<?php define('IN_ADMINCP', TRUE); define('NOROBOT', TRUE); require_once './include/common.inc.php'; $url = 'http://us.tiandaoedu.com/vi/immigration/607303.html'; $title = '/<divclass="wzlistofh"><pclass="wztityh">(.*?)<\\/p>/'; $content = '/<divclass="wzy_bot">(.*?)<\\/div>/'; $data = getTitleContentByUrl($title, $content, $url, 'gbk'); print_r($data); die; $hrefs = getAllHref($url, $keyword); $hrefs = getHrefByType($hrefs); $hrefs = $hrefs['html']; echo '<pre>'; //print_r($hrefs); foreach ($hrefs as $href) { echo $href; /* if(!MyDomIsList($href)){ $data = getTitleContentByUrl($title,$content,$href); print_r($data); } */ } echo '</pre>'; //print_r($data);
<?php define('IN_ADMINCP', TRUE); define('NOROBOT', TRUE); require_once './include/common.inc.php'; $sql = "SELECT * FROM col_contentrule where rid = 4"; $query = $db->query($sql); $data = $db->fetch_assoc($query); $hrefs = getAllHref($data['listurl'], $data['keyword']); $hrefs = getHrefByType($hrefs); //var_dump($hrefs); foreach ($hrefs['html'] as $href) { if (!MyDomIsList($href)) { echo $href; echo '<br>'; $da = getTitleContentByUrl($data['title'], $data['content'], $href, $data['charset']); print_r($da); } }