Skip to content

CHANGELOG.md

Latest commit

 

History

History
386 lines (296 loc) · 12.6 KB

CHANGELOG.md

File metadata and controls

386 lines (296 loc) · 12.6 KB

Changelog

🇯🇵 日本語版

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.8.0] - 2025-12-11

Added

New Language Support (4 languages)

  • Kotlin (kotlin.py): Classes, interfaces, objects, functions, properties
    • Supported extensions: .kt, .kts
    • tree-sitter-kotlin >= 1.0.0
  • Swift (swift.py): Classes, structs, protocols, functions, extensions
    • Supported extensions: .swift
    • tree-sitter-swift >= 0.0.1
  • Scala (scala.py): Classes, traits, objects, functions
    • Supported extensions: .scala, .sc
    • tree-sitter-scala >= 0.20.0
  • Lua (lua.py): Functions, local functions, table assignments
    • Supported extensions: .lua
    • tree-sitter-lua >= 0.1.0

Changed

  • Total supported languages: 16 (was 12)
  • Updated parser.py with LANGUAGE_EXTENSIONS for new languages
  • Updated __init__.py with new extractor registrations

Tests

  • Added 26 new tests for Kotlin, Swift, Scala, Lua extractors
  • Total: 334 passed, 1 skipped

[0.7.1] - 2025-11-27

Added

C Language Support

  • Added .c file extension support (pure C language)
  • C files are parsed using tree-sitter-cpp parser
  • Supported extensions: .c, .cpp, .cc, .cxx, .h, .hpp, .hxx
  • Total supported languages: 12 (was 11)

Qiita Article Metadata

  • Added Qiita frontmatter (tags, private, updated_at, etc.)

Changed

  • Updated cpp.py extractor docstring to "C/C++-specific"
  • Updated parser.py LANGUAGE_EXTENSIONS with C/C++ mappings

0.7.0 - 2025-11-27

Added

File Watching (watch command)

  • codegraph-mcp watch: Watch repository and auto-reindex on file changes
    • Real-time file monitoring with configurable debounce
    • --debounce option for delay between changes (default: 1.0s)
    • --community flag to run community detection after each reindex
    • Filters to supported language files only
    • Graceful shutdown with Ctrl+C

GitHub Actions CI/CD

  • CI Workflow (.github/workflows/ci.yml):
    • Runs on push/PR to main branch
    • Tests on Python 3.11 and 3.12
    • Linting with ruff, type checking with mypy
    • Coverage reporting to Codecov
    • Build verification
  • Release Workflow (.github/workflows/release.yml):
    • Triggers on version tags (v*)
    • Runs tests before release
    • Creates GitHub Release with artifacts
    • Auto-publishes to PyPI

Tests

  • Added 8 new CLI tests for watch command
  • Total: 308 passed, 1 skipped

0.6.2 - 2025-11-27

Added

Entity ID Partial Matching

  • resolve_entity_id(): Resolve partial entity IDs to full IDs
    • Exact match, name match, qualified_name suffix match
    • file::name pattern support (e.g., linux.rs::hashmap_random_keys)
  • search_entities(): Pattern-based entity search
  • Partial ID support in find_callers(), find_callees(), find_dependencies()

Auto Community Detection

  • --community flag (default): Auto-detect communities after indexing
  • --no-community flag: Skip community detection for large repositories
  • Displays community count and modularity in index results

Enhanced query_codebase

  • Relevance scoring: Exact match (1.0), starts with (0.8), contains (0.6)
  • include_related: Include related entities in results
  • include_community: Include community information
  • entity_types filter: Filter by function, class, method, etc.
  • Score and community_id included in JSON output

Incremental Community Update

  • update_incremental(): Reassign changed entities to best-fit communities
  • 20% change threshold triggers full re-detection
  • IndexResult.changed_entity_ids for tracking changes

