git-codereview/app/lib/diff.js

/** Parse unified git diffs into per-file hunks with old/new line maps. */

const FILE_HEADER = /^diff --git "?a\/(.+?)"? "?b\/(.+?)"?$/;
const HUNK_HEADER = /^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@/;

function stripPrefixQuotes(value) {
  if (value.startsWith('"') && value.endsWith('"')) {
    return value.slice(1, -1).replace(/\\(.)/g, "$1");
  }
  return value;
}

/**
 * Parse a unified diff string.
 * Returns a Map of newPath -> { oldPath, status, hunks: [{ oldStart, oldLines,
 * newStart, newLines, newLineNumbers: number[], lines: string[] }] }.
 * `newLineNumbers[i]` is the new-file line number for hunk line i, or 0 for
 * removed lines / "\ No newline" markers.
 */
export function parseUnifiedDiff(diffText) {
  const files = new Map();
  if (!diffText) return files;

  let current = null;
  let hunk = null;
  let newLine = 0;
  let oldLine = 0;

  for (const raw of diffText.split("\n")) {
    const header = FILE_HEADER.exec(raw);
    if (header) {
      const newPath = stripPrefixQuotes(header[2]);
      current = {
        oldPath: stripPrefixQuotes(header[1]),
        newPath,
        status: "modified",
        hunks: [],
      };
      files.set(newPath, current);
      hunk = null;
      continue;
    }
    if (!current) continue;

    if (raw.startsWith("new file mode")) { current.status = "added"; continue; }
    if (raw.startsWith("deleted file mode")) { current.status = "deleted"; continue; }
    if (raw.startsWith("rename to ")) { current.newPath = stripPrefixQuotes(raw.slice(10).trim()); continue; }
    if (raw.startsWith("--- ") || raw.startsWith("+++ ") || raw.startsWith("index ")
        || raw.startsWith("similarity index") || raw.startsWith("rename from")
        || raw.startsWith("old mode") || raw.startsWith("new mode")) {
      continue;
    }

    const h = HUNK_HEADER.exec(raw);
    if (h) {
      oldLine = Number(h[1]);
      newLine = Number(h[3]);
      hunk = {
        oldStart: oldLine,
        oldLines: h[2] === undefined ? 1 : Number(h[2]),
        newStart: newLine,
        newLines: h[4] === undefined ? 1 : Number(h[4]),
        newLineNumbers: [],
        lines: [],
      };
      current.hunks.push(hunk);
      continue;
    }
    if (!hunk) continue;

    hunk.lines.push(raw);
    if (raw.startsWith("+")) {
      hunk.newLineNumbers.push(newLine);
      newLine += 1;
    } else if (raw.startsWith("-")) {
      hunk.newLineNumbers.push(0);
      oldLine += 1;
    } else if (raw.startsWith("\\")) {
      hunk.newLineNumbers.push(0);
    } else {
      hunk.newLineNumbers.push(newLine);
      newLine += 1;
      oldLine += 1;
    }
  }

  return files;
}

/** All new-file line numbers present in the diff for a given path. */
export function addedLineNumbers(fileEntry) {
  const added = new Set();
  for (const hunk of fileEntry.hunks) {
    for (let i = 0; i < hunk.lines.length; i += 1) {
      if (hunk.lines[i].startsWith("+")) {
        const line = hunk.newLineNumbers[i];
        if (line > 0) added.add(line);
      }
    }
  }
  return added;
}

/** All new-file line numbers in any hunk (added + context) for a path. */
export function diffLineNumbers(fileEntry) {
  const lines = new Set();
  for (const hunk of fileEntry.hunks) {
    for (let i = 0; i < hunk.lines.length; i += 1) {
      const line = hunk.newLineNumbers[i];
      if (line > 0) lines.add(line);
    }
  }
  return lines;
}

/**
 * Choose the best inline anchor for a finding.
 * Prefers a line inside the diff (Gitea renders those inline); falls back to
 * the start line so the finding is still recorded on the pull request.
 */
export function pickAnchorLine(fileEntry, startLine, endLine) {
  const inDiff = diffLineNumbers(fileEntry);
  const added = addedLineNumbers(fileEntry);
  const lo = Math.max(1, Math.min(startLine || endLine || 1, endLine || startLine || 1));
  const hi = Math.max(startLine || endLine || 1, endLine || startLine || 1);
  for (let line = lo; line <= hi; line += 1) {
    if (added.has(line)) return { line, inDiff: true };
  }
  for (let line = lo; line <= hi; line += 1) {
    if (inDiff.has(line)) return { line, inDiff: true };
  }
  return { line: lo, inDiff: false };
}