Fixing undefined variables

[squirrelmail.git] / functions / url_parser.php

diff --git a/functions/url_parser.php b/functions/url_parser.php
index ba037388285cd93b3296c91039818ef9488105d3..ca1b52d10ce155a86eecc772c24952e0ad8a9eb0 100644 (file)
--- a/functions/url_parser.php
+++ b/functions/url_parser.php
@@ -9,7 +9,7 @@
  * This code provides various string manipulation functions that are
  * used by the rest of the Squirrelmail code.
  *
- * $Id$
+ * @version $Id$
  * @package squirrelmail
  */
 
@@ -26,6 +26,10 @@ function replaceBlock (&$in, $replace, $start, $end) {
  * to be made to the pattern
  * Make sure that the expression is evaluated case insensitively
  *
+ * RFC2822 (and RFC822) defines the left side of an email address as (roughly):
+ *  1*atext *("." 1*atext)
+ * where atext is: a-zA-Z0-9!#$%&'*+-/=?^_`{|}~
+ *
  * Here's pretty sophisticated IP matching:
  * $IPMatch = '(2[0-5][0-9]|1?[0-9]{1,2})';
  * $IPMatch = '\[?' . $IPMatch . '(\.' . $IPMatch . '){3}\]?';
@@ -35,8 +39,10 @@ global $IP_RegExp_Match, $Host_RegExp_Match, $Email_RegExp_Match;
 $IP_RegExp_Match = '\\[?[0-9]{1,3}(\\.[0-9]{1,3}){3}\\]?';
 $Host_RegExp_Match = '(' . $IP_RegExp_Match .
     '|[0-9a-z]([-.]?[0-9a-z])*\\.[a-z][a-z]+)';
-$Email_RegExp_Match = '[0-9a-z]([-_.+]?[0-9a-z])*(%' . $Host_RegExp_Match .
-    ')?@' . $Host_RegExp_Match;
+$atext = '([a-z0-9!#$&%*+/=?^_`{|}~-]|&)';
+$dot_atom = $atext . '+(\.' . $atext . '+)*';
+$Email_RegExp_Match = $dot_atom . '(%' . $Host_RegExp_Match . ')?@' .
+                      $Host_RegExp_Match;
 
 /**
  * Parses a body and converts all found email addresses to clickable links.
@@ -51,14 +57,14 @@ function parseEmail (&$body) {
 
     /* Find all the email addresses in the body */
     while(eregi($Email_RegExp_Match, $sbody, $regs)) {
-        $addresses[$regs[0]] = $regs[0];
+        $addresses[$regs[0]] = strtr($regs[0], array('&' => '&'));
         $start = strpos($sbody, $regs[0]) + strlen($regs[0]);
         $sbody = substr($sbody, $start);
     }
     /* Replace each email address with a compose URL */
-    foreach ($addresses as $email) {
-        $comp_uri = makeComposeLink('src/compose.php?send_to='.urlencode($email), $email);
-        $body = str_replace($email, $comp_uri, $body);
+    foreach ($addresses as $text => $email) {
+        $comp_uri = makeComposeLink('src/compose.php?send_to='.urlencode($email), $text);
+        $body = str_replace($text, $comp_uri, $body);
     }
     /* Return number of unique addresses found */
     return count($addresses);
@@ -79,8 +85,8 @@ $url_parser_url_tokens = array(
     'news://');
 
 global $url_parser_poss_ends;
-$url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n", 
-    '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<', 
+$url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
+    '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
     ']', '[', '{', '}', "\240", ', ', '. ', ",\n", ",\r");
 
 
@@ -111,7 +117,7 @@ function parseUrl (&$body) {
         /* Find the first token to replace */
         foreach ($url_parser_url_tokens as $the_token) {
             $pos = strpos(strtolower($body), $the_token, $start);
-            if (is_int($pos) && $pos < $blength) {
+            if (is_int($pos) && $pos < $target_pos) {
                 $target_pos   = $pos;
                 $target_token = $the_token;
             }
@@ -191,5 +197,5 @@ function parseUrl (&$body) {
         $start   = $target_pos;
         $blength = strlen($body);
     }
-} 
-?>
+}
+?>
\ No newline at end of file