Update main.py

main.py

@@ -1,14 +1,15 @@
-from fastapi import FastAPI, HTTPException, WebSocket, WebSocketDisconnect
+from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
-from gradio_client import Client, handle_file
-import tempfile
-import wave
+from gradio_client import Client
 import base64
 import os
-import json
 
 app = FastAPI()
 
+@app.get("/")
+async def root():
+    return {"message": "16 族語 AI 大師雲端大腦正在運行中！請透過 Chrome 套件呼叫 API。"}
+
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -17,16 +18,12 @@ app.add_middleware(
 )
 
 # ==========================================
-# 🔗 連結核心大腦 (依照您最順的 16 族配置)
+# 🔗 連結原語會 AI 實驗室 (16 族雙大腦)
 # ==========================================
-# 1. 翻譯大腦
 trans_client = Client("https://ai-labs.ilrdf.org.tw/kari-seejiq-tnpusu-ai-hmjil/")
-# 2. 語音合成大腦
 tts_client = Client("https://ai-labs.ilrdf.org.tw/hnang-kari-ai-asi-sluhay/")
-# 3. 語音辨識大腦 (最新 Kaldi 規格)
-asr_client = Client("https://ai-labs.ilrdf.org.tw/sapolita-kaldi/")
 
-# 🛠️ 您最順的解析工具
+# 🛠️ 解析字典檔的小工具
 def parse_dialect(dialect_result):
     if isinstance(dialect_result, dict) and 'value' in dialect_result:
         return dialect_result['value']
@@ -34,109 +31,118 @@ def parse_dialect(dialect_result):
         return dialect_result[0]
     return dialect_result
 
-@app.get("/")
-async def root():
-    return {"message": "16 族全能雙語神獸正在運行中！"}
-
-# ==========================================
-# 🐉 核心進化：雅美語即時雙語字幕 (WebSocket)
-# ==========================================
-@app.websocket("/ws/subtitle")
-async def websocket_subtitle(websocket: WebSocket):
-    await websocket.accept()
-    audio_buffer = bytearray()
-    # ✨ 鎖定剛才測試成功的正確代碼
-    TAO_MT_CODE = "tao_Yami" 
-    
-    try:
-        while True:
-            audio_chunk = await websocket.receive_bytes()
-            audio_buffer.extend(audio_chunk)
-            
-            # 累積約 2.5 秒音訊進行辨識
-            if len(audio_buffer) > 110000:
-                with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_wav:
-                    with wave.open(temp_wav.name, 'wb') as wav_file:
-                        wav_file.setnchannels(1)
-                        wav_file.setsampwidth(2)
-                        wav_file.setframerate(44100)
-                        wav_file.writeframes(audio_buffer)
-                    temp_path = temp_wav.name
-
-                ind_text, zh_text = "", ""
-                try:
-                    # 1. ASR 辨識
-                    asr_res = asr_client.predict(
-                        dialect_id="formosan_tao", 
-                        audio_data=handle_file(temp_path),
-                        api_name="/automatic_speech_recognition"
-                    )
-                    ind_text = str(asr_res).strip()
-
-                    # 2. MT 翻譯 (鎖定 tao_Yami)
-                    if ind_text and ind_text != "...":
-                        zh_res = trans_client.predict(
-                            text=ind_text, 
-                            src_lang=TAO_MT_CODE, 
-                            tgt_lang="zho_Hant", 
-                            api_name="/translate"
-                        )
-                        zh_text = str(zh_res)
-                except Exception as e:
-                    print(f"⚠️ 即時處理錯誤: {e}")
-
-                await websocket.send_json({
-                    "status": "recognizing",
-                    "indigenous": ind_text if ind_text else "...",
-                    "chinese": zh_text
-                })
-                
-                audio_buffer = bytearray()
-                if os.path.exists(temp_path): os.remove(temp_path)
-            else:
-                if len(audio_buffer) % 40 == 0:
-                    await websocket.send_json({"status": "buffering", "text": "🟢 神獸正在聽..."})
-    except Exception: pass
-
 # ==========================================
-# 📚 保留您原本最順暢的 REST 功能 (16 族)
+# 📚 功能 A：16 族雙向文字翻譯 (✨ 已升級支援 16 族)
 # ==========================================
 @app.post("/translate")
 async def translate(data: dict):
     source_text = data.get("text")
-    direction = data.get("direction", "zh2indigenous")
-    ethnicity = data.get("ethnicity", "太魯閣")
+    direction = data.get("direction", "zh2indigenous") # 改為更通用的命名
+    ethnicity = data.get("ethnicity", "太魯閣") # ✨ 關鍵升級：動態接收族別
+    
     try:
+        # 相容舊的 zh2trv 參數，確保原本的右鍵選單不會壞掉
         if direction in ["zh2trv", "zh2indigenous", "中翻族"]:
-            dialect_code = parse_dialect(trans_client.predict(ethnicity=ethnicity, api_name="/lambda_1"))
-            result = trans_client.predict(text=source_text, src_lang="zho_Hant", tgt_lang=dialect_code, api_name="/translate_1")
+            # 【中翻族】
+            dialect_result = trans_client.predict(ethnicity=ethnicity, api_name="/lambda_1")
+            dialect_code = parse_dialect(dialect_result)
+            
+            result = trans_client.predict(
+                text=source_text,
+                src_lang="zho_Hant",
+                tgt_lang=dialect_code,
+                api_name="/translate_1"
+            )
         else:
-            dialect_code = parse_dialect(trans_client.predict(ethnicity=ethnicity, api_name="/lambda"))
-            result = trans_client.predict(text=source_text, src_lang=dialect_code, tgt_lang="zho_Hant", api_name="/translate")
+            # 【族翻中】
+            dialect_result = trans_client.predict(ethnicity=ethnicity, api_name="/lambda")
+            dialect_code = parse_dialect(dialect_result)
+            
+            result = trans_client.predict(
+                text=source_text,
+                src_lang=dialect_code,
+                tgt_lang="zho_Hant",
+                api_name="/translate"
+            )
+            
         return {"result": result}
-    except Exception as e: return {"result": str(e)}
+        
+    except Exception as e:
+        print(f"❌ {ethnicity} 翻譯發生錯誤: {e}")
+        return {"result": f"API 呼叫失敗: {str(e)}"}
 
