Update app.py

app.py

@@ -1,12 +1,14 @@
 import os
 
-# ✅ 修正為 /tmp 可寫入的路徑，解決 PermissionError
+# ✅ Hugging Face 建議路徑（防止 cache 錯誤）
 os.environ["TRANSFORMERS_CACHE"] = "/tmp/transformers"
 os.environ["HF_HOME"] = "/tmp/huggingface"
 os.environ["TORCH_HOME"] = "/tmp/torch"
 
 from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
+from fastapi.staticfiles import StaticFiles
+from fastapi.responses import FileResponse, JSONResponse
 from pydantic import BaseModel
 from datetime import datetime
 from typing import Optional, List
@@ -18,12 +20,14 @@ import json
 import requests
 import torch
 
+# ✅ 初始化 FastAPI
 app = FastAPI(
     title="詐騙訊息辨識 API",
     description="使用 BERT 模型分析輸入文字是否為詐騙內容",
     version="1.0.0"
 )
 
+# ✅ 跨域處理
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -32,18 +36,15 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-class TextAnalysisRequest(BaseModel):
-    text: str
-    user_id: Optional[str] = None
+# ✅ 掛載靜態檔案：支援 script.js / style.css
+app.mount("/static", StaticFiles(directory="."), name="static")
 
-class TextAnalysisResponse(BaseModel):
-    status: str
-    confidence: float
-    suspicious_keywords: List[str]
-    analysis_timestamp: datetime
-    text_id: str
+# ✅ 回傳首頁 index.html
+@app.get("/", response_class=FileResponse)
+async def serve_index():
+    return FileResponse("index.html")
 
-# 初始化 Firebase 使用環境變數
+# ✅ Firebase 初始化
 try:
     cred_data = os.getenv("FIREBASE_CREDENTIALS")
     if not cred_data:
@@ -54,76 +55,79 @@ try:
 except Exception as e:
     print(f"Firebase 初始化錯誤: {e}")
 
-# 從 Google Drive 載入 model.pth
-def load_model_from_drive():
+# ✅ 從 Hugging Face Hub 載入模型（改為 /tmp）
+def load_model_from_hub():
     model_url = "https://huggingface.co/jerrynnms/scam-model/resolve/main/model.pth"
+    model_path = "/tmp/model.pth"
     response = requests.get(model_url)
     if response.status_code == 200:
-        with open("model.pth", "wb") as f:
+        with open(model_path, "wb") as f:
             f.write(response.content)
-        return True
-    return False
+        return model_path
+    raise FileNotFoundError("❌ 無法從 Hugging Face 載入 model.pth")
 
-if not os.path.exists("model.pth"):
-    if not load_model_from_drive():
-        raise FileNotFoundError("無法從 Google Drive 載入 model.pth")
+model_path = "/tmp/model.pth"
+if not os.path.exists(model_path):
+    model_path = load_model_from_hub()
 
 from AI_Model_architecture import BertLSTM_CNN_Classifier
 model = BertLSTM_CNN_Classifier()
-model.load_state_dict(torch.load("model.pth", map_location="cpu"))
+model.load_state_dict(torch.load(model_path, map_location="cpu"))
 model.eval()
 
-@app.get("/")
-async def root():
-    return {"message": "詐騙文字辨識 API 已啟動", "version": "1.0.0", "status": "active", "docs": "/docs"}
+# ✅ 資料格式
+class TextAnalysisRequest(BaseModel):
+    text: str
+    user_id: Optional[str] = None
+
+class TextAnalysisResponse(BaseModel):
+    status: str
+    confidence: float
+    suspicious_keywords: List[str]
+    analysis_timestamp: datetime
+    text_id: str
 
+# ✅ /predict API
 @app.post("/predict", response_model=TextAnalysisResponse)
 async def analyze_text_api(request: TextAnalysisRequest):
     try:
         tz = pytz.timezone("Asia/Taipei")
-        taiwan_now = datetime.now(tz)
-        collection_name = taiwan_now.strftime("%Y%m%d")
-        document_id = taiwan_now.strftime("%Y%m%dT%H%M%S")
-        timestamp_str = taiwan_now.strftime("%Y-%m-%d %H:%M:%S")
+        now = datetime.now(tz)
+        doc_id = now.strftime("%Y%m%dT%H%M%S")
+        date_str = now.strftime("%Y-%m-%d %H:%M:%S")
+        collection = now.strftime("%Y%m%d")
 
         result = bert_analyze_text(request.text)
 
         record = {
-            "text_id": document_id,
+            "text_id": doc_id,
             "text": request.text,
             "user_id": request.user_id,
-            "analysis_result": {
-                "status": result["status"],
-                "confidence": result["confidence"],
-                "suspicious_keywords": result["suspicious_keywords"],
-            },
-            "timestamp": timestamp_str,
+            "analysis_result": result,
+            "timestamp": date_str,
             "type": "text_analysis"
         }
 
-        db.collection(collection_name).document(document_id).set(record)
+        db.collection(collection).document(doc_id).set(record)
 
         return TextAnalysisResponse(
             status=result["status"],
             confidence=result["confidence"],
             suspicious_keywords=result["suspicious_keywords"],
-            analysis_timestamp=taiwan_now,
-            text_id=document_id
+            analysis_timestamp=now,
+            text_id=doc_id
         )
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
-        
-from fastapi.responses import JSONResponse
+
+# ✅ /feedback API
 @app.post("/feedback")
 async def save_user_feedback(feedback: dict):
     try:
         tz = pytz.timezone("Asia/Taipei")
-        taiwan_now = datetime.now(tz)
-        timestamp_str = taiwan_now.strftime("%Y-%m-%d %H:%M:%S")
-
-        feedback["used_in_training"] = False
+        timestamp_str = datetime.now(tz).strftime("%Y-%m-%d %H:%M:%S")
         feedback["timestamp"] = timestamp_str
-
+        feedback["used_in_training"] = False
         db.collection("user_feedback").add(feedback)
         return {"message": "✅ 已記錄使用者回饋"}
     except Exception as e: