当前位置：Gxlcms > PHP教程 > 解析HTML标签，并实现快速查找节点，获取节点信息

解析HTML标签，并实现快速查找节点，获取节点信息

时间：2021-07-01 10:21:17 帮助过：25人阅读

详细介绍和使用请点击源码出处。
  1. /**
  2. * html标签解析包
  3. *
  4. * @category TagParse
  5. * @package TagParse
  6. * @author kun
  7. * @copyright 2014 kun
  8. * @license http://www.php.com/license/3_01.txt PHP License 3.01
  9. * @version 1.0
  10. * @link http://www.blogkun.com
  11. * @since 1.0
  12. */
  13. namespace TagParse;
  15. /**
  16. * TagDomRoot
  17. *
  18. * @category TagParse
  19. * @package TagParse
  20. * @author kun
  21. * @copyright 2014 kun
  22. * @license http://www.php.com/license/3_01.txt PHP License 3.01
  23. * @version 1.0
  24. * @link http://www.blogkun.com
  25. * @since 1.0
  26. */
  27. class TagDomRoot
  28. {
  29. public $tag = 'root';
  30. public $plaintext;
  31. public $child = array();
  32. public $level = 0;
  33. public static $TagParseError = false;
  34. protected static $TagSet = array();
  35. protected static $FoundNode = array();
  36. public static $ErrorTag = array();
  38. /**
  39. * initProperty
  40. *
  41. * @access public
  42. *
  43. * @return null
  44. */
  45. public function initProperty()
  46. {
  47. $TagParseError = false;
  48. $TagSet = array();
  49. $FoundNode = array();
  50. $DumpScriptCode = array();
  51. $ErrorTag = array();
  52. }
  54. /**
  55. * __construct
  56. *
  57. * @param string $str The tag string to be parse.
  58. *
  59. * @access public
  60. *
  61. * @return TagDomRoot
  62. */
  63. public function __construct($str)
  64. {
  65. $this->_removeNoise($str);
  66. if ($str === null) {
  67. self::$TagParseError = true;
  68. } else {
  69. $l = strpos($str, '<');
  70. if ($l !== false) {
  71. $this->plaintext = substr($str, 0, $l);
  72. }
  73. $res = preg_match_all('~>(.*?)<~s', $str, $matches);
  74. if ($res !== false && $res > 0) {
  75. $this->plaintext .= implode($matches[1]);
  76. }
  77. $r = strrpos($str, '>');
  78. if ($r !== false) {
  79. $this->plaintext .= substr($str, $r+1);
  80. }
  82. $tagCollect = array();
  83. $attrCollect = array();
  84. $innerContentCollect = array();
  86. if ($this->parseTag($str, $tagCollect, $attrCollect, $innerContentCollect) === false) {
  87. self::$TagParseError = true;
  88. }
  90. foreach ($tagCollect as $index => $tag) {
  91. $this->child[] = new TagDomNode($tag, $this, $attrCollect[$index], $innerContentCollect[$index], $this->level+1);
  92. }
  93. }
  94. }
  96. /**
  97. * parseTag
  98. *
  99. * @param mixed $str Description.
  100. * @param mixed &$tagCollect Description.
  101. * @param mixed &$attrCollect Description.
  102. * @param mixed &$innerContentCollect Description.
  103. *
  104. * @access protected
  105. *
  106. * @return boolean Value.
