Parse HTML tags, quickly find nodes, and obtain node information

WBOY
Release: 2016-07-25 08:48:49
Original
1414 people have browsed it
For detailed introduction and usage, please click on the source code.
  1. /**
  2. * html tag parsing package
  3. *
  4. * @category TagParse
  5. * @package TagParse
  6. * @author kun
  7. * @copyright 2014 kun
  8. * @license http://www.php.net /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;
  14. /**
  15. * TagDomRoot
  16. *
  17. * @category TagParse
  18. * @package TagParse
  19. * @author kun
  20. * @copyright 2014 kun
  21. * @license http://www.php.net/license/3_01.txt PHP License 3.01
  22. * @version 1.0
  23. * @link http://www.blogkun.com
  24. * @since 1.0
  25. */
  26. class TagDomRoot
  27. {
  28. public $tag = 'root';
  29. public $plaintext;
  30. public $child = array();
  31. public $level = 0;
  32. public static $TagParseError = false;
  33. protected static $TagSet = array();
  34. protected static $FoundNode = array();
  35. public static $ErrorTag = array();
  36. /**
  37. * initProperty
  38. *
  39. * @access public
  40. *
  41. * @return null
  42. */
  43. public function initProperty()
  44. {
  45. $TagParseError = false;
  46. $TagSet = array();
  47. $FoundNode = array();
  48. $DumpScriptCode = array();
  49. $ErrorTag = array();
  50. }
  51. /**
  52. * __construct
  53. *
  54. * @param string $str The tag string to be parse.
  55. *
  56. * @access public
  57. *
  58. * @return TagDomRoot
  59. */
  60. public function __construct($str)
  61. {
  62. $this->_removeNoise($str);
  63. if ($str === null) {
  64. self::$TagParseError = true;
  65. } else {
  66. $l = strpos($str, '<');
  67. if ($l !== false) {
  68. $this->plaintext = substr($str, 0, $l);
  69. }
  70. $res = preg_match_all('~>(.*?)<~s', $str, $matches);
  71. if ($res !== false && $res > 0) {
  72. $this->plaintext .= implode($matches[1]);
  73. }
  74. $r = strrpos($str, '>');
  75. if ($r !== false) {
  76. $this->plaintext .= substr($str, $r+1);
  77. }
  78. $tagCollect = array();
  79. $attrCollect = array();
  80. $innerContentCollect = array();
  81. if ($this->parseTag($str, $tagCollect, $attrCollect, $innerContentCollect) === false) {
  82. self::$TagParseError = true;
  83. }
  84. foreach ($tagCollect as $index => $tag) {
  85. $this->child[] = new TagDomNode($tag, $this, $attrCollect[$index], $innerContentCollect[$index], $this->level+1);
  86. }
  87. }
  88. }
  89. /**
  90. * parseTag
  91. *
  92. * @param mixed $str Description.
  93. * @param mixed &$tagCollect Description.
  94. * @param mixed &$attrCollect Description.
  95. * @param mixed &$innerContentCollect Description.
  96. *
  97. * @access protected
  98. *
  99. * @return boolean Value.
