最近采集写的一个超简单实用的HTML解析类

$xp = new xf_HtmlDom(); $xp->loadHtml('http://dealer.bitauto.com/100040078/cars.html'); $rows = $xp->find('dl/dd/a', 0)->innertext; print_r($rows);

1. $oldSetting = libxml_use_internal_errors( true );
2. libxml_clear_errors();
3. /**
4. *
5. * -+-----------------------------------
6. * |PHP5 Framework - 2011
7. * |Web Site: www.iblue.cc
8. * |E-mail: [e-mail]
9. * |Date: 2012-10-12
10. * -+-----------------------------------
11. *
12. * @desc HTML解析器
13. * @author jingke
14. */
15. class XF_HtmlDom
16. {
17. private $_xpath = null;
18. private $_nodePath = '';
20. public function __construct($xpath = null, $nodePath = '')
21. {
22. $this->_xpath = $xpath;
23. $this->_nodePath = $nodePath;
24. }
26. public function loadHtml($url)
27. {
28. ini_set('user_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');
29. $content = '';
30. if(strpos(strtolower($url), 'http')===false)
31. {
32. $content = file_get_contents($url);
33. }
34. else
35. {
36. $ch = curl_init();
37. $user_agent = "Baiduspider+(+http://www.baidu.com/search/spider.htm)";
38. $user_agent1='Mozilla/5.0 (Windows NT 5.1; rv:6.0) Gecko/20100101 Firefox/6.0';
39. curl_setopt($ch, CURLOPT_URL, $url);
40. curl_setopt($ch, CURLOPT_HEADER, false);
41. curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
42. curl_setopt($ch, CURLOPT_REFERER, $url);
43. curl_setopt($ch, CURLOPT_USERAGENT, $user_agent1);
44. curl_setopt($ch, CURLOPT_FOLLOWLOCATION,1);
45. $content =curl_exec($ch);
46. curl_close($ch);
47. }
49. $html = new DOMDocument();
50. $html->loadHtml($content);
51. $this->_xpath = new DOMXPath( $html );
52. return $this;
54. }
56. public function find($query, $index = null)
57. {
58. if($this->_nodePath == '')
59. $this->_nodePath = '//';
60. else
61. $this->_nodePath .= '/';
63. $nodes = $this->_xpath->query($this->_nodePath.$query);
64. if ($index == null && !is_numeric($index))
65. {
66. $tmp = array();
67. foreach ($nodes as $node)
68. {
69. $tmp[] = new XF_HtmlDom($this->_xpath, $node->getNodePath());
70. }
71. return $tmp;
72. }
73. return new XF_HtmlDom($this->_xpath,$this->_xpath->query($this->_nodePath.$query)->item($index)->getNodePath());
74. }
76. /**
77. * 获取内容
78. */
79. public function text()
80. {
81. if ($this->_nodePath != '' && $this->_xpath != null )
82. return $this->_xpath->query($this->_nodePath)->item(0)->textContent;
83. else
84. return false;
85. }
87. /**
88. * 获取属性值
89. */
90. public function getAttribute($name)
91. {
92. if ($this->_nodePath != '' && $this->_xpath != null )
93. return $this->_xpath->query($this->_nodePath)->item(0)->getAttribute($name);
94. else
95. return false;
96. }
98. public function __get($name)
99. {
100. if($name == 'innertext')
101. return $this->text();
102. else
103. return $this->getAttribute($name);
104. }
106. }

复制代码