Spaces:
Running
Running
pre-loaded models
Browse files
main.py
CHANGED
|
@@ -1,19 +1,95 @@
|
|
| 1 |
-
# main.py
|
| 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
import os
|
| 4 |
import importlib
|
| 5 |
import uvicorn
|
| 6 |
import shutil
|
|
|
|
| 7 |
from datetime import datetime
|
| 8 |
-
from
|
|
|
|
| 9 |
from fastapi.middleware.cors import CORSMiddleware
|
| 10 |
|
| 11 |
-
#
|
| 12 |
-
|
| 13 |
-
#
|
| 14 |
-
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
origins = ["*"]
|
| 18 |
app.add_middleware(
|
| 19 |
CORSMiddleware,
|
|
@@ -23,35 +99,33 @@ app.add_middleware(
|
|
| 23 |
allow_headers=["*"],
|
| 24 |
)
|
| 25 |
|
| 26 |
-
#
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
os.makedirs(TEMP_DIR)
|
| 32 |
|
|
|
|
| 33 |
def get_analyzer_module(language: str):
|
| 34 |
if language in ANALYZERS:
|
| 35 |
return ANALYZERS[language]
|
| 36 |
|
|
|
|
|
|
|
| 37 |
try:
|
| 38 |
-
print(f"首次請求 '{language}',正在動態載入模組...")
|
| 39 |
analyzer_module = importlib.import_module(f"analyzer.ASR_{language}")
|
| 40 |
-
|
| 41 |
-
print(f"正在呼叫 analyzer.ASR_{language}.load_model()...")
|
| 42 |
analyzer_module.load_model()
|
| 43 |
-
|
| 44 |
ANALYZERS[language] = analyzer_module
|
| 45 |
-
print(f"'{language}'
|
| 46 |
return analyzer_module
|
| 47 |
except ImportError:
|
| 48 |
-
print(f"
|
| 49 |
-
raise HTTPException(status_code=400, detail=f"
|
| 50 |
except Exception as e:
|
| 51 |
-
print(f"
|
| 52 |
-
raise HTTPException(status_code=500, detail=f"
|
| 53 |
|
| 54 |
-
#
|
| 55 |
@api_router.post("/recognize")
|
| 56 |
async def recognize_speech_api(
|
| 57 |
file: UploadFile = File(...),
|
|
@@ -59,31 +133,32 @@ async def recognize_speech_api(
|
|
| 59 |
language: str = Form(...)
|
| 60 |
):
|
| 61 |
analyzer_module = get_analyzer_module(language)
|
| 62 |
-
|
| 63 |
base_filename = os.path.basename(file.filename)
|
| 64 |
temp_file_path = os.path.join(TEMP_DIR, f"{datetime.now().strftime('%Y%m%d%H%M%S')}-{base_filename}")
|
| 65 |
-
|
| 66 |
-
print(f"使用 '{language}' 分析器處理檔案: {temp_file_path}")
|
| 67 |
-
|
| 68 |
try:
|
| 69 |
with open(temp_file_path, "wb") as buffer:
|
| 70 |
shutil.copyfileobj(file.file, buffer)
|
| 71 |
-
|
| 72 |
result = analyzer_module.analyze(temp_file_path, target_sentence)
|
| 73 |
return result
|
| 74 |
except Exception as e:
|
| 75 |
-
print(f"
|
| 76 |
raise HTTPException(status_code=500, detail=str(e))
|
| 77 |
finally:
|
| 78 |
if os.path.exists(temp_file_path):
|
| 79 |
os.remove(temp_file_path)
|
| 80 |
|
|
|
|
| 81 |
app.include_router(api_router)
|
| 82 |
|
| 83 |
-
#
|
|
|
|
|
|
|
|
|
|
| 84 |
if __name__ == "__main__":
|
| 85 |
print("="*60)
|
| 86 |
-
print("
|
| 87 |
-
print("
|
| 88 |
print("="*60)
|
|
|
|
| 89 |
uvicorn.run("main:app", host="0.0.0.0", port=8000, reload=True)
|
|
|
|
| 1 |
+
# main.py
|
| 2 |
|
| 3 |
+
# =======================================================================
|
| 4 |
+
# 1. 匯入區 (Imports)
|
| 5 |
+
# 所有需要的模組都放在檔案的最頂部。
|
| 6 |
+
# =======================================================================
|
| 7 |
import os
|
| 8 |
import importlib
|
| 9 |
import uvicorn
|
| 10 |
import shutil
|
| 11 |
+
import re # <--- 【【新增】】為了動態掃描檔案
|
| 12 |
