scripts/bus-factor.mjs

@@ -0,0 +1,213 @@
+#!/usr/bin/env node
+/**
+ * Bus-Factor Analyzer
+ * Analyzes code ownership concentration by examining git commit history.
+ * Usage: node scripts/bus-factor.js [--json] [--min-commits N] [--threshold N] [--top N]
+ */
+
+import { execSync } from 'child_process';
+import { fileURLToPath } from 'url';
+import path from 'path';
+
+// --- Pure analysis functions (exported for testing) ---
+
+/**
+ * Parse raw `git log --numstat` output into a map of file -> author -> commitCount.
+ * @param {string} rawLog - Output from git log --numstat
+ * @returns {Object} { [filePath]: { [author]: number } }
+ */
+export function parseGitLog(rawLog) {
+  const ownership = {};
+  const lines = rawLog.split('\n');
+  let currentAuthor = null;
+
+  for (const line of lines) {
+    // Commit header line: "commit <hash>"
+    if (line.startsWith('commit ')) {
+      currentAuthor = null;
+      continue;
+    }
+
+    // Author line: "Author: Name <email>"
+    const authorMatch = line.match(/^Author:\s+(.+?)\s+<[^>]+>/);
+    if (authorMatch) {
+      currentAuthor = authorMatch[1].trim();
+      continue;
+    }
+
+    // Numstat line: "<added>\t<deleted>\t<filename>"
+    if (currentAuthor && /^\d+\t\d+\t/.test(line)) {
+      const parts = line.split('\t');
+      if (parts.length < 3) continue;
+
+      // Handle rename: "old/path => new/path" or "{old => new}/suffix"
+      let filePath = parts[2];
+      if (filePath.includes('{') && filePath.includes('=>')) {
+        filePath = filePath.replace(/\{([^}]*?)\s*=>\s*([^}]*?)\}/g, '$2').replace(/\s+/g, '');
+      } else if (filePath.includes(' => ')) {
+        filePath = filePath.split(' => ')[1].trim();
+      }
+
+      filePath = filePath.trim();
+      if (!filePath) continue;
+
+      if (!ownership[filePath]) ownership[filePath] = {};
+      ownership[filePath][currentAuthor] = (ownership[filePath][currentAuthor] || 0) + 1;
+    }
+  }
+
+  return ownership;
+}
+
+/**
+ * Compute ownership metrics for a single file.
+ * @param {Object} authorCounts - { [author]: commitCount }
+ * @param {number} ownershipThreshold - min fraction to count toward bus-factor (default 0.1)
+ * @returns {Object} { totalCommits, authors, busFactor, primaryOwner }
+ */
+export function computeOwnership(authorCounts, ownershipThreshold = 0.1) {
+  const entries = Object.entries(authorCounts).sort((a, b) => b[1] - a[1]);
+  const totalCommits = entries.reduce((sum, [, n]) => sum + n, 0);
+
+  const authors = entries.map(([name, commits]) => ({
+    name,
+    commits,
+    pct: totalCommits > 0 ? commits / totalCommits : 0,
+  }));
+
+  const busFactor = authors.filter(a => a.pct >= ownershipThreshold).length;
+  const primaryOwner = authors[0] || null;
+
+  return { totalCommits, authors, busFactor, primaryOwner };
+}
+
+/**
+ * Score all files and return sorted results.
+ * @param {Object} ownership - Output from parseGitLog
+ * @param {Object} options
+ * @returns {Array} Sorted file entries with ownership metrics
+ */
+export function scoreFiles(ownership, { minCommits = 2, ownershipThreshold = 0.1 } = {}) {
+  const results = [];
+
+  for (const [filePath, authorCounts] of Object.entries(ownership)) {
+    const metrics = computeOwnership(authorCounts, ownershipThreshold);
+    if (metrics.totalCommits < minCommits) continue;
+
+    results.push({ file: filePath, ...metrics });
+  }
+
+  // Sort: lowest bus-factor first, then most commits (highest risk first)
+  results.sort((a, b) => {
+    if (a.busFactor !== b.busFactor) return a.busFactor - b.busFactor;
+    return b.totalCommits - a.totalCommits;
+  });
+
+  return results;
+}
+
+/**
+ * Compute overall repo stats (weighted average bus-factor, high-risk count).
+ */
+export function repoStats(scoredFiles) {
