API tools faq
paste
Guest User

Untitled

a guest
Dec 14th, 2010
295
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 32.82 KB | None | 0 0
raw download clone embed print report
  1. <?php
  2. /**
  3. * PHP IMDB.com SCRAPER
  4. *
  5. * This class can be used to retrieve data from IMDB.com with PHP.
  6. *
  7. * The technique used is called “web scraping”
  8. * (see http://en.wikipedia.org/wiki/Web_scraping for details).
  9. * Which means: If IMDB changes *anything* on their HTML, the script is going to
  10. * fail (even a single space might be enough).
  11. *
  12. * You might not know, but there is an IMDB API available. The problem?
  13. * You will have to pay at least $15.000 to use it. Great, thank you.
  14. *
  15. *
  16. * If you want to thank me for my work and the support, feel free to do this
  17. * through PayPal (use [email protected] as payment destination) or just
  18. * buy me a book at Amazon (http://www.amazon.de/wishlist/3IAUEEEY6GD20)
  19. * – thank you! :-)
  20. *
  21. *
  22. * @link http://fabian-beiner.de
  23. * @copyright 2010 Fabian Beiner
  24. * @author Fabian Beiner ([email protected])
  25. * @license MIT License
  26. *
  27. * @version 5.2.4 (December 4th, 2010)
  28. */
  29.  
  30. class IMDBException extends Exception {}
  31.  
  32. class IMDB {
  33. // Please set this to 'true' for debugging purposes only.
  34. const IMDB_DEBUG = false;
  35. // Define a timeout for the request of the IMDB page.
  36. const IMDB_TIMEOUT = 15;
  37.  
  38. // Regular expressions, I would not touch them. :)
  39. const IMDB_BUDGET = '~Budget:</h4> (.*)\(estimated\)~Ui';
  40. const IMDB_CAST = '~<td class="name">\s+<a\s+href="/name/nm(\d+)/">(.*)</a>\s+</td~Ui';
  41. //const IMDB_CHAR = '~<td class="character">\s+<div>\s+(.*)\s+</div>\s+</td~Ui';
  42. const IMDB_CHAR = '~<td class="character">(.*)</td~Ui';
  43. const IMDB_COUNTRY = '~<a href="/country/(\w+)">(.*)</a>~Ui';
  44. const IMDB_CREATOR = '~<h4 class="inline">\s+(Creator|Creators):\s+</h4>(.*)</div><div~Ui';
  45. const IMDB_DIRECTOR = '~<h4 class="inline">\s+(Director|Directors):\s+</h4>(.*)</div><div~Ui';
  46. const IMDB_GENRE = '~<a href="/genre/(.*)"~Ui';
  47. const IMDB_LANGUAGES = '~<a href="/language/(\w+)">(.*)</a>~Ui';
  48. const IMDB_LOCATION = '~<h4 class="inline">Filming Locations:</h4> <a href="/search/title\?locations=(.*)">(.*)</a>~Ui';
  49. const IMDB_MPAA = '~<h4>Motion Picture Rating \(<a href="/mpaa">MPAA</a>\)</h4>(.*) <span~Ui';
  50. const IMDB_NAME = '~href="/name/nm(\d+)/">(.*)</a>~Ui';
  51. const IMDB_PLOT = '~<h2>Storyline</h2><p>(.*)(<em class="nobr">|</p>)~Ui';
  52. const IMDB_POSTER = '~href="/media/(.*)"\s+><img src="(.*)"~Ui';
  53. const IMDB_RATING = '~<span class="rating-rating">(\d+\.\d+)<span>~Ui';
  54. const IMDB_REDIRECT = '~Location:\s(.*)~';
  55. const IMDB_RELEASE_DATE = '~Release Date:</h4>(.*)(<span|</div>)~Ui';
  56. const IMDB_RUNTIME = '~(\d+)\smin~Uis';
  57. const IMDB_SEASONS = '~<h4 class="inline">Season: </h4><span class="see-more inline">(.*)</div><div~Ui';
  58. const IMDB_SEARCH = '~<b>Media from&nbsp;<a href="/title/tt(\d+)/"~i';
  59. const IMDB_TAGLINE = '~<h4 class="inline">Taglines:</h4>(.*)(<[^>]+>)~Ui';
  60. const IMDB_TITLE = '~<title>(.*) \((.*)\).*~Ui';
  61. const IMDB_TITLE_ORIG = '~<span class="title-extra">(.*) <i>\(original title\)</i></span>~Ui';
  62. const IMDB_URL = '~http://(.*\.|.*)imdb.com/(t|T)itle(\?|/)(..\d+)~i';
  63. const IMDB_VOTES = '~>(\d+|\d+,\d+) votes</a>\)~Ui';
  64. const IMDB_WRITER = '~<h4 class="inline">\s+(Writer|Writers):(.*)</div><div~Ui';
  65.  
