You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

855 lines
30 KiB

4 years ago
4 years ago
6 years ago
  1. <?php
  2. namespace p3k\XRay\Formats;
  3. use HTMLPurifier, HTMLPurifier_Config;
  4. class Mf2 extends Format {
  5. use Mf2Feed;
  6. public static function matches_host($url) {
  7. return true;
  8. }
  9. public static function matches($url) {
  10. return true;
  11. }
  12. public static function parse($mf2, $url, $http, $opts=[]) {
  13. if(count($mf2['items']) == 0)
  14. return false;
  15. // If they are expecting a feed, always return a feed or an error
  16. if(isset($opts['expect']) && $opts['expect'] == 'feed') {
  17. return self::parseAsHFeed($mf2, $http, $url);
  18. }
  19. // If there is only one item on the page, just use that
  20. if(count($mf2['items']) == 1) {
  21. $item = $mf2['items'][0];
  22. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  23. #Parse::debug("mf2:0: Recognized $url as an h-entry it is the only item on the page");
  24. return self::parseAsHEntry($mf2, $item, $http, $url);
  25. }
  26. if(in_array('h-event', $item['type'])) {
  27. #Parse::debug("mf2:0: Recognized $url as an h-event it is the only item on the page");
  28. return self::parseAsHEvent($mf2, $item, $http, $url);
  29. }
  30. if(in_array('h-review', $item['type'])) {
  31. #Parse::debug("mf2:0: Recognized $url as an h-review it is the only item on the page");
  32. return self::parseAsHReview($mf2, $item, $http, $url);
  33. }
  34. if(in_array('h-recipe', $item['type'])) {
  35. #Parse::debug("mf2:0: Recognized $url as an h-recipe it is the only item on the page");
  36. return self::parseAsHRecipe($mf2, $item, $http, $url);
  37. }
  38. if(in_array('h-product', $item['type'])) {
  39. #Parse::debug("mf2:0: Recognized $url as an h-product it is the only item on the page");
  40. return self::parseAsHProduct($mf2, $item, $http, $url);
  41. }
  42. if(in_array('h-item', $item['type'])) {
  43. #Parse::debug("mf2:0: Recognized $url as an h-product it is the only item on the page");
  44. return self::parseAsHItem($mf2, $item, $http, $url);
  45. }
  46. if(in_array('h-card', $item['type'])) {
  47. #Parse::debug("mf2:0: Recognized $url as an h-card it is the only item on the page");
  48. return self::parseAsHCard($item, $http, $url, $url);
  49. }
  50. if(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  51. #Parse::debug("mf2:0: Recognized $url as an h-feed because it is the only item on the page");
  52. return self::parseAsHApp($mf2, $item, $http, $url);
  53. }
  54. if(in_array('h-feed', $item['type'])) {
  55. #Parse::debug("mf2:0: Recognized $url as an h-feed because it is the only item on the page");
  56. return self::parseAsHFeed($mf2, $http, $url);
  57. }
  58. }
  59. // Check the list of items on the page to see if one matches the URL of the page,
  60. // and treat as a permalink for that object if so.
