fd-mcp

# fd-mcp Benchmark Suite This directory contains benchmarking tools to demonstrate the performance advantages of fd-mcp tools over traditional bash commands. ## Quick Start ```bash # 1. Generate test data (default: 100,000 files) ./setup.sh # 2. Run benchmarks ./run_benchmark.sh ``` ## Setup Options The `setup.sh` script supports configurable file counts: ```bash # Default: 100,000 files ./setup.sh # Custom file count ./setup.sh -n 10000 # 10,000 files ./setup.sh -n 1000000 # 1 million files # Show help ./setup.sh --help ``` ### Minimum Requirements - **File count**: Minimum 100 files - **Python 3**: Required for file generation - **Disk space**: Approximately 1MB per 100 files ## File Distribution The benchmark creates a realistic mix of file types: | Type | Percentage | Description | |------|------------|-------------| | Python source | 15% | Module files with classes | | Python helpers | 5% | Helper function files | | Test files | 15% | Unit tests | | JavaScript components | 10% | React-like components | | JavaScript utilities | 5% | Utility functions | | JSON config | 5% | Configuration files | | YAML config | 5% | Settings files | | Documentation | 10% | Markdown docs | | Shell scripts | 8% | Bash scripts | | Python libraries | 4% | Library modules | | JavaScript libraries | 3% | Common utilities | | Examples | 10% | Example scripts | | Text files | 3% | Data files | | Environment files | 1% | .env files | | README files | 1% | Documentation | | Mixed files | ~5% | Legacy and integration tests | ## Benchmark Tests The benchmark suite tests the following operations: 1. **Find all Python files** - Testing file discovery by extension 2. **Find test files by pattern** - Pattern matching performance 3. **Search TODO in Python files** - Content search in specific file types 4. **Search FIXME with context** - Context-aware content search 5. **Count all files** - Simple file counting 6. **Count Python files** - Filtered file counting 7. **Find recently modified files** - Time-based filtering 8. **Find class definitions** - Code pattern matching 9. **Search TODO in JavaScript** - Cross-language content search 10. **Search for function definitions** - Complex regex patterns 11. **Find config files** - Multiple extension matching 12. **Search in documentation** - Markdown file searches ## Performance Expectations Based on the default 100,000 file dataset: | Operation | Traditional (find/grep) | fd-mcp | Expected Speedup | |-----------|------------------------|---------|------------------| | File by extension | ~2-5s | ~0.3-0.5s | 5-10x | | Content search | ~10-30s | ~0.5-1s | 10-100x | | Recent files | ~3-8s | ~0.5-1s | 3-5x | | File counting | ~1-3s | ~0.2-0.4s | 5-8x | *Note: Actual performance varies based on hardware, especially disk I/O speed.* ## Implementation Details ### File Generation The `setup.sh` script is a wrapper that: 1. Parses command-line arguments 2. Validates inputs 3. Delegates to `generate_files.py` for fast, reliable file creation The Python implementation (`generate_files.py`) is used for: - **Speed**: Much faster than bash loops for large file counts - **Reliability**: No shell quirks or loop issues - **Maintainability**: Easier to read and modify ### File Content All generated files contain realistic code patterns: - TODO and FIXME comments for search testing - Class and function definitions for pattern matching - Proper syntax for relevant file types - Numbered identifiers for uniqueness ## Troubleshooting ### Slow File Generation If setup is slow: - Check available disk space - Consider using a faster filesystem (SSD vs HDD) - Reduce file count for testing ### Benchmark Fails If benchmarks fail: - Ensure `fd` (or `fdfind`) is installed - Ensure `ripgrep` (`rg`) is installed - Check that test_data directory exists and is populated ### Installation ```bash # Ubuntu/Debian sudo apt install fd-find ripgrep # macOS brew install fd ripgrep ``` ## Cleaning Up ```bash # Remove test data rm -rf test_data/ # Or regenerate with different count ./setup.sh -n 10000 ``` ## Contributing To modify the benchmark: 1. Edit `generate_files.py` for file generation changes 2. Edit `run_benchmark.sh` for new benchmark tests 3. Test with small file counts first (`-n 1000`) 4. Verify results with the default 100,000 files