From 543845c673a9fd24e2e95ee9fd2acc5a70f8051b Mon Sep 17 00:00:00 2001
From: Sam Saffron <sam.saffron@gmail.com>
Date: Mon, 11 Feb 2013 11:43:07 +1100
Subject: [PATCH] rel nofollow, on by default to protect forums from spam etc.
 we should consider lifting it at high trust by default.

---
 app/models/site_setting.rb          |  1 +
 config/locales/en.yml               |  1 +
 lib/pretty_text.rb                  | 28 +++++++++++++++++++++++++++-
 spec/components/pretty_text_spec.rb | 19 +++++++++++++++++++
 4 files changed, 48 insertions(+), 1 deletion(-)

diff --git a/app/models/site_setting.rb b/app/models/site_setting.rb
index 65c3b93b7..5dcf86d25 100644
--- a/app/models/site_setting.rb
+++ b/app/models/site_setting.rb
@@ -91,6 +91,7 @@ class SiteSetting < ActiveRecord::Base
 
   setting(:allow_duplicate_topic_titles, false)
 
+  setting(:add_rel_nofollow_to_user_content, true)
   setting(:post_excerpt_maxlength, 300)
   setting(:post_onebox_maxlength, 500)
   setting(:best_of_score_threshold, 15)
diff --git a/config/locales/en.yml b/config/locales/en.yml
index 0a3331ef4..8c4ab9945 100644
--- a/config/locales/en.yml
+++ b/config/locales/en.yml
@@ -219,6 +219,7 @@ en:
     max_image_width: "maximum width for an image in a post"
     category_featured_topics: "number of topics displayed in the category list"
     popup_delay: "Length of time in ms before popups appear on the screen"
+    add_rel_nofollow_to_user_content: "Add rel nofollow to all submitted user content, except for internal links (including parent domains) changing this requires you update all your baked markdown"
     post_excerpt_maxlength: "Maximum length in chars of a post's excerpt."
     post_onebox_maxlength: "Maximum length of a oneboxed discourse post."
     category_post_template: "The post template that appears once you create a category"
diff --git a/lib/pretty_text.rb b/lib/pretty_text.rb
index 4842cd817..cb28ebac6 100644
--- a/lib/pretty_text.rb
+++ b/lib/pretty_text.rb
@@ -172,7 +172,33 @@ module PrettyText
     cloned = opts.dup
     # we have a minor inconsistency
     cloned[:topicId] = opts[:topic_id]
-    Sanitize.clean(markdown(text.dup, cloned), PrettyText.whitelist)    
+    sanitized = Sanitize.clean(markdown(text.dup, cloned), PrettyText.whitelist)    
+    if SiteSetting.add_rel_nofollow_to_user_content
+      sanitized = add_rel_nofollow_to_user_content(sanitized) 
+    end
+    sanitized
+  end
+  
+  def self.add_rel_nofollow_to_user_content(html)
+    site_uri = nil
+    doc = Nokogiri::HTML.fragment(html)
+    doc.css("a").each do |l|
+      href = l["href"].to_s
+      begin 
+        uri = URI(href)
+        site_uri ||= URI(Discourse.base_url)
+        
+        if !uri.host.present? || uri.host.ends_with?(site_uri.host)
+          # we are good no need for nofollow
+        else
+          l["rel"] = "nofollow"
+        end
+      rescue URI::InvalidURIError
+        # add a nofollow anyway 
+        l["rel"] = "nofollow"
+      end
+    end
+    doc.to_html
   end
 
   def self.extract_links(html)
diff --git a/spec/components/pretty_text_spec.rb b/spec/components/pretty_text_spec.rb
index cb8536fd6..430b8ac62 100644
--- a/spec/components/pretty_text_spec.rb
+++ b/spec/components/pretty_text_spec.rb
@@ -75,6 +75,24 @@ test
         .should == "<pre><code>```\nhello\n```\n</code></pre>"
     end
   end
+  
+  describe "rel nofollow" do
+    before do 
+      SiteSetting.stubs(:add_rel_nofollow_to_user_content).returns(true)
+    end
+
+    it "should inject nofollow in all user provided links" do 
+      PrettyText.cook('<a href="http://cnn.com">cnn</a>').should =~ /nofollow/
+    end
+    
+    it "should not inject nofollow in all local links" do 
+      (PrettyText.cook("<a href='#{Discourse.base_url}/test.html'>cnn</a>") !~ /nofollow/).should be_true  
+    end
+    
+    it "should not inject nofollow in all subdomain links" do 
+      (PrettyText.cook("<a href='#{Discourse.base_url.sub('http://', 'http://bla.')}/test.html'>cnn</a>") !~ /nofollow/).should be_true  
+    end
+  end
 
   describe "Excerpt" do 
     it "should preserve links" do 
@@ -130,6 +148,7 @@ test
     end
   end
 
+
   describe "apply cdn" do 
     it "should detect bare links to images and apply a CDN" do
       PrettyText.apply_cdn("<a href='/hello.png'>hello</a><img src='/a.jpeg'>","http://a.com").should ==