+  if (scoredFiles.length === 0) return { avgBusFactor: 0, highRiskCount: 0, totalFiles: 0 };
+
+  const totalCommits = scoredFiles.reduce((s, f) => s + f.totalCommits, 0);
+  const weightedBf = scoredFiles.reduce((s, f) => s + f.busFactor * f.totalCommits, 0);
+  const avgBusFactor = totalCommits > 0 ? weightedBf / totalCommits : 0;
+  const highRiskCount = scoredFiles.filter(f => f.busFactor === 1).length;
+
+  return { avgBusFactor, highRiskCount, totalFiles: scoredFiles.length };
+}
+
+// --- CLI ---
+
+function collectGitLog(repoRoot) {
+  const dirs = ['server/src', 'client/src', 'db/migrations'];
+  const cmd = `git -C "${repoRoot}" log --numstat -- ${dirs.join(' ')}`;
+  return execSync(cmd, { maxBuffer: 50 * 1024 * 1024 }).toString();
+}
+
+function formatReport(scoredFiles, stats, topN = 10) {
+  const lines = [];
+  lines.push('');
+  lines.push('=== Bus-Factor Analysis ===');
+  lines.push('');
+  lines.push(`Files analyzed : ${stats.totalFiles}`);
+  lines.push(`High-risk files: ${stats.highRiskCount} (bus-factor = 1)`);
+  lines.push(`Avg bus-factor : ${stats.avgBusFactor.toFixed(2)} (weighted by commits)`);
+  lines.push('');
+
+  const highRisk = scoredFiles.filter(f => f.busFactor === 1);
+  if (highRisk.length === 0) {
+    lines.push('No high-risk files found.');
+  } else {
+    lines.push(`--- Top ${Math.min(topN, highRisk.length)} High-Risk Files (bus-factor = 1) ---`);
+    lines.push('');
+    for (const f of highRisk.slice(0, topN)) {
+      const owner = f.primaryOwner;
+      lines.push(`  ${f.file}`);
+      lines.push(`    commits: ${f.totalCommits}  owner: ${owner.name} (${(owner.pct * 100).toFixed(0)}%)`);
+      if (f.authors.length > 1) {
+        const others = f.authors.slice(1, 3).map(a => `${a.name} ${(a.pct * 100).toFixed(0)}%`).join(', ');
+        lines.push(`    others: ${others}`);
+      }
+      lines.push('');
+    }
+  }
+
+  lines.push('--- Author Contribution Summary ---');
+  lines.push('');
+  const authorTotals = {};
+  for (const f of scoredFiles) {
+    for (const a of f.authors) {
+      authorTotals[a.name] = (authorTotals[a.name] || 0) + a.commits;
+    }
+  }
+  const totalAll = Object.values(authorTotals).reduce((s, n) => s + n, 0);
+  const sorted = Object.entries(authorTotals).sort((a, b) => b[1] - a[1]);
+  for (const [name, commits] of sorted) {
+    const pct = totalAll > 0 ? (commits / totalAll * 100).toFixed(1) : '0.0';
+    lines.push(`  ${name.padEnd(30)} ${String(commits).padStart(5)} commits  (${pct}%)`);
+  }
+  lines.push('');
+
+  return lines.join('\n');
+}
+
+// Detect if running as main script (ESM equivalent of require.main === module)
+const isMain = process.argv[1] === fileURLToPath(import.meta.url);
+
+if (isMain) {
+  const args = process.argv.slice(2);
+  const jsonMode = args.includes('--json');
+
+  const minCommitsIdx = args.indexOf('--min-commits');
+  const minCommits = minCommitsIdx !== -1 ? parseInt(args[minCommitsIdx + 1], 10) : 2;
+
+  const thresholdIdx = args.indexOf('--threshold');
+  const threshold = thresholdIdx !== -1 ? parseFloat(args[thresholdIdx + 1]) : 0.1;
+
+  const topIdx = args.indexOf('--top');
+  const topN = topIdx !== -1 ? parseInt(args[topIdx + 1], 10) : 10;
+
+  const repoRoot = path.resolve(fileURLToPath(import.meta.url), '..', '..');
+
+  let rawLog;
+  try {
+    rawLog = collectGitLog(repoRoot);
+  } catch (err) {
+    process.stderr.write(`Error running git log: ${err.message}\n`);
+    process.exit(1);
+  }
+
+  const ownership = parseGitLog(rawLog);
+  const scoredFiles = scoreFiles(ownership, { minCommits, ownershipThreshold: threshold });
+  const stats = repoStats(scoredFiles);
+
+  if (jsonMode) {
+    process.stdout.write(JSON.stringify({ stats, files: scoredFiles }, null, 2) + '\n');
+  } else {
+    process.stdout.write(formatReport(scoredFiles, stats, topN));
+  }
+}

