FuturaSciencesBridge.php 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173
  1. <?php
  2. class FuturaSciencesBridge extends FeedExpander {
  3. const MAINTAINER = 'ORelio';
  4. const NAME = 'Futura-Sciences Bridge';
  5. const URI = 'http://www.futura-sciences.com/';
  6. const DESCRIPTION = 'Returns the newest articles.';
  7. const PARAMETERS = array( array(
  8. 'feed' => array(
  9. 'name' => 'Feed',
  10. 'type' => 'list',
  11. 'values' => array(
  12. 'Les flux multi-magazines' => array(
  13. 'Les dernières actualités de Futura-Sciences' => 'actualites',
  14. 'Les dernières définitions de Futura-Sciences' => 'definitions',
  15. 'Les dernières photos de Futura-Sciences' => 'photos',
  16. 'Les dernières questions - réponses de Futura-Sciences' => 'questions-reponses',
  17. 'Les derniers dossiers de Futura-Sciences' => 'dossiers'
  18. ),
  19. 'Les flux Services' => array(
  20. 'Les cartes virtuelles de Futura-Sciences' => 'services/cartes-virtuelles',
  21. 'Les fonds d\'écran de Futura-Sciences' => 'services/fonds-ecran'
  22. ),
  23. 'Les flux Santé' => array(
  24. 'Les dernières actualités de Futura-Santé' => 'sante/actualites',
  25. 'Les dernières définitions de Futura-Santé' => 'sante/definitions',
  26. 'Les dernières questions-réponses de Futura-Santé' => 'sante/question-reponses',
  27. 'Les derniers dossiers de Futura-Santé' => 'sante/dossiers'
  28. ),
  29. 'Les flux High-Tech' => array(
  30. 'Les dernières actualités de Futura-High-Tech' => 'high-tech/actualites',
  31. 'Les dernières astuces de Futura-High-Tech' => 'high-tech/question-reponses',
  32. 'Les dernières définitions de Futura-High-Tech' => 'high-tech/definitions',
  33. 'Les derniers dossiers de Futura-High-Tech' => 'high-tech/dossiers'
  34. ),
  35. 'Les flux Espace' => array(
  36. 'Les dernières actualités de Futura-Espace' => 'espace/actualites',
  37. 'Les dernières définitions de Futura-Espace' => 'espace/definitions',
  38. 'Les dernières questions-réponses de Futura-Espace' => 'espace/question-reponses',
  39. 'Les derniers dossiers de Futura-Espace' => 'espace/dossiers'
  40. ),
  41. 'Les flux Environnement' => array(
  42. 'Les dernières actualités de Futura-Environnement' => 'environnement/actualites',
  43. 'Les dernières définitions de Futura-Environnement' => 'environnement/definitions',
  44. 'Les dernières questions-réponses de Futura-Environnement' => 'environnement/question-reponses',
  45. 'Les derniers dossiers de Futura-Environnement' => 'environnement/dossiers'
  46. ),
  47. 'Les flux Maison' => array(
  48. 'Les dernières actualités de Futura-Maison' => 'maison/actualites',
  49. 'Les dernières astuces de Futura-Maison' => 'maison/question-reponses',
  50. 'Les dernières définitions de Futura-Maison' => 'maison/definitions',
  51. 'Les derniers dossiers de Futura-Maison' => 'maison/dossiers'
  52. ),
  53. 'Les flux Nature' => array(
  54. 'Les dernières actualités de Futura-Nature' => 'nature/actualites',
  55. 'Les dernières définitions de Futura-Nature' => 'nature/definitions',
  56. 'Les dernières questions-réponses de Futura-Nature' => 'nature/question-reponses',
  57. 'Les derniers dossiers de Futura-Nature' => 'nature/dossiers'
  58. ),
  59. 'Les flux Terre' => array(
  60. 'Les dernières actualités de Futura-Terre' => 'terre/actualites',
  61. 'Les dernières définitions de Futura-Terre' => 'terre/definitions',
  62. 'Les dernières questions-réponses de Futura-Terre' => 'terre/question-reponses',
  63. 'Les derniers dossiers de Futura-Terre' => 'terre/dossiers'
  64. ),
  65. 'Les flux Matière' => array(
  66. 'Les dernières actualités de Futura-Matière' => 'matiere/actualites',
  67. 'Les dernières définitions de Futura-Matière' => 'matiere/definitions',
  68. 'Les dernières questions-réponses de Futura-Matière' => 'matiere/question-reponses',
  69. 'Les derniers dossiers de Futura-Matière' => 'matiere/dossiers'
  70. ),
  71. 'Les flux Mathématiques' => array(
  72. 'Les dernières actualités de Futura-Mathématiques' => 'mathematiques/actualites',
  73. 'Les derniers dossiers de Futura-Mathématiques' => 'mathematiques/dossiers'
