Autolink URLs and emails in blobs and diffs

lib/banzai/filter/autolink_filter.rb

@@ -54,15 +54,13 @@ module Banzai
       #
       # `@doc` will be re-parsed with the HTML String from Rinku.
       def rinku_parse
-        # Convert the options from a Hash to a String that Rinku expects
-        options = tag_options(link_options)
-
         # NOTE: We don't parse email links because it will erroneously match
         # external Commit and CommitRange references.
         #
         # The final argument tells Rinku to link short URLs that don't include a
         # period (e.g., http://localhost:3000/)
-        rinku = Rinku.auto_link(html, :urls, options, IGNORE_PARENTS.to_a, 1)
+        mode = context[:autolink_emails] ? :all : :urls
+        rinku = Rinku.auto_link(html, mode, tag_options(link_options), IGNORE_PARENTS.to_a, 1)
         return if rinku == html
@@ -111,9 +109,9 @@ module Banzai
         # order to be output literally rather than escaped.
         match.gsub!(/((?:&[\w#]+;)+)\z/, '')
         dropped = ($1 || '').html_safe
-        options = link_options.merge(href: match)
-        content_tag(:a, match, options) + dropped
+        match = ERB::Util.html_escape_once(match)
+        %{<a href="#{match}" #{tag_options(link_options)}>#{match}</a>#{dropped}}.html_safe
       end
       def autolink_filter(text)

lib/banzai/pipeline/autolink_pipeline.rb

+module Banzai
+  module Pipeline
+    class AutolinkPipeline < BasePipeline
+      def self.filters
+        @filters ||= FilterArray[
+          Filter::AutolinkFilter,
+          Filter::ExternalLinkFilter
+        ]
+      end
+    end
+  end
+end

lib/gitlab/highlight.rb

@@ -25,7 +25,7 @@ module Gitlab
     def highlight(text, continue: true, plain: false)
       highlighted_text = highlight_text(text, continue: continue, plain: plain)
       highlighted_text = link_dependencies(text, highlighted_text) if blob_name
-      highlighted_text
+      autolink_strings(highlighted_text)
     end
     def lexer
@@ -67,5 +67,23 @@ module Gitlab
     def link_dependencies(text, highlighted_text)
       Gitlab::DependencyLinker.link(blob_name, text, highlighted_text)
     end
+
+    def autolink_strings(highlighted_text)
+      doc = Nokogiri::HTML::DocumentFragment.parse(highlighted_text)
+
+      # Files without highlighting have all text in `span.line`.
+      # Files with highlighting have strings and comments in `span`s with a
+      # `class` starting with `c` or `s`.
+      doc.xpath('.//span[@class="line" or starts-with(@class, "c") or starts-with(@class, "s")]/text()').each do |node|
+        content = node.to_html
+        html = Banzai.render(content, pipeline: :autolink, autolink_emails: true)
+
+        next if html == content
+
+        node.replace(html)
+      end
+
+      doc.to_html.html_safe
+    end
   end
 end

spec/lib/gitlab/highlight_spec.rb

@@ -9,7 +9,7 @@ describe Gitlab::Highlight, lib: true do
   describe '.highlight_lines' do
     let(:lines) do
-      Gitlab::Highlight.highlight_lines(project.repository, commit.id, 'files/ruby/popen.rb')
+      described_class.highlight_lines(project.repository, commit.id, 'files/ruby/popen.rb')
     end
     it 'highlights all the lines properly' do
@@ -59,7 +59,7 @@ describe Gitlab::Highlight, lib: true do
   end
   describe '#highlight' do
-    subject { described_class.highlight(file_name, file_content, nowrap: false) }
+    subject { described_class.highlight(file_name, file_content) }
     it 'links dependencies via DependencyLinker' do
       expect(Gitlab::DependencyLinker).to receive(:link).
@@ -67,5 +67,66 @@ describe Gitlab::Highlight, lib: true do
       described_class.highlight('file.name', 'Contents')
     end
+
+    context "plain text file" do
+      let(:file_name) { "example.txt" }
+      let(:file_content) do
+        <<-CONTENT.strip_heredoc
+          URL: http://www.google.com
+          Email: hello@example.com
+        CONTENT
+      end
+
+      it "links URLs" do
+        expect(subject).to include(%{<a href="http://www.google.com" rel="nofollow noreferrer" target="_blank">http://www.google.com</a>})
+      end
+
+      it "links emails" do
+        expect(subject).to include(%{<a href="mailto:hello@example.com">hello@example.com</a>})
+      end
+    end
+
+    context "file with highlighting" do
+      let(:file_name) { "example.rb" }
+      let(:file_content) do
+        <<-CONTENT.strip_heredoc
+          # URL in comment: http://www.google.com
+          # Email in comment: hello@example.com
+
+          "URL in string: http://www.google.com"
+          "Email in string: hello@example.com"
+
+          # <http://www.google.com>
+          # <url>http://www.google.com</url>
+        CONTENT
+      end
+
+      context "in a comment" do
+        it "links URLs" do
+          expect(subject).to include(%{URL in comment: <a href="http://www.google.com" rel="nofollow noreferrer" target="_blank">http://www.google.com</a>})
+        end
+
+        it "links emails" do
+          expect(subject).to include(%{Email in comment: <a href="mailto:hello@example.com">hello@example.com</a>})
+        end
+      end
+
+      context "in a string" do
+        it "links URLs" do
+          expect(subject).to include(%{URL in string: <a href="http://www.google.com" rel="nofollow noreferrer" target="_blank">http://www.google.com</a>})
+        end
+
+        it "links emails" do
+          expect(subject).to include(%{Email in string: <a href="mailto:hello@example.com">hello@example.com</a>})
+        end
+      end
+
+      context 'in HTML/XML tags' do
+        it "links URLs" do
+          expect(subject).to include(%{&lt;<a href="http://www.google.com" rel="nofollow noreferrer" target="_blank">http://www.google.com</a>&gt;})
+          expect(subject).to include(%{&lt;url&gt;<a href="http://www.google.com" rel="nofollow noreferrer" target="_blank">http://www.google.com</a>&lt;/url&gt;})
+        end
+      end
+    end
   end
 end