CodeStat

A fast code statistics tool written in Rust. Counts lines of code and shows language distribution for Git repositories or directories.

中文文档

Features

• Fast parallel processing using Rayon
• Memory-safe Rust implementation
• Counts code, comments, and blank lines separately
• Detects 50+ programming languages
• Respects .gitignore rules
• Output in table, JSON, or CSV format
• Incremental caching for faster re-runs

Installation

cargo install codestat

Or build from source:

git clone https://github.com/Cactusinhand/codestat
cd codestat
cargo build --release

The binary will be at target/release/codestat.

Usage

Basic usage:

# Analyze current directory
codestat

# Analyze specific directory
codestat /path/to/project

# Analyze single file
codestat src/main.rs

Output formats:

# Table (default)
codestat --format table .

# JSON
codestat --format json .

# CSV
codestat --format csv .

Options:

# Show per-file statistics
codestat -F .

# Show progress
codestat --progress .

# Include hidden files
codestat --hidden .

# Filter by language
codestat --languages rust,python,go .

# Exclude patterns
codestat --exclude "*.test.js,*.min.css" .

# Disable parallel processing (less memory)
codestat --no-parallel .

# Follow symlinks
codestat --follow-links .

Example Output

Table format:

================================================================================
 Language                  Files      Lines       Code   Comments     Blanks
--------------------------------------------------------------------------------
 Rust                          9       2536       1993        215        328
 Markdown                      2        356        260          0         96
 TOML                          1         35         32          0          3
--------------------------------------------------------------------------------
 Total                        12       2927       2285        215        427
================================================================================

Statistics completed in 0.014s (857 files/s, 209071 lines/s)

JSON format:

{
  "summary": {
    "elapsed_ms": 6,
    "error_count": 0,
    "total_blanks": 405,
    "total_bytes": 84917,
    "total_code": 2235,
    "total_comments": 215,
    "total_files": 12,
    "total_lines": 2855
  },
  "languages": [
    {
      "blanks": 328,
      "bytes": 78300,
      "code": 1993,
      "comments": 215,
      "files": 9,
      "lines": 2536,
      "name": "Rust"
    },
    {
      "blanks": 74,
      "bytes": 5779,
      "code": 210,
      "comments": 0,
      "files": 2,
      "lines": 284,
      "name": "Markdown"
    },
    {
      "blanks": 3,
      "bytes": 838,
      "code": 32,
      "comments": 0,
      "files": 1,
      "lines": 35,
      "name": "TOML"
    }
  ],
  "files": null
}

Implementation Notes

Techniques used for performance:

• Memory mapping: Large files use mmap to avoid copy overhead
• Byte-level parsing: Works on &[u8] directly, no UTF-8 validation
• Parallel batching: Files sorted by size for better load balancing
• Cached detection: Language detection uses static HashMap with O(1) lookup

License

MIT