search

Home  >  Article  >  Backend Development  >  PHP uses the Tencent QQ Weibo API interface to obtain the code of Weibo content

PHP uses the Tencent QQ Weibo API interface to obtain the code of Weibo content

WBOY
WBOYOriginal
2016-07-25 08:57:471209browse

  3. /**

  4. * Use Tencent QQ Weibo API interface to obtain Weibo content
  5. * by bbs.it-home.org
  6. */
  7. define('HDOM_TYPE_ELEMENT', 1);
  8. define('HDOM_TYPE_COMMENT', 2);
  9. define('HDOM_TYPE_TEXT', 3);
  10. define('HDOM_TYPE_ENDTAG', 4);
  11. define('HDOM_TYPE_ROOT', 5);
  12. define('HDOM_TYPE_UNKNOWN', 6);
  13. define('HDOM_QUOTE_DOUBLE', 0);
  14. define('HDOM_QUOTE_SINGLE', 1);
  15. define('HDOM_QUOTE_NO', 3);
  16. define('HDOM_INFO_BEGIN', 0);
  17. define('HDOM_INFO_END', 1);
  18. define('HDOM_INFO_QUOTE', 2);
  19. define('HDOM_INFO_SPACE', 3);
  20. define('HDOM_INFO_TEXT', 4);
  21. define('HDOM_INFO_INNER', 5);
  22. define('HDOM_INFO_OUTER', 6);
  23. define('HDOM_INFO_ENDSPACE',7);
  24. // helper functions
  25. // ---------------
  26. // get html dom form file
  27. function file_get_html() {
  28. $dom = new simple_html_dom;
  29. $args = func_get_args();
  30. $dom->load(call_user_func_array('file_get_contents', $args), true);
  31. return $dom;
  32. }
  33. // get html dom form string
  34. function str_get_html($str, $lowercase=true) {
  35. $dom = new simple_html_dom;
  36. $dom->load($str, $lowercase);
  37. return $dom;
  38. }
  39. // dump html dom tree
  40. function dump_html_tree($node, $show_attr=true, $deep=0) {
  41. $lead = str_repeat(' ', $deep);
  42. echo $lead.$node->tag;
  43. if ($show_attr && count($node->attr)>0) {
  44. echo '(';
  45. foreach($node->attr as $k=>$v)
  46. echo "[$k]=>"".$node->$k.'", ';
  47. echo ')';
  48. }
  49. echo "n";
  50. foreach($node->nodes as $c)
  51. dump_html_tree($c, $show_attr, $deep+1);
  52. }
  53. // get dom form file (dePRecated)
  54. function file_get_dom() {
  55. $dom = new simple_html_dom;
  56. $args = func_get_args();
  57. $dom->load(call_user_func_array('file_get_contents', $args), true);
  58. return $dom;
  59. }
  60. // get dom form string (deprecated)
  61. function str_get_dom($str, $lowercase=true) {
  62. $dom = new simple_html_dom;
  63. $dom->load($str, $lowercase);
  64. return $dom;
  65. }
  66. // simple html dom node
  67. // ---------------
  68. class simple_html_dom_node {
  69. public $nodetype = HDOM_TYPE_TEXT;
  70. public $tag = 'text';
  71. public $attr = array();
  72. public $children = array();
  73. public $nodes = array();
  74. public $parent = null;
  75. public $_ = array();
  76. private $dom = null;
  77. function __construct($dom) {
  78. $this->dom = $dom;
  79. $dom->nodes[] = $this;
  80. }
  81. function __destruct() {
  82. $this->clear();
  83. }
  84. function __toString() {
  85. return $this->outertext();
  86. }
  87. // clean up memory due to php5 circular references memory leak...
  88. function clear() {
  89. $this->dom = null;
  90. $this->nodes = null;
  91. $this->parent = null;
  92. $this->children = null;
  93. }

