Add link statistics report to project

7 files changed, 110 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
index df16e7c9..e0403e79 100644
--- a/.gitignore
+++ b/.gitignore
@@ -32,3 +32,4 @@ tmp/
 vendor/
 node_modules/
 coverage/
+link_stats.txt
diff --git a/.gitlab/ci/test.gitlab-ci.yml b/.gitlab/ci/test.gitlab-ci.yml
index 6dc4a6f0..82e1f80f 100644
--- a/.gitlab/ci/test.gitlab-ci.yml
+++ b/.gitlab/ci/test.gitlab-ci.yml
@@ -186,6 +186,22 @@ test_external_links:
     - bundle exec nanoc check external_links
 
 #
+# Run link statistics report
+#
+link_statistics_report:
+  extends:
+    - .bundle
+    - .rules_chores
+  stage: test
+  script:
+    - bundle exec nanoc check link_stats > link_stats.txt
+  artifacts:
+    name: "link statistics report"
+    paths:
+      - link_stats.txt
+    expire_in: "1 day"
+
+#
 # Run markdownlint to find EOL whitespace to clean up
 #
 test_EOL_whitespace:
diff --git a/.rubocop.yml b/.rubocop.yml
index 371edfd4..809c0fca 100644
--- a/.rubocop.yml
+++ b/.rubocop.yml
@@ -18,6 +18,8 @@ Naming/FileName:
 # Project is not a Rails project
 Rails/NegateInclude:
   Enabled: false
+Rails/Output:
+  Enabled: false
 Rails/RakeEnvironment:
   Enabled: false
 
diff --git a/Makefile b/Makefile
index e445dc17..127c7f73 100644
--- a/Makefile
+++ b/Makefile
@@ -109,6 +109,10 @@ clean:
 	@printf "\n$(INFO)INFO: Removing tmp and public directories...$(END)\n"
 	@rm -rf tmp public
 
+clean-link-stats:
+	@printf "\n$(INFO)INFO: Removing link_stats.txt...$(END)\n"
+	@rm -rf link_stats.txt
+
 build-lunr-index:
 	@printf "\n$(INFO)INFO: Building offline search index..$(END)\n"
 	node scripts/lunr/preindex.js
@@ -133,6 +137,10 @@ external-links-check: compile
 	@printf "\n$(INFO)INFO: Checking all external links...$(END)\n"
 	@bundle exec nanoc check external_links
 
+link-stats: clone-all-docs-projects compile clean-link-stats
+	@printf "\n$(INFO)INFO: Generating link statistics report to link_stats.txt...$(END)\n"
+	@bundle exec nanoc check link_stats > link_stats.txt
+
 brew-bundle:
 	@printf "\n$(INFO)INFO: Checking Brew dependencies, if Brew is available...$(END)\n"
 	@(command -v brew > /dev/null 2>&1) && brew bundle --no-lock || true
diff --git a/doc/testing.md b/doc/testing.md
index f6d2921d..1ad63ce1 100644
--- a/doc/testing.md
+++ b/doc/testing.md
@@ -79,3 +79,17 @@ To test external links, run:
 ```shell
 make external-links-check
 ```
+
+To generate the link statistics report:
+
+- To a file, run:
+
+  ```shell
+  make link-stats
+  ```
+
+- To your terminal, run:
+
+  ```shell
+  make compile && bundle exec nanoc check link_stats
+  ```
diff --git a/lib/checks/link_stats.rb b/lib/checks/link_stats.rb
new file mode 100644
index 00000000..605d042f
--- /dev/null
+++ b/lib/checks/link_stats.rb
@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+COLOR_CODE_RESET = "\e[0m"
+COLOR_CODE_RED = "\e[31m"
+COLOR_CODE_GREEN = "\e[32m"
+
+Nanoc::Check.define(:link_stats) do
+  total_anchor_link_count = 0
+  total_external_link_count = 0
+  total_link_count = 0
+
+  puts "\n#{COLOR_CODE_GREEN}INFO: Generating links statistics report...#{COLOR_CODE_RESET}"
+  output_html_filenames.each do |file|
+    page_anchor_link_count = 0
+    page_external_link_count = 0
+    page_link_count = 0
+    puts "\n#{COLOR_CODE_GREEN}INFO:#{COLOR_CODE_RESET} Processing file: #{file}:"
+    Gitlab::Docs::Page.new(file).links.each do |link|
+      next if link.excluded_link?
+
+      page_anchor_link_count += 1 if link.internal_anchor?
+      page_external_link_count += 1 if link.external?
+      page_link_count += 1
+      puts "#{COLOR_CODE_GREEN}INFO:#{COLOR_CODE_RESET}   Found link (#{page_link_count}): #{link.href}"
+    end
+    puts "#{COLOR_CODE_GREEN}INFO:#{COLOR_CODE_RESET} Total number of page links: #{page_link_count} (including: #{page_anchor_link_count} anchors, #{page_external_link_count} external)!"
+    total_anchor_link_count += page_anchor_link_count
+    total_external_link_count += page_external_link_count
+    total_link_count += page_link_count
+  end
+  puts "\n#{COLOR_CODE_GREEN}INFO:#{COLOR_CODE_RESET} Total number of links: #{total_link_count} (including: #{total_anchor_link_count} anchors, #{total_external_link_count} external)!"
+end
diff --git a/lib/gitlab/docs/link.rb b/lib/gitlab/docs/link.rb
index d0e7b9d8..d6c983d2 100644
--- a/lib/gitlab/docs/link.rb
+++ b/lib/gitlab/docs/link.rb
@@ -32,6 +32,43 @@ module Gitlab
         @href.length.positive? && !@href.include?(':')
       end
 
+      def external?
+        @href.length.positive? && @href.include?(':')
+      end
+
+      def excluded_link?
+        @href.length.positive? &&
+          (@href.start_with?(
+            '/',
+            'https://about.gitlab.com/releases/categories/releases',
+            'https://gitlab.com/-/trial_registrations',
+            'https://customers.staging.gitlab.com/openapi_docs',
+            'https://about.gitlab.com/handbook/product/product-intelligence-guide',
+            'https://metrics.gitlab.com',
+            'https://gitlab.com/gitlab-org/gitlab-development-kit',
+            'https://design.gitlab.com/get-started/contributing',
+            'https://gitlab.com/gitlab-org/gitlab/-/issues/new',
+            '#help-feedback-content',
+            'https://about.gitlab.com/community/gitlab-first-look',
+            'https://about.gitlab.com/pricing',
+            'https://about.gitlab.com/free-trial',
+            'https://forum.gitlab.com/new-topic',
+            'https://about.gitlab.com/support',
+            'https://gitlab.com/dashboard',
+            'https://twitter.com/gitlab',
+            'https://www.facebook.com/gitlab',
+            'https://www.youtube.com/channel',
+            'https://www.linkedin.com/company/gitlab-com',
+            'https://gitlab.com/gitlab-org/gitlab-docs',
+            'https://about.gitlab.com/company',
+            'https://about.gitlab.com/terms',
+            'https://about.gitlab.com/privacy',
+            'https://about.gitlab.com/company/contact',
+            'https://gitlab.com/-/ide/project/gitlab-org/gitlab/edit',
+            'https://creativecommons.org/licenses/by-sa/4.0'
+          ) || @href == '#')
+      end
+
       def path
         @href.partition('#').first
       end