from datetime import datetime
|
| 13 |
+
from contextlib import asynccontextmanager
|
| 14 |
+
from fastapi import FastAPI, File, UploadFile, Form, HTTPException, APIRouter
|
| 15 |
from fastapi.middleware.cors import CORSMiddleware
|
| 16 |
|
| 17 |
+
# =======================================================================
|
| 18 |
+
# 2. 全域變數與配置區 (Global Variables & Config)
|
| 19 |
+
# 在應用程式啟動前,定義好所有全域會用到的變數。
|
| 20 |
+
# =======================================================================
|
| 21 |
+
# 【【修改】】不再硬編碼,讓它們在啟動時被動態填充
|
| 22 |
+
SUPPORTED_LANGUAGES = []
|
| 23 |
+
ANALYZERS = {}
|
| 24 |
|
| 25 |
+
# 【【新增】】讀取環境變數「信號旗」
|
| 26 |
+
APP_ENV = os.getenv('APP_ENV', 'development')
|
| 27 |
+
|
| 28 |
+
# 【【新增】】這個臨時目錄的定義也放在這裡,保持整潔
|
| 29 |
+
TEMP_DIR = "/tmp/temp_audio"
|
| 30 |
+
if not os.path.exists(TEMP_DIR):
|
| 31 |
+
os.makedirs(TEMP_DIR)
|
| 32 |
+
|
| 33 |
+
# =======================================================================
|
| 34 |
+
# 3. 應用程式生命週期管理區 (Application Lifespan Management)
|
| 35 |
+
# 這是最核心的新增部分,它控制著應用程式啟動和關閉時的行為。
|
| 36 |
+
# =======================================================================
|
| 37 |
+
# 【【新增】】動態掃描 analyzer/ 資料夾的輔助函數
|
| 38 |
+
def discover_supported_languages():
|
| 39 |
+
global SUPPORTED_LANGUAGES
|
| 40 |
+
analyzer_dir = 'analyzer'
|
| 41 |
+
pattern = re.compile(r"ASR_(.+)\.py$")
|
| 42 |
+
try:
|
| 43 |
+
for filename in os.listdir(analyzer_dir):
|
| 44 |
+
match = pattern.match(filename)
|
| 45 |
+
if match:
|
| 46 |
+
language_code = match.group(1)
|
| 47 |
+
SUPPORTED_LANGUAGES.append(language_code)
|
| 48 |
+
if not SUPPORTED_LANGUAGES:
|
| 49 |
+
print("WARNING: No language analyzer modules found in 'analyzer' directory.")
|
| 50 |
+
else:
|
| 51 |
+
print(f"Discovered supported languages: {', '.join(SUPPORTED_LANGUAGES)}")
|
| 52 |
+
except FileNotFoundError:
|
| 53 |
+
print(f"ERROR: The '{analyzer_dir}' directory was not found.")
|
| 54 |
+
SUPPORTED_LANGUAGES = []
|
| 55 |
+
|
| 56 |
+
# 【【新增】】FastAPI 的 lifespan 管理器
|
| 57 |
+
@asynccontextmanager
|
| 58 |
+
async def lifespan(app: FastAPI):
|
| 59 |
+
# --- 應用程式啟動時執行的程式碼 ---
|
| 60 |
+
print("="*60)
|
| 61 |
+
print(f"Application starting up in '{APP_ENV}' mode.")
|
| 62 |
+
|
| 63 |
+
discover_supported_languages() # 首先,動態發現支援的語言
|
| 64 |
+
|
| 65 |
+
if APP_ENV == "production":
|
| 66 |
+
print("Production mode detected. Eager loading all models...")
|
| 67 |
+
for lang in SUPPORTED_LANGUAGES:
|
| 68 |
+
try:
|
| 69 |
+
print(f"--- Loading model for language: {lang} ---")
|
| 70 |
+
analyzer_module = importlib.import_module(f"analyzer.ASR_{lang}")
|
| 71 |
+
analyzer_module.load_model()
|
| 72 |
+
ANALYZERS[lang] = analyzer_module
|
| 73 |
+
print(f"--- Model for {lang} loaded successfully. ---")
|
| 74 |
+
except Exception as e:
|
| 75 |
+
print(f"FATAL: Failed to load model for {lang}. Error: {e}")
|
| 76 |
+
print("All models pre-loaded. Application is ready for requests.")
|
| 77 |
+
else:
|
| 78 |
+
print("Development mode detected. Models will be loaded on-demand (lazily).")
|
| 79 |
+
|
| 80 |
+
print("="*60)
|
| 81 |
+
yield # <--- 應用程式在這裡運行
|
| 82 |
+
# --- 應用程式關閉時執行的程式碼 ---
|
| 83 |
+
print("Application shutting down.")