  66. // cURL cookie file
  67. private $_fCookie = false;
  68. // IMDB url
  69. private $_strUrl = NULL;
  70. // IMDB source
  71. private $_strSource = NULL;
  72. // IMDB cache
  73. private $_intCache = 0;
  74. // IMDB posters directory
  75. private $_bolPoster = NULL;
  76. // IMDB cache directory
  77. private $_bolCache = NULL;
  78. // Movie found?
  79. public $isReady = false;
  80.  
  81. /**
  82. * IMDB constructor.
  83. *
  84. * @param string $strSearch The movie name / IMDB url
  85. * @param integer $intCache The maximum age (in minutes) of a cache
  86. */
  87. public function __construct($strSearch, $intCache = 1440) {
  88. // Cookie path.
  89. if (function_exists('sys_get_temp_dir')) {
  90. $this->_fCookie = tempnam(sys_get_temp_dir(), 'imdb');
  91. if (IMDB::IMDB_DEBUG) echo '<b>- Path to cookie:</b> ' . $this->_fCookie . '<br>';
  92. }
  93. // Posters and cache directory existant?
  94. if (!file_exists(getcwd() . '/posters/')) {
  95. if (mkdir(getcwd() . '/posters/', 0777)) {
  96. $this->_bolPoster = true;
  97. }
  98. }
  99. elseif (is_writeable(getcwd() . '/posters/')) {
  100. $this->_bolPoster = true;
  101. }
  102. else {
  103. throw new IMDBException(getcwd() . '/posters/ is not writable!');
  104. }
  105. if (!file_exists(getcwd() . '/cache/')) {
  106. if (mkdir(getcwd() . '/cache/', 0777)) {
  107. $this->_bolCache = true;
  108. }
  109. }
  110. elseif (is_writeable(getcwd() . '/cache/')) {
  111. $this->_bolCache = true;
  112. }
  113. else {
  114. throw new IMDBException(getcwd() . '/cache/ is not writable!');
  115. }
  116. // cURL.
  117. if (!function_exists('curl_init')) {
  118. throw new IMDBException('You need PHP with cURL enabled to use this script!');
  119. }
  120. // Debug only.
  121. if (IMDB::IMDB_DEBUG) {
  122. error_reporting(E_ALL);
  123. ini_set('display_errors', '1');
  124. echo '<b>- Running:</b> IMDB::fetchUrl<br>';
  125. }
  126. // Set global cache and fetch the data.
  127. $this->_intCache = (int)$intCache;
  128. IMDB::fetchUrl($strSearch);
  129. }
  130.  
  131. /**
  132. * Regular expressions helper function.
  133. *
  134. * @param string $strContent The content to search in
  135. * @param string $strRegex The regular expression
  136. * @param integer $intIndex The index to return
  137. * @return string The match found
  138. * @return array The matches found
  139. */
  140. private function matchRegex($strContent, $strRegex, $intIndex = null) {
  141. $arrMatches = null;
  142. preg_match_all($strRegex, $strContent, $arrMatches);
  143. if ($arrMatches === FALSE) return;
  144. if ($intIndex != null && is_int($intIndex)) {
  145. if ($arrMatches[$intIndex]) {
  146. return $arrMatches[$intIndex][0];
  147. }
  148. return;
  149. }
  150. return $arrMatches;
  151. }
  152.  
  153. /**
  154. * Returns a shortened text.
  155. *
  156. * @param string $strText The text to shorten
  157. * @param integer $intLength The new length of the text
  158. */
  159. public function makeShort($strText, $intLength = 100) {
  160. $strText = trim($strText) . ' ';
  161. $strText = substr($strText, 0, $intLength);
  162. $strText = substr($strText, 0, strrpos($strText, ' '));
  163. return $strText . '&hellip;';
  164. }
  165.  
  166. /**
  167. * Fetch data from the given url.
