Exemplo n.º 1
0
function html2bbcode($message)
{
    $message = str_replace("\r", "", $message);
    $message = preg_replace_callback("|<pre><code>([^<]*)</code></pre>|ism", "_replace_code_cb", $message);
    $message = str_replace(array("<li><p>", "</p></li>"), array("<li>", "</li>"), $message);
    // remove namespaces
    $message = preg_replace('=<(\\w+):(.+?)>=', '<removeme>', $message);
    $message = preg_replace('=</(\\w+):(.+?)>=', '</removeme>', $message);
    $doc = new DOMDocument();
    $doc->preserveWhiteSpace = false;
    $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
    @$doc->loadHTML($message);
    deletenode($doc, 'style');
    deletenode($doc, 'head');
    deletenode($doc, 'title');
    deletenode($doc, 'meta');
    deletenode($doc, 'xml');
    deletenode($doc, 'removeme');
    $xpath = new DomXPath($doc);
    $list = $xpath->query("//pre");
    foreach ($list as $node) {
        $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
    }
    $message = $doc->saveHTML();
    $message = str_replace(array("\n<", ">\n", "\r", "\n", " "), array("<", ">", "<br />", " ", ""), $message);
    $message = preg_replace('= [\\s]*=i', " ", $message);
    @$doc->loadHTML($message);
    node2bbcode($doc, 'html', array(), "", "");
    node2bbcode($doc, 'body', array(), "", "");
    // Outlook-Quote - Variant 1
    node2bbcode($doc, 'p', array('class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'), '[quote]', '[/quote]');
    // Outlook-Quote - Variant 2
    node2bbcode($doc, 'div', array('style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'), '[quote]', '[/quote]');
    // MyBB-Stuff
    node2bbcode($doc, 'span', array('style' => 'text-decoration: underline;'), '[u]', '[/u]');
    node2bbcode($doc, 'span', array('style' => 'font-style: italic;'), '[i]', '[/i]');
    node2bbcode($doc, 'span', array('style' => 'font-weight: bold;'), '[b]', '[/b]');
    /*node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
    	node2bbcode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
    	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
    	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
    	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
    	node2bbcode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
    	node2bbcode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
    */
    // Untested
    //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
    //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
    //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
    node2bbcode($doc, 'span', array('style' => '/.*color:\\s*(.+?)[,;].*/'), '[color="$1"]', '[/color]');
    //node2bbcode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
    //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
    //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
    //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
    // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
    // Test
    //node2bbcode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
    node2bbcode($doc, 'span', array('class' => 'type-link'), '[class=type-link]', '[/class]');
    node2bbcode($doc, 'span', array('class' => 'type-video'), '[class=type-video]', '[/class]');
    node2bbcode($doc, 'strong', array(), '[b]', '[/b]');
    node2bbcode($doc, 'em', array(), '[i]', '[/i]');
    node2bbcode($doc, 'b', array(), '[b]', '[/b]');
    node2bbcode($doc, 'i', array(), '[i]', '[/i]');
    node2bbcode($doc, 'u', array(), '[u]', '[/u]');
    node2bbcode($doc, 'big', array(), "[size=large]", "[/size]");
    node2bbcode($doc, 'small', array(), "[size=small]", "[/size]");
    node2bbcode($doc, 'blockquote', array(), '[quote]', '[/quote]');
    node2bbcode($doc, 'br', array(), "\n", '');
    node2bbcode($doc, 'p', array('class' => 'MsoNormal'), "\n", "");
    node2bbcode($doc, 'div', array('class' => 'MsoNormal'), "\r", "");
    node2bbcode($doc, 'span', array(), "", "");
    node2bbcode($doc, 'span', array(), "", "");
    node2bbcode($doc, 'pre', array(), "", "");
    node2bbcode($doc, 'div', array(), "\r", "\r");
    node2bbcode($doc, 'p', array(), "\n", "\n");
    node2bbcode($doc, 'ul', array(), "[list]", "[/list]");
    node2bbcode($doc, 'ol', array(), "[list=1]", "[/list]");
    node2bbcode($doc, 'li', array(), "[*]", "");
    node2bbcode($doc, 'hr', array(), "[hr]", "");
    node2bbcode($doc, 'table', array(), "", "");
    node2bbcode($doc, 'tr', array(), "\n", "");
    node2bbcode($doc, 'td', array(), "\t", "");
    //node2bbcode($doc, 'table', array(), "[table]", "[/table]");
    //node2bbcode($doc, 'th', array(), "[th]", "[/th]");
    //node2bbcode($doc, 'tr', array(), "[tr]", "[/tr]");
    //node2bbcode($doc, 'td', array(), "[td]", "[/td]");
    //node2bbcode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
    //node2bbcode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
    //node2bbcode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
    //node2bbcode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
    //node2bbcode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
    //node2bbcode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
    node2bbcode($doc, 'h1', array(), "\n\n[h1]", "[/h1]\n");
    node2bbcode($doc, 'h2', array(), "\n\n[h2]", "[/h2]\n");
    node2bbcode($doc, 'h3', array(), "\n\n[h3]", "[/h3]\n");
    node2bbcode($doc, 'h4', array(), "\n\n[h4]", "[/h4]\n");
    node2bbcode($doc, 'h5', array(), "\n\n[h5]", "[/h5]\n");
    node2bbcode($doc, 'h6', array(), "\n\n[h6]", "[/h6]\n");
    node2bbcode($doc, 'a', array('href' => '/mailto:(.+)/'), '[mail=$1]', '[/mail]');
    node2bbcode($doc, 'a', array('href' => '/(.+)/'), '[url=$1]', '[/url]');
    node2bbcode($doc, 'img', array('src' => '/(.+)/', 'width' => '/(\\d+)/', 'height' => '/(\\d+)/'), '[img=$2x$3]$1', '[/img]');
    node2bbcode($doc, 'img', array('src' => '/(.+)/'), '[img]$1', '[/img]');
    node2bbcode($doc, 'video', array('src' => '/(.+)/'), '[video]$1', '[/video]');
    node2bbcode($doc, 'audio', array('src' => '/(.+)/'), '[audio]$1', '[/audio]');
    node2bbcode($doc, 'iframe', array('src' => '/(.+)/'), '[iframe]$1', '[/iframe]');
    node2bbcode($doc, 'code', array(), '[code]', '[/code]');
    node2bbcode($doc, 'key', array(), '[code]', '[/code]');
    $message = $doc->saveHTML();
    // I'm removing something really disturbing
    // Don't know exactly what it is
    $message = str_replace(chr(194) . chr(160), ' ', $message);
    $message = str_replace("&nbsp;", " ", $message);
    // removing multiple DIVs
    $message = preg_replace('=\\r *\\r=i', "\n", $message);
    $message = str_replace("\r", "\n", $message);
    call_hooks('html2bbcode', $message);
    $message = strip_tags($message);
    $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
    $message = str_replace(array("<"), array("&lt;"), $message);
    // remove quotes if they don't make sense
    $message = preg_replace('=\\[/quote\\][\\s]*\\[quote\\]=i', "\n", $message);
    $message = preg_replace('=\\[quote\\]\\s*=i', "[quote]", $message);
    $message = preg_replace('=\\s*\\[/quote\\]=i', "[/quote]", $message);
    do {
        $oldmessage = $message;
        $message = str_replace("\n \n", "\n\n", $message);
    } while ($oldmessage != $message);
    do {
        $oldmessage = $message;
        $message = str_replace("\n\n\n", "\n\n", $message);
    } while ($oldmessage != $message);
    do {
        $oldmessage = $message;
        $message = str_replace(array("[/size]\n\n", "\n[hr]", "[hr]\n", "\n[list", "[/list]\n", "\n[/", "[list]\n", "[list=1]\n", "\n[*]"), array("[/size]\n", "[hr]", "[hr]", "[list", "[/list]", "[/", "[list]", "[list=1]", "[*]"), $message);
    } while ($message != $oldmessage);
    $message = str_replace(array('[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'), array('[b]', '[/b]', '[i]', '[/i]'), $message);
    // Handling Yahoo style of mails
    $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
    return trim($message);
}
Exemplo n.º 2
0
function html2plain($html, $wraplength = 75, $compact = false)
{
    global $lang;
    $message = str_replace("\r", "", $html);
    $doc = new DOMDocument();
    $doc->preserveWhiteSpace = false;
    $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
    @$doc->loadHTML($message);
    $xpath = new DomXPath($doc);
    $list = $xpath->query("//pre");
    foreach ($list as $node) {
        $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
    }
    $message = $doc->saveHTML();
    $message = str_replace(array("\n<", ">\n", "\r", "\n", " "), array("<", ">", "<br>", " ", ""), $message);
    $message = preg_replace('= [\\s]*=i', " ", $message);
    // Collecting all links
    $urls = collecturls($message);
    @$doc->loadHTML($message);
    node2bbcode($doc, 'html', array(), '', '');
    node2bbcode($doc, 'body', array(), '', '');
    // MyBB-Auszeichnungen
    /*
    node2bbcode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
    node2bbcode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
    node2bbcode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
    
    node2bbcode($doc, 'strong', array(), '*', '*');
    node2bbcode($doc, 'b', array(), '*', '*');
    node2bbcode($doc, 'i', array(), '/', '/');
    node2bbcode($doc, 'u', array(), '_', '_');
    */
    if ($compact) {
        node2bbcode($doc, 'blockquote', array(), "»", "«");
    } else {
        node2bbcode($doc, 'blockquote', array(), '[quote]', "[/quote]\n");
    }
    node2bbcode($doc, 'br', array(), "\n", '');
    node2bbcode($doc, 'span', array(), "", "");
    node2bbcode($doc, 'pre', array(), "", "");
    node2bbcode($doc, 'div', array(), "\r", "\r");
    node2bbcode($doc, 'p', array(), "\n", "\n");
    //node2bbcode($doc, 'ul', array(), "\n[list]", "[/list]\n");
    //node2bbcode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
    node2bbcode($doc, 'li', array(), "\n* ", "\n");
    node2bbcode($doc, 'hr', array(), "\n" . str_repeat("-", 70) . "\n", "");
    node2bbcode($doc, 'tr', array(), "\n", "");
    node2bbcode($doc, 'td', array(), "\t", "");
    node2bbcode($doc, 'h1', array(), "\n\n*", "*\n");
    node2bbcode($doc, 'h2', array(), "\n\n*", "*\n");
    node2bbcode($doc, 'h3', array(), "\n\n*", "*\n");
    node2bbcode($doc, 'h4', array(), "\n\n*", "*\n");
    node2bbcode($doc, 'h5', array(), "\n\n*", "*\n");
    node2bbcode($doc, 'h6', array(), "\n\n*", "*\n");
    // Problem: there is no reliable way to detect if it is a link to a tag or profile
    //node2bbcode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', '', true);
    node2bbcode($doc, 'a', array('href' => '/(.+)/', 'rel' => 'oembed'), ' $1 ', '', true);
    //node2bbcode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
    //node2bbcode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
    //node2bbcode($doc, 'img', array(), '', '');
    if (!$compact) {
        node2bbcode($doc, 'img', array('src' => '/(.+)/'), '[img]$1', '[/img]');
    } else {
        node2bbcode($doc, 'img', array('src' => '/(.+)/'), '', '');
    }
    node2bbcode($doc, 'iframe', array('src' => '/(.+)/'), ' $1 ', '', true);
    $message = $doc->saveHTML();
    if (!$compact) {
        $message = str_replace("[img]", "", $message);
        $message = str_replace("[/img]", "", $message);
    }
    // was ersetze ich da?
    // Irgendein stoerrisches UTF-Zeug
    $message = str_replace(chr(194) . chr(160), ' ', $message);
    $message = str_replace("&nbsp;", " ", $message);
    // Aufeinanderfolgende DIVs
    $message = preg_replace('=\\r *\\r=i', "\n", $message);
    $message = str_replace("\r", "\n", $message);
    $message = strip_tags($message);
    $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
    if (!$compact) {
        $counter = 1;
        foreach ($urls as $id => $url) {
            if (strpos($message, $url) == false) {
                $message .= "\n" . $url . " ";
            }
        }
        //$message .= "\n[".($counter++)."] ".$url;
    }
    do {
        $oldmessage = $message;
        $message = str_replace("\n\n\n", "\n\n", $message);
    } while ($oldmessage != $message);
    $message = quotelevel(trim($message), $wraplength);
    return trim($message);
}