git-tweezers/lib/core/hunk-id.js

Advanced git staging tool with hunk and line-level control

import { createHash } from 'crypto';
/**
 * Normalize a line for consistent hashing
 * - Replace tabs with spaces
 * - Trim trailing whitespace
 * - Remove carriage returns
 */
function normalizeLine(line) {
    return line
        .replace(/\r/g, '') // Remove CR
        .replace(/\t/g, ' ') // Replace tabs with spaces
        .replace(/\s+$/, ''); // Trim trailing whitespace
}
/**
 * Extract context lines before and after changes
 */
function extractContext(changes) {
    const before = [];
    const after = [];
    const actualChanges = [];
    let foundFirstChange = false;
    let lastChangeIndex = -1;
    // Find actual changes (not UnchangedLine)
    changes.forEach((change, index) => {
        if (change.type !== 'UnchangedLine') {
            if (!foundFirstChange) {
                // Collect up to 3 context lines before first change
                for (let i = Math.max(0, index - 3); i < index; i++) {
                    if (changes[i].type === 'UnchangedLine') {
                        before.push(changes[i].content);
                    }
                }
                foundFirstChange = true;
            }
            actualChanges.push(change);
            lastChangeIndex = index;
        }
    });
    // Collect up to 3 context lines after last change
    if (lastChangeIndex >= 0) {
        for (let i = lastChangeIndex + 1; i < Math.min(changes.length, lastChangeIndex + 4); i++) {
            if (changes[i].type === 'UnchangedLine') {
                after.push(changes[i].content);
            }
        }
    }
    return { before, after, actualChanges };
}
/**
 * Generate a stable content-based fingerprint for a hunk
 * Based on O3's suggestion: normalize content to be independent of +/- prefixes
 * This ensures the same content has the same ID whether staged or unstaged
 */
export function generateContentFingerprint(hunk, filePath) {
    const hash = createHash('sha256');
    // Include file path for uniqueness across files
    hash.update(filePath + '\n');
    // Extract context and changes
    const { before, after, actualChanges } = extractContext(hunk.changes);
    // Hash normalized context before (without prefixes)
    before.forEach(line => {
        hash.update(normalizeLine(line) + '\n');
    });
    // Hash actual changes WITHOUT their types (no +/- prefix)
    // This makes the fingerprint identical whether the change is staged or unstaged
    actualChanges.forEach(change => {
        // Just use the content, not the type
        hash.update(normalizeLine(change.content) + '\n');
    });
    // Hash normalized context after (without prefixes)
    after.forEach(line => {
        hash.update(normalizeLine(line) + '\n');
    });
    return hash.digest('hex');
}
/**
 * Generate a stable ID for a hunk based on its content
 * Uses content-based fingerprint to ensure stability across staging operations
 */
export function generateHunkId(hunk, filePath, existingIds) {
    const fingerprint = generateContentFingerprint(hunk, filePath);
    // Start with 4 characters, increase if collision
    let length = 4;
    let id = fingerprint.substring(0, length);
    // Handle collisions by increasing length
    while (existingIds && existingIds.has(id) && length < fingerprint.length) {
        length++;
        id = fingerprint.substring(0, length);
    }
    return id;
}
/**
 * Extract a summary from hunk changes for better identification
 */
export function getHunkSummary(hunk) {
    // Find the first meaningful change
    const meaningfulChange = hunk.changes.find(change => {
        if (change.type === 'UnchangedLine')
            return false;
        // Skip empty lines or lines with only whitespace
        const trimmed = change.content.trim();
        return trimmed.length > 0;
    });
    if (!meaningfulChange) {
        return '';
    }
    const content = meaningfulChange.content.trim();
    const maxLength = 50;
    // Truncate if too long
    if (content.length > maxLength) {
        return content.substring(0, maxLength) + '...';
    }
    return content;
}
/**
 * Get change statistics for a hunk
 */
export function getHunkStats(hunk) {
    let additions = 0;
    let deletions = 0;
    hunk.changes.forEach(change => {
        if (change.type === 'AddedLine')
            additions++;
        else if (change.type === 'DeletedLine')
            deletions++;
    });
    return { additions, deletions };
}