Extract anchors validation classes to separate files

5 files changed, 178 insertions, 0 deletions

diff --git a/lib/gitlab/docs/document.rb b/lib/gitlab/docs/document.rb
new file mode 100644
index 00000000..d4e486b2
--- /dev/null
+++ b/lib/gitlab/docs/document.rb
@@ -0,0 +1,16 @@
+module Gitlab
+  module Docs
+    class Document < Nokogiri::XML::SAX::Document
+      def initialize(page)
+        @page = page
+      end
+
+      def start_element(name, attributes = [])
+        Gitlab::Docs::Element.new(name, attributes).tap do |element|
+          @page.hrefs << element.href if element.link?
+          @page.ids << element.id if element.has_id?
+        end
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/docs/element.rb b/lib/gitlab/docs/element.rb
new file mode 100644
index 00000000..a78b9a01
--- /dev/null
+++ b/lib/gitlab/docs/element.rb
@@ -0,0 +1,32 @@
+module Gitlab
+  module Docs
+    class Element
+      def initialize(name, attributes)
+        @name = name
+        @attributes = attributes
+      end
+
+      def link?
+        @name == 'a' && !href.to_s.empty?
+      end
+
+      def has_id?
+        !id.to_s.empty?
+      end
+
+      def href
+        @href ||= attribute('href')
+      end
+
+      def id
+        @id ||= attribute('id')
+      end
+
+      private
+
+      def attribute(name)
+        @attributes.find { |attr| attr.first == name }&.last
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/docs/link.rb b/lib/gitlab/docs/link.rb
new file mode 100644
index 00000000..8509f0ca
--- /dev/null
+++ b/lib/gitlab/docs/link.rb
@@ -0,0 +1,71 @@
+module Gitlab
+  module Docs
+    class Link
+      attr_reader :link, :href, :page
+
+      def initialize(link, page)
+        @href = link
+        @page = page
+      end
+
+      def to_anchor?
+        @href.to_s.include?('#')
+      end
+
+      def anchor_name
+        raise ArgumentError unless to_anchor?
+
+        @href.to_s.partition('#').last.downcase
+      end
+
+      def internal_anchor?
+        raise ArgumentError unless to_anchor?
+
+        @href.to_s.partition('#').first.empty?
+      end
+
+      def internal?
+        @href.to_s.length > 0 && !@href.include?(':')
+      end
+
+      def path
+        @href.to_s.partition('#').first
+      end
+
+      def absolute_path
+        raise unless internal?
+
+        if @href.start_with?('/')
+          Gitlab::Docs::Nanoc.output_dir + path
+        else
+          ::File.expand_path(path, @page.directory)
+        end
+      end
+
+      def destination_page
+        if internal_anchor?
+          @page
+        else
+          Gitlab::Docs::Page.build(absolute_path)
+        end
+      end
+
+      def source_file
+        @page.file
+      end
+
+      def destination_file
+        destination_page.file
+      end
+
+      def destination_page_not_found?
+        !destination_page.exists?
+      end
+
+      def destination_anchor_not_found?
+        !destination_page.has_anchor?(anchor_name)
+      end
+    end
+  end
+end
+
diff --git a/lib/gitlab/docs/nanoc.rb b/lib/gitlab/docs/nanoc.rb
new file mode 100644
index 00000000..70b4c397
--- /dev/null
+++ b/lib/gitlab/docs/nanoc.rb
@@ -0,0 +1,13 @@
+module Gitlab
+  module Docs
+    module Nanoc
+      def self.config
+        @config ||= YAML.load(File.read('nanoc.yaml'))
+      end
+
+      def self.output_dir
+        config.fetch('output_dir')
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/docs/page.rb b/lib/gitlab/docs/page.rb
new file mode 100644
index 00000000..8bd5e9b9
--- /dev/null
+++ b/lib/gitlab/docs/page.rb
@@ -0,0 +1,46 @@
+module Gitlab
+  module Docs
+    class Page
+      attr_reader :file
+      attr_accessor :hrefs, :ids
+
+      def initialize(file)
+        @file = file
+        @hrefs = []
+        @ids = []
+
+        return unless exists?
+
+        Nokogiri::HTML::SAX::Parser
+          .new(Gitlab::Docs::Document.new(self))
+          .parse(File.read(file))
+      end
+
+      def exists?
+        File.exists?(@file)
+      end
+
+      def directory
+        File.dirname(@file)
+      end
+
+      def links
+        @links ||= @hrefs.map do |link|
+          Gitlab::Docs::Link.new(link, self)
+        end
+      end
+
+      def has_anchor?(name)
+        @ids.include?(name)
+      end
+
+      def self.build(path)
+        if path.end_with?('.html')
+          new(path)
+        else
+          new(File.join(path, 'index.html'))
+        end
+      end
+    end
+  end
+end