Merge branch 'main' into dependabot/github_actions/dot-github/workflows/actions-minor-da8be134b1

pr-checks/check-repo-size.test.ts

@@ -0,0 +1,259 @@
+#!/usr/bin/env npx tsx
+
+/*
+Tests for check-repo-size.ts.
+*/
+
+import * as assert from "node:assert/strict";
+import { execFileSync } from "node:child_process";
+import { randomBytes } from "node:crypto";
+import * as fs from "node:fs";
+import * as os from "node:os";
+import * as path from "node:path";
+import { afterEach, beforeEach, describe, it } from "node:test";
+
+import {
+  COMMENT_MARKER,
+  DEFAULT_BASE_REF,
+  buildCommentBody,
+  formatBytes,
+  formatPercent,
+  isDeltaSignificant,
+  measureArchiveSize,
+  readArgs,
+} from "./check-repo-size";
+
+describe("formatBytes", async () => {
+  const cases: Array<[number, boolean, string]> = [
+    // Unsigned values, including sub-KiB amounts which round to 0.00.
+    [0, false, "0.00 KiB"],
+    [512, false, "0.50 KiB"],
+    [1024, false, "1.00 KiB"],
+    [1024 * 1024, false, "1024.00 KiB"],
+    [2 * 1024 * 1024, false, "2048.00 KiB"],
+    // Negative values always use a leading minus.
+    [-2 * 1024 * 1024, false, "-2048.00 KiB"],
+    // signed=true prepends a + to non-negative values.
+    [0, true, "+0.00 KiB"],
+    [2 * 1024 * 1024, true, "+2048.00 KiB"],
+    [-2 * 1024 * 1024, true, "-2048.00 KiB"],
+  ];
+  for (const [bytes, signed, expected] of cases) {
+    await it(`formats ${bytes} (signed=${signed}) as ${expected}`, () => {
+      assert.equal(formatBytes(bytes, signed), expected);
+    });
+  }
+});
+
+describe("formatPercent", async () => {
+  await it("formats positive fractions with a leading +", () => {
+    assert.equal(formatPercent(0.1), "+10.00%");
+    assert.equal(formatPercent(0.0123), "+1.23%");
+  });
+
+  await it("formats negative fractions with a leading -", () => {
+    assert.equal(formatPercent(-0.1), "-10.00%");
+  });
+
+  await it("formats zero without a sign", () => {
+    assert.equal(formatPercent(0), "0.00%");
+  });
+});
+
+describe("isDeltaSignificant", async () => {
+  const cases: Array<[number, number, number, boolean]> = [
+    // At and above threshold (both signs).
+    [100, 1000, 0.1, true],
+    [101, 1000, 0.1, true],
+    [-100, 1000, 0.1, true],
+    // Below threshold (both signs, plus exact zero).
+    [99, 1000, 0.1, false],
+    [-99, 1000, 0.1, false],
+    [0, 1000, 0.1, false],
+  ];
+  for (const [delta, base, fraction, expected] of cases) {
+    await it(`returns ${expected} for delta=${delta}, base=${base}, fraction=${fraction}`, () => {
+      assert.equal(isDeltaSignificant(delta, base, fraction), expected);
+    });
+  }
+});
+
+describe("buildCommentBody", async () => {
+  await it("includes the marker, the base/PR/delta rows, and the run URL", () => {
+    const body = buildCommentBody({
+      baseRef: "main",
+      baseSize: 2_000_000,
+      prSize: 2_300_000,
+      runUrl: "https://example.test/run",
+    });
+
+    assert.match(body, new RegExp(`^${escapeRegExp(COMMENT_MARKER)}`));
+    assert.match(body, /Base \(`main`\) \| 1953\.13 KiB \(2000000 bytes\)/);
+    assert.match(body, /This PR \| 2246\.09 KiB \(2300000 bytes\)/);
+    assert.match(
+      body,
+      /\*\*Delta\*\* \| \*\*\+292\.97 KiB \(\+300000 bytes, \+15\.00%\)\*\*/,
+    );
+    assert.match(body, /\[workflow run\]\(https:\/\/example\.test\/run\)/);
+  });
+
+  await it("formats negative deltas with a leading minus and omits the run URL when missing", () => {
+    const body = buildCommentBody({
+      baseRef: "main",
+      baseSize: 2_000_000,
+      prSize: 1_800_000,
+    });
+    assert.match(
+      body,
+      /\*\*Delta\*\* \| \*\*-195\.31 KiB \(-200000 bytes, -10\.00%\)\*\*/,
+    );
+    assert.doesNotMatch(body, /workflow run/);
+  });
+});
+
+describe("readArgs", async () => {
+  await it("defaults the base ref and head commit for local runs", () => {
+    const originalEnv = process.env;
+    const originalArgv = process.argv;
+
+    try {
+      process.env = {};
+      process.argv = ["node", "check-repo-size.ts", "--output-dir", "/tmp/out"];
+
+      const args = readArgs();
+
+      assert.equal(args.baseRef, DEFAULT_BASE_REF);
+      assert.equal(args.baseCommitish, `origin/${DEFAULT_BASE_REF}`);
+      assert.equal(args.headCommitish, "HEAD");
+      assert.equal(args.outputDir, "/tmp/out");