  100. */
  101. protected function parseTag($str, array &$tagCollect, array &$attrCollect, array &$innerContentCollect)
  102. {
  103. $selfClosingTags = array('img' => 1, 'br' => 1, 'input' => 1, 'meta' => 1, 'link' => 1, 'hr' => 1, 'base' => 1, 'embed' => 1, 'spacer' => 1);
  104. $end = -2;
  105. $close = 0;
  106. $error = false;
  107. $tag = '';
  108. while (true) {
  109. $l = strpos($str, '<', $end+strlen($tag)+2);
  110. if ($l === false) {//parse end
  111. break;
  112. }
  113. if (strpos(substr($str, $l, 2), '/') !== false) {//surplus closing tag,discard
  114. $error = true;
  115. $end = $l+strlen($tag);
  116. self::$ErrorTag[] = substr($str, $l, strpos($str, '>', $l)-$l+1);
  117. continue;
  118. }
  119. $r = strpos($str, '>', $l);
  120. $tag = substr($str, $l+1, $r-$l-1);
  121. if (!ctype_alpha($tag[0]) || strpos($tag, '<') !== false) {
  122. $end = $r + 1;
  123. continue;
  124. }
  125. $tag = preg_replace("~n+~", ' ', $tag);
  126. $space = strpos($tag, ' ');
  127. if ($space !== false) {
  128. $attrCollect[] = substr($tag, $space+1);
  129. $tag = substr($tag, 0, $space);
  130. } else {
  131. $attrCollect[] = '';
  132. }
  133. $tagCollect[] = $tag;
  134. if (isset($selfClosingTags[$tag])) {
  135. $innerContentCollect[] = '';
  136. $end = $r-strlen($tag)-2;
  137. $close = $r+1;
  138. continue;
  139. }
  140. $countOpen = -1;
  141. $open = strpos($str, '<'.$tag, $close);
  142. $close = strpos($str, '', $open);
  143. if ($close === false) {//surplus opening tag
  144. $innerContentCollect[] = substr($str, $r+1);
  145. $error = true;
  146. self::$ErrorTag[] = '<'.$tag.'>';
  147. break;
  148. }
  149. $start = $open;
  150. while ($open < $close && $open !== false) {
  151. $countOpen++;
  152. $open = strpos($str, '<'.$tag, $open+strlen($tag));
  153. }
  154. while ($countOpen > 0 && $close !== false) {
  155. $open = strpos($str, '<'.$tag, $close+strlen($tag)+3);
  156. $close = strpos($str, '', $close+strlen($tag)+3);
  157. if ($close === false) {
  158. break;
  159. }
  160. $countOpen--;
  161. while ($open < $close && $open !== false) {
  162. $open = strpos($str, '<'.$tag, $open+strlen($tag)+3);
  163. $countOpen++;
  164. }
  165. }
  166. if ($close === false) {//标签闭合不配对
  167. $innerContentCollect[] = substr($str, $r+1);
  168. $error = true;
  169. break;
  170. }
  171. $end = $close;
  172. $r = strpos($str, '>', $start);
  173. $innerContentCollect[] = substr($str, $r+1, $end - $r - 1);
  174. }
  175. return !$error;
  176. }
  177. /**
  178. * _removeNoise
  179. *
  180. * @param string &$str The tag string to be parse.
  181. *
  182. * @access private
  183. *
  184. * @return string
  185. */
  186. private function _removeNoise(&$str)
  187. {
  188. $str = preg_replace('~~is', '', $str);
  189. $str = preg_replace('~~is', '', $str);
  190. $str = preg_replace('~~is', '', $str);
  191. }
  192. /**
  193. * parseSelectors
  194. *
  195. * @param string $selectors user's select condition.
  196. * @param array &$selectorsTag tags
  197. * @param array &$selectorsAttr attributes
  198. *
  199. * @access protected
  200. *
  201. * @return null
  202. */
  203. protected function parseSelectors($selectors, array &$selectorsTag, array &$selectorsAttr)
  204. {
  205. preg_match_all('~([wd]+)([[wd -="._/]+])?~', $selectors, $matches);
  206. $selectorsTag = $matches[1];
  207. foreach ($matches[2] as $key => $value) {
  208. $selectorsAttr[$key] = array();
  209. if ($value !== '') {
  210. preg_match_all('~([wd-]+)="([wd-. _/]+)"~', $value, $matches);
  211. foreach ($matches[1] as $index => $attr) {
  212. $selectorsAttr[$key][$attr] = $matches[2][$index];
  213. }
  214. }
  215. }
  216. }
  217. /**
  218. * find
  219. *
  220. * @param mixed $selectors user's select condition.
  221. * @param array $selectorsTag tags.
  222. * @param array $selectorsAttr attributes.
