comment-strip-cli/src/languages.js

A powerful CLI tool to strip comments from source code files while preserving strings and important metadata

const fs = require('fs');
const path = require('path');

const builtinLanguageConfigs = {
  // C/C++ family
  c: {
    name: 'C',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  h: {
    name: 'C Header',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  cpp: {
    name: 'C++',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  cxx: {
    name: 'C++',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  cc: {
    name: 'C++',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  hpp: {
    name: 'C++ Header',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  hxx: {
    name: 'C++ Header',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },

  // JavaScript/TypeScript family
  js: {
    name: 'JavaScript',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  mjs: {
    name: 'JavaScript Module',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  jsx: {
    name: 'React JSX',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  ts: {
    name: 'TypeScript',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  tsx: {
    name: 'TypeScript JSX',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },

  // Python
  py: {
    name: 'Python',
    hash: { type: 'hash', pattern: '#' }
  },
  pyw: {
    name: 'Python',
    hash: { type: 'hash', pattern: '#' }
  },

  // Other programming languages
  sol: {
    name: 'Solidity',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  rs: {
    name: 'Rust',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  go: {
    name: 'Go',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  java: {
    name: 'Java',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  kt: {
    name: 'Kotlin',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  kts: {
    name: 'Kotlin Script',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  swift: {
    name: 'Swift',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  dart: {
    name: 'Dart',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  scala: {
    name: 'Scala',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  sc: {
    name: 'Scala',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },

  // Web technologies
  php: {
    name: 'PHP',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' },
    hash: { type: 'hash', pattern: '#' }
  },
  phtml: {
    name: 'PHP',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' },
    hash: { type: 'hash', pattern: '#' }
  },
  css: {
    name: 'CSS',
    block: { type: 'block', pattern: '/* */' }
  },
  scss: {
    name: 'SCSS',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },

  // Scripting languages
  rb: {
    name: 'Ruby',
    hash: { type: 'hash', pattern: '#' }
  },
  rbw: {
    name: 'Ruby',
    hash: { type: 'hash', pattern: '#' }
  },
  pl: {
    name: 'Perl',
    hash: { type: 'hash', pattern: '#' }
  },
  pm: {
    name: 'Perl',
    hash: { type: 'hash', pattern: '#' }
  },
  r: {
    name: 'R',
    hash: { type: 'hash', pattern: '#' }
  },
  R: {
    name: 'R',
    hash: { type: 'hash', pattern: '#' }
  },

  // Shell scripts
  sh: {
    name: 'Shell',
    hash: { type: 'hash', pattern: '#' }
  },
  bash: {
    name: 'Bash',
    hash: { type: 'hash', pattern: '#' }
  },
  zsh: {
    name: 'Zsh',
    hash: { type: 'hash', pattern: '#' }
  },

  // Configuration files
  dockerfile: {
    name: 'Docker',
    hash: { type: 'hash', pattern: '#' }
  },
  cmake: {
    name: 'CMake',
    hash: { type: 'hash', pattern: '#' }
  },
  toml: {
    name: 'TOML',
    hash: { type: 'hash', pattern: '#' }
  },
  yml: {
    name: 'YAML',
    hash: { type: 'hash', pattern: '#' }
  },
  yaml: {
    name: 'YAML',
    hash: { type: 'hash', pattern: '#' }
  },
  ini: {
    name: 'INI Config',
    hash: { type: 'hash', pattern: '#' },
    semicolon: { type: 'semicolon', pattern: ';' }
  },
  cfg: {
    name: 'Config',
    hash: { type: 'hash', pattern: '#' },
    semicolon: { type: 'semicolon', pattern: ';' }
  },
  conf: {
    name: 'Config',
    hash: { type: 'hash', pattern: '#' },
    semicolon: { type: 'semicolon', pattern: ';' }
  },
  makefile: {
    name: 'Makefile',
    hash: { type: 'hash', pattern: '#' }
  },
  mk: {
    name: 'Makefile',
    hash: { type: 'hash', pattern: '#' }
  },

  // Assembly
  asm: {
    name: 'Assembly',
    hash: { type: 'hash', pattern: '#' },
    semicolon: { type: 'semicolon', pattern: ';' }
  },
  s: {
    name: 'Assembly',
    hash: { type: 'hash', pattern: '#' },
    semicolon: { type: 'semicolon', pattern: ';' }
  },

  // Database
  sql: {
    name: 'SQL',
    single: { type: 'single', pattern: '--' },
    block: { type: 'block', pattern: '/* */' }
  },

  // Functional languages
  lua: {
    name: 'Lua',
    single: { type: 'single', pattern: '--' },
    block: { type: 'block', pattern: '--[[ ]]' }
  },
  hs: {
    name: 'Haskell',
    single: { type: 'single', pattern: '--' },
    block: { type: 'block', pattern: '{- -}' }
  },
  lhs: {
    name: 'Haskell',
    single: { type: 'single', pattern: '--' },
    block: { type: 'block', pattern: '{- -}' }
  },

  // JSON (with comments)
  json: {
    name: 'JSON',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  },
  jsonc: {
    name: 'JSON with Comments',
    single: { type: 'single', pattern: '//' },
    block: { type: 'block', pattern: '/* */' }
  }
};

function loadCustomLanguages() {
  try {
    const configPath = path.join(__dirname, '../config/languages.json');
    if (fs.existsSync(configPath)) {
      const customConfig = JSON.parse(fs.readFileSync(configPath, 'utf8'));
      return customConfig;
    }
  } catch (error) {
    console.warn('Warning: Could not load custom language config:', error.message);
  }
  return {};
}

function detectSpecialFiles(filename) {
  const lowerName = filename.toLowerCase();
  
  if (lowerName === 'dockerfile' || lowerName.startsWith('dockerfile.')) {
    return 'dockerfile';
  }
  if (lowerName === 'makefile' || lowerName.startsWith('makefile.')) {
    return 'makefile';
  }
  if (lowerName === 'cmakelists.txt') {
    return 'cmake';
  }
  
  return null;
}

// Merge built-in and custom configurations
const customLanguages = loadCustomLanguages();
const allLanguageConfigs = { ...builtinLanguageConfigs, ...customLanguages };

function getLanguageConfig(fileExtension) {
  const ext = fileExtension.toLowerCase();
  return allLanguageConfigs[ext] || null;
}

function getLanguageConfigByFilename(filename) {
  // First try special file detection
  const specialType = detectSpecialFiles(filename);
  if (specialType) {
    return allLanguageConfigs[specialType];
  }
  
  // Then try by extension
  const ext = path.extname(filename).slice(1).toLowerCase();
  return getLanguageConfig(ext);
}

function getSupportedExtensions() {
  return Object.keys(allLanguageConfigs);
}

function isExtensionSupported(fileExtension) {
  return allLanguageConfigs.hasOwnProperty(fileExtension.toLowerCase());
}

function getAllLanguages() {
  return allLanguageConfigs;
}

module.exports = {
  getLanguageConfig,
  getLanguageConfigByFilename,
  getSupportedExtensions,
  isExtensionSupported,
  getAllLanguages
};