  61. foreach($mf2['items'] as $item) {
  62. if(array_key_exists('url', $item['properties'])) {
  63. $urls = $item['properties']['url'];
  64. $urls = array_map('\p3k\XRay\normalize_url', $urls);
  65. if(in_array($url, $urls)) {
  66. #Parse::debug("mf2:1: Recognized $url as a permalink because an object on the page matched the URL of the request");
  67. if(in_array('h-card', $item['type'])) {
  68. return self::parseAsHCard($item, $http, $url, $url);
  69. } elseif(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  70. return self::parseAsHEntry($mf2, $item, $http, $url);
  71. } elseif(in_array('h-event', $item['type'])) {
  72. return self::parseAsHEvent($mf2, $item, $http, $url);
  73. } elseif(in_array('h-review', $item['type'])) {
  74. return self::parseAsHReview($mf2, $item, $http, $url);
  75. } elseif(in_array('h-recipe', $item['type'])) {
  76. return self::parseAsHRecipe($mf2, $item, $http, $url);
  77. } elseif(in_array('h-product', $item['type'])) {
  78. return self::parseAsHProduct($mf2, $item, $http, $url);
  79. } elseif(in_array('h-item', $item['type'])) {
  80. return self::parseAsHItem($mf2, $item, $http, $url);
  81. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  82. return self::parseAsHApp($mf2, $item, $http, $url);
  83. } elseif(in_array('h-feed', $item['type'])) {
  84. return self::parseAsHFeed($mf2, $http, $url);
  85. } else {
  86. #Parse::debug('This object was not a recognized type.');
  87. return false;
  88. }
  89. }
  90. }
  91. }
  92. // Check for an h-card matching rel=author or the author URL of any h-* on the page,
  93. // and return the h-* object if so
  94. if(isset($mf2['rels']['author'])) {
  95. foreach($mf2['items'] as $card) {
  96. if(in_array('h-card', $card['type']) && array_key_exists('url', $card['properties'])) {
  97. $urls = \p3k\XRay\normalize_urls($card['properties']['url']);
  98. if(count(array_intersect($urls, \p3k\XRay\normalize_urls($mf2['rels']['author']))) > 0) {
  99. // There is an author h-card on this page
  100. // Now look for the first h-* object other than an h-card and use that as the object
  101. foreach($mf2['items'] as $item) {
  102. if(!in_array('h-card', $item['type'])) {
  103. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  104. return self::parseAsHEntry($mf2, $item, $http, $url);
  105. } elseif(in_array('h-event', $item['type'])) {
  106. return self::parseAsHEvent($mf2, $item, $http, $url);
  107. } elseif(in_array('h-review', $item['type'])) {
  108. return self::parseAsHReview($mf2, $item, $http, $url);
  109. } elseif(in_array('h-recipe', $item['type'])) {
  110. return self::parseAsHRecipe($mf2, $item, $http, $url);
  111. } elseif(in_array('h-product', $item['type'])) {
  112. return self::parseAsHProduct($mf2, $item, $http, $url);
  113. } elseif(in_array('h-item', $item['type'])) {
  114. return self::parseAsHItem($mf2, $item, $http, $url);
  115. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  116. return self::parseAsHApp($mf2, $item, $http, $url);
  117. }
  118. }
  119. }
  120. }
  121. }
  122. }
  123. }
  124. // If there was more than one h-entry on the page, treat the whole page as a feed
  125. if(count($mf2['items']) > 1) {
  126. if(count(array_filter($mf2['items'], function($item){
  127. return in_array('h-entry', $item['type']);
  128. })) > 1) {
  129. #Parse::debug("mf2:2: Recognized $url as an h-feed because there are more than one object on the page");
  130. return self::parseAsHFeed($mf2, $http, $url);
  131. }
  132. }
  133. // If the first item is an h-feed, parse as a feed
  134. $first = $mf2['items'][0];
  135. if(in_array('h-feed', $first['type'])) {
  136. #Parse::debug("mf2:3: Recognized $url as an h-feed because the first item is an h-feed");
  137. return self::parseAsHFeed($mf2, $http, $url);
  138. }
  139. // Fallback case, but hopefully we have found something before this point