+      assert.equal(args.runUrl, undefined);
+    } finally {
+      process.env = originalEnv;
+      process.argv = originalArgv;
+    }
+  });
+
+  await it("uses the base and head SHAs when provided by the workflow", () => {
+    const originalEnv = process.env;
+    const originalArgv = process.argv;
+
+    try {
+      process.env = {
+        BASE_REF: "main",
+        BASE_SHA: "abc123",
+        HEAD_SHA: "def456",
+        RUN_URL: "https://example.test/run",
+      };
+      process.argv = ["node", "check-repo-size.ts", "--output-dir", "/tmp/out"];
+
+      const args = readArgs();
+
+      assert.equal(args.baseRef, "main");
+      assert.equal(args.baseCommitish, "abc123");
+      assert.equal(args.headCommitish, "def456");
+      assert.equal(args.outputDir, "/tmp/out");
+      assert.equal(args.runUrl, "https://example.test/run");
+    } finally {
+      process.env = originalEnv;
+      process.argv = originalArgv;
+    }
+  });
+
+  await it("throws when --output-dir is missing", () => {
+    const originalEnv = process.env;
+    const originalArgv = process.argv;
+
+    try {
+      process.env = {};
+      process.argv = ["node", "check-repo-size.ts"];
+      assert.throws(() => readArgs(), /--output-dir is required/);
+    } finally {
+      process.env = originalEnv;
+      process.argv = originalArgv;
+    }
+  });
+});
+
+let repoDir: string;
+
+beforeEach(() => {
+  repoDir = fs.mkdtempSync(path.join(os.tmpdir(), "check-repo-size-test-"));
+  execFileSync("git", ["init", "--initial-branch=main", "-q"], {
+    cwd: repoDir,
+  });
+  execFileSync("git", ["config", "user.email", "test@example.test"], {
+    cwd: repoDir,
+  });
+  execFileSync("git", ["config", "user.name", "Test"], { cwd: repoDir });
+  execFileSync("git", ["config", "commit.gpgsign", "false"], { cwd: repoDir });
+});
+
+afterEach(() => {
+  fs.rmSync(repoDir, { recursive: true, force: true });
+});
+
+function commit(name: string, content: string, message: string) {
+  fs.writeFileSync(path.join(repoDir, name), content);
+  execFileSync("git", ["add", name], { cwd: repoDir });
+  execFileSync("git", ["commit", "-q", "-m", message], { cwd: repoDir });
+}
+
+describe("measureArchiveSize", async () => {
+  await it("returns a positive byte count for a non-empty repo", async () => {
+    commit("a.txt", "hello world\n", "first");
+    const size = await measureArchiveSize("HEAD", repoDir);
+    assert.ok(size > 0, `expected size > 0, got ${size}`);
+  });
+
+  await it("returns the same size on repeated runs (deterministic)", async () => {
+    commit("a.txt", "hello world\n", "first");
+    const a = await measureArchiveSize("HEAD", repoDir);
+    const b = await measureArchiveSize("HEAD", repoDir);
+    assert.equal(a, b);
+  });
+
+  await it("returns a larger size when more content is added", async () => {
+    commit("a.txt", "hello world\n", "first");
+    const small = await measureArchiveSize("HEAD", repoDir);
+
+    // Use random bytes so the new content is incompressible and the archive
+    // is guaranteed to grow even after gzip.
+    commit("b.bin", randomBytes(8192).toString("base64"), "second");
+    const big = await measureArchiveSize("HEAD", repoDir);
+    assert.ok(
+      big > small,
+      `expected ${big} > ${small} after adding more content`,
+    );
+  });
+
+  await it("ignores untracked files (e.g. node_modules)", async () => {
+    commit("a.txt", "hello\n", "first");
+    commit(".gitignore", "node_modules/\n", "ignore node_modules");
+    const sizeBefore = await measureArchiveSize("HEAD", repoDir);
+
+    fs.mkdirSync(path.join(repoDir, "node_modules"));
+    fs.writeFileSync(
+      path.join(repoDir, "node_modules", "huge.bin"),
+      "x".repeat(1_000_000),
+    );
+
+    const sizeAfter = await measureArchiveSize("HEAD", repoDir);
+    assert.equal(
+      sizeAfter,
+      sizeBefore,
+      "untracked node_modules should not affect the archive size",
+    );
+  });
+
+  await it("rejects when the ref does not exist", async () => {
+    commit("a.txt", "hello\n", "first");
+    await assert.rejects(
+      () => measureArchiveSize("does-not-exist", repoDir),
+      /git archive does-not-exist exited with code/,
+    );
+  });
+});
+
+function escapeRegExp(s: string): string {
+  return s.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}

