VideoUrlParser是一款基于PHP根据视频URL抓取视频信息的工具，支持优酷、土豆、酷六、56、乐视、搜狐、腾讯、新浪。

使用方法：

1. require_once "VideoUrlParser.class.php";
2. $url = "http://v.youku.com/v_show/id_XMjkwMzc0Njg4.html";
3. $info = VedioUrlParser::parse($url);
4. echo $info;

说明：调用该工具php文件VideoUrlParser.class.php，$url变量后面的字符串为视频页的地址，然后使用echo输出变量$info。

附：info含有的几个值，分别是img（用于视频缩略图），title（视频标题），url（地址），swf（视频swf播放地址）。我只用到了img和swf地址。具体的可以根据自己的需要进行调整。

VideoUrlParser类源码：

1. <?php
2. /**
3. * Video
4. *
5. * @package
6. * @version 1.2
7. * @copyright 2005-2011 HDJ.ME
8. * @author Dijia Huang <huangdijia@gmail.com>
9. * @license PHP Version 3.0 {@link http://www.php.net/license/3_0.txt}
10. *
11. * Usage
12. * require_once "VideoUrlParser.class.php";
13. * $urls[] = "http://v.youku.com/v_show/id_XMjI4MDM4NDc2.html";
14. * $urls[] = "http://www.tudou.com/playlist/p/l13087099.html";
15. * $urls[] = "http://www.tudou.com/programs/view/ufg-A3tlcxk/";
16. * $urls[] = "http://v.ku6.com/special/show_4926690/Klze2mhMeSK6g05X.html";
17. * $urls[] = "http://www.56.com/u68/v_NjI2NTkxMzc.html";
18. * $urls[] = "http://www.letv.com/ptv/vplay/1168109.html";
19. * $urls[] = "http://video.sina.com.cn/v/b/46909166-1290055681.html";
20. *
21. * foreach($urls as $url){
22. * $info = VideoUrlParser::parse($url);
23. * //var_dump($info);
24. * echo "<a href='{$info['url']}' target='_new'>{$info['title']}</a>";
25. * echo "<br />";
26. * echo $info['object'];
27. * echo "<br />";
28. * }
29. *
30. *
31. *
32. * //优酷
33. * http://v.youku.com/v_show/id_XMjU0NjY4OTEy.html
34. * <embed src="http://player.youku.com/player.php/sid/XMjU0NjY4OTEy/v.swf" quality="high" width="480" height="400" align="middle" allowScriptAccess="sameDomain" type="application/x-shockwave-flash"></embed>
35. *
36. * //酷六
37. * http://v.ku6.com/special/show_3917484/x0BMXAbgZdQS6FqN.html
38. * <embed src="http://player.ku6.com/refer/x0BMXAbgZdQS6FqN/v.swf" quality="high" width="480" height="400" align="middle" allowScriptAccess="always" allowfullscreen="true" type="application/x-shockwave-flash"></embed>
39. *
40. * //土豆
41. * http://www.tudou.com/playlist/p/a65929.html?iid=74905844
42. * <embed src="http://www.tudou.com/l/A_0urj-Geec/&i type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" wmode="opaque" width="480" height="400"></embed>
43. *
44. * //56
45. * http://www.56.com/u98/v_NTkyODY2NTU.html
46. * <embed src="http://player.56.com/v_NTkyODY2NTU.swf" type="application/x-shockwave-flash" width="480" height="405" allowNetworking="all" allowScriptAccess="always"></embed>
47. *
48. * //新浪播客
49. * http://video.sina.com.cn/v/b/46909166-1290055681.html
50. * <embed src="http://you.video.sina.com.cn/api/sinawebApi/outplayrefer.php/v pluginspage="http://www.macromedia.com/go/getflashplayer" type="application/x-shockwave-flash" name="ssss" allowFullScreen="true" allowScriptAccess="always" width="480" height="370"></embed>
51. *
52. * //乐视
53. * http://www.letv.com/ptv/vplay/1168109.html
54. * <embed src="http://i3.imgs.letv.com/player/swfPlayer.swf? quality="high" scale="NO_SCALE" wmode="opaque" bgcolor="#000000" width="480" height="388" name="FLV_player" align="middle" allowscriptaccess="always" allowfullscreen="true" type="application/x-shockwave-flash" pluginspage="http://www.macromedia.com/go/getflashplayer">