  107. */
  108. protected function parseTag($str, array &$tagCollect, array &$attrCollect, array &$innerContentCollect)
  109. {
  110. $selfClosingTags = array('img' => 1, 'br' => 1, 'input' => 1, 'meta' => 1, 'link' => 1, 'hr' => 1, 'base' => 1, 'embed' => 1, 'spacer' => 1);
  111. $end = -2;
  112. $close = 0;
  113. $error = false;
  114. $tag = '';
  115. while (true) {
  116. $l = strpos($str, '<', $end+strlen($tag)+2);
  117. if ($l === false) {//parse end
  118. break;
  119. }
  120. if (strpos(substr($str, $l, 2), '/') !== false) {//surplus closing tag,discard
  121. $error = true;
  122. $end = $l+strlen($tag);
  123. self::$ErrorTag[] = substr($str, $l, strpos($str, '>', $l)-$l+1);
  124. continue;
  125. }
  127. $r = strpos($str, '>', $l);
  128. $tag = substr($str, $l+1, $r-$l-1);
  129. if (!ctype_alpha($tag[0]) || strpos($tag, '<') !== false) {
  130. $end = $r + 1;
  131. continue;
  132. }
  133. $tag = preg_replace("~\n+~", ' ', $tag);
  134. $space = strpos($tag, ' ');
  135. if ($space !== false) {
  136. $attrCollect[] = substr($tag, $space+1);
  137. $tag = substr($tag, 0, $space);
  138. } else {
  139. $attrCollect[] = '';
  140. }
  141. $tagCollect[] = $tag;
  142. if (isset($selfClosingTags[$tag])) {
  143. $innerContentCollect[] = '';
  144. $end = $r-strlen($tag)-2;
  145. $close = $r+1;
  146. continue;
  147. }
  149. $countOpen = -1;
  150. $open = strpos($str, '<'.$tag, $close);
  151. $close = strpos($str, '', $open);
  152. if ($close === false) {//surplus opening tag
  153. $innerContentCollect[] = substr($str, $r+1);
  154. $error = true;
  155. self::$ErrorTag[] = '<'.$tag.'>';
  156. break;
  157. }
  158. $start = $open;
  159. while ($open < $close && $open !== false) {
  160. $countOpen++;
  161. $open = strpos($str, '<'.$tag, $open+strlen($tag));
  162. }
  164. while ($countOpen > 0 && $close !== false) {
  165. $open = strpos($str, '<'.$tag, $close+strlen($tag)+3);
  166. $close = strpos($str, '', $close+strlen($tag)+3);
  167. if ($close === false) {
  168. break;
  169. }
  170. $countOpen--;
  171. while ($open < $close && $open !== false) {
  172. $open = strpos($str, '<'.$tag, $open+strlen($tag)+3);
  173. $countOpen++;
  174. }
  175. }
  176. if ($close === false) {//标签闭合不配对
  177. $innerContentCollect[] = substr($str, $r+1);
  178. $error = true;
  179. break;
  180. }
  181. $end = $close;
  182. $r = strpos($str, '>', $start);
  183. $innerContentCollect[] = substr($str, $r+1, $end - $r - 1);
  184. }
  186. return !$error;
  187. }
  189. /**
  190. * _removeNoise
  191. *
  192. * @param string &$str The tag string to be parse.
  193. *
  194. * @access private
  195. *
  196. * @return string
  197. */
  198. private function _removeNoise(&$str)
  199. {
  200. $str = preg_replace('~~is', '', $str);
  201. $str = preg_replace('~~is', '', $str);
  202. $str = preg_replace('~~is', '', $str);
  203. }
  205. /**
  206. * parseSelectors
  207. *
  208. * @param string $selectors user's select condition.
  209. * @param array &$selectorsTag tags
  210. * @param array &$selectorsAttr attributes
  211. *
  212. * @access protected
  213. *
  214. * @return null
  215. */
  216. protected function parseSelectors($selectors, array &$selectorsTag, array &$selectorsAttr)
  217. {
  218. preg_match_all('~([\w\d]+)(\[[\w\d -="._/]+\])?~', $selectors, $matches);
  219. $selectorsTag = $matches[1];
  220. foreach ($matches[2] as $key => $value) {
  221. $selectorsAttr[$key] = array();
  222. if ($value !== '') {
  223. preg_match_all('~([\w\d-]+)="([\w\d-. _/]+)"~', $value, $matches);
  224. foreach ($matches[1] as $index => $attr) {
  225. $selectorsAttr[$key][$attr] = $matches[2][$index];
  226. }
  227. }
  228. }
  229. }
  231. /**
  232. * find
  233. *
  234. * @param mixed $selectors user's select condition.
  235. * @param array $selectorsTag tags.
  236. * @param array $selectorsAttr attributes.