pr-checks/check-repo-size.ts

@@ -0,0 +1,223 @@
+#!/usr/bin/env npx tsx
+
+/*
+Measures the difference in the `.tar.gz`'d checkout size of the repo between the PR head and the PR
+base. This size is relevant because it corresponds to the duration of the "Download action
+repository" step that happens at the start of every job that uses this Action.
+
+Writes the candidate sticky-comment body and a small metadata file to `--output-dir`. A separate
+workflow job consumes those artifacts and decides whether to create or update a PR comment.
+*/
+
+import { spawn } from "node:child_process";
+import * as fs from "node:fs";
+import * as path from "node:path";
+import { parseArgs } from "node:util";
+
+import { REPO_ROOT } from "./config";
+
+/** Hidden marker used to find the existing sticky comment on a PR. */
+export const COMMENT_MARKER = "<!-- repo-size-diff-bot -->";
+
+export const DEFAULT_BASE_REF = "main";
+
+/**
+ * Fraction of the base archive size at which a delta is considered significant enough to warrant
+ * a new sticky comment. We always update an existing comment regardless, so the comment stays in
+ * sync as the diff evolves.
+ */
+export const SIGNIFICANT_DELTA_FRACTION = 0.1;
+
+/**
+ * Stream `git archive --format=tar.gz <ref>` and count the compressed bytes.
+ *
+ * `git archive` only includes tracked files, so untracked directories like `node_modules` and
+ * `build` aren't counted in the size downloaded when starting up a CodeQL job.
+ */
+export async function measureArchiveSize(
+  ref: string,
+  cwd: string,
+): Promise<number> {
+  const git = spawn("git", ["archive", "--format=tar.gz", ref], { cwd });
+
+  let stderr = "";
+  git.stderr.on("data", (chunk: Buffer) => {
+    stderr += chunk.toString();
+  });
+
+  let size = 0;
+  git.stdout.on("data", (chunk: Buffer) => {
+    size += chunk.length;
+  });
+
+  const exitCode = await new Promise<number>((resolve, reject) => {
+    git.on("error", reject);
+    git.on("close", resolve);
+  });
+
+  if (exitCode !== 0) {
+    throw new Error(
+      `git archive ${ref} exited with code ${exitCode}: ${stderr.trim()}`,
+    );
+  }
+  return size;
+}
+
+/**
+ * Format a byte count as KiB. If `signed` is true, a leading `+` is prepended for non-negative
+ * values so gains and losses are visually distinct.
+ */
+export function formatBytes(bytes: number, signed = false): string {
+  const sign = bytes < 0 ? "-" : signed ? "+" : "";
+  const kib = Math.abs(bytes) / 1024;
+  return `${sign}${kib.toFixed(2)} KiB`;
+}
+
+/** Format a fraction as a signed percentage with 2 decimal places. */
+export function formatPercent(fraction: number): string {
+  const pct = fraction * 100;
+  const sign = pct > 0 ? "+" : "";
+  return `${sign}${pct.toFixed(2)}%`;
+}
+
+export interface CommentBodyOptions {
+  baseRef: string;
+  baseSize: number;
+  prSize: number;
+  /** Optional URL of the workflow run, included in the comment footer. */
+  runUrl?: string;
+}
+
+export function buildCommentBody(opts: CommentBodyOptions): string {
+  const { baseRef, baseSize, prSize, runUrl } = opts;
+  const delta = prSize - baseSize;
+  const signedDelta = delta >= 0 ? `+${delta}` : `${delta}`;
+  const runUrlLine = runUrl
+    ? ` See the [workflow run](${runUrl}) for details.`
+    : "";
+
+  return [
+    COMMENT_MARKER,
+    "### Repository checkout size",
+    "",
+    "| | Compressed archive size |",
+    "|---|---|",
+    `| Base (\`${baseRef}\`) | ${formatBytes(baseSize)} (${baseSize} bytes) |`,
+    `| This PR | ${formatBytes(prSize)} (${prSize} bytes) |`,