  168. *
  169. * @param string $strSearch The movie name / IMDB url
  170. * @param string $strSave The path to the file
  171. * @return boolean
  172. */
  173. private function fetchUrl($strSearch) {
  174. // Remove whitespaces.
  175. $strSearch = trim($strSearch);
  176.  
  177. // Check for a valid IMDB URL and use it, if available.
  178. if ($strId = IMDB::matchRegex($strSearch, IMDB::IMDB_URL, 4)) {
  179. $this->_strUrl = 'http://www.imdb.com/title/tt' . preg_replace('~[\D]~', '', $strId) . '/';
  180. $this->isReady = true;
  181. }
  182. // Otherwise try to find one.
  183. else {
  184. $this->_strUrl = 'http://www.imdb.com/find?s=all&q=' . str_replace(' ', '+', $strSearch);
  185. // Check for cached redirects of this search.
  186. $fRedirect = getcwd() . '/cache/' . md5($this->_strUrl);
  187. if (file_exists($fRedirect) && is_readable($fRedirect)) {
  188. $fRedirect = file_get_contents($fRedirect);
  189. if (IMDB::IMDB_DEBUG) echo '<b>- Found an old redirect:</b> ' . $fRedirect . '<br>';
  190. $this->_strUrl = $fRedirect;
  191. $this->isReady = true;
  192. }
  193. }
  194.  
  195. $fCache = getcwd() . '/cache/' . md5($this->_strUrl) . '.cache';
  196.  
  197. // Check if there is a cache we can use.
  198. $bolNewRequest = false;
  199. if (file_exists($fCache)) {
  200. $intChanged = filemtime($fCache);
  201. $intNow = time();
  202. $intDiff = $intNow - $intChanged;
  203. $intCache = $this->_intCache * 60;
  204. if ($intCache >= $intDiff) {
  205. $bolNewRequest = true;
  206. }
  207. }
  208.  
  209. if ($bolNewRequest) {
  210. if (IMDB::IMDB_DEBUG) echo '<b>- Using cache for ' . $strSearch . ' from . ' . $fCache . '</b><br>';
  211. $this->_strSource = file_get_contents($fCache);
  212. return true;
  213. }
  214. // Check if cURL is installed.
  215. else {
  216. // Initialize and run the request.
  217. if (IMDB::IMDB_DEBUG) echo '<b>- Run cURL on:</b> ' . $this->_strUrl . '<br>';
  218. $oCurl = curl_init($this->_strUrl);
  219. curl_setopt_array($oCurl, array (
  220. CURLOPT_VERBOSE => FALSE,
  221. CURLOPT_HEADER => TRUE,
  222. CURLOPT_FRESH_CONNECT => TRUE,
  223. CURLOPT_RETURNTRANSFER => TRUE,
  224. CURLOPT_TIMEOUT => IMDB::IMDB_TIMEOUT,
  225. CURLOPT_CONNECTTIMEOUT => 0,
  226. CURLOPT_REFERER => 'http://www.google.com',
  227. CURLOPT_USERAGENT, 'Googlebot/2.1 (+http://www.google.com/bot.html)',
  228. CURLOPT_FOLLOWLOCATION => FALSE,
  229. CURLOPT_COOKIEFILE => $this->_fCookie
  230. ));
  231. $strOutput = curl_exec($oCurl);
  232. $this->_strSource = $strOutput;
  233.  
  234. // Check if the request actually worked.
  235. if ($strOutput === FALSE) {
  236. if (IMDB::IMDB_DEBUG) echo '<b>! cURL error:</b> ' . $_strUrl . '<br>';
  237. if (file_exists($fCache)) {
  238. $this->_strSource = file_get_contents($fCache);
  239. return true;
  240. }
  241. return;
  242. }
  243.  
  244. // Get returned information.
  245. $arrInfo = curl_getinfo($oCurl);
  246. curl_close($oCurl);
  247.  
  248. // Check if there is a redirect given (IMDB sometimes does not return 301 for this...).
  249. if ($strMatch = $this->matchRegex($strOutput, IMDB::IMDB_REDIRECT, 1)) {
  250. if (IMDB::IMDB_DEBUG) echo '<b>- Found a redirect:</b> ' . $strMatch . '<br>';
  251. // Try to save the redirect for later usage.
