Update app.py

app.py

@@ -82,67 +82,66 @@ def sync_google_maps(url):
     try:
         yield "🚀 [1/4] 啟動瀏覽器..."
         with sync_playwright() as p:
+            # 使用更真實的 User-Agent 避免被擋
             browser = p.chromium.launch(headless=True)
-            page = browser.new_page()
-            yield "🌐 [2/4] 連線中..."
-            page.goto(clean_url, wait_until="domcontentloaded", timeout=60000)
-            yield "⏳ [3/4] 等待列表加載 (10秒)..."
-            time.sleep(10) 
+            context = browser.new_context(
+                user_agent="Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
+            )
+            page = context.new_page()
             
-            yield "📄 [4/4] 解析餐廳資訊與圖片..."
+            yield "🌐 [2/4] 連線中並等待列表出現..."
+            page.goto(clean_url, wait_until="networkidle", timeout=60000)
             
-            # 找到所有餐廳的容器 (通常在 Saved Places 中是 div[role="article"])
-            # 若失效，請檢查 Google Maps 頁面原始碼
-            items = page.locator('div[role="article"]').all()
+            # 強制等待關鍵元素出現 (Google Maps 列表項目的常見特徵)
+            try:
+                page.wait_for_selector('div[role="article"]', timeout=15000)
+            except:
+                # 如果找不到 article，嘗試找尋包含標題的 div
+                page.wait_for_selector('.fontHeadlineSmall', timeout=10000)
+
+            yield "⏳ [3/4] 正在掃描清單與圖片..."
             
-            if not items:
-                # 備用方案：尋找包含名稱的父級容器
-                items = page.locator('div.m67Hec').all() 
+            # 模擬微幅捲動以確保懶加載圖片有被讀取
+            page.mouse.wheel(0, 5000)
+            time.sleep(2) 
+
+            # 使用 JavaScript 直接提取資料，這比 Playwright Locator 更穩定
+            results = page.evaluate("""() => {
+                const items = Array.from(document.querySelectorAll('div[role="article"], .m67Hec'));
+                return items.map(item => {
+                    const name = item.querySelector('.fontHeadlineSmall')?.innerText || "";
+                    const addr = item.querySelector('.fontBodyMedium')?.innerText || "詳情見地圖";
+                    const img = item.querySelector('img')?.src || "";
+                    return { name, addr, img };
+                }).filter(x => x.name.length > 0);
+            }""")
 
             restaurant_list = []
-            for item in items:
-                try:
-                    # 提取名稱
-                    name_el = item.locator('div.fontHeadlineSmall')
-                    if name_el.count() == 0: continue
-                    name = name_el.inner_text().strip()
-                    
-                    # 提取地址
-                    addr_el = item.locator('div.fontBodyMedium')
-                    addr = addr_el.inner_text().strip() if addr_el.count() > 0 else "詳情見地圖"
-                    
-                    # 提取圖片 URL (通常是 img 標籤的 src)
-                    img_el = item.locator('img')
-                    img_url = None
-                    if img_el.count() > 0:
-                        img_url = img_el.get_attribute('src')
-                    
-                    # 處理無圖片或 Base64 縮圖的情況
-                    if not img_url or "http" not in img_url:
-                        img_url = "https://images.unsplash.com/photo-1517248135467-4c7edcad34c4?w=500" # 預設圖片
-
-                    restaurant_list.append({
-                        "Name": name,
-                        "Address": addr,
-                        "Image_URL": img_url,
-                        "RAG_Content": f"餐廳：{name}，地址：{addr}",
-                        "URL": f"https://www.google.com/search?q={urllib.parse.quote(name)}"
-                    })
-                except:
-                    continue
+            for res in results:
+                name = res['name']
+                # 建立結構化資料
+                restaurant_list.append({
+                    "Name": name,
+                    "Address": res['addr'],
+                    "Image_URL": res['img'] if res['img'].startswith('http') else "https://images.unsplash.com/photo-1517248135467-4c7edcad34c4?w=500",
+                    "RAG_Content": f"餐廳：{name}，地址：{res['addr']}",
+                    "URL": f"https://www.google.com/search?q={urllib.parse.quote(name)}"
+                })
 
             browser.close()
 
             if not restaurant_list:
-                yield "⚠️ 找不到餐廳。請確認連結已設為「共用」並包含已儲存地點。"
+                yield "⚠️ 偵測到空列表。請確認網址是否正確，或嘗試將連結改為『公開』。"
                 return
 
+            # 更新全域變數
             global_df = pd.DataFrame(restaurant_list)
             embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
             vectorstore = FAISS.from_texts(global_df['RAG_Content'].tolist(), embeddings, metadatas=global_df.to_dict('records'))
             global_retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
             rag_initialized = True
             yield f"✅ 同步成功！共有 {len(restaurant_list)} 間餐廳。"
+
     except Exception as e:
         yield f"❌ 系統錯誤：{str(e)}"