  223. *
  224. * @access public
  225. *
  226. * @return array
  227. */
  228. public function find($selectors, $selectorsTag = array(), $selectorsAttr = array())
  229. {
  230. if ($selectors !== null) {
  231. $this->parseSelectors($selectors, $selectorsTag, $selectorsAttr);
  232. }
  233. var_dump($selectorsTag, $selectorsAttr);exit();
  234. if (!empty($selectorsTag)) {
  235. $this->seek($selectorsTag, $selectorsAttr);
  236. foreach ($this->child as $key => $node) {
  237. $node->find(null, $selectorsTag, $selectorsAttr);
  238. }
  239. }
  240. if ($selectors !== null) {
  241. $res = self::$FoundNode;
  242. self::$FoundNode = array();
  243. return $res;
  244. }
  245. }
  246. /**
  247. * findGlobal
  248. *
  249. * @param string $selectors user's select condition.
  250. *
  251. * @access public
  252. *
  253. * @return array
  254. */
  255. public function findGlobal($selectors)
  256. {
  257. $space = strpos($selectors, ' ', strpos($selectors, ']'));
  258. if ($space === false) {
  259. return $this->findOneGlobal($selectors);
  260. } else {
  261. $selectorsAttr = array();
  262. $selectorsTag = array();
  263. $this->findOneGlobal(substr($selectors, 0, $space), false);
  264. $this->parseSelectors(substr($selectors, $space + 1), $selectorsTag, $selectorsAttr);
  265. if (!empty(self::$FoundNode) && !empty($selectorsTag)) {
  266. $nodes = self::$FoundNode;
  267. self::$FoundNode = array();
  268. foreach ($nodes as $key => $node) {
  269. $node->seek($selectorsTag, $selectorsAttr);
  270. }
  271. }
  272. }
  273. $res = self::$FoundNode;
  274. self::$FoundNode = array();
  275. return $res;
  276. }
  277. /**
  278. * seek
  279. *
  280. * @param array $selectorsTag tags.
  281. * @param array $selectorsAttr attributes.
  282. *
  283. * @access protected
  284. *
  285. * @return null
  286. */
  287. protected function seek($selectorsTag, $selectorsAttr)
  288. {
  289. foreach ($this->child as $key => $node) {
  290. $isFind = true;
  291. if ($node->tag === $selectorsTag[0]) {
  292. foreach ($selectorsAttr[0] as $attrName => $value) {
  293. if (isset($node->attr[$attrName])
  294. && (preg_match('~.*? '.$value.' .*?~', $node->attr[$attrName]) > 0
  295. || preg_match('~^'.$value.'$~', $node->attr[$attrName]) > 0
  296. || preg_match('~^'.$value.' ~', $node->attr[$attrName]) > 0
  297. || preg_match('~ '.$value.'$~', $node->attr[$attrName]) > 0)
  298. ) {
  299. continue;
  300. } else {
  301. $isFind = false;
  302. break;
  303. }
  304. }
  305. } else {
  306. $isFind = false;
  307. }
  308. if ($isFind) {
  309. if (count($selectorsTag) === 1) {
  310. self::$FoundNode[] = $node;
  311. } else {
  312. $node->seek(
  313. array_slice($selectorsTag, 1),
  314. array_slice($selectorsAttr, 1)
  315. );
  316. }
  317. }
  318. }
  319. }
  320. /**
  321. * findOneGlobal
  322. *
  323. * @param string $selector user's select condition.
  324. * @param bool $isReturn weather return value.
