Skip to content

Latest commit

 

History

History
138 lines (109 loc) · 5.43 KB

CHANGELOG.md

File metadata and controls

138 lines (109 loc) · 5.43 KB

Changelog

All notable changes to LayoutLens are documented in this file.

[1.4.0] - 2024-12-21

🚀 Major Changes

  • LiteLLM Integration: Complete migration to LiteLLM as the unified provider
    • Removed OpenRouter provider in favor of LiteLLM's unified interface
    • Support for OpenAI, Anthropic, Google via LiteLLM's standardized API
    • Simplified architecture with single provider handling all models
    • Model naming follows LiteLLM conventions (e.g., "anthropic/claude-3-5-sonnet")

🎯 Breaking Changes

  • No backward compatibility for OpenRouter provider
  • Removed openrouter from provider choices
  • Updated default provider to openai (via LiteLLM)
  • Changed API key environment variable references (removed OPENROUTER_API_KEY)

🔧 Updated Provider Support

  • Provider Options: openai, anthropic, google, gemini, litellm
  • Unified Interface: All providers use LiteLLM for consistent behavior
  • Model Format: LiteLLM naming convention for all models

[1.3.0] - 2024-01-21

🚀 Major Features Added

  • Multi-Provider Support: Complete plugin architecture for AI providers
    • LiteLLM integration for unified access to 25+ AI models
    • Support for OpenAI, Anthropic Claude, Google Gemini, and more
    • Factory pattern for easy provider instantiation and management
    • Backward compatibility with existing OpenAI-only code

🎯 Interactive Mode

  • Interactive CLI: New layoutlens interactive command for real-time analysis
    • Session statistics and progress tracking
    • Rich terminal formatting (optional, falls back gracefully)
    • Live progress indicators and error handling
    • Command history and help system

🔧 Enhanced CLI Experience

  • Provider Selection: --provider flag with choices (litellm, openai, anthropic, google, gemini)
  • Model Selection: --model flag for specifying exact models
  • Enhanced Info Command: Shows available providers, models, and API key status
  • Unified API Keys: Support for OPENAI_API_KEY environment variable

🏗️ Architecture Improvements

  • Provider Architecture: Abstract base classes with unified interface
    • VisionProvider, VisionProviderConfig, VisionAnalysisRequest/Response
    • LiteLLMProvider as unified gateway to multiple AI services
    • Extensible factory pattern for adding new providers

📦 Dependencies

  • Optional Rich Support: Enhanced interactive mode with pip install layoutlens[interactive]
  • OpenAI SDK: Single dependency for all provider communication via OpenRouter

🧪 Testing

  • Comprehensive Provider Tests: 40+ tests covering provider architecture
  • Integration Tests: Full API integration with provider system
  • Interactive Mode Tests: Session management and progress tracking
  • Backward Compatibility: Ensures existing code continues to work

[1.2.0] - 2024-01-20

🚀 Major Features Added

  • Async Processing: Added high-performance async analysis methods
    • analyze_async() - Single page async analysis
    • analyze_batch_async() - Concurrent batch processing with configurable limits
    • 3-5x performance improvement for batch operations
    • Semaphore-based concurrency control to prevent API overload

🔧 CLI Enhancements

  • Added --async flag to main CLI for async processing
  • Added --max-concurrent parameter for concurrency control
  • New dedicated layoutlens-async CLI with enhanced batch commands
  • Added async support to test and compare commands

📚 Documentation

  • Updated README with async examples and performance metrics
  • Added comprehensive async usage examples
  • Updated CLI help text with async command examples

🐛 Bug Fixes

  • Fixed pytest class name conflicts (TestCase → UITestCase, etc.)
  • Enhanced error handling in batch operations
  • Improved type annotations for Python 3.11+ compatibility

🔧 Developer Experience

  • Enhanced pre-commit hooks with full CI/CD integration
  • Improved GitHub Actions workflows (ci.yml, docs.yml, python-publish.yml)
  • Added performance benchmarking tests
  • Clean up of unused imports and linting improvements

[1.1.0] - 2024-01-15

✨ Features Added

  • Production-ready test suites with UITestCase and UITestSuite
  • Smart caching system with memory and file backends
  • Comprehensive exception hierarchy for better error handling
  • Enhanced CLI with regression testing commands

🔧 Improvements

  • Modernized type annotations for Python 3.11+
  • GitHub Pages documentation with Furo theme
  • Comprehensive integration tests with mocked OpenAI API
  • Pre-commit hooks and local CI/CD setup

🐛 Bug Fixes

  • Fixed CLI regression command implementation
  • Improved error handling across the codebase
  • Better resource management and cleanup

[1.0.2] - 2024-01-10

🔒 Security

  • CRITICAL: Fixed API key logging vulnerability in CLI
  • Enhanced security practices across the codebase

🐛 Bug Fixes

  • CLI no longer exposes API keys in logs
  • Improved error handling for missing dependencies

[1.0.1] - 2024-01-05

🐛 Bug Fixes

  • Fixed import issues in certain environments
  • Improved error messages for missing API keys
  • Better handling of screenshot capture failures

[1.0.0] - 2024-01-01

🎉 Initial Release

  • Core LayoutLens functionality for UI testing
  • Natural language visual analysis using GPT-4 Vision API
  • Screenshot capture with Playwright
  • Accessibility and mobile-friendly checks
  • Basic CLI commands (test, compare, generate)
  • Support for multiple viewports and queries