  252. $fRedirect = getcwd() . '/cache/' . md5($this->_strUrl);
  253. if (IMDB::IMDB_DEBUG) echo '<b>- Saved a new redirect:</b> ' . $fRedirect . '<br>';
  254. file_put_contents($fRedirect, $strMatch);
  255. // Run the cURL request again with the new url.
  256. IMDB::fetchUrl($strMatch);
  257. }
  258. // Lets assume the first search result is what we want. :)
  259. elseif ($strMatch = $this->matchRegex($strOutput, IMDB::IMDB_SEARCH, 1)) {
  260. $strMatch = 'http://www.imdb.com/title/tt' . $strMatch . '/';
  261. if (IMDB::IMDB_DEBUG) echo '<b>- Using the first search result:</b> ' . $strMatch . '<br>';
  262. // Try to save the redirect for later usage.
  263. $fRedirect = getcwd() . '/cache/' . md5($this->_strUrl);
  264. if (IMDB::IMDB_DEBUG) echo '<b>- Saved a new redirect:</b> ' . $fRedirect . '<br>';
  265. file_put_contents($fRedirect, $strMatch);
  266. // Run the cURL request again with the new url.
  267. IMDB::fetchUrl($strMatch);
  268. }
  269. // If it's not a redirect and the HTTP response is not 200 or 302, abort.
  270. elseif ($arrInfo['http_code'] != 200 && $arrInfo['http_code'] != 302) {
  271. if (IMDB::IMDB_DEBUG) echo '<b>- Wrong HTTP code received, aborting:</b> ' . $arrInfo['http_code'] . '<br>';
  272. return;
  273. }
  274.  
  275. // Set the global source.
  276. $this->_strSource = preg_replace('~(\r|\n|\r\n)~', '', $this->_strSource);
  277.  
  278. // Save cache.
  279. if (IMDB::IMDB_DEBUG) echo '<b>- Saved a new cache:</b> ' . $fCache . '<br>';
  280. file_put_contents($fCache, $this->_strSource);
  281.  
  282. return true;
  283. }
  284. return;
  285. }
  286.  
  287. /**
  288. * Save the image locally.
  289. *
  290. * @param string $_strUrl The URL to the image on imdb
  291. * @return string The local path to the image
  292. */
  293. private function saveImage($_strUrl) {
  294. $_strUrl = trim($_strUrl);
  295.  
  296. if (preg_match('/imdb-share-logo.gif/', $_strUrl)) {
  297. if (file_exists('posters/not-found.jpg')) {
  298. return 'posters/not-found.jpg';
  299. }
  300. return $_strUrl;
  301. }
  302.  
  303. $strFilename = getcwd() . '/posters/' . md5(IMDB::getTitle()) . '.jpg';
  304. if (file_exists($strFilename)) {
  305. return 'posters/' . md5(IMDB::getTitle()) . '.jpg';
  306. }
  307. if (function_exists('curl_init')) {
  308. $oCurl = curl_init($_strUrl);
  309. curl_setopt_array($oCurl, array (
  310. CURLOPT_VERBOSE => FALSE,
  311. CURLOPT_HEADER => FALSE,
  312. CURLOPT_RETURNTRANSFER => TRUE,
  313. CURLOPT_TIMEOUT => IMDB::IMDB_TIMEOUT,
  314. CURLOPT_CONNECTTIMEOUT => IMDB::IMDB_TIMEOUT,
  315. CURLOPT_REFERER => $_strUrl,
  316. CURLOPT_BINARYTRANSFER => TRUE));
  317. $sOutput = curl_exec($oCurl);
  318. curl_close($oCurl);
  319. $oFile = fopen($strFilename, 'x');
  320. fwrite($oFile, $sOutput);
  321. fclose($oFile);
  322. return 'posters/' . md5(IMDB::getTitle()) . '.jpg';
  323. } else {
  324. $oImg = imagecreatefromjpeg($_strUrl);
  325. imagejpeg($oImg, $strFilename);
  326. return 'posters/' . md5(IMDB::getTitle()) . '.jpg';
  327. }
  328. return $_strUrl;
  329. }
  330.  
  331. /**
  332. * Returns the budget.
  333. *
  334. * @return string The movie budget.
  335. */
  336. public function getBudget() {
  337. if ($this->isReady) {
  338. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_BUDGET, 1)) {
  339. return $strReturn;
  340. }
  341. return 'n/A';
  342. }
  343. return 'n/A';
  344. }
  345.  
