Skip to content

v1.18.0 Multi-file context support, enhanced search, and intelligent language detection

Choose a tag to compare

View all tags
@JochenYang JochenYang released this 24 Oct 16:17
· 97 commits to main since this release
v1.18.0
a3fd6dd

Added

  • Multi-file Context Support: Record and track contexts spanning multiple files
    • New context_files table for storing file associations
    • Support for tracking change types, line ranges, and diff statistics per file
    • Automatic migration of existing single-file contexts to new system
    • Backward compatible with single file_path parameter
  • Enhanced Search Performance: Multi-dimensional scoring and intelligent caching
    • Hybrid scoring combining semantic similarity (40%), keyword matching (30%), quality (20%), and freshness (10%)
    • LRU cache with 5-minute TTL for repeated searches
    • File-based filtering with new file_path parameter in semantic_search
    • Search result caching with use_cache parameter (default: true)
  • Parallel Embedding Generation: Significantly faster vector embedding processing
    • Configurable concurrency with concurrency parameter (default: 5)
    • Batch processing with automatic error handling
    • Performance statistics including processing speed (embeddings/sec)
    • Up to 5x speed improvement for large context sets
  • Intelligent Language Detection: Automatic project language detection for localized auto-recording
    • Analyzes code comments to determine project language (Chinese/English)
    • Falls back to README detection if comments are insufficient
    • Daemon auto-recording messages now use detected language
    • Supports bilingual projects with smart detection

Changed

  • record_context tool now supports files_changed array parameter for multi-file recording
  • semantic_search tool enhanced with file_path and use_cache parameters
  • generate_embeddings tool now supports parallel processing with concurrency parameter
  • Database schema extended with context_files table and indexes
  • DatabaseManager.createContext() now automatically creates file associations when file_path is provided

Fixed

  • Data migration now handles contexts with NULL file_path correctly
  • File watcher initialization fixed for ES modules (replaced require with dynamic import)
  • CLI and daemon now properly record file associations in context_files table
  • CLI extract command now correctly records line number information for documentation files

Technical

  • Added SearchCache class with LRU eviction and TTL support
  • Added ContextFileManager for managing file-context relationships
  • Added LanguageDetector utility for intelligent project language detection
  • Implemented multi-dimensional scoring algorithm in VectorSearchEngine
  • Added parallel embedding generation methods: batchGenerateEmbeddingsParallel() and batchUpdateEmbeddings()
  • Database migration system for seamless upgrade from v1.17.0
  • Lazy initialization pattern for ContextFileManager to avoid circular dependencies
Assets 2
Loading