Update app.py

app.py

@@ -4,10 +4,13 @@ import os
 import pandas as pd
 from dotenv import load_dotenv
 from services import GeminiService
+from huggingface_hub import HfApi, hf_hub_download
 
 # Load Env
 load_dotenv()
 SAVE_FILE = os.getenv("SAVE_FILE_NAME", "saved_professors.json")
+HF_TOKEN = os.getenv("HF_TOKEN")
+DATASET_REPO_ID = os.getenv("DATASET_REPO_ID")
 
 # Init Service
 try:
@@ -22,24 +25,55 @@ def get_key(p):
     return f"{p['name']}-{p['university']}"
 
 def load_data():
+    data = []
+    # 1. 嘗試從雲端下載
+    if HF_TOKEN and DATASET_REPO_ID:
+        try:
+            print(f"正在同步雲端資料: {DATASET_REPO_ID}...")
+            hf_hub_download(
+                repo_id=DATASET_REPO_ID,
+                filename=SAVE_FILE,
+                repo_type="dataset",
+                token=HF_TOKEN,
+                local_dir="." # 覆蓋本地檔案
+            )
+        except Exception as e:
+            print(f"雲端同步略過 (初次啟動或無權限): {e}")
+
+    # 2. 讀取檔案
     if os.path.exists(SAVE_FILE):
         try:
             with open(SAVE_FILE, 'r', encoding='utf-8') as f:
-                return json.load(f)
+                data = json.load(f)
         except:
-            return []
-    return []
+            data = []
+    return data
 
 def save_data(data):
+    # 1. 存本地
     try:
         with open(SAVE_FILE, 'w', encoding='utf-8') as f:
             json.dump(data, f, ensure_ascii=False, indent=2)
     except Exception as e:
         print(f"Save Error: {e}")
+        return
+
+    # 2. 上傳雲端
+    if HF_TOKEN and DATASET_REPO_ID:
+        try:
+            api = HfApi(token=HF_TOKEN)
+            api.upload_file(
+                path_or_fileobj=SAVE_FILE,
+                path_in_repo=SAVE_FILE,
+                repo_id=DATASET_REPO_ID,
+                repo_type="dataset",
+                commit_message="Sync data from Space"
+            )
+        except Exception as e:
+            print(f"Upload Error: {e}")
 
 def format_df(source_list, saved_list):
     if not source_list:
-        # 如果沒有來源資料，回傳空表格
         return pd.DataFrame(columns=["狀態", "姓名", "大學", "系所", "標籤"])
     
     if saved_list is None:
@@ -82,7 +116,6 @@ def search_professors(query, current_saved):
     
     try:
         results = gemini_service.search_professors(query)
-        # 搜尋時，顯示搜尋結果
         return format_df(results, current_saved), results, gr.update(visible=True)
     except Exception as e:
         raise gr.Error(f"搜尋失敗: {e}")
@@ -119,14 +152,12 @@ def select_professor_from_df(evt: gr.SelectData, search_results, saved_data, vie
     
     details_md = ""
     
-    # Check if details are cached
     if current_prof.get('details') and len(current_prof.get('details')) > 10:
         details_md = current_prof['details']
         if not saved_prof: 
             saved_data.insert(0, current_prof)
             save_data(saved_data)
     else:
-        # Call API
         gr.Info(f"正在調查 {current_prof['name']}...")
         try:
             res = gemini_service.get_professor_details(current_prof)
@@ -266,25 +297,19 @@ def remove_prof(selected_prof, saved_data, view_mode, search_results):
 
 def toggle_view(mode, search_res, saved_data):
     if mode == "搜尋結果":
-        # 如果是搜尋結果，傳入搜尋結果清單
         return format_df(search_res, saved_data), gr.update(visible=True)
     else:
-        # 如果是追蹤清單，傳入追蹤清單
         return format_df(saved_data, saved_data), gr.update(visible=False)
 
-# 新增這個函式，用來在網頁載入時初始化畫面
 def init_on_load(saved_data):
-    # 預設顯示追蹤清單，這樣使用者一重新整理就能看到資料
     return format_df(saved_data, saved_data)
 
 # --- UI Layout ---
 
 with gr.Blocks(title="Prof.404 開箱教授去哪兒？", theme=gr.themes.Soft()) as demo:
     
-    # 🌟 修正點1: 把 load_data() 的括號拿掉，傳入函式本身
-    # 這樣每次重新整理網頁時，都會重新執行 load_data，讀取最新的檔案
-    saved_state = gr.State(load_data) 
-    
+    # 🌟 使用 load_data 函式本身作為 State 初始值 (不加括號)，確保每次開啟都會重新讀取
+    saved_state = gr.State(load_data)
     search_res_state = gr.State([])
     selected_prof_state = gr.State(None)
 
@@ -294,7 +319,6 @@ with gr.Blocks(title="Prof.404 開箱教授去哪兒？", theme=gr.themes.Soft()
         search_input = gr.Textbox(label="搜尋研究領域", placeholder="例如: 大型語言模型, 後量子密碼遷移...", scale=4)
         search_btn = gr.Button("🔍 搜尋", variant="primary", scale=1)
     
-    # 🌟 修正點2: 預設選取 "追蹤清單"，讓使用者一進來就看到存檔
     with gr.Row():
         view_radio = gr.Radio(["搜尋結果", "追蹤清單"], label="顯示模式", value="追蹤清單")
     
@@ -342,7 +366,7 @@ with gr.Blocks(title="Prof.404 開箱教授去哪兒？", theme=gr.themes.Soft()
 
     # --- Wiring ---
     
-    # ���� 修正點3: 網頁載入時(Load)，自動把 saved_state 的內容顯示到 prof_df 上
+    # 🌟 啟動時自動載入資料到表格
     demo.load(init_on_load, inputs=[saved_state], outputs=[prof_df])
 
     search_btn.click(
@@ -350,7 +374,6 @@ with gr.Blocks(title="Prof.404 開箱教授去哪兒？", theme=gr.themes.Soft()
         inputs=[search_input, saved_state],
         outputs=[prof_df, search_res_state, load_more_btn]
     ).then(
-        # 搜尋後自動切換到 "搜尋結果" tab
         lambda: gr.update(value="搜尋結果"), outputs=[view_radio]
     )