From aa2cc670fe2c9de772c82d90df4ee2d8a77c23fc Mon Sep 17 00:00:00 2001
From: Robert Speicher <rspeicher@gmail.com>
Date: Mon, 27 Apr 2015 18:54:13 -0400
Subject: [PATCH] Add Gitlab::Markdown::AutolinkFilter

---
 app/helpers/gitlab_markdown_helper.rb         |  4 +-
 lib/gitlab/markdown.rb                        |  4 +-
 lib/gitlab/markdown/autolink_filter.rb        | 95 ++++++++++++++++++
 lib/redcarpet/render/gitlab_html.rb           | 18 +---
 .../gitlab/markdown/autolink_filter_spec.rb   | 98 +++++++++++++++++++
 5 files changed, 202 insertions(+), 17 deletions(-)
 create mode 100644 lib/gitlab/markdown/autolink_filter.rb
 create mode 100644 spec/lib/gitlab/markdown/autolink_filter_spec.rb

diff --git a/app/helpers/gitlab_markdown_helper.rb b/app/helpers/gitlab_markdown_helper.rb
index 48d754dcf7..7dbffaae5f 100644
--- a/app/helpers/gitlab_markdown_helper.rb
+++ b/app/helpers/gitlab_markdown_helper.rb
@@ -34,9 +34,8 @@ module GitlabMarkdownHelper
 
       # see https://github.com/vmg/redcarpet#darling-i-packed-you-a-couple-renderers-for-lunch
       rend = Redcarpet::Render::GitlabHTML.new(self, user_color_scheme_class, {
-        safe_links_only: true,
         # Handled further down the line by HTML::Pipeline::SanitizationFilter
-        escape_html:     false
+        escape_html: false
       }.merge(options))
 
       # see https://github.com/vmg/redcarpet#and-its-like-really-simple-to-use
@@ -44,7 +43,6 @@ module GitlabMarkdownHelper
         no_intra_emphasis:   true,
         tables:              true,
         fenced_code_blocks:  true,
-        autolink:            true,
         strikethrough:       true,
         lax_spacing:         true,
         space_after_headers: true,
diff --git a/lib/gitlab/markdown.rb b/lib/gitlab/markdown.rb
index ffb7e16aed..beb97bbdf4 100644
--- a/lib/gitlab/markdown.rb
+++ b/lib/gitlab/markdown.rb
@@ -30,6 +30,7 @@ module Gitlab
   #   => "<img alt=\":trollface:\" class=\"emoji\" src=\"/images/trollface.png" title=\":trollface:\" />
   module Markdown
     # Provide autoload paths for filters to prevent a circular dependency error
+    autoload :AutolinkFilter,               'gitlab/markdown/autolink_filter'
     autoload :CommitRangeReferenceFilter,   'gitlab/markdown/commit_range_reference_filter'
     autoload :CommitReferenceFilter,        'gitlab/markdown/commit_reference_filter'
     autoload :EmojiFilter,                  'gitlab/markdown/emoji_filter'
@@ -122,6 +123,7 @@ module Gitlab
 
         Gitlab::Markdown::EmojiFilter,
         Gitlab::Markdown::TableOfContentsFilter,
+        Gitlab::Markdown::AutolinkFilter,
 
         Gitlab::Markdown::UserReferenceFilter,
         Gitlab::Markdown::IssueReferenceFilter,
@@ -130,7 +132,7 @@ module Gitlab
         Gitlab::Markdown::SnippetReferenceFilter,
         Gitlab::Markdown::CommitRangeReferenceFilter,
         Gitlab::Markdown::CommitReferenceFilter,
-        Gitlab::Markdown::LabelReferenceFilter,
+        Gitlab::Markdown::LabelReferenceFilter
       ]
     end
 