|
| 84 |
+
|
| 85 |
+
# =======================================================================
|
| 86 |
+
# 4. FastAPI 應用程式實例化與中介軟體設定區
|
| 87 |
+
# 創建 FastAPI app 物件,並掛載所有需要的設定。
|
| 88 |
+
# =======================================================================
|
| 89 |
+
# 【【修改】】告訴 FastAPI 使用我們上面定義的 lifespan
|
| 90 |
+
app = FastAPI(title="Pronunciation Analysis API", lifespan=lifespan)
|
| 91 |
+
|
| 92 |
+
# CORS 中介軟體設定 (不變)
|
| 93 |
origins = ["*"]
|
| 94 |
app.add_middleware(
|
| 95 |
CORSMiddleware,
|
|
|
|
| 99 |
allow_headers=["*"],
|
| 100 |
)
|
| 101 |
|
| 102 |
+
# =======================================================================
|
| 103 |
+
# 5. 核心業務邏輯與 API 端點區
|
| 104 |
+
# 定義所有 API 的路由和處理函數。
|
| 105 |
+
# =======================================================================
|
| 106 |
+
api_router = APIRouter(prefix="/api/v1")
|
|
|
|
| 107 |
|
| 108 |
+
# 【【修改】】這是我們新的、更簡潔的 get_analyzer_module 函數
|
| 109 |
def get_analyzer_module(language: str):
|
| 110 |
if language in ANALYZERS:
|
| 111 |
return ANALYZERS[language]
|
| 112 |
|
| 113 |
+
# 這段程式碼只會在開發模式下被執行
|
| 114 |
+
print(f"'{language}' not in cache. Loading on-demand (development mode)...")
|
| 115 |
try:
|
|
|
|
| 116 |
analyzer_module = importlib.import_module(f"analyzer.ASR_{language}")
|
|
|
|
|
|
|
| 117 |
analyzer_module.load_model()
|
|
|
|
| 118 |
ANALYZERS[language] = analyzer_module
|
| 119 |
+
print(f"'{language}' analyzer loaded and cached successfully.")
|
| 120 |
return analyzer_module
|
| 121 |
except ImportError:
|
| 122 |
+
print(f"Error: Analyzer module for '{language}' not found (analyzer.ASR_{language}.py).")
|
| 123 |
+
raise HTTPException(status_code=400, detail=f"Unsupported language: {language}")
|
| 124 |
except Exception as e:
|
| 125 |
+
print(f"Error: A critical error occurred while loading the model for '{language}': {e}")
|
| 126 |
+
raise HTTPException(status_code=500, detail=f"Failed to load model for language '{language}'.")
|
| 127 |
|
| 128 |
+
# API 端點定義 (不變)
|
| 129 |
@api_router.post("/recognize")
|
| 130 |
async def recognize_speech_api(
|
| 131 |
file: UploadFile = File(...),
|
|
|
|
| 133 |
language: str = Form(...)
|
| 134 |
):
|
| 135 |
analyzer_module = get_analyzer_module(language)
|
|
|
|
| 136 |
base_filename = os.path.basename(file.filename)
|
| 137 |
temp_file_path = os.path.join(TEMP_DIR, f"{datetime.now().strftime('%Y%m%d%H%M%S')}-{base_filename}")
|
| 138 |
+
print(f"Processing file: {temp_file_path} with '{language}' analyzer.")
|
|
|
|
|
|
|
| 139 |
try:
|
| 140 |
with open(temp_file_path, "wb") as buffer:
|
| 141 |
shutil.copyfileobj(file.file, buffer)
|
|
|
|
| 142 |
result = analyzer_module.analyze(temp_file_path, target_sentence)
|
| 143 |
return result
|
| 144 |
except Exception as e:
|
| 145 |
+
print(f"An unexpected error occurred during request processing: {e}")
|
| 146 |
raise HTTPException(status_code=500, detail=str(e))
|
| 147 |
finally:
|
| 148 |
if os.path.exists(temp_file_path):
|
| 149 |
os.remove(temp_file_path)
|
| 150 |
|
| 151 |
+
# 將路由掛載到主應用上
|
| 152 |
app.include_router(api_router)
|
| 153 |
|
| 154 |
+
# =======================================================================
|
| 155 |
+
# 6. 主程式入口 (Main Entry Point)
|
| 156 |
+
# 當您直接運行 `python main.py` 時,這部分程式碼會被執行。
|
| 157 |
+
# =======================================================================
|
| 158 |
if __name__ == "__main__":
|
| 159 |
print("="*60)
|
| 160 |
+
print("Starting FastAPI server in development mode (http://localhost:8000 )...")
|
| 161 |
+
print("Run ngrok manually if needed: ngrok http 8000" )
|
| 162 |
print("="*60)
|
| 163 |
+
# uvicorn 會自動找到 app 物件並運行它
|
| 164 |
uvicorn.run("main:app", host="0.0.0.0", port=8000, reload=True)
|