  140. foreach($mf2['items'] as $item) {
  141. // Otherwise check for a recognized h-* object
  142. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  143. #Parse::debug("mf2:6: $url is falling back to the first h-entry on the page");
  144. return self::parseAsHEntry($mf2, $item, $http, $url);
  145. } elseif(in_array('h-event', $item['type'])) {
  146. #Parse::debug("mf2:6: $url is falling back to the first h-event on the page");
  147. return self::parseAsHEvent($mf2, $item, $http, $url);
  148. } elseif(in_array('h-review', $item['type'])) {
  149. #Parse::debug("mf2:6: $url is falling back to the first h-review on the page");
  150. return self::parseAsHReview($mf2, $item, $http, $url);
  151. } elseif(in_array('h-recipe', $item['type'])) {
  152. #Parse::debug("mf2:6: $url is falling back to the first h-recipe on the page");
  153. return self::parseAsHRecipe($mf2, $item, $http, $url);
  154. } elseif(in_array('h-product', $item['type'])) {
  155. #Parse::debug("mf2:6: $url is falling back to the first h-product on the page");
  156. return self::parseAsHProduct($mf2, $item, $http, $url);
  157. } elseif(in_array('h-item', $item['type'])) {
  158. #Parse::debug("mf2:6: $url is falling back to the first h-item on the page");
  159. return self::parseAsHItem($mf2, $item, $http, $url);
  160. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  161. #Parse::debug("mf2:6: $url is falling back to the first h-item on the page");
  162. return self::parseAsHApp($mf2, $item, $http, $url);
  163. }
  164. }
  165. #Parse::debug("mf2:E: No object at $url was recognized");
  166. return false;
  167. }
  168. private static function collectSingleValues($properties, $urlProperties, $item, $url, &$data) {
  169. foreach($properties as $p) {
  170. if(($v = self::getPlaintext($item, $p)) !== null) {
  171. $data[$p] = $v;
  172. }
  173. }
  174. foreach($urlProperties as $p) {
  175. if($p == 'url') {
  176. // Special handling for the 'url' property to prioritize finding the URL on the same domain
  177. if($values = self::getPlaintextValues($item, 'url')) {
  178. if(count($values) == 1) {
  179. if(self::isURL($values[0]))
  180. $data['url'] = $values[0];
  181. }
  182. else {
  183. $set = false;
  184. foreach($values as $v) {
  185. if(self::isURL($v) && parse_url($v, PHP_URL_HOST) == parse_url($url, PHP_URL_HOST)) {
  186. $set = true;
  187. $data['url'] = $v;
  188. }
  189. }
  190. if(!$set) {
  191. // Fall back to the first URL if there isn't one on the domain
  192. if(self::isURL($values[0]))
  193. $data['url'] = $values[0];
  194. }
  195. }
  196. }
  197. } else {
  198. if(($v = self::getPlaintext($item, $p)) !== null) {
  199. if(self::isURL($v))
  200. $data[$p] = $v;
  201. }
  202. }
  203. }
  204. }
  205. private static function parseHTMLValue($property, $item) {
  206. if(!array_key_exists($property, $item['properties']))
  207. return null;
  208. $textContent = false;
  209. $htmlContent = false;
  210. $content = $item['properties'][$property][0];
  211. if(is_string($content)) {
  212. $textContent = $content;
  213. } elseif(!is_string($content) && is_array($content) && array_key_exists('value', $content)) {
  214. if(array_key_exists('html', $content)) {
  215. // Only allow images in the content if there is no photo property set
  216. if(isset($item['properties']['photo']))
  217. $allowImg = false;
  218. else
  219. $allowImg = true;
  220. $htmlContent = trim(self::sanitizeHTML($content['html'], $allowImg));
  221. #$textContent = trim(str_replace("&#xD;","\r",$content['value']));
  222. $textContent = trim(self::stripHTML($htmlContent));
  223. } else {
  224. $textContent = trim($content['value']);
  225. }
  226. }
  227. if($textContent || $htmlContent) {
  228. $data = [
  229. 'text' => $textContent
  230. ];
  231. // Only add HTML content if there is actual content.
