From 2cb6d48e63b6d2e0a71fc0b3bbbcc85b8c9ee7d3 Mon Sep 17 00:00:00 2001
From: alecpl <alec@alec.pl>
Date: Wed, 17 Sep 2008 13:43:38 -0400
Subject: [PATCH] - code cleanup + fix for malformed html (#1485139)
---
program/steps/mail/func.inc | 30 +++++++++++++++++-------------
1 files changed, 17 insertions(+), 13 deletions(-)
diff --git a/program/steps/mail/func.inc b/program/steps/mail/func.inc
index 3995bff..440de92 100644
--- a/program/steps/mail/func.inc
+++ b/program/steps/mail/func.inc
@@ -597,31 +597,35 @@
}
// text/html
else if ($part->ctype_secondary == 'html') {
- // charset was converted to UTF-8 in rcube_imap::get_message_part() -> change charset specification in HTML accordingly
- $html = $part->body;
- 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);
- else {
- // add <head> for malformed messages, washtml cannot work without that
- if (!preg_match('/<head>(.*)<\\/head>/Uims', $html))
- $html = '<head></head>' . $html;
- $html = substr_replace($html, '<meta http-equiv="Content-Type" content="text/html; charset='.RCMAIL_CHARSET.'" />', intval(stripos($html, '</head>')), 0);
- }
-
- // PHP bug #32547 workaround: remove title tag
- $html = preg_replace('/<title>.*<\/title>/', '', $html);
+ $html = $part->body;
// 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)
);
$html_replace = array(
'\\1'.' '.'\\3',
'',
+ '',
+ '',
+ '',
);
$html = preg_replace($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);
+ else {
+ // add head for malformed messages, washtml cannot work without that
+ if (!preg_match('/<head[^>]*>(.*)<\/head>/Uims', $html))
+ $html = '<head></head>'. $html;
+ $html = substr_replace($html, '<meta http-equiv="Content-Type" content="text/html; charset='.RCMAIL_CHARSET.'" />', intval(stripos($html, '</head>')), 0);
+ }
+
// clean HTML with washhtml by Frederic Motte
$wash_opts = array(
'show_washed' => false,
--
Gitblit v1.9.1