php实现的一个很好用HTML解析器类可用于采集数据

复制代码 代码如下:

<?php

$oldtting = libxml_u_internal_errors( true );

libxml_clear_errors();

/**

*

* -+———————————–

* |php5 framework – 2011

* |web site: www.iblue.cc

* |e-mail: mejinke@gmail.com

* |date: 2012-10-12

* -+———————————–

*

* @desc html解析器

* @author jingke

*/

class xf_htmldom

{

private $_xpath = null;

private $_nodepath = ”;

public function __construct($xpath = null, $nodepath = ”)

{

$this->_xpath = $xpath;

$this->_nodepath = $nodepath;

}

public function loadhtml($url)

{

ini_t(‘ur_agent’, ‘mozilla/5.0 (linux; u; android 2.1; en-us; nexus one build/erd62) applewebkit/530.17 (khtml, like gecko) version/4.0 mobile safari/530.17 –nexus’);

$content = ”;

if(strpos(strtolower($url), ‘http’)===fal)

{腊肉的腌制方法

$content = file_get_contents($url);

}

el

{

$ch = curl_init();山东省简称

$ur_agent = “baiduspider+(+http://www.baidu.com/arch/spider.htm)”;

$ur_agent1=’mozilla/5.0 (windows nt 5.1; rv:6.0) gecko/20100101 firefox/6.0′;

curl_topt($ch, curlopt_url, $url);

curl_topt($ch, curlopt_header, fal);

curl_topt($ch, curlopt_returntransfer, 1);

curl_topt($ch, curlopt_referer, $url);

curl_topt($ch, curlopt_uragent, $ur_agent1);

curl_topt($ch, curlopt_followlocation,1);

$content =curl_exec($ch);

curl_clo($ch);

}

$html = new domdocument();

$html->loadhtml($content);

$this->_xpath = new domxpath( $妈妈发现儿子女友不是人html );

//return $this;

}

p献血日ublic function find($query, $index = null)

{

if($this->_nodepath == ”)

$this->_nodepath = ‘//’;

el

$this->_nodepath .= ‘/’;

$nodes = $this-&g沈阳药科大学t;_xpath->query($this->_nodepath.$query);

//echo $nodes->item(0)->getnodepath();exit;

if ($index == null && !is_numeric($index))

{

$tmp = array();

foreach ($nodes as $node)

{

$tmp[] = new xf_htmldom($this->_xpath, $node->getnodepath());

}

return $tmp;

}

return new xf_htmldom($this->_xpath,$this->_xpath->query($this->_nodepath.$query)->item($index)->getnodepath());

}

/**

* 获取内容

*/

public function text()

{

if ($this->_nodepath != ” && $this->_xpath != null )

return $this->_xpath->query($this->_nodepath)->item(0)->textcontent;

el

return fal;

}

/**

* 获取属性值

*/

public function getattribute($name)

{

if ($this->_nodepath != ” && $this->_xpath != null )

return $this->_xpath->query($this->_nodepath)->item(0)->getattribute($name);

el

return fal;

}

public function __get($name)

{

if($name == ‘innertext’)

return $this->text();

el

return $this->getattribute($name);

}

}

$xp = new xf_htmldom();

$xp->loadhtml(‘http://www.aizhan.com/siteall/www.opendir.cn/’);

$rows = $xp->find(“td[@id=’baidu’]/a”, 0)->innertext;

print_r($rows);