55. */
56. class VideoUrlParser
57. {
58. const USER_AGENT = "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/534.10 (KHTML, like Gecko)
59. Chrome/8.0.552.224 Safari/534.10";
60. const CHECK_URL_VALID = "/(youku\.com|tudou\.com|ku6\.com|56\.com|letv\.com|video\.sina\.com\.cn|(my\.)?tv\.sohu\.com|v\.qq\.com)/";
61. /**
62. * parse
63. *
64. * @param string $url
65. * @param mixed $createObject
66. * @static
67. * @access public
68. * @return void
69. */
70. static public function parse($url='', $createObject=true){
71. $lowerurl = strtolower($url);
72. preg_match(self::CHECK_URL_VALID, $lowerurl, $matches);
73. if(!$matches) return false;
74. switch($matches[1]){
75. case 'youku.com':
76. $data = self::_parseYouku($url);
77. break;
78. case 'tudou.com':
79. $data = self::_parseTudou($url);
80. break;
81. case 'ku6.com':
82. $data = self::_parseKu6($url);
83. break;
84. case '56.com':
85. $data = self::_parse56($url);
86. break;
87. case 'letv.com':
88. $data = self::_parseLetv($url);
89. break;
90. case 'video.sina.com.cn':
91. $data = self::_parseSina($url);
92. break;
93. case 'my.tv.sohu.com':
94. case 'tv.sohu.com':
95. case 'sohu.com':
96. $data = self::_parseSohu($url);
97. break;
98. case 'v.qq.com':
99. $data = self::_parseQq($url);
100. break;
101. default:
102. $data = false;
103. }
104. if($data && $createObject) $data['object'] = "<embed src=\"{$data['swf']}\" quality=\"high\" width=\"480\" height=\"400\" align=\"middle\" allowNetworking=\"all\" allowScriptAccess=\"always\" type=\"application/x-shockwave-flash\"></embed>";
105. return $data;
106. }
107. /**
108. * 腾讯视频
109. * http://v.qq.com/cover/o/o9tab7nuu0q3esh.html?vid=97abu74o4w3_0
110. * http://v.qq.com/play/97abu74o4w3.html
111. * http://v.qq.com/cover/d/dtdqyd8g7xvoj0o.html
112. * http://v.qq.com/cover/d/dtdqyd8g7xvoj0o/9SfqULsrtSb.html
113. * http://imgcache.qq.com/tencentvideo_v1/player/TencentPlayer.swf?_v=20110829&vid=97abu74o4w3&autoplay=1&list=2&showcfg=1&tpid=23&title=%E7%AC%AC%E4%B8%80%E7%8E%B0%E5%9C%BA&adplay=1&cid=o9tab7nuu0q3esh
114. */
115. private function _parseQq($url){
116. if(preg_match("/\/play\//", $url)){
117. $html = self::_fget($url);
118. preg_match("/url=[^\"]+/", $html, $matches);
119. if(!$matches); return false;
120. $url = $matches[0];
121. }
122. preg_match("/v, $url, $matches);
123. $vid = $matches[1];
124. $html = self::_fget($url);
125. // query
126. preg_match("/flashvars\s=\s\"([^;]+)/s", $html, $matches);
127. $query = $matches[1];
128. if(!$vid){
129. preg_match("/vid\s?=\s?vid\s?\|\|\s?\"(\w+)\";/i", $html, $matches);
130. $vid = $matches[1];
131. }
132. $query = str_replace('"+vid+"', $vid, $query);
133. parse_str($query, $output);
134. $data['img'] = "http://vpic.video.qq.com/{$$output['cid']}/{$vid}_1.jpg";
135. $data['url'] = $url;
136. $data['title'] = $output['title'];
137. $data['swf'] = "http://imgcache.qq.com/tencentvideo_v1/player/TencentPlayer.swf?".$query;
138. return $data;
139. }
140. /**
141. * 优酷网
142. * http://v.youku.com/v_show/id_XMjI4MDM4NDc2.html
