Add pull request size check

.github/chainguard/self.check-pull-request-size.sts.yaml

@@ -0,0 +1,11 @@
+issuer: https://token.actions.githubusercontent.com
+
+subject: repo:DataDog/dd-trace-java:pull_request
+
+claim_pattern:
+  event_name: pull_request
+  job_workflow_ref: DataDog/dd-trace-java/\.github/workflows/check-pull-request-size\.yaml@refs/(pull/[0-9]+/merge|heads/.+)
+
+permissions:
+  issues: write
+  pull_requests: write

.github/scripts/count-added-lines.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+# Counts the lines added from the merge base of two revisions, excluding moved lines and lock files.
+# Usage: count-added-lines.sh <base> [head]
+#
+# Moved lines (blocks of at least 20 alphanumeric characters deleted and added elsewhere, even
+# re-indented or in another file) are told apart from added lines using diff colors:
+# added lines are painted green and moved lines blue, then only the green lines are counted.
+set -euo pipefail
+
+BASE=${1:?Usage: $0 <base> [head]}
+HEAD=${2:-HEAD}
+
+git \
+  -c color.diff.meta=bold \
+  -c color.diff.new=green \
+  -c color.diff.newMoved=blue \
+  diff \
+  --color=always \
+  --color-moved=blocks \
+  --color-moved-ws=allow-indentation-change \
+  --ws-error-highlight=none \
+  --merge-base \
+  --find-renames \
+  --no-ext-diff \
+  "$BASE" "$HEAD" \
+  -- . ':(exclude,glob)**/gradle.lockfile' |
+  awk '/^\033\[32m\+/ { count++ } END { print count + 0 }'

.github/workflows/README.md

@@ -40,6 +40,15 @@ _Actions:_
 
 _Recovery:_ Update the pull request or add a comment to trigger the action again.
 
+### check-pull-request-size [🔗](check-pull-request-size.yaml)
+
+_Trigger:_ When creating or updating a pull request targeting `master` or a patch release (`release/vM.N.x`) branch, or when labels are updated. Draft pull requests are skipped.
+
+_Action:_ Fail the pull request if it adds 1000 lines or more, then comment asking to split it into smaller stacked pull requests.
+Deleted lines, moved code (detected using `git diff --color-moved`), and `gradle.lockfile` files are not counted.
+
+_Recovery:_ Split the pull request into smaller stacked pull requests. To override this check entirely, add the `tag: override pr size check` label to the PR. Remove the label to re-enable the check.
+
 ### enforce-datadog-merge-queue [🔗](enforce-datadog-merge-queue.yaml)
 
 _Trigger:_ When creating or updating a pull request, or when a pull request is added to GitHub merge queue.