  237. *
  238. * @access public
  239. *
  240. * @return array
  241. */
  242. public function find($selectors, $selectorsTag = array(), $selectorsAttr = array())
  243. {
  244. if ($selectors !== null) {
  245. $this->parseSelectors($selectors, $selectorsTag, $selectorsAttr);
  246. }
  247. var_dump($selectorsTag, $selectorsAttr);exit();
  248. if (!empty($selectorsTag)) {
  249. $this->seek($selectorsTag, $selectorsAttr);
  250. foreach ($this->child as $key => $node) {
  251. $node->find(null, $selectorsTag, $selectorsAttr);
  252. }
  253. }
  255. if ($selectors !== null) {
  256. $res = self::$FoundNode;
  257. self::$FoundNode = array();
  259. return $res;
  260. }
  261. }
  263. /**
  264. * findGlobal
  265. *
  266. * @param string $selectors user's select condition.
  267. *
  268. * @access public
  269. *
  270. * @return array
  271. */
  272. public function findGlobal($selectors)
  273. {
  274. $space = strpos($selectors, ' ', strpos($selectors, ']'));
  275. if ($space === false) {
  276. return $this->findOneGlobal($selectors);
  277. } else {
  278. $selectorsAttr = array();
  279. $selectorsTag = array();
  280. $this->findOneGlobal(substr($selectors, 0, $space), false);
  281. $this->parseSelectors(substr($selectors, $space + 1), $selectorsTag, $selectorsAttr);
  282. if (!empty(self::$FoundNode) && !empty($selectorsTag)) {
  283. $nodes = self::$FoundNode;
  284. self::$FoundNode = array();
  285. foreach ($nodes as $key => $node) {
  286. $node->seek($selectorsTag, $selectorsAttr);
  287. }
  288. }
  291. }
  293. $res = self::$FoundNode;
  294. self::$FoundNode = array();
  296. return $res;
  297. }
  299. /**
  300. * seek
  301. *
  302. * @param array $selectorsTag tags.
  303. * @param array $selectorsAttr attributes.
  304. *
  305. * @access protected
  306. *
  307. * @return null
  308. */
  309. protected function seek($selectorsTag, $selectorsAttr)
  310. {
  311. foreach ($this->child as $key => $node) {
  312. $isFind = true;
  313. if ($node->tag === $selectorsTag[0]) {
  314. foreach ($selectorsAttr[0] as $attrName => $value) {
  316. if (isset($node->attr[$attrName])
  317. && (preg_match('~.*? '.$value.' .*?~', $node->attr[$attrName]) > 0
  318. || preg_match('~^'.$value.'$~', $node->attr[$attrName]) > 0
  319. || preg_match('~^'.$value.' ~', $node->attr[$attrName]) > 0
  320. || preg_match('~ '.$value.'$~', $node->attr[$attrName]) > 0)
  321. ) {
  322. continue;
  323. } else {
  324. $isFind = false;
  325. break;
  326. }
  327. }
  328. } else {
  329. $isFind = false;
  330. }
  331. if ($isFind) {
  332. if (count($selectorsTag) === 1) {
  333. self::$FoundNode[] = $node;
  334. } else {
  335. $node->seek(
  336. array_slice($selectorsTag, 1),
  337. array_slice($selectorsAttr, 1)
  338. );
  339. }
  340. }
  341. }
  342. }
  344. /**
  345. * findOneGlobal
  346. *
  347. * @param string $selector user's select condition.
  348. * @param bool $isReturn weather return value.