  232. // If the text content ends up empty, then the HTML should be too
  233. // e.g. <div class="e-content"><a href=""><img src="" class="u-photo"></a></div>
  234. // should not return content of <a href=""></a>
  235. // TODO: still need to remove empty <a> tags when there is other text in the content
  236. if($htmlContent && $textContent && $textContent != $htmlContent) {
  237. $data['html'] = $htmlContent;
  238. }
  239. if(!$data['text'])
  240. return null;
  241. return $data;
  242. } else {
  243. return null;
  244. }
  245. }
  246. // Always return arrays, and may contain plaintext content
  247. // Nested objects are added to refs and the URL is used as the value if present
  248. private static function collectArrayValues($properties, $item, &$data, &$refs, &$http) {
  249. foreach($properties as $p) {
  250. if(array_key_exists($p, $item['properties'])) {
  251. foreach($item['properties'][$p] as $v) {
  252. if(is_string($v)) {
  253. if(!array_key_exists($p, $data)) $data[$p] = [];
  254. if(!in_array($v, $data[$p]))
  255. $data[$p][] = $v;
  256. } elseif(self::isMicroformat($v)) {
  257. if(($u=self::getPlaintext($v, 'url')) && self::isURL($u)) {
  258. if(!array_key_exists($p, $data)) $data[$p] = [];
  259. if(!in_array($u, $data[$p]))
  260. $data[$p][] = $u;
  261. $ref = self::parse(['items'=>[$v]], $u, $http);
  262. if($ref) {
  263. $refs[$u] = $ref['data'];
  264. }
  265. } else {
  266. if(!array_key_exists($p, $data)) $data[$p] = [];
  267. if(!in_array($v['value'], $data[$p]))
  268. $data[$p][] = $v['value'];
  269. }
  270. }
  271. }
  272. }
  273. }
  274. }
  275. private static function parseEmbeddedHCard($property, $item, &$http) {
  276. if(array_key_exists($property, $item['properties'])) {
  277. $mf2 = $item['properties'][$property][0];
  278. if(is_string($mf2) && self::isURL($mf2)) {
  279. $hcard = [
  280. 'type' => 'card',
  281. 'url' => $mf2
  282. ];
  283. return $hcard;
  284. } if(self::isMicroformat($mf2) && in_array('h-card', $mf2['type'])) {
  285. $hcard = [
  286. 'type' => 'card',
  287. ];
  288. $properties = ['name','latitude','longitude','locality','region','country','url'];
  289. foreach($properties as $p) {
  290. if($v=self::getPlaintext($mf2, $p)) {
  291. $hcard[$p] = $v;
  292. }
  293. }
  294. return $hcard;
  295. }
  296. }
  297. return false;
  298. }
  299. private static function collectArrayURLValues($properties, $item, &$data, &$refs, &$http) {
  300. $keys = [];
  301. foreach($properties as $p) {
  302. if(array_key_exists($p, $item['properties'])) {
  303. foreach($item['properties'][$p] as $v) {
  304. if(is_string($v) && self::isURL($v)) {
  305. if(!array_key_exists($p, $data)) $data[$p] = [];
  306. $data[$p][] = $v;
  307. $keys[] = $p;
  308. }
  309. elseif(self::isMicroformat($v) && ($u=self::getPlaintext($v, 'url')) && self::isURL($u)) {
  310. if(!array_key_exists($p, $data)) $data[$p] = [];
  311. $data[$p][] = $u;
  312. $keys[] = $p;
  313. // parse the object and put the result in the "refs" object
  314. $ref = self::parse(['items'=>[$v]], $u, $http);
  315. if($ref) {
  316. $refs[$u] = $ref['data'];
  317. }
  318. }
  319. }
  320. }
  321. }
  322. // Remove duplicate values
  323. foreach(array_unique($keys) as $key) {
  324. $data[$key] = array_unique($data[$key]);
  325. }
  326. }
  327. private static function determineNameAndContent($item, &$data) {
  328. // Determine if the name is distinct from the content
  329. $name = self::getPlaintext($item, 'name');
  330. $textContent = null;
  331. $htmlContent = null;
  332. $content = self::getHTMLValue($item, 'content');
  333. if(is_string($content)) {
  334. $textContent = $content;
  335. } elseif($content) {
  336. $htmlContent = array_key_exists('html', $content) ? $content['html'] : null;
  337. $textContent = array_key_exists('value', $content) ? $content['value'] : null;
  338. }
  339. $checkedname = $name;
  340. if($content) {
  341. // Trim ellipses from the name
  342. $name = preg_replace('/ ?(\.\.\.|…)$/', '', $name);
  343. // Remove all whitespace when checking equality
  344. $nameCompare = preg_replace('/\s/','',trim($name));
  345. $contentCompare = preg_replace('/\s/','',trim($textContent));
  346. // Check if the name is a prefix of the content
  347. if($contentCompare && $nameCompare && strpos($contentCompare, $nameCompare) === 0) {
  348. $checkedname = null;
  349. }
  350. }
  351. if($checkedname) {
  352. $data['name'] = $checkedname;
  353. }
  354. // If there is content, always return the plaintext content, and return HTML content if it's different
  355. if($content) {
  356. $content = self::parseHTMLValue('content', $item);
  357. if($content['text']) {
  358. $data['content']['text'] = $content['text'];
  359. if(isset($content['html']))
  360. $data['content']['html'] = $content['html'];
  361. } else {
  362. // If the content text was blank because the img was removed and that was the only content,
  363. // then put the name back as the name if it was previously set.
