From 151680eef93cd0b9dd6930ae5bafef7ec7d78ae4 Mon Sep 17 00:00:00 2001
From: Hao Sun <131156805+surbadest@users.noreply.github.com>
Date: Tue, 22 Sep 2026 01:40:46 +0000
Subject: [PATCH] ci: add deterministic pull request size labels (#42615)

Co-authored-by: surbadest <surbadest@users.noreply.github.com>
---
 .github/scripts/size-labeler.test.cjs | 94 +++++++++++++++++++++++++++
 .github/workflows/size-labeler.yml    | 55 ++++++++++++++++
 2 files changed, 149 insertions(+)
 create mode 100644 .github/scripts/size-labeler.test.cjs
 create mode 100644 .github/workflows/size-labeler.yml

diff --git a/.github/scripts/size-labeler.test.cjs b/.github/scripts/size-labeler.test.cjs
new file mode 100644
index 00000000000..947df4dc996
--- /dev/null
+++ b/.github/scripts/size-labeler.test.cjs
@@ -0,0 +1,94 @@
+const fs = require('node:fs')
+const assert = require('node:assert/strict')
+const test = require('node:test')
+const source = fs.readFileSync(
+  require('node:path').join(__dirname, '../workflows/size-labeler.yml'),
+  'utf8',
+)
+const script = source
+  .replace(/\r\n/g, '\n')
+  .split('          script: |\n')[1]
+  .split('\n')
+  .map((l) => l.replace(/^            /, ''))
+  .join('\n')
+const run = new (Object.getPrototypeOf(async function () {}).constructor)(
+  'github',
+  'context',
+  'core',
+  script,
+)
+async function scenario(files, existing = [], changed = files.length) {
+  const calls = []
+  const warnings = []
+  const api = {
+    pulls: {
+      get: async () => ({ data: { changed_files: changed } }),
+      listFiles: 'files',
+    },
+    issues: {
+      listLabelsOnIssue: 'labels',
+      addLabels: async (p) => calls.push(['add', p.labels[0]]),
+      removeLabel: async (p) => calls.push(['remove', p.name]),
+    },
+  }
+  await run(
+    {
+      rest: api,
+      paginate: async (endpoint) =>
+        endpoint === 'files' ? files : existing.map((name) => ({ name })),
+    },
+    {
+      repo: { owner: 'test', repo: 'repo' },
+      payload: { pull_request: { number: 1 } },
+    },
+    { warning: (s) => warnings.push(s) },
+  )
+  return { calls, warnings }
+}
+for (const [n, label] of [
+  [0, 'XS'],
+  [9, 'XS'],
+  [10, 'S'],
+  [29, 'S'],
+  [30, 'M'],
+  [99, 'M'],
+  [100, 'L'],
+  [499, 'L'],
+  [500, 'XL'],
+  [999, 'XL'],
+  [1000, 'XXL'],
+]) {
+  test(`threshold ${n}`, async () =>
+    assert.deepEqual((await scenario([{ filename: 'a.py', additions: n, deletions: 0 }])).calls, [
+      ['add', `size:${label}`],
+    ]))
+}
+test('count deletions', async () =>
+  assert.deepEqual((await scenario([{ filename: 'a.py', additions: 0, deletions: 30 }])).calls, [
+    ['add', 'size:M'],
+  ]))
+test('ignore nested lockfiles', async () =>
+  assert.deepEqual(
+    (
+      await scenario([
+        { filename: 'api/uv.lock', additions: 9000, deletions: 5 },
+        { filename: 'web/pnpm-lock.yaml', additions: 5000, deletions: 0 },
+      ])
+    ).calls,
+    [['add', 'size:XS']],
+  ))
+test('replace stale size only', async () =>
+  assert.deepEqual(
+    (await scenario([{ filename: 'a.py', additions: 10, deletions: 0 }], ['size:L', 'bug', 'web']))
+      .calls,
+    [
+      ['add', 'size:S'],
+      ['remove', 'size:L'],
+    ],
+  ))
+test('idempotent', async () => assert.deepEqual((await scenario([], ['size:XS', 'bug'])).calls, []))
+test('incomplete file list', async () => {
+  const r = await scenario([], ['size:L'], 3001)
+  assert.deepEqual(r.calls, [])
+  assert.equal(r.warnings.length, 1)
+})
diff --git a/.github/workflows/size-labeler.yml b/.github/workflows/size-labeler.yml
new file mode 100644
index 00000000000..b6807655b1f
--- /dev/null
+++ b/.github/workflows/size-labeler.yml
@@ -0,0 +1,55 @@
+name: Pull Request Size Labeler
+
+on:
+  pull_request_target:
+    types: [opened, reopened, synchronize, ready_for_review, edited]
+
+permissions:
+  pull-requests: write
+
+concurrency:
+  group: size-labeler-${{ github.event.pull_request.number }}
+  cancel-in-progress: false
+
+jobs:
+  label:
+    runs-on: depot-ubuntu-24.04
+    steps:
+      # Only read PR metadata. Never check out or execute the PR's code here.
+      - uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
+        with:
+          script: |
+            const { owner, repo } = context.repo
+            const issue_number = context.payload.pull_request.number
+            const { data: pr } = await github.rest.pulls.get({ owner, repo, pull_number: issue_number })
+            const files = await github.paginate(github.rest.pulls.listFiles, {
+              owner, repo, pull_number: issue_number, per_page: 100,
+            })
+            // GitHub caps this endpoint at 3,000 files; avoid labeling a partial count.
+            if (files.length !== pr.changed_files) {
+              core.warning('Skipping size label because the changed-file list is incomplete')
+              return
+            }
+            const ignoredNames = new Set(['uv.lock', 'pnpm-lock.yaml', 'package-lock.json', 'yarn.lock'])
+            const changes = files
+              .filter(file => !ignoredNames.has(file.filename.split('/').pop()))
+              .reduce((total, file) => total + file.additions + file.deletions, 0)
+            // Match the descriptions of the repository's existing size labels.
+            const buckets = [
+              [10, 'size:XS'], [30, 'size:S'], [100, 'size:M'],
+              [500, 'size:L'], [1000, 'size:XL'], [Infinity, 'size:XXL'],
+            ]
+            const label = buckets.find(([limit]) => changes < limit)[1]
+            const sizeLabels = new Set(buckets.map(([, name]) => name))
+            const labels = await github.paginate(github.rest.issues.listLabelsOnIssue, {
+              owner, repo, issue_number, per_page: 100,
+            })
+            // Add first so an API error cannot leave the PR without a size label.
+            if (!labels.some(item => item.name === label)) {
+              await github.rest.issues.addLabels({ owner, repo, issue_number, labels: [label] })
+            }
+            for (const item of labels) {
+              if (sizeLabels.has(item.name) && item.name !== label) {
+                await github.rest.issues.removeLabel({ owner, repo, issue_number, name: item.name })
+              }
+            }