143. * http://player.youku.com/player.php/sid/XMjU0NjI2Njg4/v.swf
144. */
145. private function _parseYouku($url){
146. preg_match("#id\_(\w+)#", $url, $matches);
147. if (emptyempty($matches)){
148. preg_match("#v_playlist\/#", $url, $mat);
149. if(!$mat) return false;
150. $html = self::_fget($url);
151. preg_match("#videoId2\s*=\s*\'(\w+)\'#", $html, $matches);
152. if(!$matches) return false;
153. }
154. $link = "http://v.youku.com/player/getPlayList/VideoIDS/{$matches[1]}/timezone/+08/version/5/source/out?password=&ran=2513&n=3";
155. $retval = self::_cget($link);
156. if ($retval) {
157. $json = json_decode($retval, true);
158. $data['img'] = $json['data'][0]['logo'];
159. $data['title'] = $json['data'][0]['title'];
160. $data['url'] = $url;
161. $data['swf'] = "http://player.youku.com/player.php/sid/{$matches[1]}/v.swf";
162. return $data;
163. } else {
164. return false;
165. }
166. }
167. /**
168. * 土豆网
169. * http://www.tudou.com/programs/view/Wtt3FjiDxEE/
170. * http://www.tudou.com/v/Wtt3FjiDxEE/v.swf
171. *
172. * http://www.tudou.com/playlist/p/a65718.html?iid=74909603
173. * http://www.tudou.com/l/G5BzgI4lAb8/&iid=74909603/v.swf
174. */
175. private function _parseTudou($url){
176. preg_match("#view/([-\w]+)/#", $url, $matches);
177. if (emptyempty($matches)) {
178. if (strpos($url, "/playlist/") == false) return false;
179. if(strpos($url, 'iid=') !== false){
180. $quarr = explode("iid=", $lowerurl);
181. if (emptyempty($quarr[1])) return false;
182. }elseif(preg_match("#p\/l(\d+).#", $lowerurl, $quarr)){
183. if (emptyempty($quarr[1])) return false;
184. }
185. $html = self::_fget($url);
186. $html = iconv("GB2312", "UTF-8", $html);
187. preg_match("/lid_code\s=\slcode\s=\s[\'\"]([^\'\"]+)/s", $html, $matches);
188. $icode = $matches[1];
189. preg_match("/iid\s=\s.*?\|\|\s(\d+)/sx", $html, $matches);
190. $iid = $matches[1];
191. preg_match("/listData\s=\s(\[\{.*\}\])/sx", $html, $matches);
192. $find = array("/\n/", '/\s/', "/:[^\d\"]\w+[^\,]*,/i", "/(\{|,)(\w+):/");
193. $replace = array("", "", ':"",', '\\1"\\2":');
194. $str = preg_replace($find, $replace, $matches[1]);
195. //var_dump($str);
196. $json = json_decode($str);
197. //var_dump($json);exit;
198. if(is_array($json) || is_object($json) && !emptyempty($json)){
199. foreach ($json as $val) {
200. if ($val->iid == $iid) {
201. break;
202. }
203. }
204. }
205. $data['img'] = $val->pic;
206. $data['title'] = $val->title;
207. $data['url'] = $url;
208. $data['swf'] = "http://www.tudou.com/l/{$icode}/&i;
209. return $data;
210. }
211. $host = "www.tudou.com";
212. $path = "/v/{$matches[1]}/v.swf";
213. $ret = self::_fsget($path, $host);
214. if (preg_match("#\nLocation: (.*)\n#", $ret, $mat)) {
215. parse_str(parse_url(urldecode($mat[1]), PHP_URL_QUERY));
216. $data['img'] = $snap_pic;
217. $data['title'] = $title;
218. $data['url'] = $url;
219. $data['swf'] = "http://www.tudou.com/v/{$matches[1]}/v.swf";
220. return $data;
221. }
222. return false;
223. }
224. /**
225. * 酷6网
226. * http://v.ku6.com/film/show_520/3X93vo4tIS7uotHg.html
227. * http://v.ku6.com/special/show_4926690/Klze2mhMeSK6g05X.html