+    `| **Delta** | **${formatBytes(delta, true)} (${signedDelta} bytes, ${formatPercent(delta / baseSize)})** |`,
+    "",
+    "Sizes are measured by streaming `git archive --format=tar.gz <ref>`, " +
+      "which includes tracked files and excludes untracked files such as " +
+      "`node_modules`. The compressed checkout is " +
+      "downloaded by every consumer of this Action, so changes here directly " +
+      `affect Action download time.${runUrlLine}`,
+  ].join("\n");
+}
+
+/**
+ * Returns true when the absolute delta is at least `fraction` of the base size. Both increases and
+ * decreases are considered significant, so we report wins as well as losses.
+ */
+export function isDeltaSignificant(
+  delta: number,
+  baseSize: number,
+  fraction: number,
+): boolean {
+  return Math.abs(delta) >= baseSize * fraction;
+}
+
+interface MainArgs {
+  /** Base ref of the PR. Defaults to `main`. Used as the label in the PR comment. */
+  baseRef: string;
+  /** Base commit-ish to archive. Defaults to `origin/<baseRef>` for local runs. */
+  baseCommitish: string;
+  /** Head commit-ish to archive. Defaults to `HEAD` for local runs. */
+  headCommitish: string;
+  /** Optional URL of the workflow run, surfaced in the comment footer. */
+  runUrl?: string;
+  /** Directory where `body.md` and `metadata.json` are written. */
+  outputDir: string;
+}
+
+export function readArgs(): MainArgs {
+  const { values } = parseArgs({
+    options: {
+      "output-dir": { type: "string" },
+    },
+    strict: true,
+  });
+
+  const outputDir = values["output-dir"];
+  if (!outputDir) {
+    throw new Error("--output-dir is required");
+  }
+
+  const baseRef = process.env.BASE_REF ?? DEFAULT_BASE_REF;
+  const baseCommitish = process.env.BASE_SHA ?? `origin/${baseRef}`;
+  const headCommitish = process.env.HEAD_SHA ?? "HEAD";
+
+  return {
+    baseRef,
+    baseCommitish,
+    headCommitish,
+    runUrl: process.env.RUN_URL,
+    outputDir,
+  };
+}
+
+async function main(): Promise<number> {
+  const args = readArgs();
+
+  console.log(`Measuring base archive size for ${args.baseCommitish}...`);
+  const baseSize = await measureArchiveSize(args.baseCommitish, REPO_ROOT);
+  console.log(`  ${baseSize} bytes`);
+
+  console.log(`Measuring PR archive size for ${args.headCommitish}...`);
+  const prSize = await measureArchiveSize(args.headCommitish, REPO_ROOT);
+  console.log(`  ${prSize} bytes`);
+
+  const delta = prSize - baseSize;
+  const significant = isDeltaSignificant(
+    delta,
+    baseSize,
+    SIGNIFICANT_DELTA_FRACTION,
+  );
+  console.log(
+    `Delta: ${delta} bytes (significant=${significant}, threshold=${(
+      SIGNIFICANT_DELTA_FRACTION * 100
+    ).toFixed(2)}%)`,
+  );
+
+  const body = buildCommentBody({
+    baseRef: args.baseRef,
+    baseSize,
+    prSize,
+    runUrl: args.runUrl,
+  });
+
+  fs.mkdirSync(args.outputDir, { recursive: true });
+  fs.writeFileSync(path.join(args.outputDir, "body.md"), body);
+  fs.writeFileSync(
+    path.join(args.outputDir, "metadata.json"),
+    `${JSON.stringify(
+      { significant, baseRef: args.baseRef, baseSize, prSize, delta },
+      null,
+      2,
+    )}\n`,
+  );
+  console.log(`Wrote body.md and metadata.json to ${args.outputDir}.`);
+  return 0;
+}
+
+async function run(): Promise<void> {
+  try {
+    process.exit(await main());
+  } catch (err) {
+    console.error(err instanceof Error ? err.message : String(err));
+    process.exit(1);
+  }
+}
+
+if (require.main === module) {
+  void run();
+}

