'commit'

2026-01-13 10:57:30 +08:00
parent 7c1864b995
commit 3978123b88
10 changed files with 73 additions and 29 deletions
--- a/Apps/XinDianTu/Config/Setting.py
+++ b/Apps/XinDianTu/Config/Setting.py
@@ -1,8 +1,8 @@

 # 采集配置
 # 滑动距离比例 (0.1 ~ 0.9)，数值越大滑动幅度越大
-# 参考场站卡片高度（约屏幕 18-20%），设置为 0.2 以便每次精确翻页一个场站
-SCROLL_DISTANCE_RATIO = 0.22
+# 使用“小步快跑”策略，设置为 0.15 以避开广告并防止跳过场站
+SCROLL_DISTANCE_RATIO = 0.15
 # 最大滑动/翻页次数，达到此次数后停止采集
 MAX_SCROLLS = 100
 # 默认抓取半径（公里），当检测到场站距离超过此值时停止采集
--- a/Apps/XinDianTu/Config/pycache/Setting.cpython-310.pyc
+++ b/Apps/XinDianTu/Config/pycache/Setting.cpython-310.pyc
--- a/Apps/XinDianTu/Crawler.py
+++ b/Apps/XinDianTu/Crawler.py
@@ -164,9 +164,11 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
        image_uuid = str(uuid.uuid4())
        screenshot_path = take_screenshot(d, image_uuid, save_dir=TEMP_IMAGE_DIR)

-        # 检查是否存在广告 (VL) - 仅处理非兔子广告
+        # 检查是否存在广告 (VL)
        logger.info("检查是否存在广告弹窗...")
        ad_res = await ReadImageKit.detect_ad_popup(screenshot_path, device_info=device_info)
+        
+        ad_top_y_norm = 0.90  # 默认使用一个更宽松的底部边界 (0.90)，防止点到导航栏
        if ad_res:
            ad_type = ad_res.get("ad_type")
            if ad_type != "rabbit":
@@ -178,7 +180,10 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
                if os.path.exists(screenshot_path): os.remove(screenshot_path)
                screenshot_path = take_screenshot(d, image_uuid, save_dir=TEMP_IMAGE_DIR)
            else:
-                logger.info(">>> 检测到兔子广告。根据策略不再处理，直接开始识别场站。")
+                # 发现兔子广告，不再尝试关闭，直接计算它的顶边界 (归一化 0-1)
+                # 我们取 0.90 作为安全阈值，低于此 Y 坐标的场站认为被遮挡
+                ad_top_y_norm = 0.90
+                logger.info(f">>> 发现兔子广告，设定安全边界为 Y_norm < {ad_top_y_norm}")
        
        # 检查是否已经滚动到底部 (排除状态栏后，内容与上次一致)
        current_md5 = Kit.get_image_content_md5(
@@ -234,12 +239,19 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
                        logger.info(f"场站 {station_name} 匹配到已处理记录，跳过。")
                        continue

-                    # 【策略调整】每页仅处理第一个识别到的新场站，确保其不被底部广告遮挡
-                    logger.info(f">>> 发现新场站 '{station_name}'。根据最新策略，每页仅处理第一个场站以避开广告。")
+                    # 【优化】检查是否被遮挡或太靠近底部
+                    # card["bounds_norm"] 是 {left, top, right, bottom}
+                    card_bottom = card["bounds_norm"]["bottom"]
+                    if card_bottom > ad_top_y_norm:
+                        logger.warning(f"场站 '{station_name}' 被遮挡或太靠近底部 (Bottom {card_bottom:.2f} > {ad_top_y_norm})，留待下次滚动处理。")
+                        continue
+
+                    # 正常处理新场站
+                    logger.info(f">>> 发现新场站 '{station_name}'，开始处理...")
                    new_stations_processed += 1
                    
                    click_x, click_y = card["click_point"]
-                    logger.info(f"准备处理第 {idx + 1} 个场站: {station_name}, 点击坐标: ({click_x}, {click_y})")
+                    logger.info(f"准备处理场站: {station_name}, 点击坐标: ({click_x}, {click_y})")
                    
                    d.click(int(click_x), int(click_y))

@@ -248,11 +260,26 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
                    
                    should_back_to_list = True

-
                    # 截取二级页面图
                    detail_uuid = str(uuid.uuid4())
                    detail_path = take_screenshot(d, detail_uuid, save_dir=TEMP_IMAGE_DIR)