  364. // See https://github.com/aaronpk/XRay/issues/57
  365. if($name) {
  366. $data['name'] = $name;
  367. }
  368. }
  369. }
  370. }
  371. private static function parseAsHEntry($mf2, $item, $http, $url) {
  372. $data = [
  373. 'type' => 'entry'
  374. ];
  375. $refs = [];
  376. // Single plaintext and URL values
  377. self::collectSingleValues(['published','summary','rsvp','swarm-coins'], ['url','featured'], $item, $url, $data);
  378. if(isset($data['rsvp']))
  379. $data['rsvp'] = strtolower($data['rsvp']);
  380. // These properties are always returned as arrays and may contain plaintext content
  381. // First strip leading hashtags from category values if present
  382. if(array_key_exists('category', $item['properties'])) {
  383. foreach($item['properties']['category'] as $i=>$c) {
  384. if(is_string($c))
  385. $item['properties']['category'][$i] = ltrim($c, '#');
  386. }
  387. }
  388. self::collectArrayValues(['category','invitee'], $item, $data, $refs, $http);
  389. // These properties are always returned as arrays and always URLs
  390. // If the value is an h-* object with a URL, the URL is used and a "ref" is added as well
  391. self::collectArrayURLValues(['photo','video','audio','syndication','in-reply-to','like-of','repost-of','bookmark-of'], $item, $data, $refs, $http);
  392. self::determineNameAndContent($item, $data);
  393. if($author = self::findAuthor($mf2, $item, $http, $url))
  394. $data['author'] = $author;
  395. if($checkin = self::parseEmbeddedHCard('checkin', $item, $http))
  396. $data['checkin'] = $checkin;
  397. $response = [
  398. 'data' => $data
  399. ];
  400. if(count($refs)) {
  401. $response['data']['refs'] = $refs;
  402. }
  403. return $response;
  404. }
  405. private static function parseAsHReview($mf2, $item, $http, $url) {
  406. $data = [
  407. 'type' => 'review'
  408. ];
  409. $refs = [];
  410. self::collectSingleValues(['summary','published','rating','best','worst'], ['url'], $item, $url, $data);
  411. // Fallback for Mf1 "description" as content. The PHP parser does not properly map this to "content"
  412. $description = self::parseHTMLValue('description', $item);
  413. if($description) {
  414. $data['content'] = $description;
  415. }
  416. self::collectArrayValues(['category'], $item, $data, $refs, $http);
  417. self::collectArrayURLValues(['item'], $item, $data, $refs, $http);
  418. self::determineNameAndContent($item, $data);
  419. if($author = self::findAuthor($mf2, $item, $http, $url))
  420. $data['author'] = $author;
  421. $response = [
  422. 'data' => $data
  423. ];
  424. if(count($refs)) {
  425. $response['data']['refs'] = $refs;
  426. }
  427. return $response;
  428. }
  429. private static function parseAsHRecipe($mf2, $item, $http, $url) {
  430. $data = [
  431. 'type' => 'recipe'
  432. ];
  433. $refs = [];
  434. self::collectSingleValues(['name','summary','published','duration','yield','nutrition'], ['url'], $item, $url, $data);
  435. $instructions = self::parseHTMLValue('instructions', $item);
  436. if($instructions) {
  437. $data['instructions'] = $instructions;
  438. }
  439. self::collectArrayValues(['category','ingredient'], $item, $data, $refs, $http);
  440. self::collectArrayURLValues(['photo'], $item, $data, $refs, $http);
  441. if($author = self::findAuthor($mf2, $item, $http, $url))
  442. $data['author'] = $author;
  443. $response = [
  444. 'data' => $data
  445. ];
  446. if(count($refs)) {
  447. $response['data']['refs'] = $refs;
  448. }
