diff-word-list.js

const childProcess = require("child_process");

function gitCmd(args) {
  const cmd = childProcess.spawnSync("/usr/local/bin/git", args);
  return cmd.stdout.toString().split("\n");
}

function commitsForPath(path) {
  return gitCmd(["rev-list", "--format=oneline", "--all", path])
    .map(x => [x.substr(0, 40), x.substr(41)])
    .reverse();
}

function diffs(commits) {
  const out = [];
  for (var i = 0; i < commits.length - 1; i++) {
    const changes = diff(commits[i][0], commits[i + 1][0]);
    if (changes.changeCount > 0) {
      out.push({
        commit: commits[i + 1][0],
        message: commits[i + 1][1],
        ...changes
      });
    }
  }
  return out;
}

function diff(hash1, hash2) {
  let changes = gitCmd(["diff", "-U0", `${hash1}..${hash2}`, "words"]).filter(
    l => l.match(/^[+-]([A-z0-9_-])+$/)
  );
  let additions = new Set(
    changes.filter(l => l[0] == "+").map(l => l.substr(1))
  );
  let removals = new Set(
    changes.filter(l => l[0] == "-").map(l => l.substr(1))
  );
  let netAdditions = setDifference(additions, removals);
  let netRemovals = setDifference(removals, additions);
  return {
    changeCount: netAdditions.size + netRemovals.size,
    removals: Array.from(netRemovals),
    additions: Array.from(netAdditions),
    moves: Array.from(setIntersection(additions, removals))
  };
}

function setDifference(a, b) {
  const difference = new Set(a);
  for (let el of b) {
    difference.delete(el);
  }
  return difference;
}

function setIntersection(a, b) {
  const intersection = new Set();
  for (let el of a) {
    if (b.has(el)) {intersection.add(el)}
  }
  return intersection;
}

// usage: node diff.js | jq '.[] | select(.removals | length > 0) | select(.removals | length < 100) | {commit: .commit, message: .message, removals: .removals}'
console.log(JSON.stringify(diffs(commitsForPath(process.argv[2] || 'words')), null, 2));