  94. // dump node's tree

  95. function dump($show_attr=true) {
  96. dump_html_tree($this, $show_attr);
  97. }
  98. // returns the parent of node
  99. function parent() {
  100. return $this->parent;
  101. }
  102. // returns children of node
  103. function children($idx=-1) {
  104. if ($idx===-1) return $this->children;
  105. if (isset($this->children[$idx])) return $this->children[$idx];
  106. return null;
  107. }
  108. // returns the first child of node
  109. function first_child() {
  110. if (count($this->children)>0) return $this->children[0];
  111. return null;
  112. }
  113. // returns the last child of node
  114. function last_child() {
  115. if (($count=count($this->children))>0) return $this->children[$count-1];
  116. return null;
  117. }
  118. // returns the next sibling of node
  119. function next_sibling() {
  120. if ($this->parent===null) return null;
  121. $idx = 0;
  122. $count = count($this->parent->children);
  123. while ($idx<$count && $this!==$this->parent->children[$idx])
  124. ++$idx;
  125. if (++$idx>=$count) return null;
  126. return $this->parent->children[$idx];
  127. }
  128. // returns the previous sibling of node
  129. function prev_sibling() {
  130. if ($this->parent===null) return null;
  131. $idx = 0;
  132. $count = count($this->parent->children);
  133. while ($idx<$count && $this!==$this->parent->children[$idx])
  134. ++$idx;
  135. if (--$idx<0) return null;
  136. return $this->parent->children[$idx];
  137. }
  138. // get dom node's inner html
  139. function innertext() {
  140. if (isset($this->_[HDOM_INFO_INNER])) return $this->_[HDOM_INFO_INNER];
  141. if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
  142. $ret = '';
  143. foreach($this->nodes as $n)
  144. $ret .= $n->outertext();
  145. return $ret;
  146. }
  147. // get dom node's outer text (with tag)
  148. function outertext() {
  149. if ($this->tag==='root') return $this->innertext();
  150. // trigger callback
  151. if ($this->dom->callback!==null)
  152. call_user_func_array($this->dom->callback, array($this));
  153. if (isset($this->_[HDOM_INFO_OUTER])) return $this->_[HDOM_INFO_OUTER];
  154. if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
  155. // render begin tag
  156. $ret = $this->dom->nodes[$this->_[HDOM_INFO_BEGIN]]->makeup();
  157. // render inner text
  158. if (isset($this->_[HDOM_INFO_INNER]))
  159. $ret .= $this->_[HDOM_INFO_INNER];
  160. else {
  161. foreach($this->nodes as $n)
  162. $ret .= $n->outertext();
  163. }
  164. // render end tag
  165. if(isset($this->_[HDOM_INFO_END]) && $this->_[HDOM_INFO_END]!=0)
  166. $ret .= 'tag.'>';
  167. return $ret;
  168. }
  169. // get dom node's plain text
  170. function text() {
  171. if (isset($this->_[HDOM_INFO_INNER])) return $this->_[HDOM_INFO_INNER];
  172. switch ($this->nodetype) {
  173. case HDOM_TYPE_TEXT: return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
  174. case HDOM_TYPE_COMMENT: return '';
  175. case HDOM_TYPE_UNKNOWN: return '';
  176. }
  177. if (strcasecmp($this->tag, 'script')===0) return '';
  178. if (strcasecmp($this->tag, 'style')===0) return '';
  179. $ret = '';
  180. foreach($this->nodes as $n)
  181. $ret .= $n->text();
  182. return $ret;
  183. }