  346. /**
  347. * Returns the cast.
  348. *
  349. * @return array The movie cast (default limited to 20).
  350. */
  351. public function getCast($intLimit = 20, $bolMore = true) {
  352. if ($this->isReady) {
  353. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_CAST);
  354. if (count($arrReturned[2])) {
  355. foreach ($arrReturned[2] as $i => $strName) {
  356. if ($i >= $intLimit) break;
  357. $arrReturn[] = $strName;
  358. }
  359. return implode(' / ', $arrReturn) . ($bolMore && (count($arrReturned[2]) > $intLimit) ? '&hellip;' : '');
  360. }
  361. return 'n/A';
  362. }
  363. return 'n/A';
  364. }
  365.  
  366. /**
  367. * Returns the cast as URL.
  368. *
  369. * @return array The movie cast as URL (default limited to 20).
  370. */
  371. public function getCastAsUrl($intLimit = 20, $bolMore = true) {
  372. if ($this->isReady) {
  373. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_CAST);
  374. if (count($arrReturned[2])) {
  375. foreach ($arrReturned[2] as $i => $strName) {
  376. if ($i >= $intLimit) break;
  377. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  378. }
  379. return implode(' / ', $arrReturn) . ($bolMore && (count($arrReturned[2]) > $intLimit) ? '&hellip;' : '');
  380. }
  381. return 'n/A';
  382. }
  383. return 'n/A';
  384. }
  385.  
  386. /**
  387. * Returns the cast and character.
  388. *
  389. * @return array The movie cast and character (default limited to 20).
  390. */
  391. public function getCastAndCharacter($intLimit = 20, $bolMore = true) {
  392. if ($this->isReady) {
  393. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_CAST);
  394. $arrChar = $this->matchRegex($this->_strSource, IMDB::IMDB_CHAR);
  395. if (count($arrReturned[2])) {
  396. foreach ($arrReturned[2] as $i => $strName) {
  397. if ($i >= $intLimit) break;
  398. $arrChar[1][$i] = trim(preg_replace('~\((.*)\)~Ui', '', strip_tags($arrChar[1][$i])));
  399. if ($arrChar[1][$i]) {
  400. $arrReturn[] = $strName . ' as ' . $arrChar[1][$i];
  401. }
  402. else {
  403. $arrReturn[] = $strName;
  404. }
  405. }
  406. return implode(' / ', $arrReturn) . ($bolMore && (count($arrReturned[2]) > $intLimit) ? '&hellip;' : '');
  407. }
  408. return 'n/A';
  409. }
  410. return 'n/A';
  411. }
  412.  
  413. /**
  414. * Returns the cast and character as URL .
  415. *
  416. * @return array The movie cast and character as URL (default limited to 20).
  417. */
  418. public function getCastAndCharacterAsUrl($intLimit = 20, $bolMore = true) {
  419. if ($this->isReady) {
  420. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_CAST);
  421. $arrChar = $this->matchRegex($this->_strSource, IMDB::IMDB_CHAR);
  422. if (count($arrReturned[2])) {
  423. foreach ($arrReturned[2] as $i => $strName) {
  424. if ($i >= $intLimit) break;
  425. $arrChar[1][$i] = trim(preg_replace('~\((.*)\)~Ui', '', $arrChar[1][$i]));
  426. //print_r($arrChar[1][$i]);
  427. preg_match_all('~<a href="/character/ch(\d+)/">(.*)</a>~Ui', $arrChar[1][$i], $arrMatches);
  428. if ($arrMatches[1][0] && $arrMatches[2][0]) {
  429. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a> as <a href="http://www.imdb.com/character/ch' . $arrMatches[1][0] . '/">' . $arrMatches[2][0] . '</a>';
  430. }
  431. else {
  432. if ($arrChar[1][$i]) {
  433. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a> as ' . strip_tags($arrChar[1][$i]);
  434. }
  435. else {
  436. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  437. }
  438. }
  439. }
  440. return implode(' / ', $arrReturn) . ($bolMore && (count($arrReturned[2]) > $intLimit) ? '&hellip;' : '');
  441. }
  442. return 'n/A';
  443. }
  444. return 'n/A';
  445. }
  446.  
  447.  
  448. /**
  449. * Returns the countr(y|ies).
  450. *
  451. * @return array The movie countr(y|ies).
