import type { DiffHunk, DiffLine, ParsedDiff, ParsedDiffFile } from '../../mcp/types/codeReview.js';
/**
* Parse a unified diff into the ParsedDiff structure used by the existing code review system.
*
* Intentionally matches the legacy behavior of CodeReviewService.parseDiff() so we can reuse
* parsing across multiple reviewers without changing outputs.
*/
export function parseUnifiedDiff(diffContent: string): ParsedDiff {
const files: ParsedDiffFile[] = [];
let totalAdded = 0;
let totalRemoved = 0;
// Split by file headers
const fileRegex = /^diff --git a\/(.+?) b\/(.+?)$/gm;
const matches = [...diffContent.matchAll(fileRegex)];
for (let i = 0; i < matches.length; i++) {
const match = matches[i];
const oldPath = match[1];
const newPath = match[2];
const startIdx = match.index ?? 0;
const endIdx = matches[i + 1]?.index ?? diffContent.length;
const fileSection = diffContent.slice(startIdx, endIdx);
const file = parseFileSection(fileSection, oldPath, newPath);
files.push(file);
totalAdded += file.hunks.reduce((sum, h) => sum + h.lines.filter(l => l.type === 'add').length, 0);
totalRemoved += file.hunks.reduce((sum, h) => sum + h.lines.filter(l => l.type === 'remove').length, 0);
}
return {
files,
lines_added: totalAdded,
lines_removed: totalRemoved,
};
}
function parseFileSection(section: string, oldPath: string, newPath: string): ParsedDiffFile {
const isNew = section.includes('new file mode');
const isDeleted = section.includes('deleted file mode');
const isBinary = section.includes('Binary files');
const hunks: DiffHunk[] = [];
const changedLines = new Set<number>();
// Parse hunks
const hunkRegex = /^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@/gm;
const hunkMatches = [...section.matchAll(hunkRegex)];
for (let i = 0; i < hunkMatches.length; i++) {
const hunkMatch = hunkMatches[i];
const hunkStart = hunkMatch.index ?? 0;
const hunkEnd = hunkMatches[i + 1]?.index ?? section.length;
const hunkContent = section.slice(hunkStart, hunkEnd);
const oldStart = parseInt(hunkMatch[1], 10);
const oldLines = parseInt(hunkMatch[2] || '1', 10);
const newStart = parseInt(hunkMatch[3], 10);
const newLines = parseInt(hunkMatch[4] || '1', 10);
const lines = parseHunkLines(hunkContent, oldStart, newStart, changedLines);
hunks.push({
old_start: oldStart,
old_lines: oldLines,
new_start: newStart,
new_lines: newLines,
lines,
});
}
return {
old_path: oldPath,
new_path: newPath,
is_new: isNew,
is_deleted: isDeleted,
is_binary: isBinary,
hunks,
changed_lines: changedLines,
};
}
function parseHunkLines(
hunkContent: string,
oldStart: number,
newStart: number,
changedLines: Set<number>
): DiffLine[] {
const lines: DiffLine[] = [];
const contentLines = hunkContent.split('\n').slice(1); // Skip the @@ header
let newLineNum = newStart;
let oldLineNum = oldStart; // Properly track old file line numbers
for (const line of contentLines) {
if (line.startsWith('+')) {
lines.push({
type: 'add',
content: line.slice(1),
new_line_number: newLineNum,
});
changedLines.add(newLineNum);
newLineNum++;
} else if (line.startsWith('-')) {
lines.push({
type: 'remove',
content: line.slice(1),
old_line_number: oldLineNum,
});
oldLineNum++;
} else if (line.startsWith(' ') || line === '') {
lines.push({
type: 'context',
content: line.slice(1) || '',
old_line_number: oldLineNum,
new_line_number: newLineNum,
});
oldLineNum++;
newLineNum++;
}
}
return lines;
}