scripts/bus-factor.test.mjs

@@ -0,0 +1,202 @@
+import { describe, it, expect } from 'vitest';
+import { parseGitLog, computeOwnership, scoreFiles, repoStats } from './bus-factor.mjs';
+
+// --- parseGitLog ---
+
+describe('parseGitLog', () => {
+  it('parses a single commit with one file', () => {
+    const raw = `commit abc123
+Author: Alice <alice@example.com>
+Date:   Mon Jan 1 00:00:00 2024
+
+    Initial commit
+
+5\t2\tserver/src/index.js
+`;
+    const result = parseGitLog(raw);
+    expect(result['server/src/index.js']).toEqual({ Alice: 1 });
+  });
+
+  it('accumulates multiple commits by the same author', () => {
+    const raw = `commit aaa
+Author: Alice <alice@example.com>
+Date:   Mon Jan 1 00:00:00 2024
+
+    First
+
+3\t0\tserver/src/app.js
+
+commit bbb
+Author: Alice <alice@example.com>
+Date:   Tue Jan 2 00:00:00 2024
+
+    Second
+
+1\t1\tserver/src/app.js
+`;
+    const result = parseGitLog(raw);
+    expect(result['server/src/app.js']['Alice']).toBe(2);
+  });
+
+  it('tracks multiple authors for the same file', () => {
+    const raw = `commit aaa
+Author: Alice <alice@example.com>
+Date:   Mon Jan 1 00:00:00 2024
+
+    Alice commit
+
+2\t0\tclient/src/App.jsx
+
+commit bbb
+Author: Bob <bob@example.com>
+Date:   Tue Jan 2 00:00:00 2024
+
+    Bob commit
+
+1\t0\tclient/src/App.jsx
+`;
+    const result = parseGitLog(raw);
+    expect(result['client/src/App.jsx']['Alice']).toBe(1);
+    expect(result['client/src/App.jsx']['Bob']).toBe(1);
+  });
+
+  it('handles multiple files per commit', () => {
+    const raw = `commit aaa
+Author: Alice <alice@example.com>
+Date:   Mon Jan 1 00:00:00 2024
+
+    Multi-file commit
+
+2\t0\tserver/src/a.js
+3\t1\tserver/src/b.js
+`;
+    const result = parseGitLog(raw);
+    expect(result['server/src/a.js']['Alice']).toBe(1);
+    expect(result['server/src/b.js']['Alice']).toBe(1);
+  });
+
+  it('handles rename syntax (old => new)', () => {
+    const raw = `commit aaa
+Author: Alice <alice@example.com>
+Date:   Mon Jan 1 00:00:00 2024
+
+    Rename
+
+2\t0\told/path.js => new/path.js
+`;
+    const result = parseGitLog(raw);
+    expect(result['new/path.js']).toBeDefined();
+    expect(result['old/path.js']).toBeUndefined();
+  });
+
+  it('returns empty object for empty log', () => {
+    expect(parseGitLog('')).toEqual({});
+  });
+});
+
+// --- computeOwnership ---
+
+describe('computeOwnership', () => {
+  it('computes bus-factor of 1 for a solo author', () => {
+    const result = computeOwnership({ Alice: 10 });
+    expect(result.busFactor).toBe(1);
+    expect(result.totalCommits).toBe(10);
+    expect(result.primaryOwner.name).toBe('Alice');
+    expect(result.primaryOwner.pct).toBe(1);
+  });
+
+  it('computes bus-factor of 2 when two authors each own >= 10%', () => {
+    const result = computeOwnership({ Alice: 8, Bob: 2 });
+    expect(result.busFactor).toBe(2);
+  });
+
+  it('does not count authors below the threshold', () => {
+    // Bob has 5% — below default 10% threshold
+    const result = computeOwnership({ Alice: 19, Bob: 1 });
+    expect(result.busFactor).toBe(1);
+  });
+
+  it('respects a custom ownership threshold', () => {
+    // With 20% threshold, Bob (10%) doesn't count