+                    # 【新增】二级页面广告检测 (如：免费停车提示)
+                    logger.info("检查详情页是否存在干扰弹窗...")
+                    detail_ad_res = await ReadImageKit.detect_ad_popup(detail_path, device_info=device_info)
+                    if detail_ad_res:
+                        dad_type = detail_ad_res.get("ad_type")
+                        if dad_type == "rabbit":
+                            logger.info(">>> 详情页检测到 rabbit 广告，判定为误报，忽略。")
+                        else:
+                            dx, dy = detail_ad_res["x"], detail_ad_res["y"]
+                            logger.info(f"检测到详情页弹窗: {dad_type}，正在点击关闭 ({dx}, {dy})...")
+                            d.click(dx, dy)
+                            await asyncio.sleep(1.5)
+                            # 重新截图，确保后续流程使用的是清理后的界面
+                            if os.path.exists(detail_path): os.remove(detail_path)
+                            detail_path = take_screenshot(d, detail_uuid, save_dir=TEMP_IMAGE_DIR)
+
                    # 【优化】后台解析详情页地址，直接传本地路径，避免等待上传
                    logger.info(f"已启动后台分析详情页: {station_name}")
                    task_addr = asyncio.create_task(service.process_station_address(station_name, detail_path, device_info=device_info))
@@ -351,17 +378,13 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
                    # 从二级页面返回 (仅当确实需要返回时)
                    if should_back_to_list:
                        d.press("back")
-                        logger.info(f"等待 {WAIT_BACK_TO_LIST} 秒返回列表...")
-                        await asyncio.sleep(WAIT_BACK_TO_LIST)
+                        logger.info(f"等待 {WAIT_BACK_TO_LIST + 1} 秒返回列表...")
+                        await asyncio.sleep(WAIT_BACK_TO_LIST + 1)

                    # 记录 Redis 去重 (仅按名称去重)
                    cleaned = Kit.clean_station_name(station_name)
                    await redis_kit.set_data(f"crawled:xdt:{cleaned}", "1", expire=REDIS_STATION_EXPIRE)

-                    # 【策略执行】处理完第一个新场站后立即退出循环，触发翻页
-                    logger.info(f"已完成本页首个新场站 '{station_name}' 的处理，准备翻页。")
-                    break
-
        # 清理已完成的后台任务
        done_tasks = [t for t in background_tasks if t.done()]
        for t in done_tasks:
@@ -379,7 +402,9 @@ async def get_station_list(d, service, uploader, max_scrolls=MAX_SCROLLS):
            no_new_data_count = 0

        # 6. 翻页
-        logger.info("执行翻页滑动...")
+        # 【优化】使用“小步快跑”策略，减小滑动距离以避免跳过场站，并能更平滑地躲避广告
+        # 步长已在 Setting.py 中统一配置为 SCROLL_DISTANCE_RATIO
+        logger.info(f"执行翻页滑动 (步长: {SCROLL_DISTANCE_RATIO})...")
        d.swipe_ext("up", scale=SCROLL_DISTANCE_RATIO)
        await asyncio.sleep(WAIT_AFTER_SCROLL)

--- a/Apps/XinDianTu/Kit.py
+++ b/Apps/XinDianTu/Kit.py
@@ -204,8 +204,8 @@ def detect_rabbit_ad_close(image_path, debug_dir=None):
                        cX = int(M["m10"] / M["m00"]) + roi_x1
                        cY = int(M["m01"] / M["m00"]) + roi_y1
                        
-                        norm_x = int(cX / w * 1000)
-                        norm_y = int(cY / h * 1000)
+                        norm_x = cX / w
+                        norm_y = cY / h
                        
                        # 避免重复
                        if not any(abs(cX - c[0]) < 15 and abs(cY - c[1]) < 15 for c in candidates):
@@ -220,17 +220,23 @@ def detect_rabbit_ad_close(image_path, debug_dir=None):
        has_x = c[5]
        # 基础分：如果有 X，大幅加分
        score = 1000 if has_x else 0
-        # 距离分：越靠近预期的 (93, 830) 分越高
-        dist = np.sqrt((c[3] - 93)**2 + (c[4] - 830)**2)
-        score -= dist * 2
+        # 距离分：越靠近预期的 (0.094, 0.830) 分越高
+        dist = np.sqrt((c[3] - 0.094)**2 + (c[4] - 0.830)**2)
+        score -= dist * 2000  # 归一化后距离变小，需加大权重
        # 面积分：理想面积在 500-1500 之间
        if 500 < c[2] < 1500: score += 200
        return score

    candidates.sort(key=score_candidate, reverse=True)
    best = candidates[0]
+    best_score = score_candidate(best)
    
-    logger.info(f"CV detected rabbit ad close button at Norm({best[3]}, {best[4]}) with score {score_candidate(best):.2f}")
+    logger.info(f"CV detected rabbit ad close button at Norm({best[3]:.3f}, {best[4]:.3f}) with score {best_score:.2f}")
+    
+    # 【优化】如果得分太低 (低于 850)，说明误判概率较大，不予返回
+    if best_score < 850:
+        logger.info(f"Score {best_score:.2f} is below threshold 850, ignoring candidate.")
+        return None
    
    if debug_dir:
        os.makedirs(debug_dir, exist_ok=True)
@@ -749,13 +755,19 @@ def crop_cards_from_image(img_path, output_dir=None, save_debug=True):
            # 使用 Median 比 Max 更稳健，避免被单个超高卡片(如广告)带偏
            threshold_h = median_h * 0.70

