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.

934 lines
33 KiB

4 years ago
4 years ago
6 years ago
  1. <?php
  2. namespace p3k\XRay\Formats;
  3. class Mf2 extends Format {
  4. use Mf2Feed;
  5. public static function matches_host($url) {
  6. return true;
  7. }
  8. public static function matches($url) {
  9. return true;
  10. }
  11. public static function parse($http_response, $http, $opts=[]) {
  12. $mf2 = $http_response['body'];
  13. $url = $http_response['url'];
  14. if(!isset($mf2['items']) || count($mf2['items']) == 0)
  15. return false;
  16. // If they are expecting a feed, always return a feed or an error
  17. if(isset($opts['expect']) && $opts['expect'] == 'feed') {
  18. return self::parseAsHFeed($mf2, $http, $url);
  19. }
  20. // Remove h-breadcrumb since we never use it and it causes problems determining
  21. // whether a page is a feed or permalink
  22. $mf2['items'] = array_values(array_filter($mf2['items'], function($item){
  23. return !in_array('h-breadcrumb', $item['type']);
  24. }));
  25. $items = $mf2['items'];
  26. // If there is more than one item on the page, it may be a feed.
  27. // Remove an h-card if there is one that doesn't match the page URL, then try again.
  28. // (Don't modify the actual tree, but compare on the modified tree)
  29. if(count($items) > 1) {
  30. $tmpmf2 = array_filter($items, function($item) use($url){
  31. return !(in_array('h-card', $item['type']) && isset($item['properties']['url'][0]) && $item['properties']['url'][0] != $url);
  32. });
  33. $items = array_values($tmpmf2);
  34. }
  35. // If there is only one item left on the page, it's a permalink, and just use that
  36. if(count($items) == 1) {
  37. $item = $items[0];
  38. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  39. #Parse::debug("mf2:0: Recognized $url as an h-entry it is the only item on the page");
  40. return self::parseAsHEntry($mf2, $item, $http, $url);
  41. }
  42. if(in_array('h-event', $item['type'])) {
  43. #Parse::debug("mf2:0: Recognized $url as an h-event it is the only item on the page");
  44. return self::parseAsHEvent($mf2, $item, $http, $url);
  45. }
  46. if(in_array('h-review', $item['type'])) {
  47. #Parse::debug("mf2:0: Recognized $url as an h-review it is the only item on the page");
  48. return self::parseAsHReview($mf2, $item, $http, $url);
  49. }
  50. if(in_array('h-recipe', $item['type'])) {
  51. #Parse::debug("mf2:0: Recognized $url as an h-recipe it is the only item on the page");
  52. return self::parseAsHRecipe($mf2, $item, $http, $url);
  53. }
  54. if(in_array('h-product', $item['type'])) {
  55. #Parse::debug("mf2:0: Recognized $url as an h-product it is the only item on the page");
  56. return self::parseAsHProduct($mf2, $item, $http, $url);
  57. }
  58. if(in_array('h-item', $item['type'])) {
  59. #Parse::debug("mf2:0: Recognized $url as an h-product it is the only item on the page");
  60. return self::parseAsHItem($mf2, $item, $http, $url);
  61. }
  62. if(in_array('h-card', $item['type'])) {
  63. #Parse::debug("mf2:0: Recognized $url as an h-card it is the only item on the page");
  64. return self::parseAsHCard($item, $http, $url, $url);
  65. }
  66. if(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  67. #Parse::debug("mf2:0: Recognized $url as an h-feed because it is the only item on the page");
  68. return self::parseAsHApp($mf2, $item, $http, $url);
  69. }
  70. if(in_array('h-feed', $item['type'])) {
  71. #Parse::debug("mf2:0: Recognized $url as an h-feed because it is the only item on the page");
  72. return self::parseAsHFeed($mf2, $http, $url);
  73. }
  74. }
  75. // Check the list of items on the page to see if one matches the URL of the page,
  76. // and treat as a permalink for that object if so.