diff --git a/lib/gitlab/markdown/autolink_filter.rb b/lib/gitlab/markdown/autolink_filter.rb
new file mode 100644
index 0000000000..97cfc1893a
--- /dev/null
+++ b/lib/gitlab/markdown/autolink_filter.rb
@@ -0,0 +1,95 @@
+require 'html/pipeline/filter'
+require 'uri'
+
+module Gitlab
+  module Markdown
+    # HTML Filter for auto-linking URLs in HTML.
+    #
+    # Based on HTML::Pipeline::AutolinkFilter
+    #
+    # Context options:
+    #   :autolink  - Boolean, skips all processing done by this filter when false
+    #   :link_attr - Hash of attributes for the generated links
+    #
+    class AutolinkFilter < HTML::Pipeline::Filter
+      include ActionView::Helpers::TagHelper
+
+      # Pattern to match text that should be autolinked.
+      #
+      # A URI scheme begins with a letter and may contain letters, numbers,
+      # plus, period and hyphen. Schemes are case-insensitive but we're being
+      # picky here and allowing only lowercase for autolinks.
+      #
+      # See http://en.wikipedia.org/wiki/URI_scheme
+      #
+      # The negative lookbehind ensures that users can paste a URL followed by a
+      # period or comma for punctuation without those characters being included
+      # in the generated link.
+      SCHEME_PATTERN = %r{([a-z][a-z0-9_\+\.\-]+:\/\/[^\s]+)(?<!,|\.)}
+
+      # Text matching SCHEME_PATTERN inside these elements will not be linked
+      IGNORE_PARENTS = %w(a code kbd pre script style).to_set
+
+      def call
+        return doc if context[:autolink] == false
+
+        rinku_parse
+        text_parse
+      end
+
+      private
+
+      # Run the text through Rinku as a first pass
+      #
+      # This will quickly autolink http(s) and ftp links.
+      #
+      # `@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.
+        rinku = Rinku.auto_link(html, :urls, options, IGNORE_PARENTS.to_a)
+
+        # Rinku returns a String, so parse it back to a Nokogiri::XML::Document
+        # for further processing.
+        @doc = parse_html(rinku)
+      end
+
+      # Autolinks any text matching SCHEME_PATTERN that Rinku didn't already
+      # replace
+      def text_parse
+        search_text_nodes(doc).each do |node|
+          content = node.to_html
+
+          next if has_ancestor?(node, IGNORE_PARENTS)
+          next unless content.match(SCHEME_PATTERN)
+
+          # If Rinku didn't link this, there's probably a good reason, so we'll
+          # skip it too
+          next if content.start_with?(*%w(http https ftp))
+
+          html = autolink_filter(content)
+
+          next if html == content
+
+          node.replace(html)
+        end
+
+        doc
+      end
+
+      def autolink_filter(text)
+        text.gsub(SCHEME_PATTERN) do |match|
+          options = link_options.merge(href: match)
+          content_tag(:a, match, options)
+        end
+      end
+
+      def link_options
+        @link_options ||= context[:link_attr] || {}
+      end
+    end
+  end
+end
diff --git a/lib/redcarpet/render/gitlab_html.rb b/lib/redcarpet/render/gitlab_html.rb
index dc5fbe3c8e..321be9202c 100644
--- a/lib/redcarpet/render/gitlab_html.rb
+++ b/lib/redcarpet/render/gitlab_html.rb
@@ -8,13 +8,8 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
     @color_scheme = color_scheme
     @project = @template.instance_variable_get("@project")
     @options = options.dup
-    super options
-  end
 
-  def preprocess(full_document)
-    # Redcarpet doesn't allow SMB links when `safe_links_only` is enabled.
-    # FTP links are allowed, so we trick Redcarpet.
-    full_document.gsub("smb://", "ftp://smb:")
+    super(options)
   end
 
   # If project has issue number 39, apostrophe will be linked in
@@ -25,6 +20,7 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
   # This only influences regular text, code blocks are untouched.
   def normal_text(text)
     return text unless text.present?
+
     text.gsub("'", "&rsquo;")
   end
 
@@ -37,7 +33,7 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
     # so we assume you're not using leading spaces that aren't tabs,
     # and just replace them here.
     if lexer.tag == 'make'
-      code.gsub! /^    /, "\t"
+      code.gsub!(/^    /, "\t")
     end
 
     formatter = Rugments::Formatters::HTML.new(
@@ -46,17 +42,13 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
     formatter.format(lexer.lex(code))
   end
 
