NL2SQL API

FastAPI REST service for the NL2SQL engine (nl2sql-engine). It builds one NL2SQL engine at startup, from the same env file and configs as the CLI, and serves it under /api/v1.

Running the API

pip install nl2sql-api
ENV=demo NL2SQL_API_ROLE=admin nl2sql-api --host 127.0.0.1 --port 8000 [--reload]
# or: python -m nl2sql_api.server ..., or: uvicorn nl2sql_api.main:app

Start it from the folder holding the env file and configs (the demo's use relative paths). Interactive docs: Swagger UI at /docs, ReDoc at /redoc, the schema at /openapi.json.

The API does not authenticate callers and never takes the RBAC role from the request body. Set one of: NL2SQL_API_ROLE_HEADER (a header a trusted proxy sets), NL2SQL_API_ROLE (one static role) or, for local testing only, NL2SQL_API_TRUST_BODY_ROLE=true (the body's user_context; logs a warning). With none, /api/v1/query answers HTTP 401.

Endpoints

• POST /api/v1/query - Ask a question as the configured role
• GET /api/v1/health - Liveness check
• GET /api/v1/ready - Readiness check (does not yet check dependencies)
• POST /api/v1/datasource - Register a datasource in the running process
• GET /api/v1/datasource - List registered datasource ids
• GET /api/v1/datasource/{datasource_id} - Check a datasource is registered
• DELETE /api/v1/datasource/{datasource_id} - Not supported yet (answers success: false)
• POST /api/v1/llm - Configure an LLM in the running process
• GET /api/v1/llm - List configured LLMs
• GET /api/v1/llm/{llm_name} - Get one configured LLM
• POST /api/v1/index/{datasource_id} - Index one datasource's schema
• POST /api/v1/index-all - Index every registered datasource
• DELETE /api/v1/index - Clear the vector store
• GET /api/v1/index/status - Index health (status, entries by type, embedding model, problems)

Reference: https://github.com/nadeem4/nl2sql/blob/main/docs/api/rest/index.md.