@regele/devtools/bin/test-clean.js

A collection of developer utilities for code processing and text analysis

#!/usr/bin/env node

try {
  // Direct implementation of comment removal
  const fs = require('fs');
  const path = require('path');
  const { Command } = require('commander');
  const chalk = require('chalk');
  const ora = require('ora');
  const fg = require('fast-glob');
  
  // Ensure ora is properly imported
  const createSpinner = ora.default || ora;

  // Function to remove comments from JavaScript code
  function removeJavaScriptComments(code) {
    let result = code;

    try {
      // First, preserve strings and regexes to avoid modifying comments inside them
      const preservedItems = [];
      let preservedCount = 0;
      
      // Temporarily replace strings and regexes with placeholders
      result = result.replace(
        /(["'`])(?:\\[\s\S]|(?!\1)[^\\])*\1|\/(?![*+?])(?:[^\r\n\[/\\]|\\.|\[(?:[^\r\n\]\\]|\\.)*\])+\/(?:[gimuy]+\b)?/g,
        (match) => {
          const placeholder = `__PRESERVED_ITEM_${preservedCount}__`;
          preservedItems[preservedCount] = match;
          preservedCount++;
          return placeholder;
        }
      );

      // Remove JSX comments
      result = result.replace(/\{\/\*[\s\S]*?\*\/\}/g, '');

      // Remove multi-line comments
      result = result.replace(/\/\*[\s\S]*?\*\//g, '');

      // Remove single-line comments
      // Handle comments at start of line
      result = result.replace(/^[ \t]*\/\/.*$/gm, '');
      // Handle comments after code (but not in URLs)
      result = result.replace(/([^:/])\/\/.*$/gm, '$1');

      // Restore preserved strings and regexes
      for (let i = 0; i < preservedCount; i++) {
        const placeholder = `__PRESERVED_ITEM_${i}__`;
        const regex = new RegExp(placeholder, 'g');
        result = result.replace(regex, () => preservedItems[i]);
      }

      // Remove empty lines
      // Remove lines that contain only whitespace
      result = result.replace(/^\s*[\r\n]/gm, '');
      // Remove consecutive empty lines (preserving indentation)
      result = result.replace(/\n(\s*)\n(\s*)\n/g, '\n$2\n');
      // Remove empty lines at the beginning and end of the file
      result = result.replace(/^(\s*\n)+/, '');
      result = result.replace(/(\n\s*)+$/, '');

      return result.trim();
    } catch (error) {
      console.error(`JavaScript comment removal failed: ${error.message}`);
      return code;
    }
  }

  // Test function
  async function testGlob(pattern) {
    try {
      console.log(`Testing pattern: ${pattern}`);
      
      // Normalize pattern for Windows paths
      const normalizedPattern = pattern.replace(/\\/g, '/');
      
      // Find files
      const files = await fg(normalizedPattern, {
        onlyFiles: true,
        absolute: true
      });
      
      console.log(`Found ${files.length} files`);
      
      // Print first 10 files
      if (files.length > 0) {
        console.log('First 10 files:');
        files.slice(0, 10).forEach(file => {
          console.log(`- ${file}`);
        });
        
        // Process the first file as a test
        if (files.length > 0) {
          const testFile = files[0];
          console.log(`\nProcessing test file: ${testFile}`);
          
          // Read the file
          const content = fs.readFileSync(testFile, 'utf8');
          console.log(`Original size: ${content.length} bytes`);
          
          // Process based on extension
          const ext = path.extname(testFile).toLowerCase();
          let newContent;
          
          if (['.js', '.jsx', '.ts', '.tsx', '.json'].includes(ext)) {
            newContent = removeJavaScriptComments(content);
          } else if (['.html', '.htm', '.xml'].includes(ext)) {
            // For HTML files, remove HTML comments
            newContent = content.replace(/<!--[\s\S]*?-->/g, '');
            // Also process script and style tags for JS and CSS comments
            newContent = newContent.replace(/(<script\b[^>]*>)([\s\S]*?)(<\/script>)/gi, (match, openTag, content, closeTag) => {
              return openTag + removeJavaScriptComments(content) + closeTag;
            });
            newContent = newContent.replace(/(<style\b[^>]*>)([\s\S]*?)(<\/style>)/gi, (match, openTag, content, closeTag) => {
              return openTag + removeJavaScriptComments(content) + closeTag;
            });
          } else if (['.css', '.scss', '.less'].includes(ext)) {
            // For CSS files, remove CSS comments
            newContent = content.replace(/\/\*[\s\S]*?\*\//g, '');
            // Remove empty lines
            newContent = newContent.replace(/^\s*[\r\n]/gm, '');
            newContent = newContent.replace(/\n\s*\n/g, '\n');
          } else {
            newContent = removeJavaScriptComments(content);
          }
          
          console.log(`New size: ${newContent.length} bytes`);
          console.log(`Removed: ${content.length - newContent.length} bytes (${((content.length - newContent.length) / content.length * 100).toFixed(2)}%)`);
          
          // Write to a test file
          const testOutputFile = testFile + '.cleaned';
          fs.writeFileSync(testOutputFile, newContent, 'utf8');
          console.log(`Cleaned file written to: ${testOutputFile}`);
        }
      }
    } catch (error) {
      console.error('Error:', error.message);
    }
  }

  // Get pattern from command line
  const pattern = process.argv[2] || 'C:/Users/darw1/Downloads/testing/**/*.{js,ts,jsx,tsx,css,html,json}';
  testGlob(pattern);
  
} catch (error) {
  console.error('Error:', error.message);
  process.exit(1);
}