  74. )
  75. )
  76. )
  77. ));
  78. public function collectData(){
  79. $url = self::URI . 'rss/' . $this->getInput('feed') . '.xml';
  80. $this->collectExpandableDatas($url, 10);
  81. }
  82. protected function parseItem($newsItem){
  83. $item = parent::parseItem($newsItem);
  84. $item['uri'] = str_replace('#xtor=RSS-8', '', $item['uri']);
  85. $article = getSimpleHTMLDOMCached($item['uri'])
  86. or returnServerError('Could not request Futura-Sciences: ' . $item['uri']);
  87. $item['content'] = $this->extractArticleContent($article);
  88. $author = $this->extractAuthor($article);
  89. $item['author'] = empty($author) ? $item['author'] : $author;
  90. return $item;
  91. }
  92. private function stripWithDelimiters($string, $start, $end){
  93. while(strpos($string, $start) !== false) {
  94. $section_to_remove = substr($string, strpos($string, $start));
  95. $section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
  96. $string = str_replace($section_to_remove, '', $string);
  97. } return $string;
  98. }
  99. private function stripRecursiveHTMLSection($string, $tag_name, $tag_start){
  100. $open_tag = '<' . $tag_name;
  101. $close_tag = '</' . $tag_name . '>';
  102. $close_tag_length = strlen($close_tag);
  103. if(strpos($tag_start, $open_tag) === 0) {
  104. while(strpos($string, $tag_start) !== false) {
  105. $max_recursion = 100;
  106. $section_to_remove = null;
  107. $section_start = strpos($string, $tag_start);
  108. $search_offset = $section_start;
  109. do {
  110. $max_recursion--;
  111. $section_end = strpos($string, $close_tag, $search_offset);
  112. $search_offset = $section_end + $close_tag_length;
  113. $section_to_remove = substr($string, $section_start, $section_end - $section_start + $close_tag_length);
  114. $open_tag_count = substr_count($section_to_remove, $open_tag);
  115. $close_tag_count = substr_count($section_to_remove, $close_tag);
  116. } while ($open_tag_count > $close_tag_count && $max_recursion > 0);
  117. $string = str_replace($section_to_remove, '', $string);
  118. }
  119. }
  120. return $string;
  121. }
  122. private function extractArticleContent($article){
  123. $contents = $article->find('section.article-text-classic', 0)->innertext;
  124. $headline = trim($article->find('p.description', 0)->plaintext);
  125. if(!empty($headline))
  126. $headline = '<p><b>' . $headline . '</b></p>';
  127. foreach (array(
  128. '<div class="clear',
  129. '<div class="sharebar2',
  130. '<div class="diaporamafullscreen"',
  131. '<div class="module social-button',
  132. '<div style="margin-bottom:10px;" class="noprint"',
  133. '<div class="ficheprevnext',
  134. '<div class="bar noprint',
  135. '<div class="toolbar noprint',
  136. '<div class="addthis_toolbox',
  137. '<div class="noprint',
  138. '<div class="bg bglight border border-full noprint',
  139. '<div class="httplogbar-wrapper noprint',
  140. '<div id="forumcomments',
  141. '<div ng-if="active"'
  142. ) as $div_start) {
  143. $contents = $this->stripRecursiveHTMLSection($contents, 'div', $div_start);
  144. }
  145. $contents = $this->stripWithDelimiters($contents, '<hr ', '/>');
  146. $contents = $this->stripWithDelimiters($contents, '<p class="content-date', '</p>');
  147. $contents = $this->stripWithDelimiters($contents, '<h1 class="content-title', '</h1>');
  148. $contents = $this->stripWithDelimiters($contents, 'fs:definition="', '"');
  149. $contents = $this->stripWithDelimiters($contents, 'fs:xt:clicktype="', '"');
  150. $contents = $this->stripWithDelimiters($contents, 'fs:xt:clickname="', '"');
  151. $contents = $this->stripWithDelimiters($contents, '<script ', '</script>');
  152. return $headline . trim($contents);
  153. }
  154. // Extracts the author from an article or element
  155. private function extractAuthor($article){
  156. $article_author = $article->find('h3.epsilon', 0);
  157. if($article_author) {
  158. return trim(str_replace(', Futura-Sciences', '', $article_author->plaintext));
  159. }
  160. return '';
  161. }
  162. }