  184. function xmltext() {

  185. $ret = $this->innertext();
  186. $ret = str_ireplace(' $ret = str_replace(']]>', '', $ret);
  187. return $ret;
  188. }
  189. // build node's text with tag
  190. function makeup() {
  191. // text, comment, unknown
  192. if (isset($this->_[HDOM_INFO_TEXT])) return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
  193. $ret = '<'.$this->tag;
  194. $i = -1;
  195. foreach($this->attr as $key=>$val) {
  196. ++$i;
  197. // skip removed attribute
  198. if ($val===null || $val===false)
  199. continue;
  200. $ret .= $this->_[HDOM_INFO_SPACE][$i][0];
  201. //no value attr: nowrap, checked selected...
  202. if ($val===true)
  203. $ret .= $key;
  204. else {
  205. switch($this->_[HDOM_INFO_QUOTE][$i]) {
  206. case HDOM_QUOTE_DOUBLE: $quote = '"'; break;
  207. case HDOM_QUOTE_SINGLE: $quote = '''; break;
  208. default: $quote = '';
  209. }
  210. $ret .= $key.$this->_[HDOM_INFO_SPACE][$i][1].'='.$this->_[HDOM_INFO_SPACE][$i][2].$quote.$val.$quote;
  211. }
  212. }
  213. $ret = $this->dom->restore_noise($ret);
  214. return $ret . $this->_[HDOM_INFO_ENDSPACE] . '>';
  215. }
  216. // find elements by CSS selector
  217. function find($selector, $idx=null) {
  218. $selectors = $this->parse_selector($selector);
  219. if (($count=count($selectors))===0) return array();
  220. $found_keys = array();
  221. // find each selector
  222. for ($c=0; $c<$count; ++$c) {
  223. if (($levle=count($selectors[0]))===0) return array();
  224. if (!isset($this->_[HDOM_INFO_BEGIN])) return array();
  225. $head = array($this->_[HDOM_INFO_BEGIN]=>1);
  226. // handle descendant selectors, no recursive!
  227. for ($l=0; $l<$levle; ++$l) {
  228. $ret = array();
  229. foreach($head as $k=>$v) {
  230. $n = ($k===-1) ? $this->dom->root : $this->dom->nodes[$k];
  231. $n->seek($selectors[$c][$l], $ret);
  232. }
  233. $head = $ret;
  234. }
  235. foreach($head as $k=>$v) {
  236. if (!isset($found_keys[$k]))
  237. $found_keys[$k] = 1;
  238. }
  239. }
  240. // sort keys
  241. ksort($found_keys);
  242. $found = array();
  243. foreach($found_keys as $k=>$v)
  244. $found[] = $this->dom->nodes[$k];
  245. // return nth-element or array
  246. if (is_null($idx)) return $found;
  247. else if ($idx<0) $idx = count($found) + $idx;
  248. return (isset($found[$idx])) ? $found[$idx] : null;
  249. }
  250. // seek for given conditions
  251. protected function seek($selector, &$ret) {
  252. list($tag, $key, $val, $exp, $no_key) = $selector;
  253. // xpath index
  254. if ($tag && $key && is_numeric($key)) {
  255. $count = 0;
  256. foreach ($this->children as $c) {
  257. if ($tag==='*' || $tag===$c->tag) {
  258. if (++$count==$key) {
  259. $ret[$c->_[HDOM_INFO_BEGIN]] = 1;
  260. return;
  261. }
  262. }
  263. }
  264. return;
  265. }
  266. $end = (!empty($this->_[HDOM_INFO_END])) ? $this->_[HDOM_INFO_END] : 0;
  267. if ($end==0) {
  268. $parent = $this->parent;
  269. while (!isset($parent->_[HDOM_INFO_END]) && $parent!==null) {
  270. $end -= 1;
  271. $parent = $parent->parent;
  272. }
  273. $end += $parent->_[HDOM_INFO_END];
  274. }
  275. for($i=$this->_[HDOM_INFO_BEGIN]+1; $i<$end; ++$i) {
  276. $node = $this->dom->nodes[$i];
  277. $pass = true;
  278. if ($tag==='*' && !$key) {
  279. if (in_array($node, $this->children, true))
  280. $ret[$i] = 1;
  281. continue;
  282. }
  283. // compare tag
  284. if ($tag && $tag!=$node->tag && $tag!=='*') {$pass=false;}