  452. */
  453. public function getCountry() {
  454. if ($this->isReady) {
  455. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_COUNTRY);
  456. if (count($arrReturned[2])) {
  457. foreach ($arrReturned[2] as $strName) {
  458. $arrReturn[] = $strName;
  459. }
  460. return implode(' / ', $arrReturn);
  461. }
  462. return 'n/A';
  463. }
  464. return 'n/A';
  465. }
  466.  
  467. /**
  468. * Returns the countr(y|ies) as URL
  469. *
  470. * @return array The movie countr(y|ies) as URL.
  471. */
  472. public function getCountryAsUrl() {
  473. if ($this->isReady) {
  474. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_COUNTRY);
  475. if (count($arrReturned[2])) {
  476. foreach ($arrReturned[2] as $i => $strName) {
  477. $arrReturn[] = '<a href="http://www.imdb.com/country/' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  478. }
  479. return implode(' / ', $arrReturn);
  480. }
  481. return 'n/A';
  482. }
  483. return 'n/A';
  484. }
  485.  
  486. /**
  487. * Returns the director(s).
  488. *
  489. * @return array The movie director(s).
  490. */
  491. public function getDirector() {
  492. if ($this->isReady) {
  493. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_DIRECTOR, 2);
  494. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  495. if (count($arrReturned[2])) {
  496. foreach ($arrReturned[2] as $i => $strName) {
  497. $arrReturn[] = $strName;
  498. }
  499. return implode(' / ', $arrReturn);
  500. }
  501. return 'n/A';
  502. }
  503. return 'n/A';
  504. }
  505.  
  506. /**
  507. * Returns the director(s) as URL.
  508. *
  509. * @return array The movie director(s) as URL.
  510. */
  511. public function getDirectorAsUrl() {
  512. if ($this->isReady) {
  513. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_DIRECTOR, 2);
  514. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  515. if (count($arrReturned[2])) {
  516. foreach ($arrReturned[2] as $i => $strName) {
  517. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  518. }
  519. return implode(' / ', $arrReturn);
  520. }
  521. return 'n/A';
  522. }
  523. return 'n/A';
  524. }
  525.  
  526.  
  527. /**
  528. * Returns the creator(s).
  529. *
  530. * @return array The movie creator(s).
  531. */
  532. public function getCreator() {
  533. if ($this->isReady) {
  534. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_CREATOR, 2);
  535. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  536. if (count($arrReturned[2])) {
  537. foreach ($arrReturned[2] as $i => $strName) {
  538. $arrReturn[] = $strName;
  539. }
  540. return implode(' / ', $arrReturn);
  541. }
  542. return 'n/A';
  543. }
  544. return 'n/A';
  545. }
  546.  
  547. /**
  548. * Returns the creator(s) as URL.
  549. *
  550. * @return array The movie creator(s) as URL.
  551. */
  552. public function getCreatorAsUrl() {
  553. if ($this->isReady) {
  554. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_CREATOR, 2);
  555. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  556. if (count($arrReturned[2])) {
  557. foreach ($arrReturned[2] as $i => $strName) {
  558. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  559. }
  560. return implode(' / ', $arrReturn);
  561. }
  562. return 'n/A';
  563. }
  564. return 'n/A';
  565. }
  566.  
  567. /**
  568. * Returns the genre(s).
  569. *
  570. * @return array The movie genre(s).
  571. */
  572. public function getGenre() {
  573. if ($this->isReady) {
  574. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_GENRE);
  575. if (count($arrReturned[1])) {
  576. foreach ($arrReturned[1] as $strName) {
  577. $arrReturn[] = $strName;
  578. }
  579. return implode(' / ', $arrReturn);
  580. }
  581. return 'n/A';
  582. }
  583. return 'n/A';
  584. }
  585.  
  586. /**
  587. * Returns the genres as URL.
  588. *
  589. * @return array The movie genre as URL.
  590. */
  591. public function getGenreAsUrl() {
  592. if ($this->isReady) {
  593. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_GENRE);
  594. if (count($arrReturned[1])) {
  595. foreach ($arrReturned[1] as $i => $strName) {
  596. $arrReturn[] = '<a href="http://www.imdb.com/genre/' . $strName . '/">' . $strName . '</a>';
  597. }
  598. return implode(' / ', $arrReturn);
  599. }
  600. return 'n/A';
  601. }
  602. return 'n/A';
  603. }
  604.  
