From 5a173887269708ee3f913a869741e03b9aceeb41 Mon Sep 17 00:00:00 2001 From: thomascube Date: Wed, 19 Nov 2008 07:32:40 +0000 Subject: Remove BOM in html message body + don't require quotes in meta tag specifying charset --- program/steps/mail/func.inc | 28 ++++++++++++++++------------ 1 file changed, 16 insertions(+), 12 deletions(-) (limited to 'program/steps/mail/func.inc') diff --git a/program/steps/mail/func.inc b/program/steps/mail/func.inc index 1debf875b..2a917b170 100644 --- a/program/steps/mail/func.inc +++ b/program/steps/mail/func.inc @@ -597,24 +597,28 @@ function rcmail_print_body($part, $p = array()) // special replacements (not properly handled by washtml class) $html_search = array( - '/(<\/nobr>)(\s+)()/i', // space(s) between - '/(<[\/]*st1:[^>]+>)/i', // Microsoft's Smart Tags - '/.*<\/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'.'   '.'\\3', - '', - '', - '', - '', + '\\1'.'   '.'\\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 if (!preg_match('/<head[^>]*>(.*)<\/head>/Uims', $html)) -- cgit v1.2.3