pr-checks/config.ts

@@ -6,14 +6,17 @@ export const OLDEST_SUPPORTED_MAJOR_VERSION = 3;
 /** The `pr-checks` directory. */
 export const PR_CHECKS_DIR = __dirname;
 
+/** The repository root. */
+export const REPO_ROOT = path.join(PR_CHECKS_DIR, "..");
+
 /** The path of the file configuring which checks shouldn't be required. */
 export const PR_CHECK_EXCLUDED_FILE = path.join(PR_CHECKS_DIR, "excluded.yml");
 
 /** The path to the esbuild metadata file. */
-export const BUNDLE_METADATA_FILE = path.join(PR_CHECKS_DIR, "..", "meta.json");
+export const BUNDLE_METADATA_FILE = path.join(REPO_ROOT, "meta.json");
 
 /** The `src` directory. */
-const SOURCE_ROOT = path.join(PR_CHECKS_DIR, "..", "src");
+const SOURCE_ROOT = path.join(REPO_ROOT, "src");
 
 /** The path to the built-in languages file. */
 export const BUILTIN_LANGUAGES_FILE = path.join(

pr-checks/excluded.yml

@@ -1,16 +1,17 @@
 # PR checks to exclude from required checks
 contains:
-  - "https://"
-  - "Update"
   - "ESLint"
-  - "update"
+  - "https://"
   - "test-setup-python-scripts"
+  - "update"
+  - "Update"
 is:
+  - "Agent"
+  - "check-expected-release-files"
+  - "Cleanup artifacts"
   - "CodeQL"
   - "Dependabot"
-  - "check-expected-release-files"
-  - "Agent"
-  - "Cleanup artifacts"
+  - "Label PR with size"
+  - "Post repo size comment"
   - "Prepare"
   - "Upload results"
-  - "Label PR with size"

pr-checks/package.json

@@ -7,10 +7,10 @@
     "@octokit/core": "^7.0.6",
     "@octokit/plugin-paginate-rest": ">=9.2.2",
     "@octokit/plugin-rest-endpoint-methods": "^17.0.0",
-    "yaml": "^2.8.4"
+    "yaml": "^2.9.0"
   },
   "devDependencies": {
-    "@types/node": "^20.19.39",
+    "@types/node": "^20.19.41",
     "tsx": "^4.21.0"
   }
 }