Weave Code
Client Laravel Package
google-gemini-php/client
Community-maintained PHP client for the Google Gemini API. Send text, images, and video; run multi-turn chat with streaming; generate images and speech; structured output, function calling, code execution, grounding/search, token counting, plus file and cached-content management.
⚡️ Gemini PHP is a community-maintained PHP API client that allows you to interact with the Gemini AI API.
Frequently asked questions about Client
- How do I install and configure google-gemini-php/client in a Laravel 10 project?
- Run `composer require google-gemini-php/client` to install. Configure your API key in `.env` under `GEMINI_API_KEY`. Register the service provider in `config/app.php` and bind the client to Laravel’s container for dependency injection. Use `GeminiClient::generate()` in your code with your desired model and prompt.
- Can I use Laravel queues to handle long-running Gemini API calls asynchronously?
- Yes. Dispatch a job (e.g., `GeminiJob`) that uses the client to generate content. Laravel’s queue system will process it in the background, avoiding timeouts. Use `GeminiJob::dispatch()` in your controller or command. For streaming responses, ensure your job handles chunked data appropriately.
- What Laravel versions and PHP versions does this package support?
- The package supports Laravel 9+ (PHP 8.1+) and Laravel 10 (PHP 8.2+). It’s fully compatible with Laravel’s latest features, including dependency injection and HTTP client abstractions. Check the package’s `composer.json` for exact PHP version requirements, which align with Laravel’s supported versions.
- How can I test my Laravel app’s Gemini integration without hitting real API costs?
- Use Laravel’s Mockery or Pest to mock the `GeminiClient` interface. Stub API responses in your tests with `GeminiClient::shouldReceive('generate')->andReturn($mockResponse)`. For HTTP contract testing, use tools like `vcr` or Pest’s HTTP testing to record and replay Gemini API calls deterministically.
- Does this package support streaming responses from Gemini, and how do I handle them in Laravel?
- Yes, the package supports streaming responses. Use the `stream()` method on the client to receive chunks incrementally. In Laravel, you can process each chunk in real-time (e.g., for chat apps) or buffer them for later use. For async processing, dispatch a job to handle the streamed data as it arrives.
- How can I enforce API rate limits and avoid hitting Gemini’s usage quotas in production?
- Implement Laravel’s retry helper with exponential backoff for failed requests. Use middleware like `GeminiRateLimitMiddleware` to throttle requests based on your quota. Log API usage in a database table (e.g., `gemini_calls`) and set up alerts via Laravel Nova or a custom notification system when approaching limits.
- Can I integrate Gemini with Laravel Scout for semantic search or vector embeddings?
- While the package primarily focuses on text and multimodal generation, you can use Gemini to generate embeddings for custom search logic. Store embeddings in a vector database (e.g., Pinecone, Weaviate) and query them separately. For Laravel Scout, you’d need to extend its engine to work with your custom embedding storage.
- What are some alternatives to this package if I need more advanced AI features or better Laravel integration?
- Consider `laravel-ai` for a higher-level Laravel wrapper around multiple AI providers, or `php-ai` for a broader set of AI tools. For Gemini-specific features, check community forks or custom wrappers that extend this package. If you need fine-tuned control, use Laravel’s HTTP client directly with Gemini’s API endpoints.
- How do I handle authentication dynamically, such as using different API keys per user or fetching keys from AWS Secrets Manager?
- Store API keys in Laravel’s `.env` or a secrets manager like AWS Secrets Manager. Use Laravel’s `config()` helper or the `env()` function to fetch keys dynamically. For per-user keys, extend the `GeminiClient` to accept a key parameter or use Laravel’s `Auth` facade to resolve keys based on the authenticated user.
- Are there any security concerns I should address when using Gemini in Laravel, such as prompt injection or data leakage?
- Sanitize user inputs before passing them to Gemini using Laravel’s `Str::of()` or `Validator`. Avoid logging raw prompts or responses if they contain sensitive data. For regulated environments (e.g., HIPAA), implement audit logging via Laravel’s `Log` facade or a dedicated table to track all Gemini interactions. Use Laravel’s encryption for sensitive prompts.
Weaver
How can I help you explore Laravel packages today?