+# ==========================================
+# 📋 功能 B：獲取 16 族配音員名單
+# ==========================================
 @app.post("/get_speakers")
 async def get_speakers(data: dict):
+    ethnicity = data.get("ethnicity", "太魯閣")
     try:
-        result = tts_client.predict(ethnicity=data.get("ethnicity", "太魯閣"), api_name="/lambda")
-        speakers = [c[0] if isinstance(c, list) else c for c in result['choices']] if isinstance(result, dict) else result
+        result = tts_client.predict(ethnicity=ethnicity, api_name="/lambda")
+        if isinstance(result, dict) and 'choices' in result:
+            speakers = [c[0] if isinstance(c, list) else c for c in result['choices']]
+        else:
+            speakers = result
         return {"speakers": speakers}
-    except Exception as e: return {"error": str(e)}
+    except Exception as e:
+        print(f"❌ 獲取名單失敗: {e}")
+        return {"error": str(e)}
 
+# ==========================================
+# 🎵 功能 C：16 族核心語音合成
+# ==========================================
 @app.post("/synthesize")
 async def synthesize(data: dict):
+    text = data.get("text", "")
+    ethnicity = data.get("ethnicity", "太魯閣")
+    requested_speaker = data.get("speaker", "太魯閣_男聲")
+    
+    if not text:
+        raise HTTPException(status_code=400, detail="請提供文字")
+
+    sanitized_text = text.replace("！", "!").replace("？", "?").replace("，", ",").replace("。", ".")
+    sanitized_text = sanitized_text.replace("：", ":").replace("（", "(").replace("）", ")")
+    
     try:
-        text, ethnicity, req_speaker = data.get("text", ""), data.get("ethnicity", "太魯閣"), data.get("speaker", "太魯閣_男聲")
+        print(f"🌍 處理族別：{ethnicity}，選定：{requested_speaker}")
+        
         speaker_choices = tts_client.predict(ethnicity=ethnicity, api_name="/lambda")
         full_list = [c[0] if isinstance(c, list) else c for c in speaker_choices['choices']]
-        target_speaker = req_speaker if req_speaker in full_list else full_list[0]
-        audio_path = tts_client.predict(ref=target_speaker, gen_text_input=text[:300], api_name="/default_speaker_tts")
-        with open(audio_path, "rb") as f: encoded_audio = base64.b64encode(f.read()).decode('utf-8')
-        os.remove(audio_path)
-        return {"audio_base64": encoded_audio, "mime_type": "audio/wav"}
-    except Exception as e: raise HTTPException(status_code=500, detail=str(e))
+        
+        if requested_speaker in full_list:
+            target_speaker = requested_speaker
+        else:
+            gender_keyword = "男聲" if "男聲" in requested_speaker else "女聲"
+            matches = [s for s in full_list if gender_keyword in s]
+            target_speaker = matches[0] if matches else full_list[0]
+
+        audio_filepath = tts_client.predict(
+            ref=target_speaker,
+            gen_text_input=sanitized_text[:300], 
+            api_name="/default_speaker_tts"
+        )
+        
+        if not os.path.exists(audio_filepath):
+            raise Exception("音檔生成失敗")
+
+        with open(audio_filepath, "rb") as audio_file:
+            encoded_audio = base64.b64encode(audio_file.read()).decode('utf-8')
+            
+        try: os.remove(audio_filepath)
+        except: pass
+            
+        return {
+            "audio_base64": encoded_audio,
+            "mime_type": "audio/wav",
+            "speaker_used": target_speaker
+        }
+        
+    except Exception as e:
+        print(f"❌ 合成錯誤: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
 
 if __name__ == "__main__":
     import uvicorn
-    uvicorn.run(app, host="0.0.0.0", port=7860)
+    port = int(os.environ.get("PORT", 8000))
+    print(f"🎬 正在啟動 16 族全能超級大腦 (Port {port})...")
+    uvicorn.run(app, host="0.0.0.0", port=port)