avoid auto_linking already linked emails; more robust detection of linked URLs

References #1523  [#1862 state:resolved]  [#3591 state:resolved]

Add test that shows how link text can contain HTML if needed:
the trick is using block form in combination with `raw`.
Let link text be automatically HTML-escaped

[#2017 state:resolved]

actionpack/lib/action_view/helpers/text_helper.rb

@@ -539,7 +539,12 @@ module ActionView
         AUTO_LINK_RE = %r{
             (?: ([\w+.:-]+:)// | www\. )
             [^\s<]+
-          }x unless const_defined?(:AUTO_LINK_RE)
+          }x
+
+        # regexps for determining context, used high-volume
+        AUTO_LINK_CRE = [/<[^>]+$/, /^[^>]*>/, /<a\b.*?>/i, /<\/a>/i]
+
+        AUTO_EMAIL_RE = /[\w.!#\$%+-]+@[\w-]+(?:\.[\w-]+)+/
 
         BRACKETS = { ']' => '[', ')' => '(', '}' => '{' }
 
@@ -550,9 +555,8 @@ module ActionView
           text.gsub(AUTO_LINK_RE) do
             scheme, href = $1, $&
             punctuation = []
-            left, right = $`, $'
+            
-            # detect already linked URLs and URLs in the middle of a tag
+            if auto_linked?($`, $')
-            if left =~ /<[^>]+$/ && right =~ /^[^>]*>/
               # do not change string; URL is already linked
               href
             else
@@ -568,7 +572,7 @@ module ActionView
               link_text = block_given?? yield(href) : href
               href = 'http://' + href unless scheme
 
-              content_tag(:a, h(link_text), link_attributes.merge('href' => href)) + punctuation.reverse.join('')
+              content_tag(:a, link_text, link_attributes.merge('href' => href)) + punctuation.reverse.join('')
             end
           end
         end
@@ -576,11 +580,10 @@ module ActionView
         # Turns all email addresses into clickable links.  If a block is given,
         # each email is yielded and the result is used as the link text.
         def auto_link_email_addresses(text, html_options = {})
-          body = text.dup
+          text.gsub(AUTO_EMAIL_RE) do
-          text.gsub(/([\w\.!#\$%\-+]+@[A-Za-z0-9\-]+(\.[A-Za-z0-9\-]+)+)/) do
+            text = $&
-            text = $1
 
-            if body.match(/<a\b[^>]*>(.*)(#{Regexp.escape(text)})(.*)<\/a>/)
+            if auto_linked?($`, $')
               text
             else
               display_text = (block_given?) ? yield(text) : text
@@ -588,6 +591,12 @@ module ActionView
             end
           end
         end
+        
+        # Detects already linked context or position in the middle of a tag
+        def auto_linked?(left, right)
+          (left =~ AUTO_LINK_CRE[0] and right =~ AUTO_LINK_CRE[1]) or
+            (left.rindex(AUTO_LINK_CRE[2]) and $' !~ AUTO_LINK_CRE[3])
+        end
     end
   end
 end

actionpack/test/template/text_helper_test.rb

@@ -304,6 +304,7 @@ class TextHelperTest < ActionView::TestCase
     assert_equal %(<p>Link #{link_result_with_options}</p>), auto_link("<p>Link #{link_raw}</p>", :all, {:target => "_blank"})
     assert_equal %(Go to #{link_result}.), auto_link(%(Go to #{link_raw}.))
     assert_equal %(<p>Go to #{link_result}, then say hello to #{email_result}.</p>), auto_link(%(<p>Go to #{link_raw}, then say hello to #{email_raw}.</p>))
+    assert_equal %(#{link_result} #{link_result}), auto_link(%(#{link_result} #{link_raw}))
 
     email2_raw    = '+david@loudthinking.com'
     email2_result = %{<a href="mailto:#{email2_raw}">#{email2_raw}</a>}
@@ -390,9 +391,24 @@ class TextHelperTest < ActionView::TestCase
 
   def test_auto_link_already_linked
     linked1 = generate_result('Ruby On Rails', 'http://www.rubyonrails.com')
-    linked2 = generate_result('www.rubyonrails.com', 'http://www.rubyonrails.com')
+    linked2 = %('<a href="http://www.example.com">www.example.com</a>')
+    linked3 = %('<a href="http://www.example.com" rel="nofollow">www.example.com</a>')
+    linked4 = %('<a href="http://www.example.com"><b>www.example.com</b></a>')
+    linked5 = %('<a href="#close">close</a> <a href="http://www.example.com"><b>www.example.com</b></a>')
     assert_equal linked1, auto_link(linked1)
     assert_equal linked2, auto_link(linked2)
+    assert_equal linked3, auto_link(linked3)
+    assert_equal linked4, auto_link(linked4)
+    assert_equal linked5, auto_link(linked5)
+    
+    linked_email = %Q(<a href="mailto:david@loudthinking.com">Mail me</a>)
+    assert_equal linked_email, auto_link(linked_email)
+  end
+
+  def test_auto_link_within_tags
+    link_raw    = 'http://www.rubyonrails.org/images/rails.png'
+    link_result = %Q(<img src="#{link_raw}" />)
+    assert_equal link_result, auto_link(link_result)
   end
 
   def test_auto_link_with_brackets
@@ -412,12 +428,6 @@ class TextHelperTest < ActionView::TestCase
     assert_equal "{link: #{link3_result}}", auto_link("{link: #{link3_raw}}")
   end
 
-  def test_auto_link_in_tags
-    link_raw    = 'http://www.rubyonrails.org/images/rails.png'
-    link_result = %Q(<img src="#{link_raw}" />)
-    assert_equal link_result, auto_link(link_result)
-  end
-
   def test_auto_link_at_eol
     url1 = "http://api.rubyonrails.com/Foo.html"
     url2 = "http://www.ruby-doc.org/core/Bar.html"
@@ -432,6 +442,19 @@ class TextHelperTest < ActionView::TestCase
     assert_equal %(<p><a href="#{url}">#{url[0...7]}...</a><br /><a href="mailto:#{email}">#{email[0...7]}...</a><br /></p>), auto_link("<p>#{url}<br />#{email}<br /></p>") { |url| truncate(url, :length => 10) }
   end
   
+  def test_auto_link_with_block_with_html
+    pic = "http://example.com/pic.png"
+    url = "http://example.com/album?a&b=c"
+    
+    assert_equal %(My pic: <a href="#{pic}"><img src="#{pic}" width="160px"></a> -- full album here #{generate_result(url)}), auto_link("My pic: #{pic} -- full album here #{url}") { |link|
+      if link =~ /\.(jpg|gif|png|bmp|tif)$/i
+        raw %(<img src="#{link}" width="160px">)
+      else
+        link
+      end
+    }
+  end
+
   def test_auto_link_with_options_hash
     assert_dom_equal 'Welcome to my new blog at <a href="http://www.myblog.com/" class="menu" target="_blank">http://www.myblog.com/</a>. Please e-mail me at <a href="mailto:me@email.com" class="menu" target="_blank">me@email.com</a>.',
       auto_link("Welcome to my new blog at http://www.myblog.com/. Please e-mail me at me@email.com.",