  285. // compare key
  286. if ($pass && $key) {
  287. if ($no_key) {
  288. if (isset($node->attr[$key])) $pass=false;
  289. }
  290. else if (!isset($node->attr[$key])) $pass=false;
  291. }
  292. // compare value
  293. if ($pass && $key && $val && $val!=='*') {
  294. $check = $this->match($exp, $val, $node->attr[$key]);
  295. // handle multiple class
  296. if (!$check && strcasecmp($key, 'class')===0) {
  297. foreach(explode(' ',$node->attr[$key]) as $k) {
  298. $check = $this->match($exp, $val, $k);
  299. if ($check) break;
  300. }
  301. }
  302. if (!$check) $pass = false;
  303. }
  304. if ($pass) $ret[$i] = 1;
  305. unset($node);
  306. }
  307. }
  308. protected function match($exp, $pattern, $value) {
  309. switch ($exp) {
  310. case '=':
  311. return ($value===$pattern);
  312. case '!=':
  313. return ($value!==$pattern);
  314. case '^=':
  315. return preg_match("/^".preg_quote($pattern,'/')."/", $value);
  316. case '$=':
  317. return preg_match("/".preg_quote($pattern,'/')."$/", $value);
  318. case '*=':
  319. if ($pattern[0]=='/')
  320. return preg_match($pattern, $value);
  321. return preg_match("/".$pattern."/i", $value);
  322. }
  323. return false;
  324. }
  325. protected function parse_selector($selector_string) {
  326. // pattern of CSS selectors, modified from mootools
  327. $pattern = "/([w-:*]*)(?:#([w-]+)|.([w-]+))?(?:[@?(!?[w-]+)(?:([!*^$]?=)["']?(.*?)["']?)?])?([/, ]+)/is";
  328. preg_match_all($pattern, trim($selector_string).' ', $matches, PREG_SET_ORDER);
  329. $selectors = array();
  330. $result = array();
  331. //print_r($matches);
  332. foreach ($matches as $m) {
  333. $m[0] = trim($m[0]);
  334. if ($m[0]==='' || $m[0]==='/' || $m[0]==='//') continue;
  335. // for borwser grnreated xpath
  336. if ($m[1]==='tbody') continue;
  337. list($tag, $key, $val, $exp, $no_key) = array($m[1], null, null, '=', false);
  338. if(!empty($m[2])) {$key='id'; $val=$m[2];}
  339. if(!empty($m[3])) {$key='class'; $val=$m[3];}
  340. if(!empty($m[4])) {$key=$m[4];}
  341. if(!empty($m[5])) {$exp=$m[5];}
  342. if(!empty($m[6])) {$val=$m[6];}
  343. // convert to lowercase
  344. if ($this->dom->lowercase) {$tag=strtolower($tag); $key=strtolower($key);}
  345. //elements that do NOT have the specified attribute
  346. if (isset($key[0]) && $key[0]==='!') {$key=substr($key, 1); $no_key=true;}
  347. $result[] = array($tag, $key, $val, $exp, $no_key);
  348. if (trim($m[7])===',') {
  349. $selectors[] = $result;
  350. $result = array();
  351. }
  352. }
  353. if (count($result)>0)
  354. $selectors[] = $result;
  355. return $selectors;
  356. }
  357. function __get($name) {
  358. if (isset($this->attr[$name])) return $this->attr[$name];
  359. switch($name) {
  360. case 'outertext': return $this->outertext();
  361. case 'innertext': return $this->innertext();
  362. case 'plaintext': return $this->text();
  363. case 'xmltext': return $this->xmltext();
  364. default: return array_key_exists($name, $this->attr);
  365. }
  366. }
  367. function __set($name, $value) {
  368. switch($name) {
  369. case 'outertext': return $this->_[HDOM_INFO_OUTER] = $value;
  370. case 'innertext':
  371. if (isset($this->_[HDOM_INFO_TEXT])) return $this->_[HDOM_INFO_TEXT] = $value;
  372. return $this->_[HDOM_INFO_INNER] = $value;
  373. }
  374. if (!isset($this->attr[$name])) {
  375. $this->_[HDOM_INFO_SPACE][] = array(' ', '', '');
  376. $this->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_DOUBLE;
