分享下页面关键字抓取www.icbase.com站点代码(带asp.net参数的)_php实例

/**
 * HOST: www.icbase.com
 */
//set_time_limit(0);
// base function
function curl_get($url, $data = array(), $header = array(), $timeout = 15, $port = 80, $reffer = '', $proxy = '')
{
 $ch = curl_init();
 if (!empty($data)) {
 $data = is_array($data)?http_build_query($data): $data;
 $url .= (strpos($url,'?')? '&': "?") . $data;
 }
 curl_setopt($ch, CURLOPT_URL, $url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
 curl_setopt($ch, CURLOPT_POST, 0);
 curl_setopt($ch, CURLOPT_PORT, $port);
 curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
 curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1); //是否抓取跳转后的页面
 $reffer && curl_setopt($ch, CURLOPT_REFERER, $reffer);
 if($proxy) {
 curl_setopt($ch, CURLOPT_PROXY, $proxy);
 curl_setopt($ch, CURLOPT_PROXYPORT, 1723);
 curl_setopt($ch, CURLOPT_PROXYUSERPWD,"andhm001:andhm123");
 }

$result = array();
 $result['result'] = curl_exec($ch);
 if (0 != curl_errno($ch)) {
 $result['error'] = "Error:\n" . curl_error($ch);

}
 curl_close($ch);
 return $result;
}

function curl_post($url, $data = array(), $header = array(), $timeout = 5, $port = 80)
{
 $ch = curl_init();
 curl_setopt($ch, CURLOPT_URL, $url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
 //curl_setopt($ch, CURLOPT_PORT, $port);
 !empty ($header) && curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
 curl_setopt($ch, CURLOPT_POST, 1);
 curl_setopt($ch, CURLOPT_POSTFIELDS, $data);

$result = array();
 $result['result'] = curl_exec($ch);
 if (0 != curl_errno($ch)) {
 $result['error'] = "Error:\n" . curl_error($ch);

}
 curl_close($ch);

return $result;
}

/**
 * 获取列表页的html源码
 * @param string $keywords 搜索关键字
 * @param int $page 页数
 * @return boolean|array
 */
function getListHtml($keywords, $page=1)
{
 if ($page  {
 return false;
 }
 $page = $page == 0 ? 1 : intval($page);
 if ($page == 1)
 {
 $result = curl_get('http://www.icbase.com/ProResult.aspx', array('ProKey' => $keywords));
 if ( isset($result['error']) )
 {
 return false;
 //exit($result['error']);
 }
 $result = $result['result'];

 // asp.net post提交数据
 if(! defined('__VIEWSTATE') && preg_match('/ {
 define('__VIEWSTATE', $matches[1]);
 } else {
 return false;
 }

 if(! defined('__PREVIOUSPAGE') && preg_match('/ {
 define('__PREVIOUSPAGE', $matches[1]);
 } else {
 return false;
 }

 if(! defined('__EVENTVALIDATION') && preg_match('/ {
 define('__EVENTVALIDATION', $matches[1]);
 } else {
 return false;
 }

 return $result;
 }
 $data = array(
 '__EVENTTARGET' => 'pager',
 '__EVENTARGUMENT' => $page,
 '__VIEWSTATE' => __VIEWSTATE,
 '__PREVIOUSPAGE' => __PREVIOUSPAGE,
 '__EVENTVALIDATION' => __EVENTVALIDATION,
 );
 $result = curl_post('http://www.icbase.com/ProResult.aspx?ProKey=' . $keywords, $data);
 if ( isset($result['error']) )
 {
 return false;
 //exit($result['error']);
 }
 $result = $result['result'];
 return $result;
}

/**
 * 获取列表页 a链接的url
 * @param string $html html源码
 * @return array
 */
function getListHref($html)
{
 $pattern = '/[\s\n]*]\/>/isU';
 if (preg_match_all($pattern, $html, $matches))
 {
 return $matches[1];
 } else {
 // 没有匹配项
 return array();
 }
}

/**
 * 获取下一页数字
 * @param string $html html源码
 * @return number
 */
function getListNextPage($html)
{
 $pattern = '/