Changed

Community Detection Performance

  • Batch graph building: add_nodes_from() / add_edges_from() for speed
  • Batch DB writes: executemany() for community storage
  • Large graph sampling: max_nodes=50000 with degree-based sampling
  • Successfully processes 230K+ entity repositories

Tests

  • Added 6 new tests for scoring and community integration
  • Total: 300 passed, 1 skipped

0.6.1 - 2025-11-27

Fixed

SSE Transport

  • Fixed SSE endpoint configuration using Mount instead of Route for /messages/
  • Added proper Response() return to avoid NoneType errors on client disconnect
  • Follows official MCP SSE implementation pattern

CLI Unicode Compatibility

  • Removed Unicode emoji characters that caused encoding errors on some terminals
  • Removed SpinnerColumn to avoid surrogate encoding issues
  • Added legacy_windows=True to Rich console for broader compatibility

Tested

  • Successfully indexed Rust compiler repository (230K+ entities, 34K+ files)
  • Verified all 14 MCP tools working via SSE transport

0.6.0 - 2025-11-27

Added

Background Server Management

  • codegraph-mcp start: Start MCP server in background (daemon mode)
    • Default SSE transport on port 8080
    • PID file stored at ~/.codegraph/server.pid
    • Logs written to ~/.codegraph/server.log
  • codegraph-mcp stop: Stop background MCP server gracefully
  • codegraph-mcp status: Check server status with recent log output

Changed

  • codegraph-mcp serve now explicitly documented as foreground mode
  • Background mode uses SSE transport by default (stdio for foreground)

0.5.0 - 2025-11-27

Changed

Performance Optimization - Batch Database Writes

  • 47x faster indexing: Implemented batch write operations for entities and relations
    • Before: 29.47s for 67 files (32 entities/sec)
    • After: 0.63s for 67 files (1495 entities/sec)
  • Reduced database commits from ~5700 per repository to 3
  • Added add_entities_batch() and add_relations_batch() methods to GraphEngine
  • Updated Indexer to collect all parse results before batch writing

Technical Details

  • Uses SQLite executemany() for bulk inserts
  • Single commit per batch instead of per-entity/relation
  • Batch file tracking updates

0.4.0 - 2025-11-27

Added

CLI Enhancement

  • Rich Progress Display: Added animated progress bar for codegraph-mcp index command
    • Spinner animation with real-time progress bar
    • File-by-file processing display
    • Results table with entity/relation counts and duration
    • Color-coded status messages

Performance Metrics (Measured)

  • Indexing speed: 32 entities/second
  • File processing: 0.44 seconds/file
  • Incremental indexing: < 2 seconds
  • Query response: < 2ms

Changed

  • Added rich>=13.0.0 dependency for CLI progress display
  • Updated Indexer.index_repository() with optional progress_callback parameter

0.3.0 - 2025-11-27

Added

Language Support - 5 New Languages

  • PHP Language Support: Full AST parsing for PHP source files

    • Class, interface, and trait extraction
    • Method and function extraction
    • Namespace handling
    • Inheritance and implements relation detection

  • C# Language Support: Comprehensive C# parsing

    • Class, struct, interface, and enum extraction
    • Method, constructor, and property extraction
    • Namespace handling
    • Inheritance relation detection
    • Using directive handling

  • C++ Language Support: Full C++ parsing

    • Class and struct extraction
    • Function and method extraction (including header declarations)
    • Namespace handling
    • Include directive handling
    • Inheritance relation detection
    • Template class support

  • HCL (Terraform) Language Support: Infrastructure as Code parsing

    • Resource and data source extraction
    • Variable and output extraction
    • Module and locals block extraction
    • Provider block extraction

  • Ruby Language Support: Full Ruby parsing

    • Class and module extraction
    • Method and singleton method extraction
    • Inheritance relation detection
    • require/require_relative handling
    • Module include/extend detection