228. * http://v.ku6.com/show/7US-kDXjyKyIInDevhpwHg...html
229. * http://player.ku6.com/refer/3X93vo4tIS7uotHg/v.swf
230. */
231. private function _parseKu6($url){
232. if(preg_match("/show\_/", $url)){
233. preg_match("#/([-\w]+)\.html#", $url, $matches);
234. $url = "http://v.ku6.com/fetchVideo4Player/{$matches[1]}.html";
235. $html = self::_fget($url);
236. if ($html) {
237. $json = json_decode($html, true);
238. if(!$json) return false;
239. $data['img'] = $json['data']['picpath'];
240. $data['title'] = $json['data']['t'];
241. $data['url'] = $url;
242. $data['swf'] = "http://player.ku6.com/refer/{$matches[1]}/v.swf";
243. return $data;
244. } else {
245. return false;
246. }
247. }elseif(preg_match("/show\//", $url, $matches)){
248. $html = self::_fget($url);
249. preg_match("/ObjectInfo\s?=\s?([^\n]*)};/si", $html, $matches);
250. $str = $matches[1];
251. // img
252. preg_match("/cover\s?:\s?\"([^\"]+)\"/", $str, $matches);
253. $data['img'] = $matches[1];
254. // title
255. preg_match("/title\"?\s?:\s?\"([^\"]+)\"/", $str, $matches);
256. $jsstr = "{\"title\":\"{$matches[1]}\"}";
257. $json = json_decode($jsstr, true);
258. $data['title'] = $json['title'];
259. // url
260. $data['url'] = $url;
261. // query
262. preg_match("/\"(v]+)\"\sname=\"flashVars\"/s", $html, $matches);
263. $query = str_replace("&amp;", '&', $matches[1]);
264. preg_match("/\/\/player\.ku6cdn\.com[^\"\']+/", $html, $matches);
265. $data['swf'] = 'http:'.$matches[0].'?'.$query;
266. return $data;
267. }
268. }
269. /**
270. * 56网
271. * http://www.56.com/u73/v_NTkzMDcwNDY.html
272. * http://player.56.com/v_NTkzMDcwNDY.swf
273. */
274. private function _parse56($url){
275. preg_match("#/v_(\w+)\.html#", $url, $matches);
276. if (emptyempty($matches)) return false;
277. $link="http://vxml.56.com/json/{$matches[1]}/?src=out";
278. $retval = self::_cget($link);
279. if ($retval) {
280. $json = json_decode($retval, true);
281. $data['img'] = $json['info']['img'];
282. $data['title'] = $json['info']['Subject'];
283. $data['url'] = $url;
284. $data['swf'] = "http://player.56.com/v_{$matches[1]}.swf";
285. return $data;
286. } else {
287. return false;
288. }
289. }
290. /**
291. * 乐视网
292. * http://www.letv.com/ptv/vplay/1168109.html
293. * http://www.letv.com/player/x1168109.swf
294. */
295. private function _parseLetv($url){
296. $html = self::_fget($url);
297. preg_match("#http://v.t.sina.com.cn/([^'\"]*)#", $html, $matches);
298. parse_str(parse_url(urldecode($matches[0]), PHP_URL_QUERY));
299. preg_match("#vplay/(\d+)#", $url, $matches);
300. $data['img'] = $pic;
301. $data['title'] = $title;
302. $data['url'] = $url;
303. $data['swf'] = "http://www.letv.com/player/x{$matches[1]}.swf";
304. return $data;
305. }
306. // 搜狐TV http://my.tv.sohu.com/u/vw/5101536
307. private function _parseSohu($url){
308. $html = self::_fget($url);
309. $html = iconv("GB2312", "UTF-8", $html);
310. preg_match_all("/og:(?:title|image|videosrc)\"\scontent=\"([^\"]+)\"/s", $html, $matches);
311. $data['img'] = $matches[1][1];
312. $data['title'] = $matches[1][0];
313. $data['url'] = $url;
314. $data['swf'] = $matches[1][2];
315. return $data;
316. }
317. /*
318. * 新浪播客