+    const result = computeOwnership({ Alice: 9, Bob: 1 }, 0.2);
+    expect(result.busFactor).toBe(1);
+  });
+
+  it('sorts authors by commit count descending', () => {
+    const result = computeOwnership({ Alice: 3, Bob: 7, Carol: 5 });
+    expect(result.authors[0].name).toBe('Bob');
+    expect(result.authors[1].name).toBe('Carol');
+    expect(result.authors[2].name).toBe('Alice');
+  });
+
+  it('handles empty author counts gracefully', () => {
+    const result = computeOwnership({});
+    expect(result.totalCommits).toBe(0);
+    expect(result.busFactor).toBe(0);
+    expect(result.primaryOwner).toBeNull();
+  });
+});
+
+// --- scoreFiles ---
+
+describe('scoreFiles', () => {
+  const ownership = {
+    'server/src/risk.js': { Alice: 9, Bob: 1 },       // bus-factor 1 (Bob < 10%)
+    'server/src/shared.js': { Alice: 5, Bob: 5 },      // bus-factor 2
+    'server/src/tiny.js': { Alice: 1 },                // below minCommits=2, filtered
+  };
+
+  it('filters files below minCommits', () => {
+    const results = scoreFiles(ownership, { minCommits: 2 });
+    expect(results.find(f => f.file === 'server/src/tiny.js')).toBeUndefined();
+  });
+
+  it('includes files at or above minCommits', () => {
+    const results = scoreFiles(ownership, { minCommits: 2 });
+    const files = results.map(f => f.file);
+    expect(files).toContain('server/src/risk.js');
+    expect(files).toContain('server/src/shared.js');
+  });
+
+  it('sorts lowest bus-factor first', () => {
+    const results = scoreFiles(ownership, { minCommits: 2 });
+    expect(results[0].file).toBe('server/src/risk.js');
+    expect(results[1].file).toBe('server/src/shared.js');
+  });
+
+  it('returns empty array for empty ownership', () => {
+    expect(scoreFiles({}, {})).toEqual([]);
+  });
+});
+
+// --- repoStats ---
+
+describe('repoStats', () => {
+  it('returns zeros for empty input', () => {
+    const stats = repoStats([]);
+    expect(stats.avgBusFactor).toBe(0);
+    expect(stats.highRiskCount).toBe(0);
+    expect(stats.totalFiles).toBe(0);
+  });
+
+  it('counts high-risk files (busFactor === 1)', () => {
+    const files = [
+      { busFactor: 1, totalCommits: 10, authors: [] },
+      { busFactor: 2, totalCommits: 5, authors: [] },
+      { busFactor: 1, totalCommits: 3, authors: [] },
+    ];
+    const stats = repoStats(files);
+    expect(stats.highRiskCount).toBe(2);
+    expect(stats.totalFiles).toBe(3);
+  });
+
+  it('computes weighted average bus-factor', () => {
+    const files = [
+      { busFactor: 1, totalCommits: 10, authors: [] },
+      { busFactor: 3, totalCommits: 10, authors: [] },
+    ];
+    const stats = repoStats(files);
+    // (1*10 + 3*10) / 20 = 2
+    expect(stats.avgBusFactor).toBe(2);
+  });
+});

scripts/package.json

@@ -0,0 +1,12 @@
+{
+  "name": "budget-scripts",
+  "version": "1.0.0",
+  "scripts": {
+    "test": "vitest run",
+    "test:watch": "vitest",
+    "bus-factor": "node bus-factor.mjs"
+  },
+  "devDependencies": {
+    "vitest": "^4.1.0"
+  }
+}

scripts/vitest.config.mjs

@@ -0,0 +1,8 @@
+import { defineConfig } from 'vitest/config';
+
+export default defineConfig({
+  test: {
+    globals: true,
+    environment: 'node',
+  },
+});