remove-comment-edits.ts

import fs from 'fs';

interface Bug {
    repo: string;
    prNumber: number;
    diffHunk: string;
    commentLink: string;
    commentBody: string;
    filePath: string;
    [key: string]: any;
}

// Helper function to check if a line is a comment based on common comment syntaxes
function isCommentLine(line: string): boolean {
    // Trim whitespace
    line = line.trim();
    
    // Common comment patterns across languages
    const commentPatterns = [
        /^\/\//, // C-style single line comments
        /^\/\*/, // C-style multi-line comments start
        /^\*/, // C-style multi-line comment continuation
        /^\#/, // Python/Ruby/Shell style comments
        /^--/, // SQL/Haskell style comments
        /^;/, // Lisp/Assembly style comments
        /^<!--/, // HTML/XML style comments
        /^%/, // LaTeX/Matlab style comments
        /^\/\/\//, // Triple slash comments (e.g., TypeScript)
        /^'''/, // Python triple quotes
        /^"""/, // Python triple double quotes
        /^<!--/, // HTML comments
    ];

    return commentPatterns.some(pattern => pattern.test(line));
}

// Function to check if diff only changes comments
function isCommentOnlyChange(diffHunk: string): boolean {
    // Split diff into lines
    const lines = diffHunk.split('\n');
    
    // Track if we've seen any non-comment changes
    let hasNonCommentChanges = false;
    
    for (const line of lines) {
        // Skip diff metadata lines (starting with @@)
        if (line.startsWith('@@')) continue;
        
        // Only look at added/removed lines
        if (!line.startsWith('+') && !line.startsWith('-')) continue;
        
        // Remove the +/- prefix
        const codeLine = line.slice(1);
        
        // If this changed line is not a comment, we found a non-comment change
        if (!isCommentLine(codeLine)) {
            hasNonCommentChanges = true;
            break;
        }
    }
    
    return !hasNonCommentChanges;
}

async function main() {
    console.log('Starting to process bugs.json...');
    
    // Read bugs.json
    const bugsData = JSON.parse(fs.readFileSync('bugs.json', 'utf8'));
    const bugs: Bug[] = Array.isArray(bugsData) ? bugsData : Object.values(bugsData);
    
    console.log(`Total bugs before filtering: ${bugs.length}`);
    
    const filteredBugs: Bug[] = [];
    let commentEditCount = 0;
    
    // Process each bug
    for (const bug of bugs) {
        if (!isCommentOnlyChange(bug.diffHunk)) {
            filteredBugs.push(bug);
        } else {
            commentEditCount++;
            console.log(`Found comment-only change in ${bug.repo}#${bug.prNumber}:`);
            console.log(`  File: ${bug.filePath}`);
            console.log(`  Comment: ${bug.commentBody.slice(0, 100)}...`);
            console.log(`  Link: ${bug.commentLink}`);
            console.log(`  Diff:\n${bug.diffHunk}\n`);
        }
    }

    console.log('\nSummary:');
    console.log(`Total bugs: ${bugs.length}`);
    console.log(`Comment-only changes found: ${commentEditCount}`);
    console.log(`Remaining bugs: ${filteredBugs.length}`);

    // Backup original file
    const timestamp = new Date().toISOString().replace(/[:.]/g, '-');
    fs.copyFileSync('bugs.json', `bugs.backup.${timestamp}.json`);
    console.log(`\nCreated backup at bugs.backup.${timestamp}.json`);

    // Write filtered data back to bugs.json
    const outputData = Array.isArray(bugsData) 
        ? filteredBugs 
        : Object.fromEntries(filteredBugs.map((bug, index) => [index.toString(), bug]));
    
    fs.writeFileSync('bugs.json', JSON.stringify(outputData, null, 2));
    console.log('Updated bugs.json with comment-only changes removed');
}

main().catch(console.error);