cc.ja.300.binをFastAPI起動時にダウンロードするコードを追加

api/main.py

@@ -11,6 +11,7 @@ from pydantic import BaseModel
 from dotenv import load_dotenv
 
 from utils.logger import setup_logger
+from utils.json import get_file_path_from_config
 from api.search.engine import SearchEngine
 import schemas.projects as schema_projects
 
@@ -32,12 +33,42 @@ async def get_api_key(key: str = Security(api_key_header)):
 
 
 # --- App ---
+engine = SearchEngine()
+
+
+def download_embeddings() -> None:
+    """Hugging Face Datasets から Embeddings をダウンロードする"""
+    from huggingface_hub import hf_hub_download
+
+    # フォルダがなかったら新規作成
+    target_dir = "resources/embeddings"
+    os.makedirs(target_dir, exist_ok=True)
+
+    # ダウンロード
+    path = hf_hub_download(
+        repo_id=os.getenv("HF_EMBEDDINGS_REPO_ID"),
+        repo_type="dataset",
+        filename="cc.ja.300.bin",
+        local_dir=target_dir,
+        local_dir_use_symlinks=False,
+        token=os.getenv("HF_TOKEN"),
+    )
+
+    log.info(f"Embeddings downloaded: {path}")
+    return
+
+
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     """アプリケーションの起動時と終了時に実行されるコード"""
     # --- 起動時処理 ---
     log.info("Initializing search engine and loading assets...")
     load_dotenv()
+    if os.path.exists(get_file_path_from_config("embeddings.fasttext_bin")):
+        log.info("FastText embeddings found locally.")
+    else:
+        log.info("FastText embeddings not found locally. Downloading...")
+        download_embeddings()
     engine.initialize()
     log.info("Initialization complete.")
     yield
@@ -46,7 +77,6 @@ async def lifespan(app: FastAPI):
     log.info("Shutting down search engine...")
 
 
-engine = SearchEngine()
 app = FastAPI(lifespan=lifespan)