  77. foreach($mf2['items'] as $item) {
  78. if(array_key_exists('url', $item['properties'])) {
  79. $urls = $item['properties']['url'];
  80. $urls = array_map('\p3k\XRay\normalize_url', $urls);
  81. if(in_array($url, $urls)) {
  82. #Parse::debug("mf2:1: Recognized $url as a permalink because an object on the page matched the URL of the request");
  83. if(in_array('h-card', $item['type'])) {
  84. return self::parseAsHCard($item, $http, $url, $url);
  85. } elseif(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  86. return self::parseAsHEntry($mf2, $item, $http, $url);
  87. } elseif(in_array('h-event', $item['type'])) {
  88. return self::parseAsHEvent($mf2, $item, $http, $url);
  89. } elseif(in_array('h-review', $item['type'])) {
  90. return self::parseAsHReview($mf2, $item, $http, $url);
  91. } elseif(in_array('h-recipe', $item['type'])) {
  92. return self::parseAsHRecipe($mf2, $item, $http, $url);
  93. } elseif(in_array('h-product', $item['type'])) {
  94. return self::parseAsHProduct($mf2, $item, $http, $url);
  95. } elseif(in_array('h-item', $item['type'])) {
  96. return self::parseAsHItem($mf2, $item, $http, $url);
  97. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  98. return self::parseAsHApp($mf2, $item, $http, $url);
  99. } elseif(in_array('h-feed', $item['type'])) {
  100. return self::parseAsHFeed($mf2, $http, $url);
  101. } else {
  102. #Parse::debug('This object was not a recognized type.');
  103. return false;
  104. }
  105. }
  106. }
  107. }
  108. // Check for an h-card matching rel=author or the author URL of any h-* on the page,
  109. // and return the h-* object if so
  110. if(isset($mf2['rels']['author'])) {
  111. foreach($mf2['items'] as $card) {
  112. if(in_array('h-card', $card['type']) && array_key_exists('url', $card['properties'])) {
  113. $urls = \p3k\XRay\normalize_urls($card['properties']['url']);
  114. if(count(array_intersect($urls, \p3k\XRay\normalize_urls($mf2['rels']['author']))) > 0) {
  115. // There is an author h-card on this page
  116. // Now look for the first h-* object other than an h-card and use that as the object
  117. foreach($mf2['items'] as $item) {
  118. if(!in_array('h-card', $item['type'])) {
  119. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  120. return self::parseAsHEntry($mf2, $item, $http, $url);
  121. } elseif(in_array('h-event', $item['type'])) {
  122. return self::parseAsHEvent($mf2, $item, $http, $url);
  123. } elseif(in_array('h-review', $item['type'])) {
  124. return self::parseAsHReview($mf2, $item, $http, $url);
  125. } elseif(in_array('h-recipe', $item['type'])) {
  126. return self::parseAsHRecipe($mf2, $item, $http, $url);
  127. } elseif(in_array('h-product', $item['type'])) {
  128. return self::parseAsHProduct($mf2, $item, $http, $url);
  129. } elseif(in_array('h-item', $item['type'])) {
  130. return self::parseAsHItem($mf2, $item, $http, $url);
  131. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  132. return self::parseAsHApp($mf2, $item, $http, $url);
  133. }
  134. }
  135. }
  136. }
  137. }
  138. }
  139. }
  140. // At this point, if there are any h-entrys left on the page, it's probably a feed.
  141. if(count($items) > 0) {
  142. if(count(array_filter($items, function($item){
  143. return in_array('h-entry', $item['type']);
  144. })) > 0) {
  145. #Parse::debug("mf2:2: Recognized $url as an h-feed because there are more than one object on the page");
  146. return self::parseAsHFeed($mf2, $http, $url);
  147. }
  148. }
  149. // If the first item is an h-feed, parse as a feed
  150. $first = $items[0];
  151. if(in_array('h-feed', $first['type'])) {
  152. #Parse::debug("mf2:3: Recognized $url as an h-feed because the first item is an h-feed");
  153. return self::parseAsHFeed($mf2, $http, $url);
  154. }
  155. // Fallback case, but hopefully we have found something before this point
  156. foreach($mf2['items'] as $item) {
