English | 中文

CompressKit

Reference implementations and evaluation benchmarks of classic lossless compression (entropy coding) algorithms in modern C++17.

CompressKit covers Huffman coding, Arithmetic Coding, Range Coding, RLE and LZSS. It provides a unified binary stream specification (magic number + frequency table + CRC-32 checksum), strict safety boundary checks, and reproducible evaluation benchmarks.

Zero external dependencies, implemented purely with the modern C++ standard library.

📖 Online documentation: https://build-workbench.github.io/compress-kit/

The documentation site contains algorithm details, the binary format specification, API references and interactive benchmark charts. For local preview:

cd docs && npm ci && npm run dev

Goals & Positioning

• Algorithm learning and implementation reference: most existing arithmetic/range coding implementations online are legacy code from years ago (lacking modern conventions and abstractions). CompressKit provides a modern C++ implementation with a clean code structure, clear comments and RAII compliance, suitable as learning material or a dependency-free code reference.
• Evaluation comparison and benchmark (Baseline): provides a unified CLI contract and test corpus, making it easy to intuitively compare the compression ratios and throughput of different classic algorithms across various data distributions when developing or evaluating new coding strategies.
• Explicit non-goals (Non-goals): this project focuses on standard implementations and comparative verification of classic algorithms, and is not an industrial-grade general-purpose compressor. For high production throughput and LZ dictionary compression, use Zstandard or libdeflate.

What's Included

Algorithm	Magic Number	Characteristics & Applicable Scenarios
Huffman coding	HFM2	Optimal prefix code based on symbol frequencies, suitable for general text and prefix code principle verification
Arithmetic coding	AEN2	Interval-division bitstream coding that approaches the Shannon entropy limit, suitable for understanding high-compression-ratio entropy coding principles
Range coding	RCN2	An integer byte-level variant of arithmetic coding, suitable for comparing throughput and engineering implementation differences
RLE run-length coding	RLE2	Minimal coding for continuously repeated data, with an intuitive format and low overhead
LZSS dictionary coding	LZS2	Sliding-window-based LZ dictionary coding, the LZ foundation of gzip-family algorithms, suitable for data with repeated segments

All command-line tools follow:

<binary> <encode|decode> <input> <output>

Quick Start

git clone https://github.com/build-workbench/compress-kit.git
cd compress-kit

make build
make test

Quick round-trip verification:

printf "Hello CompressKit\n" > input.txt
./build/huffman_cpp encode input.txt output.huf
./build/huffman_cpp decode output.huf restored.txt
diff input.txt restored.txt

Repository Structure

algorithms/
  huffman/cpp/      # Huffman 编码 CLI
  arithmetic/cpp/   # 算术编码 CLI
  range/cpp/        # 区间编码 CLI
  rle/cpp/          # RLE 行程编码 CLI
  lzss/cpp/         # LZSS 字典编码 CLI
  shared/cpp/       # 公共库（序列化、位读写、频率表、CLI 框架）
docs/               # VitePress 中文文档站
tests/              # 测试语料生成与 CLI smoke 测试

Engineering Baseline

Command	Purpose
make build	Build all C++ CLI tools (CMake)
make test	Run unit tests and CLI smoke tests
make lint	clang-format dry-run

License

MIT License · Copyright © 2026 build-workbench

---

English | 中文

CompressKit

使用现代 C++17 实现的经典无损压缩（熵编码）算法参考实现与评测基准。

CompressKit 涵盖 Huffman 编码、算术编码 (Arithmetic Coding)、区间编码 (Range Coding)、RLE 与 LZSS。提供统一的二进制流规范（魔数 + 频率表 + CRC-32 校验）、严格的安全边界检查与可复现的评测基准。

零外部依赖，纯现代 C++ 标准库实现。

📖 在线文档：https://build-workbench.github.io/compress-kit/

文档站包含算法详解、二进制格式规范、API 参考与交互式基准测试图表。如需本地预览：

cd docs && npm ci && npm run dev

目标与定位

• 算法学习与实现参考：网上现存的算术/区间编码实现多为早年遗留代码（缺乏现代规范与抽象）。CompressKit 提供代码结构规范、注释清晰、遵循 RAII 的现代 C++ 实现，适合作为学习资料或无依赖的代码参考。
• 评测对照与基准（Baseline）：提供统一的 CLI 契约与测试语料，便于在自研或对比新编码策略时，直观对照不同经典算法在各类数据分布下的压缩率与吞吐量。
• 明确非目标（Non-goals）：本项目专注于经典算法的规范实现与对比验证，非工业级通用压缩器。如需生产环境的高吞吐与 LZ 字典压缩，请使用 Zstandard 或 libdeflate。

包含内容

算法	魔数	特点与适用场景
Huffman 编码	HFM2	基于符号频率的最优前缀码，适合通用文本与前缀码原理验证
算术编码	AEN2	逼近香农熵极限的区间划分位流编码，适合理解高压缩率熵编码原理
区间编码	RCN2	算术编码的整数字节级变体，适合对比吞吐量与工程实现差异
RLE 行程编码	RLE2	针对连续重复数据的极简编码，格式直观、开销低
LZSS 字典编码	LZS2	基于滑动窗口的 LZ 字典编码，gzip 系算法的 LZ 基础，适合重复片段数据

所有命令行工具都遵循：

<binary> <encode|decode> <input> <output>

快速开始

git clone https://github.com/build-workbench/compress-kit.git
cd compress-kit

make build
make test

快速 round-trip 验证：

printf "Hello CompressKit\n" > input.txt
./build/huffman_cpp encode input.txt output.huf
./build/huffman_cpp decode output.huf restored.txt
diff input.txt restored.txt

仓库结构

algorithms/
  huffman/cpp/      # Huffman 编码 CLI
  arithmetic/cpp/   # 算术编码 CLI
  range/cpp/        # 区间编码 CLI
  rle/cpp/          # RLE 行程编码 CLI
  lzss/cpp/         # LZSS 字典编码 CLI
  shared/cpp/       # 公共库（序列化、位读写、频率表、CLI 框架）
docs/               # VitePress 中文文档站
tests/              # 测试语料生成与 CLI smoke 测试

工程基线

命令	用途
make build	构建全部 C++ CLI 工具（CMake）
make test	运行单元测试与 CLI smoke 测试
make lint	clang-format dry-run

许可证

MIT 许可证 · 版权所有 © 2026 build-workbench