319. * http://video.sina.com.cn/v/b/48717043-1290055681.html
320. * http://you.video.sina.com.cn/api/sinawebApi/outplayrefer.php/vid=48717043_1290055681_PUzkSndrDzXK+l1lHz2stqkP7KQNt6nki2O0u1ehIwZYQ0/XM5GdatoG5ynSA9kEqDhAQJA4dPkm0x4/s.swf
321. */
322. private function _parseSina($url){
323. preg_match("/(\d+)(?:\-|\_)(\d+)/", $url, $matches);
324. $url = "http://video.sina.com.cn/v/b/{$matches[1]}-{$matches[2]}.html";
325. $html = self::_fget($url);
326. preg_match("/video\s?:\s?([^<]+)}/", $html, $matches);
327. $find = array("/\n/", "/\s*/", "/\'/", "/\{([^:,]+):/", "/,([^:]+):/", "/:[^\d\"]\w+[^\,]*,/i");
328. $replace = array('', '', '"', '{"\\1":', ',"\\1":', ':"",');
329. $str = preg_replace($find, $replace, $matches[1]);
330. $arr = json_decode($str, true);
331. $data['img'] = $arr['pic'];
332. $data['title'] = $arr['title'];
333. $data['url'] = $url;
334. $data['swf'] = $arr['swfOutsideUrl'];
335. return $data;
336. }
337. /*
338. * 通过 file_get_contents 获取内容
339. */
340. private function _fget($url=''){
341. if(!$url) return false;
342. $html = file_get_contents($url);
343. // 判断是否gzip压缩
344. if($dehtml = self::_gzdecode($html))
345. return $dehtml;
346. else
347. return $html;
348. }
349. /*
350. * 通过 fsockopen 获取内容
351. */
352. private function _fsget($path='/', $host='', $user_agent=''){
353. if(!$path || !$host) return false;
354. $user_agent = $user_agent ? $user_agent : self::USER_AGENT;
355. $out = <<<HEADER
356. GET $path HTTP/1.1
357. Host: $host
358. User-Agent: $user_agent
359. Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
360. Accept-Language: zh-cn,zh;q=0.5
361. Accept-Charset: GB2312,utf-8;q=0.7,*;q=0.7\r\n\r\n
362. HEADER;
363. $fp = @fsockopen($host, 80, $errno, $errstr, 10);
364. if (!$fp) return false;
365. if(!fputs($fp, $out)) return false;
366. while ( !feof($fp) ) {
367. $html .= fgets($fp, 1024);
368. }
369. fclose($fp);
370. // 判断是否gzip压缩
371. if($dehtml = self::_gzdecode($html))
372. return $dehtml;
373. else
374. return $html;
375. }
376. /*
377. * 通过 curl 获取内容
378. */
379. private function _cget($url='', $user_agent=''){
380. if(!$url) return;
381. $user_agent = $user_agent ? $user_agent : self::USER_AGENT;
382. $ch = curl_init();
383. curl_setopt($ch, CURLOPT_URL, $url);
384. curl_setopt($ch, CURLOPT_HEADER, 0);
385. if(strlen($user_agent)) curl_setopt($ch, CURLOPT_USERAGENT, $user_agent);
386. ob_start();
387. curl_exec($ch);
388. $html = ob_get_contents();
389. ob_end_clean();
390. if(curl_errno($ch)){
391. curl_close($ch);
392. return false;
393. }
394. curl_close($ch);
395. if(!is_string($html) || !strlen($html)){
396. return false;
397. }
398. return $html;
399. // 判断是否gzip压缩
400. if($dehtml = self::_gzdecode($html))
401. return $dehtml;
402. else
403. return $html;
404. }
405. private function _gzdecode($data) {
406. $len = strlen ( $data );
407. if ($len < 18 || strcmp ( substr ( $data, 0, 2 ), "\x1f\x8b" )) {
408. return null; // Not GZIP format (See RFC 1952)
409. }
410. $method = ord ( substr ( $data, 2, 1 ) ); // Compression method
411. $flags = ord ( substr ( $data, 3, 1 ) ); // Flags
412. if ($flags & 31 != $flags) {
413. // Reserved bits are set -- NOT ALLOWED by RFC 1952
