Copyright Code Extractor

A tool to automatically extract source code from a project into the source-program identification material (程序鉴别材料) required for a China Software Copyright (软著) registration.

It scans a project directory, finds source files by extension, strips comments and blank lines, and writes an A4 DOCX with exactly 50 code lines per page, a header carrying the software name + version, and an automatic page number — matching the CPCC (中国版权保护中心) filing requirements.

Compliance checklist

Official basis: Article 10 of the Measures for the Registration of Computer Software Copyright (《计算机软件著作权登记办法》第十条, NCAC Order No. 1) — a copy is included at docs/计算机软件著作权登记办法.pdf (source).

CPCC requirement	How it is handled
First 30 + last 30 consecutive pages (60 pages total); submit everything if shorter	--all keeps the first/last 1500 lines automatically
At least 50 lines per page	Exactly 50 lines per page; long lines are hard-wrapped so pagination never drifts
Header shows software name + version, identical to the application form	--name / --soft-version (or config); centred in the page header
Page number at the top right	Word PAGE field, numbered continuously from 1
No large blank areas / blank lines	Blank lines and comments are stripped
PDF upload	Built-in: --pdf exports a PDF next to the DOCX, or -o file.pdf for PDF only

The header text must match the registration form character-for-character. If --name is omitted the project directory name is used and a warning is logged.

Installation

Quick run — no install, no clone (recommended for AI agents):

# uv downloads, caches and runs the published CLI in one step
uvx copyright-code-extractor /path/to/project --all --name "某某系统" --soft-version V1.0 --pdf -o 源代码.docx

# pipx works the same way
pipx run copyright-code-extractor --help

Standalone binaries (no Python required): download from GitHub Releases — linux-x86_64 / linux-arm64 / macos-arm64 / windows-x86_64. (Intel Macs: install from PyPI with uvx / pip above — Apple Silicon binaries don't run on Intel, and a native Intel build isn't published.)

curl -LO https://github.com/kirklin/copyright-code-extractor/releases/latest/download/copyright-code-extractor-macos-arm64
chmod +x copyright-code-extractor-macos-arm64 && ./copyright-code-extractor-macos-arm64 --help

From source:

uv sync            # development
pip install -e .   # or any environment

Usage

# Typical filing: first 30 + last 30 pages, proper header, DOCX + PDF
uv run copyright-code-extractor /path/to/your/project \
  --all \
  --name "某某管理系统" \
  --soft-version V1.0 \
  --pdf \
  -o 源代码.docx

Options:

• -n, --name TEXT: Software name for the page header (must match the application form).
• --soft-version TEXT: Software version for the page header (default: V1.0).
• --all: Extract everything, then apply the first 30 + last 30 pages rule.
• --pdf: Also export a PDF next to the DOCX (embeds a system CJK font, auto-detected; override with pdf_font_path in the config).
• -o, --output FILE_PATH: Output DOCX path.
• -l, --lines NUM_LINES: Total number of lines to extract when not using --all (default: 3000).
• -c, --config CONFIG_FILE: Path to a custom configuration file.
• -v, --verbose: Verbose logging.
• --version: Show the tool version and exit.

Configuration

Create a .copyright-extractor.json in the root of the project being analyzed. Command-line options override the config file.

{
  "software_name": "某某管理系统",
  "software_version": "V1.0",
  "output_file": "源代码.docx",
  "extract_all": true,
  "source_root": "src",
  "ignore_patterns": ["tests", "docs", "*.min.js"],
  "include_extensions": [".py", ".ts", ".tsx", ".vue", ".go"]
}

Useful defaults you can override:

• lines_per_page (50), keep_pages_head / keep_pages_tail (30 / 30)
• max_line_width (90) — wrap display width (CJK counts as 2) that keeps one code line = one printed line
• pdf_font_path — TTF/TTC with CJK glyphs to embed in the PDF when auto-detection fails
• ignore_patterns — already covers .git, node_modules, dist, build, .next, .nuxt, .output, .turbo, coverage, minified assets, etc.

See src/copyright_code_extractor/config.py for the full list.

Notes

• ~100 source extensions are extracted by default, covering every major language family — C/C++/Java/Kotlin/Go/Rust/Swift/JS/TS/Python/C#/F#/VB/PHP/Ruby/Lua/ SQL/Haskell/OCaml/Erlang/Elixir/Fortran/Pascal/PowerShell/assembly/Verilog/VHDL and more — each with its own comment syntax (e.g. -- for SQL, {- -} for Haskell, <# #> for PowerShell). Unknown extensions can be added via include_extensions and fall back to string-aware //, #, /* */ stripping.
• Comment stripping is string-aware: URLs or #fff inside string literals survive. Known limitation: an unescaped // inside a JS regex literal is treated as a comment.
• The generated DOCX uses Consolas 9pt (SimSun for CJK), single spacing, A4 with 2 cm margins, widow/orphan control disabled — so 50 lines per page is exact.

License

This project is licensed under the GNU General Public License v3.0.