(BaseCrawler):
                     entrance_point = (match_res[0], match_res[1])
 
         if not entrance_point:
-            logger.error("[详情页] 未能通过 jgxx.jpg 找到价格入口，本次流程直接中止，需人工检查详情页布局或模板。")
+            logger.error("[详情页] 未能通过 jgxx.jpg 找到价格入口，本次流程直接中止，需人工检查详情页布局或滚动距离。")
             raise RuntimeError("TeLaiDian: 价格入口模板匹配失败")
 
-        logger.info(f"[详情页] 通过 jgxx.jpg 成功找到价格入口，点击坐标: {entrance_point}，即将进入电价页")
+        before_price_path = take_screenshot(d, f"tld_detail_price_before_{int(time.time())}.jpg")
-        d.click(int(entrance_point[0]), int(entrance_point[1]))
+        before_md5 = get_image_content_md5(
+            before_price_path,
+            top_ratio=SAFE_EXCLUDE_RATIO,
+            bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO,
+        )
+
+        ex, ey = int(entrance_point[0]), int(entrance_point[1])
+        click_x = ex
+        offset_y = int(h * 0.1)
+        click_y = min(h - 10, ey + offset_y)
+        logger.info(f"[详情页] 通过 jgxx.jpg 成功找到价格入口，模板中心: ({ex}, {ey})，调整后点击坐标: ({click_x}, {click_y})，即将进入电价页")
+        d.click(click_x, click_y)
         await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
 
+        after_price_path = take_screenshot(d, f"tld_detail_price_after_{int(time.time())}.jpg")
+        after_md5 = get_image_content_md5(
+            after_price_path,
+            top_ratio=SAFE_EXCLUDE_RATIO,
+            bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO,
+        )
+
+        entered_price_page = False
+
+        if before_md5 and after_md5 and before_md5 != after_md5:
+            entered_price_page = True
+            logger.info("[电价页] 模板点击后页面内容发生变化，判定已进入电价详情页。")
+        else:
+            logger.warning("[电价页] 模板点击后页面内容无明显变化，尝试使用 VLM 兜底寻找价格入口。")
+            try:
+                vlm_res = await self.read_image_kit.find_price_entrance_vlm(before_price_path)
+            except Exception as e:
+                vlm_res = {}
+                logger.error(f"[电价页] VLM 寻找价格入口失败: {e}")
+
+            point = vlm_res.get("point") if isinstance(vlm_res, dict) else None
+            if vlm_res.get("found") and point and len(point) >= 2:
+                w, h = d.window_size()
+                vx = int(point[0] * w / 1000)
+                vy = int(point[1] * h / 1000)
+                if vy < int(h * SAFE_EXCLUDE_RATIO):
+                    vy = int(h * SAFE_EXCLUDE_RATIO)
+                if vy > int(h * (1 - BOTTOM_SAFE_EXCLUDE_RATIO)):
+                    vy = int(h * (1 - BOTTOM_SAFE_EXCLUDE_RATIO))
+                logger.info(f"[电价页] 使用 VLM 兜底点击价格入口: ({vx}, {vy})，原因: {vlm_res.get('reason')}")
+                d.click(vx, vy)
+                await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
+
+                vlm_after_path = take_screenshot(d, f"tld_detail_price_after_vlm_{int(time.time())}.jpg")
+                vlm_after_md5 = get_image_content_md5(
+                    vlm_after_path,
+                    top_ratio=SAFE_EXCLUDE_RATIO,
+                    bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO,
+                )
+                if before_md5 and vlm_after_md5 and before_md5 != vlm_after_md5:
+                    entered_price_page = True
+                    logger.info("[电价页] VLM 兜底点击后页面内容发生变化，判定已进入电价详情页。")
+            else:
+                logger.error(f"[电价页] VLM 未能找到可靠的价格入口: {vlm_res}")
+
+        if not entered_price_page:
+            logger.error("[电价页] 多轮尝试后页面仍未变化，疑似点击未生效或入口不可用。")
+            raise RuntimeError("TeLaiDian: 价格入口点击后页面未变化")
+
+        logger.info("[电价页] 已成功进入电价详情页，调试模式下在此主动停止，供人工检查页面内容。")
+        raise RuntimeError("TeLaiDian: 调试模式已在电价页入口处停止")
+
         # 4. 进入分时电价页面后的处理
         try:
             logger.info("[电价页] 已点击价格入口，等待小程序自动滚动定位完成 (4秒)...")
-            await asyncio.sleep(4.0) 
+            await asyncio.sleep(4.0)
             logger.info("[电价页] 开始执行两次大幅向下拉动，目标是回到 00:00 时段顶部")
             for idx in range(2):
                 logger.info(f"[电价页] 第 {idx+1} 次向下滑动(d.swipe_ext(\"down\", scale=0.9))")
@@ -271,11 +334,80 @@ class TeLaiDianCrawler(BaseCrawler):
         all_prices = []
         last_price_md5 = None
         price_page_count = 0
-        max_price_pages = 4 # 增加到4页确保覆盖 00:00-24:00
+        max_price_pages = 4
-        screenshot_tasks = [] # 用于异步分析图片的任务列表
+        screenshot_tasks = []
-        temp_screenshots = [] # 记录临时文件以便后续清理
+        temp_screenshots = []
 