Changed

  • Updated dependencies to include tree-sitter-php, tree-sitter-c-sharp, tree-sitter-cpp, tree-sitter-hcl, tree-sitter-ruby
  • Extended language registry with 5 new extractors

Testing

  • Added 73 new tests for PHP, C#, C++, HCL, and Ruby parsers
  • Total tests: 286 (from 212 in v0.2.0)

0.2.0 - 2025-11-27

Added

Language Support

  • Go Language Support: Full AST parsing for Go source files

    • Function and method extraction with receiver types
    • Struct and interface extraction
    • Package and import handling
    • Call relation detection

  • Java Language Support: Comprehensive Java parsing

    • Class, interface, and enum extraction
    • Method and constructor extraction
    • Inheritance (extends/implements) relation detection
    • Import statement handling

Changed

  • Updated dependencies to include tree-sitter-go and tree-sitter-java
  • Extended language registry with Go and Java extractors

Testing

  • Added 30 new tests for Go and Java parsers
  • Total tests: 212 (from 182 in v0.1.0)

0.1.0 - 2025-11-26

Added

Core Features

  • AST Parsing: Tree-sitter based multi-language code analysis

    • Python support with full class/function/method extraction
    • TypeScript support with interface and type alias handling
    • JavaScript support (ES6+, JSX, CommonJS, ESM)
    • Rust support with struct/enum/trait/impl extraction

  • Code Graph Engine: NetworkX-based graph construction

    • Entity extraction (classes, functions, methods, modules)
    • Relation detection (calls, contains, imports, implements, extends)
    • Dependency analysis with configurable depth

  • GraphRAG Integration: Graph-based Retrieval Augmented Generation

    • Community detection using Louvain algorithm
    • Global search across all code communities
    • Local search within entity neighborhoods
    • LLM integration (OpenAI, Anthropic, Ollama, Rule-based)

  • Storage Layer: SQLite-based persistence

    • Async database operations with aiosqlite
    • File-based caching for performance
    • Vector storage for semantic search

MCP Interface

  • 14 MCP Tools:

    • Graph Query: query_codebase, find_dependencies, find_callers, find_callees, find_implementations, analyze_module_structure
    • Code Retrieval: get_code_snippet, read_file_content, get_file_structure
    • GraphRAG: global_search, local_search
    • Management: suggest_refactoring, reindex_repository, execute_shell_command

  • 4 MCP Resources:

    • codegraph://entities/{id} - Entity details
    • codegraph://files/{path} - File graph information
    • codegraph://communities/{id} - Community data
    • codegraph://stats - Graph statistics

  • 6 MCP Prompts:

    • code_review - Code review assistance
    • explain_codebase - Codebase explanation
    • implement_feature - Feature implementation guidance
    • debug_issue - Debug assistance
    • refactor_guidance - Refactoring suggestions
    • test_generation - Test generation help

Transport Protocols

  • stdio transport (default) for standard MCP clients
  • SSE transport for HTTP-based integrations

CLI Commands

  • codegraph-mcp serve - Start MCP server
  • codegraph-mcp index - Index repository
  • codegraph-mcp query - Search code entities
  • codegraph-mcp stats - Display graph statistics

Documentation

  • Comprehensive API reference (docs/api.md)
  • Configuration guide (docs/configuration.md)
  • Usage examples (docs/examples.md)
  • Sample scripts in examples/ directory

Performance

  • Initial indexing: ~700 entities in 21 seconds
  • Query response: < 2ms average
  • Incremental indexing: < 2 seconds

Testing

  • 182 unit and integration tests
  • 80%+ code coverage target
  • Async test support with pytest-asyncio

Supported Platforms

  • Python 3.11+
  • Linux, macOS, Windows
  • MCP Clients: Claude Desktop, VS Code, Cursor, Windsurf

Unreleased

Planned

  • Performance optimizations for 100k+ line repositories
  • MkDocs documentation site
  • Web UI for graph visualization