  449. return $response;
  450. }
  451. private static function parseAsHProduct($mf2, $item, $http, $url) {
  452. $data = [
  453. 'type' => 'product'
  454. ];
  455. $refs = [];
  456. self::collectSingleValues(['name','identifier','price'], ['url'], $item, $url, $data);
  457. $description = self::parseHTMLValue('description', $item);
  458. if($description) {
  459. $data['description'] = $description;
  460. }
  461. self::collectArrayValues(['category','brand'], $item, $data, $refs, $http);
  462. self::collectArrayURLValues(['photo','video','audio'], $item, $data, $refs, $http);
  463. $response = [
  464. 'data' => $data
  465. ];
  466. if(count($refs)) {
  467. $response['data']['refs'] = $refs;
  468. }
  469. return $response;
  470. }
  471. private static function parseAsHItem($mf2, $item, $http, $url) {
  472. $data = [
  473. 'type' => 'item'
  474. ];
  475. $refs = [];
  476. self::collectSingleValues(['name'], ['url'], $item, $url, $data);
  477. self::collectArrayURLValues(['photo','video','audio'], $item, $data, $refs, $http);
  478. $response = [
  479. 'data' => $data
  480. ];
  481. if(count($refs)) {
  482. $response['data']['refs'] = $refs;
  483. }
  484. return $response;
  485. }
  486. private static function parseAsHApp($mf2, $item, $http, $url) {
  487. $data = [
  488. 'type' => 'app'
  489. ];
  490. self::collectSingleValues(['name'], ['url','logo'], $item, $url, $data);
  491. if(!isset($data['url']))
  492. $data['url'] = $url;
  493. $response = [
  494. 'data' => $data
  495. ];
  496. return $response;
  497. }
  498. private static function parseAsHEvent($mf2, $item, $http, $url) {
  499. $data = [
  500. 'type' => 'event'
  501. ];
  502. $refs = [];
  503. // Single plaintext and URL values
  504. self::collectSingleValues(['name','summary','published','start','end','duration'], ['url'], $item, $url, $data);
  505. // These properties are always returned as arrays and may contain plaintext content
  506. self::collectArrayValues(['category','location','attendee'], $item, $data, $refs, $http);
  507. // These properties are always returned as arrays and always URLs
  508. // If the value is an h-* object with a URL, the URL is used and a "ref" is added as well
  509. self::collectArrayURLValues(['photo','video','audio','syndication'], $item, $data, $refs, $http);
  510. // If there is a description, always return the plaintext description, and return HTML description if it's different
  511. $description = self::parseHTMLValue('description', $item);
  512. if($description) {
  513. $data['description'] = $description;
  514. }
  515. $response = [
  516. 'data' => $data
  517. ];
  518. if(count($refs)) {
  519. $response['data']['refs'] = $refs;
  520. }
  521. return $response;
  522. }
  523. private static function parseAsHCard($item, $http, $url, $authorURL=false) {
  524. $data = [
  525. 'type' => 'card',
  526. 'name' => null,
  527. 'url' => null,
  528. 'photo' => null
  529. ];
  530. $properties = ['url','name','photo'];
  531. foreach($properties as $p) {
  532. if($p == 'url' && $authorURL) {
  533. // If there is a matching author URL, use that one
  534. $found = false;
  535. foreach($item['properties']['url'] as $url) {
  536. if(self::isURL($url)) {
  537. $url = \p3k\XRay\normalize_url($url);
  538. if($url == \p3k\XRay\normalize_url($authorURL)) {
  539. $data['url'] = $url;
  540. $found = true;
  541. }
  542. }
  543. }
  544. if(!$found && self::isURL($item['properties']['url'][0])) {
