From a1013c0372c244c3722b21fee1b6bc312651669e Mon Sep 17 00:00:00 2001
From: thomascube <thomas@roundcube.net>
Date: Mon, 10 Jan 2011 09:50:48 -0500
Subject: [PATCH] Exclude punctuation from URLs within text; only link accepted url schemes

---
 program/include/rcube_string_replacer.php |   19 +++++++++++--------
 tests/mailfunc.php                        |    2 +-
 tests/src/plainbody.txt                   |    2 +-
 3 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/program/include/rcube_string_replacer.php b/program/include/rcube_string_replacer.php
index e0a7364..8fcbeca 100644
--- a/program/include/rcube_string_replacer.php
+++ b/program/include/rcube_string_replacer.php
@@ -37,13 +37,14 @@
   {
     // Simplified domain expression for UTF8 characters handling
     $utf_domain = '[^?&@"\'\\/()\s\r\t\n]+\\.[a-z]{2,5}';
-    $url = '[a-z0-9%=#@+?.:;&\\/_~\\[\\]-]+';
+    $url1 = '.:;';
+    $url2 = 'a-z0-9%=#@+?&\\/_~\\[\\]-';
 
-    $this->link_pattern = "/([\w]+:\/\/|\Wwww\.)($utf_domain($url)?)/i";
+    $this->link_pattern = "/([\w]+:\/\/|\Wwww\.)($utf_domain([$url1]?[$url2]+)*)/i";
     $this->mailto_pattern = "/("
         ."[-\w!\#\$%&\'*+~\/^`|{}=]+(?:\.[-\w!\#\$%&\'*+~\/^`|{}=]+)*"  // local-part
         ."@$utf_domain"                                                 // domain-part
-        ."(\?$url)?"                                                    // e.g. ?subject=test...
+        ."(\?[$url1$url2]+)?"                                           // e.g. ?subject=test...
         .")/i";
   }
 
@@ -88,11 +89,13 @@
       $prefix     = $m[1];
     }
 
-    $suffix = $this->parse_url_brackets($url);
-    $i = $this->add($prefix . html::a(array(
-      'href' => $url_prefix . $url,
-      'target' => '_blank'
-      ), Q($url)) . $suffix);
+    if ($url) {
+      $suffix = $this->parse_url_brackets($url);
+      $i = $this->add($prefix . html::a(array(
+          'href' => $url_prefix . $url,
+          'target' => '_blank'
+        ), Q($url)) . $suffix);
+    }
 
     // Return valid link for recognized schemes, otherwise, return the unmodified string for unrecognized schemes.
     return $i >= 0 ? $this->get_replacement($i) : $matches[0];
diff --git a/tests/mailfunc.php b/tests/mailfunc.php
index eb724b6..cc26f77 100644
--- a/tests/mailfunc.php
+++ b/tests/mailfunc.php
@@ -112,7 +112,7 @@
     $html = rcmail_print_body($part, array('safe' => true));
 
     $this->assertPattern('/<a href="mailto:nobody@roundcube.net" onclick="return rcmail.command\(\'compose\',\'nobody@roundcube.net\',this\)">nobody@roundcube.net<\/a>/', $html, "Mailto links with onclick");
-    $this->assertPattern('#<a href="http://www.apple.com/legal/privacy/" target="_blank">http://www.apple.com/legal/privacy/</a>#', $html, "Links with target=_blank");
+    $this->assertPattern('#<a href="http://www.apple.com/legal/privacy" target="_blank">http://www.apple.com/legal/privacy</a>#', $html, "Links with target=_blank");
     $this->assertPattern('#\\[<a href="http://example.com/\\?tx\\[a\\]=5" target="_blank">http://example.com/\\?tx\\[a\\]=5</a>\\]#', $html, "Links with square brackets");
   }
 
diff --git a/tests/src/plainbody.txt b/tests/src/plainbody.txt
index 5d391f5..7fba94f 100644
--- a/tests/src/plainbody.txt
+++ b/tests/src/plainbody.txt
@@ -30,7 +30,7 @@
 http://www.apple.com/enews/subscribe/
 
 Privacy Policy
-http://www.apple.com/legal/privacy/
+http://www.apple.com/legal/privacy.
 
 My Info
 https://myinfo.apple.com/cgi-bin/WebObjects/MyInfo

--
Gitblit v1.9.1