rspade_system/bin/js-sanitizer.js

#!/usr/bin/env node

/**
 * JavaScript Sanitizer for Code Quality Checks
 * 
 * This script removes comments and replaces string contents with spaces
 * while preserving line numbers and column positions for accurate violation reporting.
 * 
 * Usage: node js-sanitizer.js <input-file>
 * Output: Sanitized JavaScript to stdout
 */

const fs = require('fs');
const decomment = require('decomment');
const acorn = require('acorn');

// Get input file from command line arguments
const inputFile = process.argv[2];

if (!inputFile) {
    console.error('Usage: node js-sanitizer.js <input-file>');
    process.exit(1);
}

// Read the input file
let code;
try {
    code = fs.readFileSync(inputFile, 'utf8');
} catch (error) {
    console.error(`Error reading file: ${error.message}`);
    process.exit(1);
}

// Step 1: Remove comments while preserving spaces
let sanitized;
try {
    sanitized = decomment(code, { 
        space: true,  // Replace comments with spaces to preserve line numbers
        tolerant: true // Handle potential syntax issues
    });
} catch (error) {
    // If decomment fails, continue with original code
    sanitized = code;
}

// Step 2: Parse the code to find string literals and replace their contents
try {
    // Parse with location tracking
    const ast = acorn.parse(sanitized, {
        ecmaVersion: 'latest',
        sourceType: 'module',
        locations: true,
        ranges: true,
        allowReturnOutsideFunction: true,
        allowImportExportEverywhere: true,
        allowAwaitOutsideFunction: true,
        allowSuperOutsideMethod: true,
        allowHashBang: true,
        onComment: () => {} // Ignore comments (already removed)
    });
    
    // Convert code to array of characters for manipulation
    const chars = sanitized.split('');
    
    // Function to replace string content with spaces
    function replaceStringContent(node) {
        if (node.type === 'Literal' && typeof node.value === 'string') {
            // For string literals, replace the content (but keep the quotes)
            const start = node.range[0];
            const end = node.range[1];
            
            // Find the quote character (first non-whitespace character)
            let quoteChar = chars[start];
            let quoteEnd = chars[end - 1];
            
            // Replace everything between quotes with spaces
            for (let i = start + 1; i < end - 1; i++) {
                if (chars[i] !== '\n') {  // Preserve newlines for line counting
                    chars[i] = ' ';
                }
            }
        } else if (node.type === 'TemplateLiteral') {
            // For template literals, replace the raw text parts
            for (let quasi of node.quasis) {
                const start = quasi.range[0];
                const end = quasi.range[1];
                
                // Replace content between backticks or ${...}
                for (let i = start + 1; i < end - 1; i++) {
                    if (chars[i] !== '\n' && chars[i] !== '$' && chars[i] !== '{' && chars[i] !== '}') {
                        chars[i] = ' ';
                    }
                }
            }
        }
    }
    
    // Walk the AST to find all string literals
    function walk(node) {
        if (!node) return;
        
        // Process current node
        replaceStringContent(node);
        
        // Recursively process all child nodes
        for (let key in node) {
            if (key === 'range' || key === 'loc' || key === 'start' || key === 'end') {
                continue; // Skip location properties
            }
            
            const value = node[key];
            if (Array.isArray(value)) {
                for (let item of value) {
                    if (typeof item === 'object' && item !== null) {
                        walk(item);
                    }
                }
            } else if (typeof value === 'object' && value !== null) {
                walk(value);
            }
        }
    }
    
    // Walk the AST to process all strings
    walk(ast);
    
    // Convert back to string
    sanitized = chars.join('');
    
} catch (error) {
    // If parsing fails (e.g., syntax error), we still have comments removed
    // Just continue with the decommented version
}

// Output the sanitized code
process.stdout.write(sanitized);