  545. $data['url'] = $item['properties']['url'][0];
  546. }
  547. } else if(($v = self::getPlaintext($item, $p)) !== null) {
  548. // Make sure the URL property is actually a URL
  549. if($p == 'url' || $p == 'photo') {
  550. if(self::isURL($v))
  551. $data[$p] = $v;
  552. } else {
  553. $data[$p] = $v;
  554. }
  555. }
  556. }
  557. // If no URL property was found, use the $authorURL provided
  558. if(!$data['url'])
  559. $data['url'] = $authorURL;
  560. $response = [
  561. 'data' => $data
  562. ];
  563. return $response;
  564. }
  565. private static function findAuthor($mf2, $item, $http, $url) {
  566. $author = [
  567. 'type' => 'card',
  568. 'name' => null,
  569. 'url' => null,
  570. 'photo' => null
  571. ];
  572. // Author Discovery
  573. // http://indiewebcamp.com/authorship
  574. $authorPage = false;
  575. if(array_key_exists('author', $item['properties'])) {
  576. // Check if any of the values of the author property are an h-card
  577. foreach($item['properties']['author'] as $a) {
  578. if(self::isHCard($a)) {
  579. // 5.1 "if it has an h-card, use it, exit."
  580. return self::parseAsHCard($a, $http, $url)['data'];
  581. } elseif(is_string($a)) {
  582. if(self::isURL($a)) {
  583. // 5.2 "otherwise if author property is an http(s) URL, let the author-page have that URL"
  584. $authorPage = $a;
  585. } else {
  586. // 5.3 "otherwise use the author property as the author name, exit"
  587. // We can only set the name, no h-card or URL was found
  588. $author['name'] = self::getPlaintext($item, 'author');
  589. return $author;
  590. }
  591. } else {
  592. // This case is only hit when the author property is an mf2 object that is not an h-card
  593. $author['name'] = self::getPlaintext($item, 'author');
  594. return $author;
  595. }
  596. }
  597. }
  598. // 6. "if no author page was found" ... check for rel-author link
  599. if(!$authorPage) {
  600. if(isset($mf2['rels']) && isset($mf2['rels']['author']))
  601. $authorPage = $mf2['rels']['author'][0];
  602. }
  603. // 7. "if there is an author-page URL" ...
  604. if($authorPage) {
  605. // 7.1 "get the author-page from that URL and parse it for microformats2"
  606. $authorPageContents = self::getURL($authorPage, $http);
  607. if($authorPageContents) {
  608. foreach($authorPageContents['items'] as $i) {
  609. if(self::isHCard($i)) {
  610. // 7.2 "if author-page has 1+ h-card with url == uid == author-page's URL, then use first such h-card, exit."
  611. if(array_key_exists('url', $i['properties'])
  612. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['url']))
  613. and array_key_exists('uid', $i['properties'])
  614. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['uid']))
  615. ) {
  616. return self::parseAsHCard($i, $http, $url, $authorPage)['data'];
  617. }
  618. // 7.3 "else if author-page has 1+ h-card with url property which matches the href of a rel-me link on the author-page"
  619. $relMeLinks = (isset($authorPageContents['rels']) && isset($authorPageContents['rels']['me'])) ? $authorPageContents['rels']['me'] : [];
  620. if(count($relMeLinks) > 0
  621. and array_key_exists('url', $i['properties'])
  622. and count(array_intersect(\p3k\XRay\normalize_urls($i['properties']['url']), \p3k\XRay\normalize_urls($relMeLinks))) > 0
  623. ) {
  624. return self::parseAsHCard($i, $http, $url, $authorPage)['data'];
  625. }
  626. }
  627. }
  628. }
  629. // 7.4 "if the h-entry's page has 1+ h-card with url == author-page URL, use first such h-card, exit."
