Back to Repositories

Testing Embeddings Generation API Integration in private-gpt

This test suite validates the embeddings generation functionality in the private-gpt API, focusing on the proper handling of text-to-vector transformations. The tests ensure the embeddings endpoint correctly processes input text and returns valid vector representations.

Test Coverage Overview

The test suite covers the core embeddings generation endpoint functionality.

Key areas tested include:

POST request handling for /v1/embeddings endpoint
Request body validation with EmbeddingsBody model
Response structure validation using EmbeddingsResponse model
Vector embedding generation verification

Implementation Analysis

The implementation utilizes FastAPI’s TestClient for HTTP request simulation and response validation. The test employs a structured approach with model-based request/response handling, demonstrating clean separation of concerns and type safety through Pydantic models.

Notable patterns include:

Direct endpoint testing with TestClient
Model-driven request/response validation
Structured assertion flow

Technical Details

Testing tools and components:

FastAPI TestClient for HTTP request simulation
Pydantic models for request/response validation
Python’s built-in assertion framework
EmbeddingsBody and EmbeddingsResponse custom models

Best Practices Demonstrated

The test implementation showcases several testing best practices for API validation.

Key practices include:

Isolated test scope with clear purpose
Strong type checking through Pydantic models
Multiple assertion levels for comprehensive validation
Clean and maintainable test structure

zylon-ai/private-gpt

tests/server/embeddings/test_embedding_routes.py

            
from fastapi.testclient import TestClient

from private_gpt.server.embeddings.embeddings_router import (
    EmbeddingsBody,
    EmbeddingsResponse,
)


def test_embeddings_generation(test_client: TestClient) -> None:
    body = EmbeddingsBody(input="Embed me")
    response = test_client.post("/v1/embeddings", json=body.model_dump())

    assert response.status_code == 200
    embedding_response = EmbeddingsResponse.model_validate(response.json())
    assert len(embedding_response.data) > 0
    assert len(embedding_response.data[0].embedding) > 0