Changeset 2064 in subversion

Timestamp:

Nov 19, 2008 2:32:40 AM (5 years ago)

Author:

thomasb

Message:

Remove BOM in html message body + don't require quotes in meta tag specifying charset

File:

• trunk/roundcubemail/program/steps/mail/func.inc (modified) (1 diff)

trunk/roundcubemail/program/steps/mail/func.inc

@@ -598 +598 @@
     // special replacements (not properly handled by washtml class)
     $html_search = array(
-        '/(<\/nobr>)(\s+)(<nobr>)/i',   // space(s) between <NOBR>
-        '/(<[\/]*st1:[^>]+>)/i',        // Microsoft's Smart Tags <ST1>
-        '/<title>.*<\/title>/i',        // PHP bug #32547 workaround: remove title tag
-        '/<html[^>]*>/im',              // malformed html: remove html tags (#1485139)
-        '/<\/html>/i',                  // malformed html: remove html tags (#1485139)
+      '/(<\/nobr>)(\s+)(<nobr>)/i',     // space(s) between <NOBR>
+    '/(<[\/]*st1:[^>]+>)/i',    // Microsoft's Smart Tags <ST1>
+    '/<title>.*<\/title>/i',    // PHP bug #32547 workaround: remove title tag
+    '/<html[^>]*>/im',          // malformed html: remove html tags (#1485139)
+    '/<\/html>/i',                      // malformed html: remove html tags (#1485139)
+    '/^[\xFE\xFF\xBB\xBF\x00]+((?:<\!doctype|\<html))/im',      // remove byte-order mark (only outlook?)
     );
     $html_replace = array(
-        '\\1'.' &nbsp; '.'\\3',
-        '',
-        '',
-        '',
-        '',
+      '\\1'.' &nbsp; '.'\\3',
+      '',
+      '',
+      '',
+      '\\1',
+      '',
     );
     $html = preg_replace($html_search, $html_replace, $html);
 
     // charset was converted to UTF-8 in rcube_imap::get_message_part() -> change charset specification in HTML accordingly
-    if (preg_match('/(\s+content=[\'"]\w+\/\w+;\s*charset)=([a-z0-9-_]+)/i', $html)) 
-      $html = preg_replace('/(\s+content=[\'"]\w+\/\w+;\s*charset)=([a-z0-9-_]+)/i', '\\1='.RCMAIL_CHARSET, $html); 
+    $charset_pattern = '/(\s+content=[\'"]?\w+\/\w+;\s*charset)=([a-z0-9-_]+)/i';
+    if (preg_match($charset_pattern, $html)) {
+      $html = preg_replace($charset_pattern, '\\1='.RCMAIL_CHARSET, $html);
+    }
     else {
       // add head for malformed messages, washtml cannot work without that