tdfs4ds.serve¶

tdfs4ds.serve ¶

tdfs4ds.serve — OpenAI-compatible + LangServe HTTP server for the two agents.

Install the optional dependencies with::

pip install tdfs4ds[serve]

Run the server::

python -m tdfs4ds.serve

See :mod:tdfs4ds.serve.app for the FastAPI factory and :mod:tdfs4ds.serve.openai_router for the OpenAI surface.

all `module-attribute` ¶

__all__ = ['create_app', 'serve']

create_app ¶

create_app(run_bootstrap: bool = True, root_path: str | None = None) -> FastAPI

Build the FastAPI app.

Parameters¶

run_bootstrap: When True (default) the Teradata context + vector index are initialised on the application startup event. Pass False to build the app object for import-only checks (e.g. GET /v1/models needs no database). root_path: External URL path prefix the app is mounted under (e.g. "/myapp" behind a reverse proxy). Defaults to TDFS4DS_SERVE_ROOT_PATH (or empty). FastAPI uses it for OpenAPI/redirects; the admin shell reads it from each request so its assets / fetch calls resolve under the prefix.

serve ¶

serve(host: str = '0.0.0.0', port: int | None = None)