FilterMoreBridge.php 7.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241
  1. <?php
  2. class FilterMoreBridge extends FeedExpander {
  3. const MAINTAINER = 'boyska';
  4. const NAME = 'FilterMore';
  5. const CACHE_TIMEOUT = 2;
  6. const DESCRIPTION = 'Filters a feed of your choice';
  7. const URI = 'https://git.lattuta.net/boyska/rss-bridge';
  8. const PARAMETERS = [
  9. [
  10. 'url' => array(
  11. 'name' => 'Feed URL',
  12. 'required' => true,
  13. ),
  14. 'conj_type' => array(
  15. 'name' => 'Conjunction type type',
  16. 'type' => 'list',
  17. 'required' => false,
  18. 'values' => array(
  19. 'All conditions must be met' => 'and',
  20. 'Any condition must be met' => 'or',
  21. ),
  22. 'defaultValue' => 'permit',
  23. ),
  24. 'title_re' => array(
  25. 'name' => 'Filter item title (regular expression, see php.net/pcre_match for details)',
  26. 'required' => false,
  27. 'exampleValue' => '/breaking\ news/i',
  28. ),
  29. 'body_re' => array(
  30. 'name' => 'Filter body (regular expression)',
  31. 'required' => false,
  32. ),
  33. 'author_re' => array(
  34. 'name' => 'Filter author (regular expression)',
  35. 'required' => false,
  36. 'exampleValue' => '/(technology|politics)/i',
  37. ),
  38. 'newer_than' => array(
  39. 'name' => 'Filter date: ok if newer than the value (see php.net/strtotime for details)',
  40. 'required' => false,
  41. 'exampleValue' => '-14 days',
  42. ),
  43. 'older_than' => array(
  44. 'name' => 'Filter date: ok if older than the value (see php.net/strtotime for details)',
  45. 'required' => false,
  46. 'exampleValue' => '-1 hour',
  47. ),
  48. 'has_media' => array(
  49. 'name' => 'Has at least 1 media inside',
  50. 'type' => 'checkbox',
  51. 'required' => false,
  52. 'defaultValue' => false,
  53. ),
  54. 'invert_filter' => array(
  55. 'name' => 'Invert filter result',
  56. 'type' => 'checkbox',
  57. 'required' => false,
  58. 'defaultValue' => false,
  59. ),
  60. 'sort_by' => [
  61. 'name' => 'Sort by',
  62. 'type' => 'list',
  63. 'required' => true,
  64. 'values' => [
  65. "Don't sort" => 'none',
  66. 'Date' => 'timestamp',
  67. 'Title' => 'title',
  68. 'Random' => 'random',
  69. ],
  70. 'defaultValue' => 'date',
  71. ],
  72. 'sort_dir' => [
  73. 'name' => 'Sort direction',
  74. 'type' => 'list',
  75. 'required' => true,
  76. 'values' => [
  77. 'Ascending' => 'asc',
  78. 'Descending' => 'desc',
  79. ],
  80. 'defaultValue' => 'asc',
  81. ],
  82. ]];
  83. protected function parseItem($newItem){
  84. $item = parent::parseItem($newItem);
  85. $item['enclosures'] = [];
  86. if(isset($newItem->enclosure)) {
  87. foreach($newItem->enclosure as $encl) {
  88. $serialized = [];
  89. foreach($encl->attributes() as $key => $value) {
  90. $serialized[$key] = (string)$value;
  91. }
  92. $serialized["length"] = intval($serialized["length"]);
  93. $item['enclosures'][] = $serialized;
  94. }
  95. }
  96. if(isset($newItem->link)) {
  97. foreach($newItem->link as $el) {
  98. if(((string)$el['rel']) !== 'enclosure') continue;
  99. $serialized = [];
  100. $serialized['url'] = (string)$el['href'];
  101. $item['enclosures'][] = $serialized;
  102. }
  103. }
  104. $filters = ['filterByTitle', 'filterByBody', 'filterByAuthor', 'filterByDateNewer', 'filterByDateOlder', 'filterByMedia'];
  105. $results = [];
  106. foreach($filters as $filter) {
  107. $filter_res = $this->$filter($item);
  108. if($filter_res === null) continue;
  109. $results[] = $filter_res;
  110. }
  111. $old_enclosures = $item['enclosures'];
  112. $item['enclosures'] = [];
  113. foreach($old_enclosures as $e) {
  114. $item['enclosures'][] = $e['url'];
  115. }
  116. if(count($results) === 0) {
  117. return $item;
  118. }
  119. if($this->getConjType() === 'and') {
  120. $result = !in_array(false, $results);
  121. } else { // or
  122. $result = in_array(true, $results);
  123. }
  124. if($this->getInvertResult()) {
  125. $result = !$result;
  126. }
  127. if($result)
  128. return $item;
  129. else
  130. return null;
  131. }
  132. protected function sortItemKey($item) {
  133. $sort_by = $this->getInput('sort_by');
  134. $key = $item[$sort_by];
  135. return $key;
  136. }
  137. public function collectExpandableDatas($url, $maxItems = -1){
  138. parent::collectExpandableDatas($url, $maxItems);
  139. if($this->getInput('sort_by') === 'random') {
  140. shuffle($this->items);
  141. } elseif($this->getInput('sort_by') !== 'none') {
  142. usort($this->items, function($itemA, $itemB) {
  143. $valA = $this->sortItemKey($itemA);
  144. $valB = $this->sortItemKey($itemB);
  145. $cmp = strcmp($valA, $valB);
  146. return $cmp;
  147. });
  148. }
  149. if($this->getInput('sort_dir') === 'desc')
  150. $this->items = array_reverse($this->items);
  151. }
  152. private function cmp($a, $b) {
  153. if($a > $b) return 1;
  154. if($a < $b) return -1;
  155. return 0;
  156. }
  157. private function filterByFieldRegexp($field, $re){
  158. if($re === "") return null;
  159. if(preg_match($re, $field)) {
  160. return true;
  161. }
  162. return false;
  163. }
  164. protected function filterByTitle($item){
  165. $re = $this->getInput('title_re');
  166. return $this->filterByFieldRegexp($item['title'], $re);
  167. }
  168. protected function filterByBody($item){
  169. $re = $this->getInput('body_re');
  170. return $this->filterByFieldRegexp($item['content'], $re);
  171. }
  172. protected function filterByAuthor($item){
  173. $re = $this->getInput('author_re');
  174. return $this->filterByFieldRegexp($item['author'], $re);
  175. }
  176. private function filterByDate($item, $input, $expected){
  177. $val = $this->getInput($input);
  178. if($val === "") return null;
  179. $ts = strtotime($val);
  180. if($ts === false) {
  181. throw new Exception("Invalid time specification: " . $val);
  182. }
  183. $cmp = $this->cmp($item['timestamp'], $ts); // 1 if newer, -1 if older
  184. return $cmp === $expected;
  185. }
  186. protected function filterByDateNewer($item){
  187. return $this->filterByDate($item, 'newer_than', 1);
  188. }
  189. protected function filterByDateOlder($item){
  190. return $this->filterByDate($item, 'older_than', -1);
  191. }
  192. protected function filterByMedia($item) {
  193. if(!$this->getInput('has_media')) return null;
  194. if(count($item['enclosures']) > 0) return true;
  195. return false;
  196. }
  197. protected function getConjType(){
  198. return $this->getInput('conj_type');
  199. }
  200. protected function getInvertResult(){
  201. return $this->getInput('invert_filter');
  202. }
  203. public function getURI(){
  204. $url = $this->getInput('url');
  205. if(empty($url)) {
  206. $url = parent::getURI();
  207. }
  208. return $url;
  209. }
  210. public function collectData(){
  211. if($this->getInput('url') && substr($this->getInput('url'), 0, strlen('http')) !== 'http') {
  212. // just in case someone find a way to access local files by playing with the url
  213. returnClientError('The url parameter must either refer to http or https protocol.');
  214. }
  215. try{
  216. $this->collectExpandableDatas($this->getURI());
  217. } catch (HttpException $e) {
  218. $this->collectExpandableDatas($this->getURI());
  219. }
  220. }
  221. }