  605. /**
  606. * Returns the language(s).
  607. *
  608. * @return string The movie language(s).
  609. */
  610. public function getLanguages() {
  611. if ($this->isReady) {
  612. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_LANGUAGES);
  613. if (count($arrReturned[2])) {
  614. foreach ($arrReturned[2] as $strName) {
  615. $arrReturn[] = $strName;
  616. }
  617. return implode(' / ', $arrReturn);
  618. }
  619. return 'n/A';
  620. }
  621. return 'n/A';
  622. }
  623.  
  624. /**
  625. * Returns the language(s) as URL.
  626. *
  627. * @return string The movie language(s) as URL.
  628. */
  629. public function getLanguagesAsUrl() {
  630. if ($this->isReady) {
  631. $arrReturned = $this->matchRegex($this->_strSource, IMDB::IMDB_LANGUAGES);
  632. if (count($arrReturned[2])) {
  633. foreach ($arrReturned[2] as $i => $strName) {
  634. $arrReturn[] = '<a href="http://www.imdb.com/language/' . $arrReturned[1][$i] . '">' . $strName . '</a>';
  635. }
  636. return implode(' / ', $arrReturn);
  637. }
  638. return 'n/A';
  639. }
  640. return 'n/A';
  641. }
  642.  
  643. /**
  644. * Returns the movie location.
  645. *
  646. * @return string The location of the movie.
  647. */
  648. public function getLocation() {
  649. if ($this->isReady) {
  650. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_LOCATION, 2)) {
  651. return $strReturn;
  652. }
  653. return 'n/A';
  654. }
  655. return 'n/A';
  656. }
  657.  
  658. /**
  659. * Returns the movie location as URL.
  660. *
  661. * @return string The location of the movie as URL.
  662. */
  663. public function getLocationAsUrl() {
  664. if ($this->isReady) {
  665. $strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_LOCATION, 2);
  666. if ($strReturn) {
  667. return '<a href="http://www.imdb.com/search/title?locations=' . urlencode($strReturn) . '">' . $strReturn . '</a>';
  668. }
  669. return 'n/A';
  670. }
  671. return 'n/A';
  672. }
  673.  
  674.  
  675. /**
  676. * Returns the MPAA.
  677. *
  678. * @return string The movie MPAA.
  679. */
  680. public function getMpaa() {
  681. if ($this->isReady) {
  682. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_MPAA, 1)) {
  683. return $strReturn;
  684. }
  685. return 'n/A';
  686. }
  687. return 'n/A';
  688. }
  689.  
  690. /**
  691. * Returns the plot.
  692. *
  693. * @return string The movie plot.
  694. */
  695. public function getPlot($intLimit = 0) {
  696. if ($this->isReady) {
  697. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_PLOT, 1)) {
  698. if ($intLimit) {
  699. return $this->makeShort($strReturn, $intLimit);
  700. }
  701. return $strReturn;
  702. }
  703. return 'n/A';
  704. }
  705. return 'n/A';
  706. }
  707.  
  708. /**
  709. * Download the poster, cache it and return the local path to the image.
  710. *
  711. * @return string The path to the poster (either local or online).
  712. */
  713. public function getPoster() {
  714. if ($this->isReady) {
  715. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_POSTER, 2)) {
  716. if ($strLocal = $this->saveImage($strReturn)) {
  717. return $strLocal;
  718. }
  719. return $strReturn;
  720. }
  721. return 'n/A';
  722. }
  723. return 'n/A';
  724. }
  725.  
  726. /**
  727. * Returns the rating.
  728. *
  729. * @return string The movie rating.
  730. */
  731. public function getRating() {
  732. if ($this->isReady) {
  733. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_RATING, 1)) {
  734. return $strReturn;
  735. }
  736. return 'n/A';
  737. }
  738. return 'n/A';
  739. }
  740.  
  741. /**
  742. * Returns the release date.
  743. *
  744. * @return string The movie release date.
  745. */
  746. public function getReleaseDate() {
  747. if ($this->isReady) {
  748. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_RELEASE_DATE, 1)) {
  749. return str_replace('(', ' (', $strReturn);
  750. }
  751. return 'n/A';
  752. }
  753. return 'n/A';
  754. }
  755.  
  756. /**
  757. * Returns the runtime.
  758. *
  759. * @return string The movie runtime.