  349. *
  350. * @access public
  351. *
  352. * @return array
  353. */
  354. public function findOneGlobal($selector, $isReturn = true)
  355. {
  356. preg_match('~([\w\d]+)(\[[\w\d -="._/]+\])?~', $selector, $matches);
  357. $tag = $matches[1];
  358. $attr = array();
  359. if (isset($matches[2])) {
  360. preg_match_all('~([\w\d-]+)="([\w\d-. _/]+)"~', $matches[2], $matches);
  361. foreach ($matches[1] as $key => $value) {
  362. $attr[$value] = $matches[2][$key];
  363. }
  364. }
  365. if (isset(self::$TagSet[$tag])) {
  366. foreach (self::$TagSet[$tag] as $attrValue => $nodeArray) {
  367. $isFind = true;
  368. foreach ($attr as $attrName => $value) {
  369. if (preg_match('~'.$attrName.'=".*? '.$value.' .*?"~', $attrValue)
  370. || preg_match('~'.$attrName.'="'.$value.' .*?"~', $attrValue)
  371. || preg_match('~'.$attrName.'=".*? '.$value.'"~', $attrValue)
  372. || preg_match('~'.$attrName.'="'.$value.'"~', $attrValue)
  373. ) {
  374. continue;
  375. } else {
  376. $isFind = false;
  377. break;
  378. }
  379. }
  380. if ($isFind) {
  381. foreach ($nodeArray as $key => $node) {
  382. self::$FoundNode[] = $node;
  383. }
  384. }
  385. }
  386. }
  387. if ($isReturn) {
  388. $res = self::$FoundNode;
  389. self::$FoundNode = array();
  391. return $res;
  392. }
  393. }
  394. }
  396. /**
  397. * TagDomNode
  398. *
  399. * @uses TagDomRoot
  400. *
  401. * @category TagParse
  402. * @package TagParse
  403. * @author kun
  404. * @copyright 2014 kun
  405. * @license http://www.php.com/license/3_01.txt PHP License 3.01
  406. * @version 1.0
  407. * @link http://www.blogkun.com
  408. * @since 1.0
  409. */
  410. class TagDomNode extends TagDomRoot
  411. {
  412. public $attr = array();
  413. public $parent = null;
  415. /**
  416. * __construct
  417. *
  418. * @param mixed $tag tag.
  419. * @param mixed $parent parent node.
  420. * @param mixed $attr attribute.
  421. * @param mixed $innerContent tag content.
  422. * @param mixed $level node level.
  423. *
  424. * @access public
  425. *
  426. * @return TagDomNode
  427. */
  428. public function __construct($tag, $parent, $attr, $innerContent, $level)
  429. {
  430. $this->tag = $tag;
  431. $this->parent = $parent;
  432. $this->_parseAttr($attr);
  433. $this->level = $level;
  435. $l = strpos($innerContent, '<');
  436. if ($l !== false) {
  437. $this->plaintext = substr($innerContent, 0, $l);
  438. }
  439. $res = preg_match_all('~>(.*?)<~s', $innerContent, $matches);
  440. if ($res !== false && $res > 0) {
  441. $this->plaintext .= implode($matches[1]);
  442. } else {
  443. $this->plaintext .= $innerContent;
  444. }
  445. $r = strrpos($innerContent, '>');
  446. if ($r !== false) {
  447. $this->plaintext .= substr($innerContent, $r+1);
  448. }
  450. $tagCollect = array();
  451. $attrCollect = array();
  452. $innerContentCollect = array();
  454. if ($this->parseTag($innerContent, $tagCollect, $attrCollect, $innerContentCollect) === false) {
  455. self::$TagParseError = true;
  456. }
  458. foreach ($tagCollect as $index => $tag) {
  459. $this->child[] = new TagDomNode($tag, $this, $attrCollect[$index], $innerContentCollect[$index], $this->level+1);
  460. }
  462. if (!isset(self::$TagSet[$this->tag])) {
  463. self::$TagSet[$this->tag] = array();
  464. }
  465. if (!isset(self::$TagSet[$this->tag][$attr])) {
  466. self::$TagSet[$this->tag][$attr] = array();
  467. }
  468. self::$TagSet[$this->tag][$attr][] = &$this;
  469. }
  471. /**
  472. * _parseAttr
  473. *
  474. * @param string $str attribute string.
  475. *
  476. * @access public
  477. *
  478. * @return null
  479. */
  480. private function _parseAttr($str)
  481. {
  482. preg_match_all('~(?[\w-]+)="(?.*?)"~s', $str, $matches);
  483. foreach ($matches['attrName'] as $key => $value) {
  484. $this->attr[$value] = $matches['attrValue'][$key];
  485. }
  486. }
  487. }

人气教程排行

本站所有资源全部来源于网络，若本站发布的内容侵害到您的隐私或者利益，请联系我们删除！