Add govuk repo tags check

.github/workflows/verify_repo_tags.yml

@@ -0,0 +1,61 @@
+name: "Verify Repo Tags"
+
+on:
+  workflow_dispatch: {}
+  schedule:
+    - cron:  '00 10 * * 1-5' # Runs at 10:00 UTC, Monday through Friday.
+
+jobs:
+  verify-repo-tags:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup ruby
+        uses: ruby/setup-ruby@v1
+        with:
+          bundler-cache: true
+
+      - name: Verify Repo Tags
+        id: verify_repo_tags
+        env:
+          GITHUB_TOKEN: ${{ secrets.GH_TOKEN }}
+        run: |
+          EXIT_CODE=0
+          output=$(bundle exec rake verify_repo_tags) || EXIT_CODE=$?
+
+          echo "$output"
+
+          exit $EXIT_CODE
+
+      - name: Notify failure
+        uses: slackapi/slack-github-action@v1
+        if: ${{ failure() }}
+        with:
+          payload: |
+            {
+              "channel": "#govuk-platform-support",
+              "username": "Platform Alerts",
+              "text": "The <https://github.com/alphagov/govuk-developer-docs/blob/main/data/repos.yml|Developer Docs repo list> is out of sync with the repos tagged as 'govuk' in GitHub.",
+              "blocks": [
+                {
+                  "type": "section",
+                  "text": {
+                    "type": "mrkdwn",
+                    "text": "The <https://github.com/alphagov/govuk-developer-docs/blob/main/data/repos.yml|Developer Docs repo list> is out of sync with the repos tagged as 'govuk' in GitHub."
+                  },
+                  "accessory": {
+                    "type": "button",
+                    "text": {
+                        "type": "plain_text",
+                        "text": "Check the build logs for details"
+                    },
+                    "url": "${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}",
+                    "action_id": "button-view-workflow"
+                  }
+                }
+              ]
+            }
+        env:
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK }}

Rakefile

@@ -1,3 +1,5 @@
+require_relative "lib/validate_repos"
+
 begin
   require "rspec/core/rake_task"
 
@@ -25,6 +27,24 @@ rescue LoadError
   # no rubocop available
 end
 
+desc "Verify that GOVUK repos are tagged #govuk"
+task :verify_repo_tags do
+  validator = ValidateRepos.new
+
+  untagged_message = <<~UNTAGGED
+    The following repos in the repos.yml file in govuk-developer-docs do not have the govuk tag on GitHub:
+  UNTAGGED
+
+  falsely_tagged_message = <<~FALSETAG
+    The following repos have the govuk tag on GitHub but are not in the repos.yml file in govuk-developer-docs:
+  FALSETAG
+
+  puts "#{untagged_message}\n#{validator.untagged_repos}" unless validator.untagged_repos.empty?
+  puts "#{falsely_tagged_message}\n#{validator.falsely_tagged_repos}" unless validator.falsely_tagged_repos.empty?
+
+  exit 1 unless validator.untagged_repos.empty? && validator.falsely_tagged_repos.empty?
+end
+
 task default: %i[
   jsonlint
   rubocop

lib/validate_repos.rb

@@ -0,0 +1,46 @@
+require "json"
+require "octokit"
+require "open-uri"
+require "yaml"
+
+class ValidateRepos
+  def initialize
+    Octokit.auto_paginate = true
+    @client = Octokit::Client.new(access_token: ENV.fetch("GITHUB_TOKEN"))
+  end
+
+  def github_repos_tagged_govuk
+    @github_repos_tagged_govuk ||= repos.map { |repo| repo["name"] }
+  end
+
+  def govuk_repo_names
+    @govuk_repo_names ||= JSON.parse(Net::HTTP.get_response(URI.parse("https://docs.publishing.service.gov.uk/repos.json")).body)
+      .map { |repo| repo["app_name"] }
+      .reject { |app| ignored_apps.include?(app) }
+  end
+
+  def untagged_repos
+    (govuk_repo_names - github_repos_tagged_govuk).join("\n")
+  end
+
+  def falsely_tagged_repos
+    (github_repos_tagged_govuk - govuk_repo_names).join("\n")
+  end
+
+  def repos
+    @client
+      .org_repos("alphagov", accept: "application/vnd.github.mercy-preview+json")
+      .select { |repo| repo.topics.to_a.include?("govuk") }
+      .reject(&:archived)
+      .reject { |repo| ignored_repos.include?(repo.full_name) }
+      .sort_by { |repo| repo[:full_name] }
+  end
+
+  def ignored_repos
+    ["alphagov/licensify"] # Licensify consists of 3 apps in 1 repo
+  end
+
+  def ignored_apps
+    %w[licensify-backend]
+  end
+end

spec/validate_repos_spec.rb

@@ -0,0 +1,71 @@
+require "spec_helper"
+require_relative "../lib/validate_repos"
+
+RSpec.describe ValidateRepos do
+  before do
+    @repo_mock = double("Repo", topics: %w[govuk], archived: false, full_name: "this-is-a-govuk-repo")
+    allow(@repo_mock).to receive(:[]).with("name").and_return("this-is-a-govuk-repo")
+    allow(@repo_mock).to receive(:[]).with(:full_name).and_return("this-is-a-govuk-repo")
+    allow(ENV).to receive(:fetch).with("GITHUB_TOKEN").and_return("dummy_token")
+    allow_any_instance_of(Octokit::Client).to receive(:org_repos).and_return([@repo_mock])
+  end
+
+  it "should ignore any repos that exist in repos.json AND are tagged govuk in GitHub" do
+    repos = [{
+      "app_name" => "this-is-a-govuk-repo",
+    }]
+
+    stub_repos_json(repos)
+
+    validator = ValidateRepos.new
+
+    expect(validator.untagged_repos).to eq("")
+    expect(validator.falsely_tagged_repos).to eq("")
+  end
+
+  it "doesn't say that a repo is missing the govuk tag if it has been added to the ignore list" do
+    app_name = "this-is-a-govuk-repo"
+    allow_any_instance_of(ValidateRepos).to receive(:ignored_repos).and_return(["alphagov/#{app_name}"])
+
+    repos = [{
+      "app_name" => app_name,
+    }]
+
+    stub_repos_json(repos)
+
+    validator = ValidateRepos.new
+
+    expect(validator.untagged_repos).to eq("")
+    expect(validator.falsely_tagged_repos).to eq("")
+  end
+
+  it "should alert if it finds an untagged repo in repos.json" do
+    repos = [
+      { "app_name" => "this-is-a-govuk-repo" },
+      { "app_name" => "this-govuk-repo-is-not-tagged!" },
+    ]
+
+    stub_repos_json(repos)
+
+    validator = ValidateRepos.new
+
+    expect(validator.untagged_repos).to eq("this-govuk-repo-is-not-tagged!")
+    expect(validator.falsely_tagged_repos).to eq("")
+  end
+
+  it "should alert if it finds a repo that has falsely been tagged as govuk." do
+    repos = []
+
+    stub_repos_json(repos)
+
+    validator = ValidateRepos.new
+
+    expect(validator.falsely_tagged_repos).to eq("this-is-a-govuk-repo")
+    expect(validator.untagged_repos).to eq("")
+  end
+
+  def stub_repos_json(repos)
+    stub_request(:get, "https://docs.publishing.service.gov.uk/repos.json")
+      .to_return(status: 200, body: repos.to_json, headers: {})
+  end
+end