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]

1 files changed, 18 insertions, 9 deletions

diff --git a/actionpack/lib/action_view/helpers/text_helper.rb b/actionpack/lib/action_view/helpers/text_helper.rb
index fa930ac626..41423d4e2e 100644
--- a/actionpack/lib/action_view/helpers/text_helper.rb
+++ b/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 = $`, $'
-            # detect already linked URLs and URLs in the middle of a tag
-            if left =~ /<[^>]+$/ && right =~ /^[^>]*>/
+            
+            if auto_linked?($`, $')
               # 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(/([\w\.!#\$%\-+]+@[A-Za-z0-9\-]+(\.[A-Za-z0-9\-]+)+)/) do
-            text = $1
+          text.gsub(AUTO_EMAIL_RE) do
+            text = $&
 
-            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