  377. }
  378. $this->attr[$name] = $value;
  379. }
  380. function __isset($name) {
  381. switch($name) {
  382. case 'outertext': return true;
  383. case 'innertext': return true;
  384. case 'plaintext': return true;
  385. }
  386. //no value attr: nowrap, checked selected...
  387. return (array_key_exists($name, $this->attr)) ? true : isset($this->attr[$name]);
  388. }
  389. function __unset($name) {
  390. if (isset($this->attr[$name]))
  391. unset($this->attr[$name]);
  392. }
  393. // camel naming conventions
  394. function getAllAttributes() {return $this->attr;}
  395. function getAttribute($name) {return $this->__get($name);}
  396. function setAttribute($name, $value) {$this->__set($name, $value);}
  397. function hasAttribute($name) {return $this->__isset($name);}
  398. function removeAttribute($name) {$this->__set($name, null);}
  399. function getElementById($id) {return $this->find("#$id", 0);}
  400. function getElementsById($id, $idx=null) {return $this->find("#$id", $idx);}
  401. function getElementByTagName($name) {return $this->find($name, 0);}
  402. function getElementsByTagName($name, $idx=null) {return $this->find($name, $idx);}
  403. function parentNode() {return $this->parent();}
  404. function childNodes($idx=-1) {return $this->children($idx);}
  405. function firstChild() {return $this->first_child();}
  406. function lastChild() {return $this->last_child();}
  407. function nextSibling() {return $this->next_sibling();}
  408. function previousSibling() {return $this->prev_sibling();}
  409. }
  410. // simple html dom parser
  411. // -----------------------------------------------------------------------------
  412. class simple_html_dom {
  413. public $root = null;
  414. public $nodes = array();
  415. public $callback = null;
  416. public $lowercase = false;
  417. protected $pos;
  418. protected $doc;
  419. protected $char;
  420. protected $size;
  421. protected $cursor;
  422. protected $parent;
  423. protected $noise = array();
  424. protected $token_blank = " trn";
  425. protected $token_equal = ' =/>';
  426. protected $token_slash = " />rnt";
  427. protected $token_attr = ' >';
  428. // use isset instead of in_array, performance boost about 30%...
  429. protected $self_closing_tags = array('img'=>1, 'br'=>1, 'input'=>1, 'meta'=>1, 'link'=>1, 'hr'=>1, 'base'=>1, 'embed'=>1, 'spacer'=>1);
  430. protected $block_tags = array('root'=>1, 'body'=>1, 'form'=>1, 'div'=>1, 'span'=>1, 'table'=>1);
  431. protected $optional_closing_tags = array(
  432. 'tr'=>array('tr'=>1, 'td'=>1, 'th'=>1),
  433. 'th'=>array('th'=>1),
  434. 'td'=>array('td'=>1),
  435. 'li'=>array('li'=>1),
  436. 'dt'=>array('dt'=>1, 'dd'=>1),
  437. 'dd'=>array('dd'=>1, 'dt'=>1),
  438. 'dl'=>array('dd'=>1, 'dt'=>1),
  439. 'p'=>array('p'=>1),
  440. 'nobr'=>array('nobr'=>1),
  441. );
  442. function __construct($str=null) {
  443. if ($str) {
  444. if (preg_match("/^http:///i",$str) || is_file($str))
  445. $this->load_file($str);
  446. else
  447. $this->load($str);
  448. }
  449. }
  450. http://www.devdao.com/
  451. function __destruct() {
  452. $this->clear();
  453. }
  454. // load html from string
  455. function load($str, $lowercase=true) {
  456. // prepare
  457. $this->prepare($str, $lowercase);
  458. // strip out comments
  459. $this->remove_noise("''is");
  460. // strip out cdata
  461. $this->remove_noise("''is", true);
  462. // strip out