Automatically review default-gem pull requests (#15116)

3 files changed, 140 insertions, 0 deletions

diff --git a/.github/workflows/auto_review_pr.yml b/.github/workflows/auto_review_pr.yml
new file mode 100644
index 0000000000..c8095dfd8e
--- /dev/null
+++ b/.github/workflows/auto_review_pr.yml
@@ -0,0 +1,33 @@
+name: Auto Review PR
+on:
+  pull_request_target:
+    types: [opened, ready_for_review, reopened]
+    branches: [master]
+
+permissions:
+  contents: read
+
+jobs:
+  auto-review-pr:
+    name: Auto Review PR
+    runs-on: ubuntu-latest
+    if: ${{ github.repository == 'ruby/ruby' && github.base_ref == 'master' }}
+
+    permissions:
+      pull-requests: write
+      contents: read
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - uses: ruby/setup-ruby@ab177d40ee5483edb974554986f56b33477e21d0 # v1.265.0
+        with:
+          ruby-version: '3.4'
+          bundler: none
+
+      - name: Auto Review PR
+        run: ruby tool/auto_review_pr.rb "$GITHUB_PR_NUMBER"
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          GITHUB_PR_NUMBER: ${{ github.event.pull_request.number }}
diff --git a/tool/auto_review_pr.rb b/tool/auto_review_pr.rb
new file mode 100755
index 0000000000..a3494e4c1a
--- /dev/null
+++ b/tool/auto_review_pr.rb
@@ -0,0 +1,106 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require 'json'
+require 'net/http'
+require 'uri'
+require_relative './sync_default_gems'
+
+class GitHubAPIClient
+  def initialize(token)
+    @token = token
+  end
+
+  def get(path)
+    response = Net::HTTP.get_response(URI("https://api.github.com#{path}"), {
+      'Authorization' => "token #{@token}",
+      'Accept' => 'application/vnd.github.v3+json',
+    }).tap(&:value)
+    JSON.parse(response.body, symbolize_names: true)
+  end
+
+  def post(path, body = {})
+    body = JSON.dump(body)
+    response = Net::HTTP.post(URI("https://api.github.com#{path}"), body, {
+      'Authorization' => "token #{@token}",
+      'Accept' => 'application/vnd.github.v3+json',
+      'Content-Type' => 'application/json',
+    }).tap(&:value)
+    JSON.parse(response.body, symbolize_names: true)
+  end
+end
+
+class AutoReviewPR
+  REPO = 'ruby/ruby'
+
+  COMMENT_USER = 'github-actions[bot]'
+  COMMENT_PREFIX = 'The following files are maintained in the following upstream repositories:'
+  COMMENT_SUFFIX = 'Please file a pull request to the above instead. Thank you.'
+
+  def initialize(client)
+    @client = client
+  end
+
+  def review(pr_number)
+    comment_body = "Please file a pull request to ruby/foo instead."
+
+    # Fetch the list of files changed by the PR
+    changed_files = @client.get("/repos/#{REPO}/pulls/#{pr_number}/files").map { it.fetch(:filename) }
+
+    # Build a Hash: { upstream_repo => files, ... }
+    upstream_repos = changed_files.group_by { |file| find_upstream_repo(file) }
+    upstream_repos.delete(nil) # exclude no-upstream files
+    upstream_repos.delete('prism') if changed_files.include?('prism_compile.c') # allow prism changes in this case
+    if upstream_repos.empty?
+      puts "Skipped: The PR ##{pr_number} doesn't have upstream repositories."
+      return
+    end
+
+    # Check if the PR is already reviewed
+    existing_comments = @client.get("/repos/#{REPO}/issues/#{pr_number}/comments")
+    existing_comments.map! { [it.fetch(:user).fetch(:login), it.fetch(:body)] }
+    if existing_comments.any? { |user, comment| user == COMMENT_USER && comment.start_with?(COMMENT_PREFIX) }
+      puts "Skipped: The PR ##{pr_number} already has an automated review comment."
+      return
+    end
+
+    # Post a comment
+    comment = format_comment(upstream_repos)
+    result = @client.post("/repos/#{REPO}/issues/#{pr_number}/comments", { body: comment })
+    puts "Success: #{JSON.pretty_generate(result)}"
+  end
+
+  private
+
+  def find_upstream_repo(file)
+    SyncDefaultGems::REPOSITORIES.each do |repo_name, repository|
+      repository.mappings.each do |_src, dst|
+        if file.start_with?(dst)
+          return repo_name
+        end
+      end
+    end
+    nil
+  end
+
+  # upstream_repos: { upstream_repo => files, ... }
+  def format_comment(upstream_repos)
+    comment = +''
+    comment << "#{COMMENT_PREFIX}\n\n"
+
+    upstream_repos.each do |upstream_repo, files|
+      comment << "* https://github.com/ruby/#{upstream_repo}\n"
+      files.each do |file|
+        comment << "    * #{file}\n"
+      end
+    end
+
+    comment << "\n#{COMMENT_SUFFIX}"
+    comment
+  end
+end
+
+pr_number = ARGV[0] || abort("Usage: #{$0} <pr_number>")
+client = GitHubAPIClient.new(ENV.fetch('GITHUB_TOKEN'))
+
+AutoReviewPR.new(client).review(pr_number)
diff --git a/tool/sync_default_gems.rb b/tool/sync_default_gems.rb
index b2509f0062..997cd5eddc 100755
--- a/tool/sync_default_gems.rb
+++ b/tool/sync_default_gems.rb
@@ -61,6 +61,7 @@ module SyncDefaultGems
     ])
   end
 
+  # Note: tool/auto_review_pr.rb also depends on this constant.
   REPOSITORIES = {
     "io-console": repo("ruby/io-console", [
       ["ext/io/console", "ext/io/console"],