  157. // Otherwise check for a recognized h-* object
  158. if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
  159. #Parse::debug("mf2:6: $url is falling back to the first h-entry on the page");
  160. return self::parseAsHEntry($mf2, $item, $http, $url);
  161. } elseif(in_array('h-event', $item['type'])) {
  162. #Parse::debug("mf2:6: $url is falling back to the first h-event on the page");
  163. return self::parseAsHEvent($mf2, $item, $http, $url);
  164. } elseif(in_array('h-review', $item['type'])) {
  165. #Parse::debug("mf2:6: $url is falling back to the first h-review on the page");
  166. return self::parseAsHReview($mf2, $item, $http, $url);
  167. } elseif(in_array('h-recipe', $item['type'])) {
  168. #Parse::debug("mf2:6: $url is falling back to the first h-recipe on the page");
  169. return self::parseAsHRecipe($mf2, $item, $http, $url);
  170. } elseif(in_array('h-product', $item['type'])) {
  171. #Parse::debug("mf2:6: $url is falling back to the first h-product on the page");
  172. return self::parseAsHProduct($mf2, $item, $http, $url);
  173. } elseif(in_array('h-item', $item['type'])) {
  174. #Parse::debug("mf2:6: $url is falling back to the first h-item on the page");
  175. return self::parseAsHItem($mf2, $item, $http, $url);
  176. } elseif(in_array('h-app', $item['type']) || in_array('h-x-app', $item['type'])) {
  177. #Parse::debug("mf2:6: $url is falling back to the first h-item on the page");
  178. return self::parseAsHApp($mf2, $item, $http, $url);
  179. }
  180. }
  181. #Parse::debug("mf2:E: No object at $url was recognized");
  182. return false;
  183. }
  184. private static function collectSingleValues($properties, $urlProperties, $item, $url, &$data) {
  185. foreach($properties as $p) {
  186. if(($v = self::getPlaintext($item, $p)) !== null) {
  187. $data[$p] = $v;
  188. }
  189. }
  190. foreach($urlProperties as $p) {
  191. if($p == 'url') {
  192. // Special handling for the 'url' property to prioritize finding the URL on the same domain
  193. if($values = self::getPlaintextValues($item, 'url')) {
  194. if(count($values) == 1) {
  195. if(self::isURL($values[0]))
  196. $data['url'] = $values[0];
  197. }
  198. else {
  199. $set = false;
  200. foreach($values as $v) {
  201. if(self::isURL($v) && parse_url($v, PHP_URL_HOST) == parse_url($url, PHP_URL_HOST)) {
  202. $set = true;
  203. $data['url'] = $v;
  204. }
  205. }
  206. if(!$set) {
  207. // Fall back to the first URL if there isn't one on the domain
  208. if(self::isURL($values[0]))
  209. $data['url'] = $values[0];
  210. }
  211. }
  212. }
  213. } else {
  214. if(($v = self::getPlaintext($item, $p)) !== null) {
  215. if(self::isURL($v))
  216. $data[$p] = $v;
  217. }
  218. }
  219. }
  220. }
  221. private static function parseHTMLValue($property, $item) {
  222. if(!array_key_exists($property, $item['properties']))
  223. return null;
  224. $textContent = false;
  225. $htmlContent = false;
  226. $content = $item['properties'][$property][0];
  227. if(is_string($content)) {
  228. $textContent = $content;
  229. } elseif(!is_string($content) && is_array($content) && array_key_exists('html', $content)) {
  230. if(array_key_exists('html', $content)) {
  231. // Only allow images in the content if there is no photo property set
  232. if(isset($item['properties']['photo']))
  233. $allowImg = false;
  234. else
  235. $allowImg = true;
  236. $htmlContent = trim(self::sanitizeHTML($content['html'], $allowImg));
  237. #$textContent = trim(str_replace("&#xD;","\r",$content['value']));
  238. $textContent = trim(self::stripHTML($htmlContent));
  239. } else {
  240. if(isset($content['value']))
  241. $textContent = trim($content['value']);
  242. }
  243. }
  244. if($textContent || $htmlContent) {
  245. $data = [
  246. 'text' => $textContent
  247. ];
  248. // Only add HTML content if there is actual content.
  249. // If the text content ends up empty, then the HTML should be too
  250. // e.g. <div class="e-content"><a href=""><img src="" class="u-photo"></a></div>
  251. // should not return content of <a href=""></a>
  252. // TODO: still need to remove empty <a> tags when there is other text in the content
  253. if($htmlContent && $textContent && $textContent != $htmlContent) {
  254. $data['html'] = $htmlContent;
  255. }
  256. if(!$data['text'])
  257. return null;
  258. return $data;
  259. } else {
  260. return null;
  261. }
  262. }
  263. // Always return arrays, and may contain plaintext content
  264. // Nested objects are added to refs and the URL is used as the value if present
  265. private static function collectArrayValues($properties, $item, &$data, &$refs, &$http) {
  266. foreach($properties as $p) {
  267. if(array_key_exists($p, $item['properties'])) {
  268. foreach($item['properties'][$p] as $v) {
  269. if(is_string($v)) {
  270. if(!array_key_exists($p, $data)) $data[$p] = [];
  271. if(!in_array($v, $data[$p]))
  272. $data[$p][] = $v;
  273. } elseif(self::isMicroformat($v)) {
  274. if(($u=self::getPlaintext($v, 'url')) && self::isURL($u)) {
  275. if(!array_key_exists($p, $data)) $data[$p] = [];
  276. if(!in_array($u, $data[$p]))
  277. $data[$p][] = $u;
  278. $ref = self::parse([
  279. 'body' => ['items'=>[$v]],
  280. 'url' => $u,
  281. 'code' => null,
  282. ], $http);
  283. if($ref) {
  284. $refs[$u] = $ref['data'];
  285. }
  286. } else {
  287. if(!array_key_exists($p, $data)) $data[$p] = [];
  288. if(!in_array($v['value'], $data[$p]))
  289. $data[$p][] = $v['value'];
  290. }
  291. }
  292. }
  293. }
  294. }
  295. }
  296. private static function parseEmbeddedHCard($property, $item, &$http) {
  297. if(array_key_exists($property, $item['properties'])) {
  298. $mf2 = $item['properties'][$property][0];
  299. if(is_string($mf2) && self::isURL($mf2)) {
  300. $hcard = [
  301. 'type' => 'card',
  302. 'url' => $mf2
  303. ];
  304. return $hcard;
  305. } if(self::isMicroformat($mf2) && in_array('h-card', $mf2['type'])) {
  306. $hcard = [
  307. 'type' => 'card',
  308. ];
  309. $properties = ['name','latitude','longitude','locality','region','country','url'];
  310. foreach($properties as $p) {
  311. if($v=self::getPlaintext($mf2, $p)) {
  312. $hcard[$p] = $v;
  313. }
  314. }
  315. return $hcard;
  316. }
  317. }
  318. return false;
  319. }
  320. private static function collectArrayURLValues($properties, $item, &$data, &$refs, &$http) {
  321. $keys = [];
  322. foreach($properties as $p) {
  323. if(array_key_exists($p, $item['properties'])) {
  324. foreach($item['properties'][$p] as $v) {
  325. if(is_string($v) && self::isURL($v)) {
  326. if(!array_key_exists($p, $data)) $data[$p] = [];
  327. $data[$p][] = $v;
  328. $keys[] = $p;
  329. }
  330. elseif(self::isMicroformat($v) && ($u=self::getPlaintext($v, 'url')) && self::isURL($u)) {
  331. if(!array_key_exists($p, $data)) $data[$p] = [];
  332. $data[$p][] = $u;
  333. $keys[] = $p;
  334. // parse the object and put the result in the "refs" object
  335. $ref = self::parse([
  336. 'body' => ['items'=>[$v]],
  337. 'url' => $u,
  338. 'code' => null,
  339. ], $http);
  340. if($ref) {
  341. $refs[$u] = $ref['data'];
  342. }
  343. }
  344. }
  345. }
  346. }
  347. // Remove duplicate values
  348. foreach(array_unique($keys) as $key) {
  349. $data[$key] = array_unique($data[$key]);
  350. }
  351. }
  352. private static function determineNameAndContent($item, &$data) {
  353. // Determine if the name is distinct from the content
  354. $name = self::getPlaintext($item, 'name');
  355. $textContent = null;
  356. $htmlContent = null;
  357. $content = self::getHTMLValue($item, 'content');
  358. if(is_string($content)) {
  359. $textContent = $content;
  360. } elseif($content) {
  361. $htmlContent = array_key_exists('html', $content) ? $content['html'] : null;
  362. $textContent = array_key_exists('value', $content) ? $content['value'] : null;
  363. }
  364. $checkedname = $name;
  365. if($content) {
  366. // Trim ellipses from the name
  367. $name = preg_replace('/ ?(\.\.\.|…)$/', '', $name);
  368. // Remove all whitespace when checking equality
  369. $nameCompare = preg_replace('/\s/','',trim($name));
  370. $contentCompare = preg_replace('/\s/','',trim($textContent));
  371. // Check if the name is a prefix of the content
  372. if($contentCompare && $nameCompare && strpos($contentCompare, $nameCompare) === 0) {
  373. $checkedname = null;
  374. }
  375. }
  376. if($checkedname) {
  377. $data['name'] = $checkedname;
  378. }
  379. // If there is content, always return the plaintext content, and return HTML content if it's different
  380. if($content) {
  381. $content = self::parseHTMLValue('content', $item);
  382. if($content['text']) {
  383. $data['content']['text'] = $content['text'];
  384. if(isset($content['html']))
  385. $data['content']['html'] = $content['html'];
  386. } else {
  387. // If the content text was blank because the img was removed and that was the only content,
  388. // then put the name back as the name if it was previously set.
  389. // See https://github.com/aaronpk/XRay/issues/57
  390. if($name) {
  391. $data['name'] = $name;
  392. }
  393. }
  394. }
  395. }
  396. private static function parseAsHEntry($mf2, $item, $http, $url) {
  397. $data = [
  398. 'type' => 'entry'
  399. ];
  400. $refs = [];
  401. // Single plaintext and URL values
  402. self::collectSingleValues(['published','summary','rsvp','swarm-coins'], ['url','featured','follow-of'], $item, $url, $data);
  403. if(isset($data['rsvp']))
  404. $data['rsvp'] = strtolower($data['rsvp']);
  405. // These properties are always returned as arrays and may contain plaintext content
  406. // First strip leading hashtags from category values if present
  407. if(array_key_exists('category', $item['properties'])) {
  408. foreach($item['properties']['category'] as $i=>$c) {
  409. if(is_string($c))
  410. $item['properties']['category'][$i] = ltrim($c, '#');
  411. }
  412. }
  413. self::collectArrayValues(['category','invitee'], $item, $data, $refs, $http);
  414. // These properties are always returned as arrays and always URLs
  415. // If the value is an h-* object with a URL, the URL is used and a "ref" is added as well
  416. self::collectArrayURLValues(['photo','video','audio','syndication','in-reply-to','like-of','repost-of','bookmark-of','quotation-of'], $item, $data, $refs, $http);
  417. // Hack to make quotation-of a single value
  418. if(isset($data['quotation-of']))
  419. $data['quotation-of'] = $data['quotation-of'][0];
  420. self::determineNameAndContent($item, $data);
  421. if($author = self::findAuthor($mf2, $item, $http, $url))
  422. $data['author'] = $author;
  423. if($checkin = self::parseEmbeddedHCard('checkin', $item, $http))
  424. $data['checkin'] = $checkin;
  425. $data['post-type'] = \p3k\XRay\PostType::discover($data);
  426. $response = [
  427. 'data' => $data,
  428. ];
  429. if(count($refs)) {
  430. $response['data']['refs'] = $refs;
  431. }
  432. return $response;
  433. }
  434. private static function parseAsHReview($mf2, $item, $http, $url) {
  435. $data = [
  436. 'type' => 'review'
  437. ];
  438. $refs = [];
  439. self::collectSingleValues(['summary','published','rating','best','worst'], ['url'], $item, $url, $data);
  440. // Fallback for Mf1 "description" as content. The PHP parser does not properly map this to "content"
  441. $description = self::parseHTMLValue('description', $item);
  442. if($description) {
  443. $data['content'] = $description;
  444. }
  445. self::collectArrayValues(['category'], $item, $data, $refs, $http);
  446. self::collectArrayURLValues(['item'], $item, $data, $refs, $http);
  447. self::determineNameAndContent($item, $data);
  448. if($author = self::findAuthor($mf2, $item, $http, $url))
  449. $data['author'] = $author;
  450. $data['post-type'] = \p3k\XRay\PostType::discover($data);
  451. $response = [
  452. 'data' => $data
  453. ];
  454. if(count($refs)) {
  455. $response['data']['refs'] = $refs;
  456. }
  457. return $response;
  458. }
  459. private static function parseAsHRecipe($mf2, $item, $http, $url) {
  460. $data = [
  461. 'type' => 'recipe',
  462. ];
  463. $refs = [];
  464. self::collectSingleValues(['name','summary','published','duration','yield','nutrition'], ['url'], $item, $url, $data);
  465. $instructions = self::parseHTMLValue('instructions', $item);
  466. if($instructions) {
  467. $data['instructions'] = $instructions;
  468. }
  469. self::collectArrayValues(['category','ingredient'], $item, $data, $refs, $http);
  470. self::collectArrayURLValues(['photo'], $item, $data, $refs, $http);
  471. if($author = self::findAuthor($mf2, $item, $http, $url))
  472. $data['author'] = $author;
  473. $data['post-type'] = \p3k\XRay\PostType::discover($data);
  474. $response = [
  475. 'data' => $data
  476. ];
  477. if(count($refs)) {
  478. $response['data']['refs'] = $refs;
  479. }
  480. return $response;
  481. }
  482. private static function parseAsHProduct($mf2, $item, $http, $url) {
  483. $data = [
  484. 'type' => 'product'
  485. ];
  486. $refs = [];
  487. self::collectSingleValues(['name','identifier','price'], ['url'], $item, $url, $data);
  488. $description = self::parseHTMLValue('description', $item);
  489. if($description) {
  490. $data['description'] = $description;
  491. }
  492. self::collectArrayValues(['category','brand'], $item, $data, $refs, $http);
  493. self::collectArrayURLValues(['photo','video','audio'], $item, $data, $refs, $http);
  494. $response = [
  495. 'data' => $data
  496. ];
  497. if(count($refs)) {
  498. $response['data']['refs'] = $refs;
  499. }
  500. return $response;
  501. }
  502. private static function parseAsHItem($mf2, $item, $http, $url) {
  503. $data = [
  504. 'type' => 'item'
  505. ];
  506. $refs = [];
  507. self::collectSingleValues(['name'], ['url'], $item, $url, $data);
  508. self::collectArrayURLValues(['photo','video','audio'], $item, $data, $refs, $http);
  509. $response = [
  510. 'data' => $data
  511. ];
  512. if(count($refs)) {
  513. $response['data']['refs'] = $refs;
  514. }
  515. return $response;
  516. }
  517. private static function parseAsHApp($mf2, $item, $http, $url) {
  518. $data = [
  519. 'type' => 'app'
  520. ];
  521. self::collectSingleValues(['name'], ['url','logo'], $item, $url, $data);
  522. self::collectArrayURLValues(['redirect-uri'], $item, $data, $refs, $http);
  523. if(!isset($data['url']))
  524. $data['url'] = $url;
  525. if(isset($mf2['rels']['redirect_uri'])) {
  526. if(!isset($data['redirect-uri'])) $data['redirect-uri'] = [];
  527. $data['redirect-uri'] = array_merge($data['redirect-uri'], $mf2['rels']['redirect_uri']);
  528. }
  529. if(isset($data['redirect-uri'])) {
  530. $data['redirect-uri'] = array_values(array_unique($data['redirect-uri']));
  531. }
  532. $response = [
  533. 'data' => $data
  534. ];
  535. return $response;
  536. }
  537. private static function parseAsHEvent($mf2, $item, $http, $url) {
  538. $data = [
  539. 'type' => 'event'
  540. ];
  541. $refs = [];
  542. // Single plaintext and URL values
  543. self::collectSingleValues(['name','summary','published','start','end','duration'], ['url'], $item, $url, $data);
  544. // These properties are always returned as arrays and may contain plaintext content
  545. self::collectArrayValues(['category','location','attendee'], $item, $data, $refs, $http);
  546. // These properties are always returned as arrays and always URLs
  547. // If the value is an h-* object with a URL, the URL is used and a "ref" is added as well
  548. self::collectArrayURLValues(['photo','video','audio','syndication'], $item, $data, $refs, $http);
  549. // If there is a description, always return the plaintext content, and return HTML content if it's different
  550. $content = self::parseHTMLValue('content', $item);
  551. if($content) {
  552. $data['content'] = $content;
  553. } else {
  554. // Fall back to looking for "description"
  555. $content = self::parseHTMLValue('description', $item);
  556. if($content)
  557. $data['content'] = $content;
  558. }
  559. $data['post-type'] = \p3k\XRay\PostType::discover($data);
  560. $response = [
  561. 'data' => $data
  562. ];
  563. if(count($refs)) {
  564. $response['data']['refs'] = $refs;
  565. }
  566. return $response;
  567. }
  568. private static function parseAsHCard($item, $http, $url, $authorURL=false) {
  569. $data = [
  570. 'type' => 'card',
  571. 'name' => null,
  572. 'url' => null,
  573. 'photo' => null
  574. ];
  575. $properties = ['url','name','photo'];
  576. foreach($properties as $p) {
  577. if($p == 'url' && $authorURL) {
  578. // If there is a matching author URL, use that one
  579. $found = false;
  580. foreach($item['properties']['url'] as $url) {
  581. if(self::isURL($url)) {
  582. $url = \p3k\XRay\normalize_url($url);
  583. if($url == \p3k\XRay\normalize_url($authorURL)) {
  584. $data['url'] = $url;
  585. $found = true;
  586. }
  587. }
  588. }
  589. if(!$found && self::isURL($item['properties']['url'][0])) {
  590. $data['url'] = $item['properties']['url'][0];
  591. }
  592. } else if(($v = self::getPlaintext($item, $p)) !== null) {
  593. // Make sure the URL property is actually a URL
  594. if($p == 'url' || $p == 'photo') {
  595. if(self::isURL($v))
  596. $data[$p] = $v;
  597. } else {
  598. $data[$p] = $v;
  599. }
  600. }
  601. }
  602. // If no URL property was found, use the $authorURL provided
  603. if(!$data['url'])
  604. $data['url'] = $authorURL;
  605. $response = [
  606. 'data' => $data
  607. ];
  608. return $response;
  609. }
  610. private static function findAuthor($mf2, $item, $http, $url) {
  611. $author = [
  612. 'type' => 'card',
  613. 'name' => null,
  614. 'url' => null,
  615. 'photo' => null
  616. ];
  617. // Author Discovery
  618. // http://indiewebcamp.com/authorship
  619. $authorPage = false;
  620. if(array_key_exists('author', $item['properties'])) {
  621. // Check if any of the values of the author property are an h-card
  622. foreach($item['properties']['author'] as $a) {
  623. if(self::isHCard($a)) {
  624. // 5.1 "if it has an h-card, use it, exit."
  625. return self::parseAsHCard($a, $http, $url)['data'];
  626. } elseif(is_string($a)) {
  627. if(self::isURL($a)) {
  628. // 5.2 "otherwise if author property is an http(s) URL, let the author-page have that URL"
  629. $authorPage = $a;
  630. } else {
  631. // 5.3 "otherwise use the author property as the author name, exit"
  632. // We can only set the name, no h-card or URL was found
  633. $author['name'] = self::getPlaintext($item, 'author');
  634. return $author;
  635. }
  636. } else {
  637. // This case is only hit when the author property is an mf2 object that is not an h-card
  638. $author['name'] = self::getPlaintext($item, 'author');
  639. return $author;
  640. }
  641. }
  642. }
  643. // 6. "if no author page was found" ... check for rel-author link
  644. if(!$authorPage) {
  645. if(isset($mf2['rels']) && isset($mf2['rels']['author']))
  646. $authorPage = $mf2['rels']['author'][0];
  647. }
  648. // 7. "if there is an author-page URL" ...
  649. if($authorPage) {
  650. // 7.1 "get the author-page from that URL and parse it for microformats2"
  651. $authorPageContents = self::getURL($authorPage, $http);
  652. if($authorPageContents) {
  653. foreach($authorPageContents['items'] as $i) {
  654. if(self::isHCard($i)) {
  655. // 7.2 "if author-page has 1+ h-card with url == uid == author-page's URL, then use first such h-card, exit."
  656. if(array_key_exists('url', $i['properties'])
  657. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['url']))
  658. and array_key_exists('uid', $i['properties'])
  659. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['uid']))
  660. ) {
  661. return self::parseAsHCard($i, $http, $url, $authorPage)['data'];
  662. }
  663. // 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"
  664. $relMeLinks = (isset($authorPageContents['rels']) && isset($authorPageContents['rels']['me'])) ? $authorPageContents['rels']['me'] : [];
  665. if(count($relMeLinks) > 0
  666. and array_key_exists('url', $i['properties'])
  667. and count(array_intersect(\p3k\XRay\normalize_urls($i['properties']['url']), \p3k\XRay\normalize_urls($relMeLinks))) > 0
  668. ) {
  669. return self::parseAsHCard($i, $http, $url, $authorPage)['data'];
  670. }
  671. }
  672. }
  673. }
  674. // 7.4 "if the h-entry's page has 1+ h-card with url == author-page URL, use first such h-card, exit."
  675. foreach($mf2['items'] as $i) {
  676. if(self::isHCard($i)) {
  677. if(array_key_exists('url', $i['properties'])
  678. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['url']))
  679. ) {
  680. return self::parseAsHCard($i, $http, $url)['data'];
  681. }
  682. }
  683. // Also check the "author" property
  684. // (for finding the author of an h-feed's children when the author is the p-author property of the h-feed)
  685. if(isset($i['properties']['author'])) {
  686. foreach($i['properties']['author'] as $ic) {
  687. if(self::isHCard($ic)) {
  688. if(array_key_exists('url', $ic['properties'])
  689. and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($ic['properties']['url']))
  690. ) {
  691. return self::parseAsHCard($ic, $http, $url)['data'];
  692. }
  693. }
  694. }
  695. }
  696. }
  697. }
  698. // The below is not yet in the authorship algorithm.
  699. // If the top object is an h-feed, check for an author property there
  700. if(isset($mf2['items'][0]['type'][0]) && in_array('h-feed', $mf2['items'][0]['type'])) {
  701. if(isset($mf2['items'][0]['properties']['author'][0])) {
  702. $potentialAuthor = $mf2['items'][0]['properties']['author'][0];
  703. if(is_array($potentialAuthor['type']) && in_array('h-card', $potentialAuthor['type'])) {
  704. return self::parseAsHCard($potentialAuthor, $http, $url)['data'];
  705. }
  706. }
  707. }
  708. // If still no author is found, and this page is a feed (list of h-*),
  709. // then use the first h-card in the list of items.
  710. $items = array_filter($mf2['items'], function($item){
  711. return !in_array('h-card', $item['type']);
  712. });
  713. if(count($items) > 1) {
  714. $card = self::_findFirstOfType($mf2, 'h-card');
  715. if($card) {
  716. return self::parseAsHCard($card, $http, $url)['data'];
  717. }
  718. }
  719. if(!$author['name'] && !$author['photo'] && !$author['url'])
  720. return null;
  721. return $author;
  722. }
  723. private static function hasNumericKeys(array $arr) {
  724. foreach($arr as $key=>$val)
  725. if (is_numeric($key))
  726. return true;
  727. return false;
  728. }
  729. private static function isMicroformat($mf) {
  730. return is_array($mf)
  731. and !self::hasNumericKeys($mf)
  732. and !empty($mf['type'])
  733. and isset($mf['properties']);
  734. }
  735. private static function isHCard($mf) {
  736. return is_array($mf)
  737. and !empty($mf['type'])
  738. and is_array($mf['type'])
  739. and in_array('h-card', $mf['type']);
  740. }
  741. private static function isURL($string) {
  742. return preg_match('/^https?:\/\/.+\..+$/', $string);
  743. }
  744. // Given an array of microformats properties and a key name, return the plaintext value
  745. // at that property
  746. // e.g.
  747. // {"properties":{"published":["foo"]}} results in "foo"
  748. private static function getPlaintext($mf2, $k, $fallback=null) {
  749. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  750. // $mf2['properties'][$v] will always be an array since the input was from the mf2 parser
  751. $value = $mf2['properties'][$k][0];
  752. if(is_string($value)) {
  753. return $value;
  754. } elseif(self::isMicroformat($value) && array_key_exists('value', $value)) {
  755. return $value['value'];
  756. }
  757. }
  758. return $fallback;
  759. }
  760. private static function getHTMLValue($mf2, $k, $fallback=null) {
  761. // Return an array with html and value if the value is html, otherwise return a string
  762. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  763. // $mf2['properties'][$v] will always be an array since the input was from the mf2 parser
  764. $value = $mf2['properties'][$k][0];
  765. if(is_string($value)) {
  766. return $value;
  767. } elseif(isset($value['html'])) {
  768. return $value;
  769. }
  770. }
  771. return $fallback;
  772. }
  773. private static function getPlaintextValues($mf2, $k, $values=[]) {
  774. if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
  775. foreach($mf2['properties'][$k] as $value) {
  776. if(is_string($value)) {
  777. $values[] = $value;
  778. } elseif(self::isMicroformat($value) && array_key_exists('value', $value)) {
  779. $values[] = $value['value'];
  780. }
  781. }
  782. }
  783. return $values;
  784. }
  785. private static function getURL($url, $http) {
  786. if(!$url || !$http) return null;
  787. // TODO: consider adding caching here
  788. $result = $http->get($url);
  789. if($result['error'] || !$result['body']) {
  790. return null;
  791. }
  792. return \mf2\Parse($result['body'], $url);
  793. }
  794. }