  630. foreach($mf2['items'] as $i) {
  631. if(self::isHCard($i)) {
  632. if(array_key_exists('url', $i['properties'])
  633. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['url']))
  634. ) {
  635. return self::parseAsHCard($i, $http, $url)['data'];
  636. }
  637. }
  638. // Also check the "author" property
  639. // (for finding the author of an h-feed's children when the author is the p-author property of the h-feed)
  640. if(isset($i['properties']['author'])) {
  641. foreach($i['properties']['author'] as $ic) {
  642. if(self::isHCard($ic)) {
  643. if(array_key_exists('url', $ic['properties'])
  644. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($ic['properties']['url']))
  645. ) {
  646. return self::parseAsHCard($ic, $http, $url)['data'];
  647. }
  648. }
  649. }
  650. }
  651. }
  652. }
  653. if(!$author['name'] && !$author['photo'] && !$author['url'])
  654. return null;
  655. return $author;
  656. }
  657. private static function hasNumericKeys(array $arr) {
  658. foreach($arr as $key=>$val)
  659. if (is_numeric($key))
  660. return true;
  661. return false;
  662. }
  663. private static function isMicroformat($mf) {
  664. return is_array($mf)
  665. and !self::hasNumericKeys($mf)
  666. and !empty($mf['type'])
  667. and isset($mf['properties']);
  668. }
  669. private static function isHCard($mf) {
  670. return is_array($mf)
  671. and !empty($mf['type'])
  672. and is_array($mf['type'])
  673. and in_array('h-card', $mf['type']);
  674. }
  675. private static function isURL($string) {
  676. return preg_match('/^https?:\/\/.+\..+$/', $string);
  677. }
  678. // Given an array of microformats properties and a key name, return the plaintext value
  679. // at that property
  680. // e.g.
  681. // {"properties":{"published":["foo"]}} results in "foo"
  682. private static function getPlaintext($mf2, $k, $fallback=null) {
  683. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  684. // $mf2['properties'][$v] will always be an array since the input was from the mf2 parser
  685. $value = $mf2['properties'][$k][0];
  686. if(is_string($value)) {
  687. return $value;
  688. } elseif(self::isMicroformat($value) && array_key_exists('value', $value)) {
  689. return $value['value'];
  690. }
  691. }
  692. return $fallback;
  693. }
  694. private static function getHTMLValue($mf2, $k, $fallback=null) {
  695. // Return an array with html and value if the value is html, otherwise return a string
  696. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  697. // $mf2['properties'][$v] will always be an array since the input was from the mf2 parser
  698. $value = $mf2['properties'][$k][0];
  699. if(is_string($value)) {
  700. return $value;
  701. } elseif(isset($value['html'])) {
  702. return $value;
  703. }
  704. }
  705. return $fallback;
  706. }
  707. private static function getPlaintextValues($mf2, $k, $values=[]) {
  708. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  709. foreach($mf2['properties'][$k] as $value) {
  710. if(is_string($value)) {
  711. $values[] = $value;
  712. } elseif(self::isMicroformat($value) && array_key_exists('value', $value)) {
  713. $values[] = $value['value'];
  714. }
  715. }
  716. }
  717. return $values;
  718. }
  719. private static function getURL($url, $http) {
  720. if(!$url || !$http) return null;
  721. // TODO: consider adding caching here
  722. $result = $http->get($url);
  723. if($result['error'] || !$result['body']) {
  724. return null;
  725. }
  726. return \mf2\Parse($result['body'], $url);
  727. }
  728. }