-            for (y1, y2), h in zip(temp_valid_segments, heights):
-                if h < threshold_h:
+            for (y1, y2), card_h in zip(temp_valid_segments, heights):
+                # 【优化】过滤顶部非场站区域 (例如会员图标、搜索栏等)
+                # 场站列表通常在屏幕 35% 高度以后
+                if y1 / h < 0.35:
+                    logger.info(f"  Filtering out segment Y={y1}-{y2} because it's too high up (Top {y1/h:.2f} < 0.35).")
+                    continue
+
+                if card_h < threshold_h:
                    logger.info(
-                        f"  Filtering out segment Y={y1}-{y2} (H={h}) because it's too short (Threshold={threshold_h:.1f}).")
+                        f"  Filtering out segment Y={y1}-{y2} (H={card_h}) because it's too short (Threshold={threshold_h:.1f}).")
                else:
                    final_cards.append((y1, y2, final_x1, final_x2))
-                    logger.info(f"  Card: Y={y1}-{y2}, X={final_x1}-{final_x2}, H={h}")
+                    logger.info(f"  Card: Y={y1}-{y2}, X={final_x1}-{final_x2}, H={card_h}")

    # 4. 保存结果
    if output_dir is None:
@@ -812,6 +824,12 @@ def crop_cards_from_image(img_path, output_dir=None, save_debug=True):
        card_info = {
            "id": idx + 1,
            "rect": [x1, y1, x2, y2],
+            "bounds_norm": {
+                "left": x1 / w,
+                "top": y1 / h,
+                "right": x2 / w,
+                "bottom": y2 / h
+            },
            "click_point": [click_x, click_y]
        }
        json_data["cards"].append(card_info)
--- a/Apps/XinDianTu/Opener.py
+++ b/Apps/XinDianTu/Opener.py
@@ -55,7 +55,7 @@ async def check_and_close_ad(d):
                logger.info(f"检测到广告关闭按钮: ({x}, {y}) [Type: {ad_type}]")
                
                if ad_type == "rabbit":
-                    logger.info(">>> 检测到兔子广告。根据最新策略，我们不再尝试关闭它，直接继续。")
+                    logger.info(">>> 检测到兔子广告。根据策略，我们不再尝试关闭它，直接继续。")
                    if os.path.exists(screenshot_path): os.remove(screenshot_path)
                    return True
                else:
--- a/Apps/XinDianTu/ReadImageKit.py
+++ b/Apps/XinDianTu/ReadImageKit.py
@@ -89,6 +89,7 @@ class ReadImageKit:
        cv_rabbit_point = detect_rabbit_ad_close(image_path, debug_dir=TEMP_IMAGE_DIR)
        if cv_rabbit_point:
            logger.info(f">>> 图形学算法精准捕捉到兔子广告关闭按钮: {cv_rabbit_point}")
+            # 注意：Kit.py 的 detect_rabbit_ad_close 现在返回的是 0-1.0 归一化坐标
            norm_point = cv_rabbit_point
            ad_type = "rabbit"
            
@@ -97,8 +98,8 @@ class ReadImageKit:
                device_info = cls._FALLBACK_DEVICE_INFO
            w = device_info.get("displayWidth", FALLBACK_WIDTH)
            h = device_info.get("displayHeight", FALLBACK_HEIGHT)
-            x = int(norm_point[0] / 1000 * w)
-            y = int(norm_point[1] / 1000 * h)
+            x = int(norm_point[0] * w)
+            y = int(norm_point[1] * h)
            return {"x": x, "y": y, "ad_type": ad_type}

        # 2. 如果图形学没找到，再请求大模型进行全量检测
@@ -110,7 +111,7 @@ class ReadImageKit:
        prompt = (
            "请仔细检查这张图片中是否存在**弹窗广告**或**悬浮广告**。\n"
            "广告可能有以下几种形式：\n"
-            "1. **屏幕中央的大型弹窗广告**：通常遮挡了页面内容，内容多为优惠券、活动推广等。\n"
+            "1. **屏幕中央的大型弹窗广告**：通常遮挡了页面内容，内容多为优惠券、活动推广、或“免费停车”类提示（例如：现在可免费停车N小时了！）。\n"
            "2. **'新电兔AI'专属悬浮广告**（高频）：一个卡通兔子头，带有'新电兔AI'字样，关闭按钮是一个黑色圆圈内含白色X。\n"
            "3. **底部的横幅广告**：带有明显的关闭按钮。\n\n"
            "请返回关闭按钮的中心坐标。\n\n"
--- a/Apps/XinDianTu/pycache/Crawler.cpython-310.pyc
+++ b/Apps/XinDianTu/pycache/Crawler.cpython-310.pyc
--- a/Apps/XinDianTu/pycache/Kit.cpython-310.pyc
+++ b/Apps/XinDianTu/pycache/Kit.cpython-310.pyc
--- a/Apps/XinDianTu/pycache/Opener.cpython-310.pyc
+++ b/Apps/XinDianTu/pycache/Opener.cpython-310.pyc
--- a/Apps/XinDianTu/pycache/ReadImageKit.cpython-310.pyc
+++ b/Apps/XinDianTu/pycache/ReadImageKit.cpython-310.pyc