-  def link(link, title, content)
-    h.link_to_gfm(content, link, title: title)
-  end
-
   def postprocess(full_document)
-    full_document.gsub!("ftp://smb:", "smb://")
-
     full_document.gsub!("&rsquo;", "'")
+
     unless @template.instance_variable_get("@project_wiki") || @project.nil?
       full_document = h.create_relative_links(full_document)
     end
+
     h.gfm_with_options(full_document, @options)
   end
 end
diff --git a/spec/lib/gitlab/markdown/autolink_filter_spec.rb b/spec/lib/gitlab/markdown/autolink_filter_spec.rb
new file mode 100644
index 0000000000..83fee12333
--- /dev/null
+++ b/spec/lib/gitlab/markdown/autolink_filter_spec.rb
@@ -0,0 +1,98 @@
+require 'spec_helper'
+
+module Gitlab::Markdown
+  describe AutolinkFilter do
+    let(:link) { 'http://about.gitlab.com/' }
+
+    def filter(html, options = {})
+      described_class.call(html, options)
+    end
+
+    it 'does nothing when :autolink is false' do
+      exp = act = link
+      expect(filter(act, autolink: false).to_html).to eq exp
+    end
+
+    it 'does nothing with non-link text' do
+      exp = act = 'This text contains no links to autolink'
+      expect(filter(act).to_html).to eq exp
+    end
+
+    context 'Rinku schemes' do
+      it 'autolinks http' do
+        doc = filter("See #{link}")
+        expect(doc.at_css('a').text).to eq link
+        expect(doc.at_css('a')['href']).to eq link
+      end
+
+      it 'autolinks https' do
+        link = 'https://google.com/'
+        doc = filter("See #{link}")
+
+        expect(doc.at_css('a').text).to eq link
+        expect(doc.at_css('a')['href']).to eq link
+      end
+
+      it 'autolinks ftp' do
+        link = 'ftp://ftp.us.debian.org/debian/'
+        doc = filter("See #{link}")
+
+        expect(doc.at_css('a').text).to eq link
+        expect(doc.at_css('a')['href']).to eq link
+      end
+
+      it 'accepts link_attr options' do
+        doc = filter("See #{link}", link_attr: {class: 'custom'})
+
+        expect(doc.at_css('a')['class']).to eq 'custom'
+      end
+
+      described_class::IGNORE_PARENTS.each do |elem|
+        it "ignores valid links contained inside '#{elem}' element" do
+          exp = act = "<#{elem}>See #{link}</#{elem}>"
+          expect(filter(act).to_html).to eq exp
+        end
+      end
+    end
+
+    context 'other schemes' do
+      let(:link) { 'foo://bar.baz/' }
+
+      it 'autolinks smb' do
+        link = 'smb:///Volumes/shared/foo.pdf'
+        doc = filter("See #{link}")
+
+        expect(doc.at_css('a').text).to eq link
+        expect(doc.at_css('a')['href']).to eq link
+      end
+
+      it 'autolinks irc' do
+        link = 'irc://irc.freenode.net/git'
+        doc = filter("See #{link}")
+
+        expect(doc.at_css('a').text).to eq link
+        expect(doc.at_css('a')['href']).to eq link
+      end
+
+      it 'does not include trailing punctuation' do
+        doc = filter("See #{link}.")
+        expect(doc.at_css('a').text).to eq link
+
+        doc = filter("See #{link}, ok?")
+        expect(doc.at_css('a').text).to eq link
+      end
+
+      it 'accepts link_attr options' do
+        doc = filter("See #{link}", link_attr: {class: 'custom'})
+        expect(doc.at_css('a')['class']).to eq 'custom'
+      end
+
+      described_class::IGNORE_PARENTS.each do |elem|
+        it "ignores valid links contained inside '#{elem}' element" do
+          exp = act = "<#{elem}>See #{link}</#{elem}>"
+          expect(filter(act).to_html).to eq exp
+        end
+      end
+    end
+  end
+end