--- /notifications/notifications_mail2web/notifications_mail2web.module +++ notifications_mail2web.module @@ -151,6 +151,8 @@ $info += array('glue' => "\n"); $message['body']['#prefix'] = implode($info['glue'], $prefix); } + //this is a dirty solution to crappy webmail clients that strip out the messageid and in-reply-to headers, got a better way? show me, please. + $message['body']['#footer'] .= t(' MessageID='.$message['params']['mail']['headers']['Message-ID']); } } @@ -160,6 +162,9 @@ */ function notifications_mail2web_mailhandler($node, $result, $i, $header, $mailbox) { // The In-reply-to header is cleaned and passed in $node->threading + if (!$node->threading) { + $node->threading=notifications_mail2web_get_backup_messageid($node->body); + } if ($node->threading && ($params = notifications_mail2web_check_messageparams($node->threading, $header)) && empty($params['error'])) { // Now check user id , just go ahead if they match and it is a valid user if ($node->uid && $node->uid == $params['uid']) { @@ -169,7 +174,7 @@ $node->type = 'comment'; $node->nid = $params['nid']; $node->pid = $params['cid']; - + $mimeparts=$node->mimeparts; // Now trim out the resf of the message if separator text exists // @ TODO May fail for html mails if ($marker = variable_get('mail2web_reply_text', t('((( Reply ABOVE this LINE to POST a COMMENT )))'))) { @@ -177,9 +182,32 @@ $pos = strpos($node->body, $marker); if ($pos !== FALSE) { $node->body = substr($node->body, 0, $pos); - // Something specifically for mac mail clients out there - $split = preg_split("/On [A-Z][a-z]{2} [0-9]{1,2}, [0-9]{4}, at [0-9]{1,2}:[0-9]{2} [AP]M, [a-zA-Z0-9._%-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4} wrote/",$node->body); - $split = preg_split("/\>/",$split[0]); + + if ($mimeparts[1]) { + //if there is a second mimepart, its html, am I wrong... + //check for html body + $node->body=notifications_mail2web_clean_html_message_body($node->body); + //this expression is messed up because of html email in the on.. wrote... + $split = preg_split('/
On/', $node->body); + //$node->body = preg_replace('/On [A-Z][a-z]{2} [0-9]{1,2}, [0-9]{4}, at [0-9]{1,2}:[0-9]{2} [AP]M, [a-zA-Z0-9._%-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4} wrote/','', $node->body); + $node->body = $split[0]; + + } else { + //the old way that seems to work for plain text + // Something specifically for mac mail clients out there + $split = preg_split("/On [A-Z][a-z]{2} [0-9]{1,2}, [0-9]{4}, at [0-9]{1,2}:[0-9]{2} [AP]M, [a-zA-Z0-9._%-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4} wrote/",$node->body); + $split = preg_split("/\>/",$split[0]); + $node->body = $split[0]; + + } + //for the old school webmail + $split = preg_split("/Original Message/",$node->body); + $node->body = $split[0]; + //for gmail + $split = preg_split("/\
On/",$node->body); + $node->body = $split[0]; + //for old yahoo + $split = preg_split("/\\-\-\- On/",$node->body); $node->body = $split[0]; } } @@ -200,6 +228,33 @@ } /** + * Find the message id in the body if yahoo was arrogant enough to strip it out of the headers. + * + * yahoo is out to destroy all that is not yahoo groups. + * + * @param $body + * raw email body + * + */ +function notifications_mail2web_get_backup_messageid($body) { + $res = preg_match_all( + "/[a-z0-9]+([_\\.-][a-z0-9]+)*@([a-z0-9]+([\.-][a-z0-9]+)*)+\\.[a-z]{2,}/i", + $body, + $matches + ); + + if ($res) { + foreach(array_unique($matches[0]) as $email) { + if (ereg(variable_get('mail2web_server_string', 'example.com'), $email)) { + $backup_messageid=$email; + } + } + } + print 'backup'. $backup_messageid; + return $backup_messageid; + } + +/** * Handle errors and bounce mail when authentication or validation fail * * We handle the incoming email carefully and don't add any user data in the response @@ -353,3 +408,62 @@ return md5(implode('-', $params)); } +/** This function is ripped straight out of og2list. god bless the work that has gone into this. + * Clean HTML Content-type: text/html message bodies. + * There are 2 main tasks performed. First we remove any extra linefeeds + * from the body, placing it all on one line. This is because Drupal will + * start inserting linefeeds in the post at the end of each HTML line. + * the length of these lines is determined by the email client as it writes + * out the HTML and has nothing to do with the line or paragraph breaks + * in the post itself. + * Finally, there is the Outlook babysitting. Predictably, Outlook produces + * mangled HTML which ends up showing up in the Drupal message body. We do some + * cleanup of the body's HTML at the end to make sure none of this affects the final + * post content. +*/ +function notifications_mail2web_clean_html_message_body($body) { + + $body = preg_replace('/[\r\n]{2,}/','ABCD_PLACEHOLDER_DCBA', $body); // save para breaks + $body = preg_replace('/[\r\n]+/',' ', $body); + $body = str_replace('ABCD_PLACEHOLDER_DCBA',"\n\n",$body); // put the para breaks back in + + // MS-Outlook HTML cleanup: + // Stylesheet commenting: + // Outlook's initial + // + // we fix this to have the comment inside the