.github/workflows/check-pull-request-size.yaml

@@ -0,0 +1,110 @@
+name: Check pull request size
+on:
+  pull_request:
+    types: [opened, reopened, edited, ready_for_review, labeled, unlabeled, synchronize]
+    branches:
+      - master
+      - "release/v*"
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  check_pull_request_size:
+    name: Check pull request size
+    # Skip draft pull requests
+    if: ${{ !github.event.pull_request.draft }}
+    permissions:
+      contents: read # Required to check out the pull request changes
+      id-token: write # Required for OIDC token federation
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout pull request changes
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # 7.0.1
+        with:
+          fetch-depth: 2 # Fetch the pull request merge commit and its parents
+          filter: blob:none
+          sparse-checkout: .github/scripts
+          persist-credentials: false
+      - name: Count added lines
+        id: count-added-lines
+        run: |
+          # The first parent of the pull request merge commit is the base branch
+          ADDED_LINES=$(.github/scripts/count-added-lines.sh HEAD^1 HEAD)
+          echo "Added lines: $ADDED_LINES"
+          echo "added_lines=$ADDED_LINES" >> "$GITHUB_OUTPUT"
+      - name: Get GitHub Token via dd-octo-sts
+        id: generate-token
+        uses: DataDog/dd-octo-sts-action@c85802a58e202b7e764b5441e4378c76043c8108 # v1.0.5
+        with:
+          scope: DataDog/dd-trace-java
+          policy: self.check-pull-request-size
+      - name: Check pull request size
+        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # 9.0.0
+        env:
+          ADDED_LINES: ${{ steps.count-added-lines.outputs.added_lines }}
+          MAX_ADDED_LINES: 1000
+        with:
+          github-token: ${{ steps.generate-token.outputs.token }}
+          script: |
+            const managedMarker = '<!-- dd-trace-java-check-pull-request-size -->'
+            const overrideLabel = 'tag: override pr size check'
+            const addedLines = parseInt(process.env.ADDED_LINES, 10)
+            const maxAddedLines = parseInt(process.env.MAX_ADDED_LINES, 10)
+
+            // Look for the previous managed comment
+            const comments = await github.paginate(github.rest.issues.listComments, {
+              issue_number: context.payload.pull_request.number,
+              owner: context.repo.owner,
+              repo: context.repo.repo
+            })
+            const existingComment = comments.find(comment => comment.body?.includes(managedMarker))
+
+            const deleteManagedComment = async () => {
+              if (existingComment) {
+                await github.rest.issues.deleteComment({
+                  comment_id: existingComment.id,
+                  owner: context.repo.owner,
+                  repo: context.repo.repo
+                })
+              }
+            }
+
+            // Check for override label — skip the check if label present
+            const labels = context.payload.pull_request.labels.map(l => l.name)
+            if (labels.includes(overrideLabel)) {
+              await deleteManagedComment()
+              console.log(`${overrideLabel} label detected — skipping the check.`)
+              return
+            }
+
+            if (addedLines < maxAddedLines) {
+              await deleteManagedComment()
+              console.log(`Pull request adds ${addedLines} lines, below the ${maxAddedLines} lines limit.`)
+              return
+            }
+
+            const body = `**❌ Pull Request Too Large**\n\n` +
+              `This pull request adds **${addedLines}** lines while the limit is **${maxAddedLines}** lines.\n` +
+              `Deleted lines, moved code, and \`gradle.lockfile\` files are not counted.\n\n` +
+              `Please split it into smaller stacked pull requests, each one targeting the branch of the previous one. ` +
+              `See the [pull request size guidelines](https://github.com/DataDog/dd-trace-java/blob/master/CONTRIBUTING.md#pull-request-size) for more details.\n\n` +
+              `If this PR cannot be split, exceptionnaly add the \`${overrideLabel}\` label to bypass this workflow.\n\n` +
+              managedMarker
+            if (existingComment) {
+              await github.rest.issues.updateComment({
+                comment_id: existingComment.id,
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                body
+              })
+            } else {
+              await github.rest.issues.createComment({
+                issue_number: context.payload.pull_request.number,
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                body
+              })
+            }
+            core.setFailed(`Pull request adds ${addedLines} lines, above the ${maxAddedLines} lines limit. See PR comment for details. To bypass this workflow, add the '${overrideLabel}' label.`)

AGENTS.md

@@ -71,6 +71,7 @@ docs/                     Developer documentation (see below)
 ## PR conventions
 
 - Title: imperative verb sentence describing user-visible change (e.g. "Fix span sampling rule parsing")
+- Size: keep PRs small (under 1000 added lines), split larger changes into stacked PRs.
 - Labels: always add `tag: ai generated` and at least one `comp:` or `inst:` label + one `type:` label
 - Use `tag: no release notes` for internal/refactoring changes
 - Open as draft first, convert to ready when reviewable

CONTRIBUTING.md

@@ -104,6 +104,17 @@ and include them in the same PR.
 This applies to changes in build scripts and `gradle/libs.versions.toml`.
 Follow the [lockfile update workflow](docs/how_to_work_with_gradle.md#updating-lock-files) before requesting review.
 
+### Pull request size
+
+Pull requests should add less than 1000 lines to remain reviewable.
+Deleted lines, moved code, and `gradle.lockfile` files are not counted.
+Run `.github/scripts/count-added-lines.sh origin/master HEAD` to check the added line count of your branch.
+
+Split larger changes into stacked pull requests: each pull request targets the branch of the previous one, and they are reviewed and merged in order.
+
+>[!NOTE]
+> If a pull request cannot be split (e.g. a large test migration), add the [`tag: override pr size check` label](#labels) to bypass the check.
+
 ### Title format
 
 Pull request titles should briefly describe the proposed changes in a way that makes sense for the users.
@@ -140,6 +151,7 @@ Labels are used to not only categorize but also alter the continuous integration
   * Refactoring pull requests
   * CI and build tools improvements
   * Minor changes like typo
+* `tag: override pr size check` to bypass [the pull request size check](#pull-request-size) when a pull request cannot be split
 * [The `run-tests:` category](https://github.com/DataDog/dd-trace-java/labels?q=run-tests%3A) to run continuous integration tests on a specific JVM in case of JVM specific changes
 * `run-tests: flaky` to run the flaky tests on continuous integration as they are disabled by default