libloot/docs/scripts/licenses.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# This script expects cargo-attribution
# <https://github.com/ameknite/cargo-attribution> to be installed.

import argparse
import datetime
import json
import os
import re
import shutil
import subprocess
import tomllib

def process_licenses(source_dir, destination_dir):
    # CC0, Unlicense and Zlib don't need licenses to be included.
    # MPL doesn't need the license included with binary distributions, and
    # allows distribution under a compatible license.
    # cargo attribution's GPL-3.0 download doesn't contain the license text.
    # The ISC license is skipped because its dependency gets filtered out, same
    # with the exceptions.
    skip_entries = [
        'exceptions',
        'CC0-1.0',
        'GPL-3.0',
        'ISC',
        'MPL-2.0',
        'Unlicense',
        'Zlib'
    ]

    for entry in os.scandir(source_dir):
        if entry.name not in skip_entries:
            destination_path = os.path.join(destination_dir, entry.name)
            if entry.is_dir():
                shutil.copytree(entry.path, destination_path, dirs_exist_ok=True)
            else:
                shutil.copy2(entry.path, destination_path)

def recurse_dependencies(packages, package, dependency_names):
    for dependency in package['dependencies']:
        name = dependency['name']
        if name in dependency_names:
            continue

        dependency_names.add(name)
        for package in packages:
            if package['name'] == name:
                recurse_dependencies(packages, package, dependency_names)

def get_target_dependency_names(target_package_name):
    result = subprocess.run(
        [
            'cargo',
            'metadata',
            '--manifest-path',
            cargo_toml_path,
            '--filter-platform',
            'x86_64-pc-windows-msvc',
            '--format-version',
            '1'
        ],
        capture_output=True,
        text=True,
        check=True
    )

    json_output = json.loads(result.stdout[:-1])
    packages = json_output['packages']

    target_dependency_names = set()
    for package in packages:
        if package['name'] == 'libloot-cpp':
            recurse_dependencies(packages, package, target_dependency_names)

    return target_dependency_names

if __name__ == "__main__":
    target_package_name = 'libloot-cpp'
    cargo_toml_path = os.path.join('..', 'cpp', 'Cargo.toml')
    attribution_dir = os.path.join('build', 'attribution')
    output_dir = os.path.join('api', 'copyright')

    subprocess.run(
        [
            'cargo',
            'attribution',
            '--manifest-path',
            cargo_toml_path,
            '--output-dir',
            attribution_dir,
            '--filter-platform',
            'x86_64-pc-windows-msvc',
            '--only-normal-dependencies'
        ],
        check=True
    )

    process_licenses(os.path.join(attribution_dir, 'licenses'), os.path.join(output_dir, 'licenses'))

    target_dependency_names = get_target_dependency_names(target_package_name)

    dependencies_toml_path = os.path.join(attribution_dir, 'dependencies.toml')
    with open(dependencies_toml_path, 'rb') as f:
        data = tomllib.load(f)
        dependencies = data['dependencies']

        # esplugin's and libloadorder's notices are nonsense, but they're my
        # libraries so I give myself permission to skip providing the notices.
        skip_dependencies = [
            'libloot',
            'libloot-ffi-errors',
            'libloot-cpp',
            'libloot-nodejs',
            'libloot-python',
            'array-parameterized-test',
            'esplugin',
            'libloadorder'
        ]

        notices_rst = f'.. This file was generated by scripts/licenses.py at {datetime.datetime.now().isoformat()}.\n\n'

        heading = 'Dependency Copyright Notices'
        notices_rst += heading
        notices_rst += '\n'
        notices_rst += '=' * len(heading)
        notices_rst += '\n\n'

        for dependency in dependencies:
            if dependency['name'] not in target_dependency_names:
                continue

            if dependency['name'] in skip_dependencies:
                continue

            if 'license' not in dependency:
                print(f'Found dependency with no license: {dependency['name']}')
                exit(1)

            if 'ISC' in dependency['license']:
                print(f'Found dependency that may require the ISC license: {dependency['name']}')
                exit(1)

            if 'exception' in dependency['license']:
                print(f'Found dependency that may require an exception: {dependency['name']}')
                exit(1)

            if 'CC0-1.0' in dependency['license']:
                print(f'Skipping {dependency['name']} because it uses the CC0-1.0 license: {dependency['license']}')
                continue

            if 'Unlicense' in dependency['license']:
                print(f'Skipping {dependency['name']} because it uses the Unlicense license: {dependency['license']}')
                continue

            if 'Zlib' in dependency['license']:
                print(f'Skipping {dependency['name']} because it uses the Zlib license: {dependency['license']}')
                continue

            if 'notices' not in dependency:
                print(f'Skipping dependency with no notices: {dependency['name']}')
                continue

            if 'repository' not in dependency:
                print(f'Found dependency with no repository: {dependency['name']}')
                exit(1)

            if dependency['name'] == 'hashlink':
                # The notice for hashlink is mangled so correcting it here.
                dependency['notices'] = ['Copyright (c) 2015 The Rust Project Developers']

            link_rst = f'`{dependency['name']} v{dependency['version']} <{dependency['repository']}>`_'
            underline_rst = '-' * len(link_rst)
            dep_notices_rst = '\n    '.join(dependency['notices'])

            section_rst = f'{link_rst}\n{underline_rst}\n\n::\n\n    {dep_notices_rst}\n\n'

            notices_rst += section_rst

        notices_rst_path = os.path.join(output_dir, 'dependency-notices.rst')
        with open(notices_rst_path, 'w', encoding="utf-8") as outfile:
            outfile.write(notices_rst)