         logger.info("[电价页] 开始循环截图（UI操作优先，后台并行分析）...")
+        try:
+            while price_page_count < max_price_pages:
+                price_screen_path = take_screenshot(d, f"tld_detail_price_{int(time.time())}_{price_page_count}.jpg")
+
+                curr_md5 = get_image_content_md5(price_screen_path, top_ratio=0.2, bottom_ratio=0.2)
+                if curr_md5 == last_price_md5:
+                    logger.info("价格页面内容无变化，判定已触底")
+                    if os.path.exists(price_screen_path):
+                        os.remove(price_screen_path)
+                    break
+                last_price_md5 = curr_md5
+                temp_screenshots.append(price_screen_path)
+
+                logger.info(f"已截取价格详情页第 {price_page_count + 1} 页: {price_screen_path}，启动后台异步分析")
+                task = asyncio.create_task(self.read_image_kit.analyze_detail_price(price_screen_path))
+                screenshot_tasks.append(task)
+
+                logger.info("向上滚动列表，准备截取下一屏价格...")
+                d.swipe_ext("up", scale=0.8)
+                await asyncio.sleep(1.0)
+                price_page_count += 1
+
+            if screenshot_tasks:
+                logger.info(f"UI 操作已完成，等待 {len(screenshot_tasks)} 个后台分析任务结束...")
+                results = await asyncio.gather(*screenshot_tasks, return_exceptions=True)
+
+                for res in results:
+                    if isinstance(res, Exception):
+                        logger.error(f"后台分析任务出错: {res}")
+                        continue
+                    if res:
+                        for p in res:
+                            is_duplicate = False
+                            for existing in all_prices:
+                                if p.get('start') == existing.get('start') and p.get('end') == existing.get('end'):
+                                    is_duplicate = True
+                                    for key in ['price', 'plus_price', 'market_price', 'elec_price', 'service_price']:
+                                        if p.get(key) is not None and (existing.get(key) is None or existing.get(key) == 0):
+                                            existing[key] = p[key]
+                                    break
+                            if not is_duplicate:
+                                all_prices.append(p)
+        except Exception as e:
+            logger.error(f"抓取价格详情过程中发生异常: {e}")
+        finally:
+            for task in screenshot_tasks:
+                if not task.done():
+                    task.cancel()
+
+            for path in temp_screenshots:
+                if os.path.exists(path):
+                    try:
+                        os.remove(path)
+                    except:
+                        pass
+
+        # 6. 保存数据
+        if all_prices:
+            station_name_clean = clean_station_name(station_name)
+            try:
+                all_prices.sort(key=lambda x: x.get('start', '00:00'))
+            except:
+                pass
+            logger.info(f"✅ 场站 {station_name_clean} 共提取到 {len(all_prices)} 条价格信息，准备保存...")
+            await self.service.save_station_data(station_name_clean, address, all_prices)
+        else:
+            logger.warning(f"❌ 未能提取到任何价格信息，请检查页面识别逻辑")
+        if os.path.exists(price_screen_path):
+            os.remove(price_screen_path)
         
         try:
             while price_page_count < max_price_pages: