Регулярка на рушку, кто правил?

[Dr_Brown]

Кто может к обычному парсеру поделиться регуляркой, перестало брать заголовок и торрент файл, описание берет нормально.

 

[Virtuoz]

Заголовок

preg_match_all ('#data-share_title="([\s\S]*?)">#', $text, $source, PREG_SET_ORDER);

 

[Dr_Brown]

Virtuoz спасибо

 

[vitalix]

Заменяем тут

function rutracker($text, $mode=false)
{
if($mode == 'title')
{
//новая регулярка для заголовка рутрекер
preg_match_all ('#data-share_title="([\s\S]*?)">#', $text, $source, PREG_SET_ORDER);
// конец новой регулярки
//preg_match_all ('#<h1 class="maintitle"><a.*?href=".*?">([\s\S]*?)</a></h1>#', $text, $source, PREG_SET_ORDER);

 

[Dr_Brown]

на торрент


preg_match_all ('#<a href="dl.php\?t=(.*?)" class#', $text, $source, PREG_SET_ORDER);


 

[vitalix]

меняем тут

elseif($mode == 'torrent')
{
// новая регулярка для торрент файла рутрекер
preg_match_all ('#<a href="dl.php\?t=(.*?)" class#', $text, $source, PREG_SET_ORDER);
// конец новой регулярки
//preg_match_all ('#<a href="

Please Login or Register to view hidden text.

" class#', $text, $source, PREG_SET_ORDER);
$text = $source[0][1];

 

[vitalix]

опять помер. только заголовок берёт - есть кто правил ?

 

[dexer7]

vitalix
Меняем это

        $pos = strpos($content, '<p style="padding-top: 6px;"');

на это

  $pos = strpos($content, '<span id="tor-status-resp"');

и это

 $pos = strpos($text, '<div class="clear"></div>');

на это

 $pos = strpos($text, '<div class="clear" style="height: 8px;"></div>');

 

[vitalix]

vitalix
Меняем это

        $pos = strpos($content, '<p style="padding-top: 6px;"');

на это

  $pos = strpos($content, '<span id="tor-status-resp"');

и это

 $pos = strpos($text, '<div class="clear"></div>');

на это

 $pos = strpos($text, '<div class="clear" style="height: 8px;"></div>');

я про ручной парсер.
этого нет в нём (

Спойлер: тут

if($tracker == 'rutracker')
    {
        $snoopy->host = "rutracker.org";
        $snoopy->agent = "opera";
        $snoopy->rawheaders["Pragma"] = "no-cache";

        $snoopy->_submit_method = "POST";

        $submit_vars = array (
            'login_username' => $auth['rutracker']['login'],
            'login_password' => $auth['rutracker']['pass'],
            'login'          => true,
        );
        $snoopy->submit("http://rutracker.org/forum/login.php", $submit_vars);

        $snoopy->_httpmethod = "GET";

        $snoopy->fetch($url);
        $content = iconv('windows-1251', 'UTF-8', $snoopy->results);
        $pos = strpos($content, 'humn_size');
        $content = substr($content, 0, $pos);

        if (!$content)
        {
            meta_refresh('release.php', '2');
            bb_die('false content');
        }

        function rutracker($text, $mode=false)
    {
       if($mode == 'title')
       {
           //нова¤ регул¤рка дл¤ заголовка рутрекер
         preg_match_all ('#data-share_title="([\s\S]*?)">#', $text, $source, PREG_SET_ORDER);
         // конец новой регул¤рки
        //preg_match_all ('#<h1 class="maintitle"><a.*?href=".*?">([\s\S]*?)</a></h1>#', $text, $source, PREG_SET_ORDER);
        $text = @$source[0][1];
        $text = str_replace('<wbr>', '', $text);
        }
            elseif($mode == 'torrent')
            {
                // нова¤ регул¤рка дл¤ торрент файла рутрекер
                preg_match_all ('#<a href="dl.php\?t=(.*?)" class#', $text, $source, PREG_SET_ORDER);
                // конец новой регул¤рки
                //preg_match_all ('#<a href="http://dl.rutracker.org/forum/dl.php\?t=(.*?)" class#', $text, $source, PREG_SET_ORDER);
                $text = $source[0][1];
            }
            else
            {
                $pos = strpos($text, '<div class="post_body"');
                $text = substr($text, $pos);
                $pos = strpos($text, '<div class="clear"></div>');
                $text = substr($text, 0, $pos);
                $text = preg_replace('/<div class="post_body" id=".*?">/', '', $text);

                $text = preg_replace('#<img class="smile" src=".*?" align="absmiddle" border="0" />#', '', $text);
                $text = preg_replace('#\n<h3 class="sp-title">.*?</h3>#', '', $text);
                $text = str_replace('<div class="q-wrap">', '', $text);
                $text = str_replace('<div class="sp-wrap">', '', $text);
                $text = str_replace('<div class="c-wrap">', '', $text);
                $text = preg_replace('#<a name=".*?"></a>#', '', $text);
                $text = preg_replace('#<a class="postLink-name" href=".*?">([\s\S]*?)</a>#', '$1', $text);
                $text = str_replace('<span class="post-hr">-</span>', "\n[hr]\n", $text);
                $text = preg_replace('#<pre class="post-pre">([\s\S]*?)</pre>#', '$1', $text);
                $text = preg_replace('#<div style="margin-.*?">([\s\S]*?)</div>#', '$1', $text);
                $text = preg_replace('/<var class="postImg" title="(.*?)">&#10;<\/var>/', '[img]$1[/img]', $text);
                $text = preg_replace('/<var class="postImg postImgAligned img-(.*?)" title="(.*?)">&#10;<\/var>/', "[img=\\1]\\2[/img]\n", $text);
                $text = str_replace('<span class="post-br"><br /></span>', "\n\n", $text);

                $text = preg_replace('/<br.*?>/', '', $text);
                $text = preg_replace('/<ul type="(.*?)">/', '[list=$1]', $text);
                $text = str_replace('<ul>', '[list]', $text);
                $text = str_replace('</ul>', '[/list]', $text);
                $text = str_replace('<li>', "\n[*]", $text);
                $text = str_replace('</li>', '', $text);

                $text = str_replace('<pre>', '[pre]', $text);
                $text = str_replace('</pre>', '[/pre]', $text);

                for ($i=0; $i<=20; $i++)
                {
                    $text = preg_replace('/<span class="post-b">([^<]*?)<(?=\/)\/span>/', '[b]$1[/b]', $text);
                    $text = preg_replace('/<span class="post-u">([^<]*?)<(?=\/)\/span>/', '[u]$1[/u]', $text);
                    $text = preg_replace('/<span class="post-i">([^<]*?)<(?=\/)\/span>/', '[i]$1[/i]', $text);
                    $text = preg_replace('/<span class="post-s">([^<]*?)<(?=\/)\/span>/', '[s]$1[/s]', $text);
                    $text = preg_replace('/<span style="font-size: ([^<]*?)px; line-height: normal;">([^<]*?)<(?=\/)\/span>/', "[size=\\1]\\2[/size]", $text);
                    $text = preg_replace('/<span style="font-family: ([^<]*?);">([^<]*?)<(?=\/)\/span>/', "[font=\"\\1\"]\\2[/font]", $text);
                    //$text = preg_replace('#<a href="http://www.imdb.com/title/(.*?)/" class=".*?">.*?</a>#', "[imdb]http://www.imdb.com/title/$1[/imdb]", $text);
                    //$text = preg_replace('#<a href="http://www.kinopoisk.ru/film/(.*?)/" class=".*?">.*?</a>#', "[kp]http://www.kinopoisk.ru/film/$1[/kp]", $text);
                    //$text = preg_replace('#<a href="http://www.kinopoisk.ru/level/(.*?)/" class=".*?">.*?</a>#', "[kp]http://www.kinopoisk.ru/level/$1[/kp]", $text);
                    $text = preg_replace('/<span class="post-align" style="text-align: ([^<]*?);">([^<]*?)<(?=\/)\/span>/', "[align=\\1]\n\\2\n[/align]", $text);
                    $text = preg_replace('/<span class="p-color" style="color: ([^<]*?);">([^<]*?)<(?=\/)\/span>/', '[color=$1]$2[/color]', $text);
                    $text = preg_replace('#<a href="http://rutracker.org.*?".*?>.*?</a>#', '', $text);
                    $text = preg_replace('#<a href="http://www.rutracker.org.*?".*?>.*?</a>#', '', $text);
                    $text = preg_replace('#<a href="http://nnmclub.to.*?".*?>.*?</a>#', '', $text);
                    $text = preg_replace('#<a href="http://www.nnmclub.to.*?".*?>.*?</a>#', '', $text);
                    $text = preg_replace('/<a href="([^<]*?)" class="postLink">([^<]*?)<(?=\/)\/a>/', '[url=$1]$2[/url]', $text);
                    $text = preg_replace('/<div class="sp-head folded">([^<]*?)<\/div>([^<]*?)<(?=\/)\/div>([\s\S]*?)<div class="sp-body">([^<]*?)<(?=\/)\/div>([\s\S]*?)<([^<]*?)\/div>/', "[spoiler=\"\\1\"]\n\\2\n[/spoiler]", $text);
                
                    $text = preg_replace('/<div class="q-head">([^<]*?)<\/div>([^<]*?)<(?=\/)\/div>([\s\S]*?)<div class="q">([^<]*?)<(?=\/)\/div>([^<]*?)<([^<]*?)\/div>/', "[quote=\"\\1\"]\n\\2\n[/quote]", $text);
                    $text = preg_replace('/<div class="q-head">(.*?)<\/div>([\s\S]*?)<([^<]*?)\/div>([^<]*?)<([^<]*?)\/div>/', "[quote=\"\\1\"]\n\\2\n[/quote]", $text);
                    $text = preg_replace('/<div class="c-head">[\s\S]*?<div class="c-body">([^<]*?)<(?=\/)\/div>([\s\S]*?)<([^<]*?)\/div>/', "[code]\n\\1\n

", $text);
$text = preg_replace('/<div style="text-align[^<]*?)">([^<]*?)<(?=\/)\/div>/', '<div>[align=$1]$2[/align]</div>', $text);
}

$text = preg_replace('/<div class="sp-head folded">(.*?)<\/div>([\s\S]*?)<([^<]*?)\/div>([^<]*?)<([^<]*?)\/div>/', "

Спойлер: \"\\1\"

\n\\2\n

", $text);

$text = trim(strip_tags($text));
$text = str_replace('&#039;', "'", $text);
$text = str_replace('&nbsp;', ' ', $text);
$text = str_replace('&gt;', '>', $text);
$text = str_replace('&lt;', '<', $text);
$text = html_entity_decode($text);
}
return $text;
}

if ($message = rutracker($content))
{
$id = rutracker($content, 'torrent');;

$snoopy->submit("

Please Login or Register to view hidden text.

". $id);
$torrent = $snoopy->results;
$tor = bdecode($torrent);
if(count($tor))
{
$new_name = md5($torrent);
$file = fopen("$attach_dir/$new_name.torrent", 'w');
fputs($file, $torrent);
fclose($file);

$hidden_form_fields .= '<input type="hidden" name="add_attachment_body" value="0" />';
$hidden_form_fields .= '<input type="hidden" name="posted_attachments_body" value="0" />';
$hidden_form_fields .= '<input type="hidden" name="attachment_list[]" value="'. $attach_dir .'/'. $new_name .'.torrent" />';
$hidden_form_fields .= '<input type="hidden" name="filename_list[]" value="torrent" />';
$hidden_form_fields .= '<input type="hidden" name="extension_list[]" value="torrent" />';
$hidden_form_fields .= '<input type="hidden" name="mimetype_list[]" value="application/x-bittorrent" />';
$hidden_form_fields .= '<input type="hidden" name="filesize_list[]" value="'. filesize("$attach_dir/$new_name.torrent") .'" />';
$hidden_form_fields .= '<input type="hidden" name="filetime_list[]" value="'. TIMENOW .'" />';
$hidden_form_fields .= '<input type="hidden" name="attach_id_list[]" value="" />';
$hidden_form_fields .= '<input type="hidden" name="attach_thumbnail_list[]" value="0" />';

}
}
$subject = rutracker($content, 'title');
}[/CODE]

 

[dexer7]

vitalix
Походу это

 $pos = strpos($content, 'humn_size');

на

 $pos = strpos($content, '<span id="tor-status-resp"');

и опять же здесь

$pos = strpos($text, '<div class="clear"></div>');

на это

$pos = strpos($text, '<div class="clear" style="height: 8px;"></div>');

Пробуем.