  325. *
  326. * @access public
  327. *
  328. * @return array
  329. */
  330. public function findOneGlobal($selector, $isReturn = true)
  331. {
  332. preg_match('~([wd]+)([[wd -="._/]+])?~', $selector, $matches);
  333. $tag = $matches[1];
  334. $attr = array();
  335. if (isset($matches[2])) {
  336. preg_match_all('~([wd-]+)="([wd-. _/]+)"~', $matches[2], $matches);
  337. foreach ($matches[1] as $key => $value) {
  338. $attr[$value] = $matches[2][$key];
  339. }
  340. }
  341. if (isset(self::$TagSet[$tag])) {
  342. foreach (self::$TagSet[$tag] as $attrValue => $nodeArray) {
  343. $isFind = true;
  344. foreach ($attr as $attrName => $value) {
  345. if (preg_match('~'.$attrName.'=".*? '.$value.' .*?"~', $attrValue)
  346. || preg_match('~'.$attrName.'="'.$value.' .*?"~', $attrValue)
  347. || preg_match('~'.$attrName.'=".*? '.$value.'"~', $attrValue)
  348. || preg_match('~'.$attrName.'="'.$value.'"~', $attrValue)
  349. ) {
  350. continue;
  351. } else {
  352. $isFind = false;
  353. break;
  354. }
  355. }
  356. if ($isFind) {
  357. foreach ($nodeArray as $key => $node) {
  358. self::$FoundNode[] = $node;
  359. }
  360. }
  361. }
  362. }
  363. if ($isReturn) {
  364. $res = self::$FoundNode;
  365. self::$FoundNode = array();
  366. return $res;
  367. }
  368. }
  369. }
  370. /**
  371. * TagDomNode
  372. *
  373. * @uses TagDomRoot
  374. *
  375. * @category TagParse
  376. * @package TagParse
  377. * @author kun
  378. * @copyright 2014 kun
  379. * @license http://www.php.net/license/3_01.txt PHP License 3.01
  380. * @version 1.0
  381. * @link http://www.blogkun.com
  382. * @since 1.0
  383. */
  384. class TagDomNode extends TagDomRoot
  385. {
  386. public $attr = array();
  387. public $parent = null;
  388. /**
  389. * __construct
  390. *
  391. * @param mixed $tag tag.
  392. * @param mixed $parent parent node.
  393. * @param mixed $attr attribute.
  394. * @param mixed $innerContent tag content.
  395. * @param mixed $level node level.
  396. *
  397. * @access public
  398. *
  399. * @return TagDomNode
  400. */
  401. public function __construct($tag, $parent, $attr, $innerContent, $level)
  402. {
  403. $this->tag = $tag;
  404. $this->parent = $parent;
  405. $this->_parseAttr($attr);
  406. $this->level = $level;
  407. $l = strpos($innerContent, '<');
  408. if ($l !== false) {
  409. $this->plaintext = substr($innerContent, 0, $l);
  410. }
  411. $res = preg_match_all('~>(.*?)<~s', $innerContent, $matches);
  412. if ($res !== false && $res > 0) {
  413. $this->plaintext .= implode($matches[1]);
  414. } else {
  415. $this->plaintext .= $innerContent;
  416. }
  417. $r = strrpos($innerContent, '>');
  418. if ($r !== false) {
  419. $this->plaintext .= substr($innerContent, $r+1);
  420. }
  421. $tagCollect = array();
  422. $attrCollect = array();
  423. $innerContentCollect = array();
  424. if ($this->parseTag($innerContent, $tagCollect, $attrCollect, $innerContentCollect) === false) {
  425. self::$TagParseError = true;
  426. }
  427. foreach ($tagCollect as $index => $tag) {
  428. $this->child[] = new TagDomNode($tag, $this, $attrCollect[$index], $innerContentCollect[$index], $this->level+1);
  429. }
  430. if (!isset(self::$TagSet[$this->tag])) {
  431. self::$TagSet[$this->tag] = array();
  432. }
  433. if (!isset(self::$TagSet[$this->tag][$attr])) {
  434. self::$TagSet[$this->tag][$attr] = array();
  435. }
  436. self::$TagSet[$this->tag][$attr][] = &$this;
  437. }
  438. /**
  439. * _parseAttr
  440. *
  441. * @param string $str attribute string.
  442. *
  443. * @access public
  444. *
  445. * @return null
  446. */
  447. private function _parseAttr($str)
  448. {
  449. preg_match_all('~(?[w-]+)="(?.*?)"~s', $str, $matches);
  450. foreach ($matches['attrName'] as $key => $value) {
  451. $this->attr[$value] = $matches['attrValue'][$key];
  452. }
  453. }
  454. }
复制代码


source:php.cn
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template