414. return null;
415. }
416. // NOTE: $mtime may be negative (PHP integer limitations)
417. $mtime = unpack ( "V", substr ( $data, 4, 4 ) );
418. $mtime = $mtime [1];
419. $xfl = substr ( $data, 8, 1 );
420. $os = substr ( $data, 8, 1 );
421. $headerlen = 10;
422. $extralen = 0;
423. $extra = "";
424. if ($flags & 4) {
425. // 2-byte length prefixed EXTRA data in header
426. if ($len - $headerlen - 2 < 8) {
427. return false; // Invalid format
428. }
429. $extralen = unpack ( "v", substr ( $data, 8, 2 ) );
430. $extralen = $extralen [1];
431. if ($len - $headerlen - 2 - $extralen < 8) {
432. return false; // Invalid format
433. }
434. $extra = substr ( $data, 10, $extralen );
435. $headerlen += 2 + $extralen;
436. }
437. $filenamelen = 0;
438. $filename = "";
439. if ($flags & 8) {
440. // C-style string file NAME data in header
441. if ($len - $headerlen - 1 < 8) {
442. return false; // Invalid format
443. }
444. $filenamelen = strpos ( substr ( $data, 8 + $extralen ), chr ( 0 ) );
445. if ($filenamelen === false || $len - $headerlen - $filenamelen - 1 < 8) {
446. return false; // Invalid format
447. }
448. $filename = substr ( $data, $headerlen, $filenamelen );
449. $headerlen += $filenamelen + 1;
450. }
451. $commentlen = 0;
452. $comment = "";
453. if ($flags & 16) {
454. // C-style string COMMENT data in header
455. if ($len - $headerlen - 1 < 8) {
456. return false; // Invalid format
457. }
458. $commentlen = strpos ( substr ( $data, 8 + $extralen + $filenamelen ), chr ( 0 ) );
459. if ($commentlen === false || $len - $headerlen - $commentlen - 1 < 8) {
460. return false; // Invalid header format
461. }
462. $comment = substr ( $data, $headerlen, $commentlen );
463. $headerlen += $commentlen + 1;
464. }
465. $headercrc = "";
466. if ($flags & 1) {
467. // 2-bytes (lowest order) of CRC32 on header present
468. if ($len - $headerlen - 2 < 8) {
469. return false; // Invalid format
470. }
471. $calccrc = crc32 ( substr ( $data, 0, $headerlen ) ) & 0xffff;
472. $headercrc = unpack ( "v", substr ( $data, $headerlen, 2 ) );
473. $headercrc = $headercrc [1];
474. if ($headercrc != $calccrc) {
475. return false; // Bad header CRC
476. }
477. $headerlen += 2;
478. }
479. // GZIP FOOTER - These be negative due to PHP's limitations
480. $datacrc = unpack ( "V", substr ( $data, - 8, 4 ) );
481. $datacrc = $datacrc [1];
482. $isize = unpack ( "V", substr ( $data, - 4 ) );
483. $isize = $isize [1];
484. // Perform the decompression:
485. $bodylen = $len - $headerlen - 8;
486. if ($bodylen < 1) {
487. // This should never happen - IMPLEMENTATION BUG!
488. return null;
489. }
490. $body = substr ( $data, $headerlen, $bodylen );
491. $data = "";
492. if ($bodylen > 0) {
493. switch ($method) {
494. case 8 :
495. // Currently the only supported compression method:
496. $data = gzinflate ( $body );
497. break;
498. default :
499. // Unknown compression method
500. return false;
501. }
502. } else {
503. //...
504. }
505. if ($isize != strlen ( $data ) || crc32 ( $data ) != $datacrc) {
506. // Bad format! Length or CRC doesn't match!
507. return false;
508. }
509. return $data;
510. }
511. }