  760. */
  761. public function getRuntime() {
  762. if ($this->isReady) {
  763. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_RUNTIME, 1)) {
  764. return $strReturn;
  765. }
  766. return 'n/A';
  767. }
  768. return 'n/A';
  769. }
  770.  
  771. /**
  772. * Returns the tagline.
  773. *
  774. * @return string The movie tagline.
  775. */
  776. public function getTagline() {
  777. if ($this->isReady) {
  778. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_TAGLINE, 1)) {
  779. return $strReturn;
  780. }
  781. return 'n/A';
  782. }
  783. return 'n/A';
  784. }
  785.  
  786. /**
  787. * Return the title.
  788. *
  789. * @return string The movie title.
  790. */
  791. public function getTitle() {
  792. if ($this->isReady) {
  793. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_TITLE_ORIG, 1)) {
  794. return $strReturn;
  795. }
  796. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_TITLE, 1)) {
  797. return $strReturn;
  798. }
  799. return 'n/A';
  800. }
  801. return 'n/A';
  802. }
  803.  
  804. /**
  805. * Returns the URL.
  806. *
  807. * @return string The movie URL.
  808. */
  809. public function getUrl() {
  810. return $this->_strUrl;
  811. }
  812.  
  813. /**
  814. * Returns the votes.
  815. *
  816. * @return string The votes of the movie.
  817. */
  818. public function getVotes() {
  819. if ($this->isReady) {
  820. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_VOTES, 1)) {
  821. return $strReturn;
  822. }
  823. return 'n/A';
  824. }
  825. return 'n/A';
  826. }
  827.  
  828. /**
  829. * Returns the writer(s).
  830. *
  831. * @return array The movie writer(s).
  832. */
  833. public function getWriter() {
  834. if ($this->isReady) {
  835. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_WRITER, 2);
  836. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  837. if (count($arrReturned[2])) {
  838. foreach ($arrReturned[2] as $i => $strName) {
  839. $arrReturn[] = $strName;
  840. }
  841. return implode(' / ', $arrReturn);
  842. }
  843. return 'n/A';
  844. }
  845. return 'n/A';
  846. }
  847.  
  848. /**
  849. * Returns the writer(s) as URL.
  850. *
  851. * @return array The movie writer(s) as URL.
  852. */
  853. public function getWriterAsUrl() {
  854. if ($this->isReady) {
  855. $strContainer = $this->matchRegex($this->_strSource, IMDB::IMDB_WRITER, 2);
  856. $arrReturned = $this->matchRegex($strContainer, IMDB::IMDB_NAME);
  857. if (count($arrReturned[2])) {
  858. foreach ($arrReturned[2] as $i => $strName) {
  859. $arrReturn[] = '<a href="http://www.imdb.com/name/nm' . $arrReturned[1][$i] . '/">' . $strName . '</a>';
  860. }
  861. return implode(' / ', $arrReturn);
  862. }
  863. return 'n/A';
  864. }
  865. return 'n/A';
  866. }
  867.  
  868. /**
  869. * Returns the movie year.
  870. *
  871. * @return string The year of the movie.
  872. */
  873. public function getYear() {
  874. if ($this->isReady) {
  875. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_TITLE, 2)) {
  876. return substr(preg_replace('~[\D]~', '', $strReturn), 0, 4);
  877. }
  878. return 'n/A';
  879. }
  880. return 'n/A';
  881. }
  882.  
  883. /**
  884. * Returns the seasons.
  885. *
  886. * @return string The movie seasons.
  887. */
  888. public function getSeasons() {
  889. if ($this->isReady) {
  890. if ($strReturn = $this->matchRegex($this->_strSource, IMDB::IMDB_SEASONS)) {
  891. $strReturn = strip_tags(implode($strReturn[1]));
  892. $strFind = array('&raquo;', '&nbsp;', 'Full episode list', ' ');
  893. $strReturn = str_replace($strFind, '', $strReturn);
  894. $arrReturn = explode('|', $strReturn);
  895. unset($arrReturn[(count($arrReturn)-1)]);
  896. if ($arrReturn) {
  897. return implode(' / ', $arrReturn);
  898. }
  899. }
  900. return 'n/A';
  901. }
  902. return 'n/A';
  903. }
  904. }
Advertisement
Add Comment
Please, Sign In to add comment
Public Pastes
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!