FuturaSciencesBridge.php 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219
  1. <?php
  2. class FuturaSciencesBridge extends BridgeAbstract {
  3. public function loadMetadatas() {
  4. $this->maintainer = 'ORelio';
  5. $this->name = $this->getName();
  6. $this->uri = $this->getURI();
  7. $this->description = 'Returns the newest articles.';
  8. $this->update = '2016-03-20';
  9. $this->parameters[] =
  10. '[
  11. {
  12. "name" : "Feed",
  13. "type" : "list",
  14. "identifier" : "feed",
  15. "values" :
  16. [
  17. { "name" : "---- Select ----", "value" : "" },
  18. { "name" : "", "value" : "" },
  19. { "name" : "Les flux multi-magazines", "value" : "" },
  20. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Sciences", "value" : "actualites" },
  21. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Sciences", "value" : "definitions" },
  22. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières photos de Futura-Sciences", "value" : "photos" },
  23. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions - réponses de Futura-Sciences", "value" : "questions-reponses" },
  24. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Sciences", "value" : "dossiers" },
  25. { "name" : "", "value" : "" },
  26. { "name" : "Les flux Services", "value" : "" },
  27. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les cartes virtuelles de Futura-Sciences", "value" : "services/cartes-virtuelles" },
  28. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les fonds d\'écran de Futura-Sciences", "value" : "services/fonds-ecran" },
  29. { "name" : "", "value" : "" },
  30. { "name" : "Les flux Santé", "value" : "" },
  31. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Santé", "value" : "sante/actualites" },
  32. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Santé", "value" : "sante/definitions" },
  33. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions-réponses de Futura-Santé", "value" : "sante/questions-reponses" },
  34. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Santé", "value" : "sante/dossiers" },
  35. { "name" : "", "value" : "" },
  36. { "name" : "Les flux High-Tech", "value" : "" },
  37. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura High-Tech", "value" : "high-tech/actualites" },
  38. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières astuces de Futura High-Tech", "value" : "high-tech/questions-reponses" },
  39. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura High-Tech", "value" : "high-tech/definitions" },
  40. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura High-Tech", "value" : "high-tech/dossiers" },
  41. { "name" : "", "value" : "" },
  42. { "name" : "Les flux Espace", "value" : "" },
  43. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Espace", "value" : "espace/actualites" },
  44. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Espace", "value" : "espace/definitions" },
  45. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions-réponses de Futura-Espace", "value" : "espace/questions-reponses" },
  46. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Espace", "value" : "espace/dossiers" },
  47. { "name" : "", "value" : "" },
  48. { "name" : "Les flux Environnement", "value" : "" },
  49. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Environnement", "value" : "environnement/actualites" },
  50. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Environnement", "value" : "environnement/definitions" },
  51. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions - réponses de Futura-Environnement", "value" : "environnement/questions-reponses" },
  52. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Environnement", "value" : "environnement/dossiers" },
  53. { "name" : "", "value" : "" },
  54. { "name" : "Les flux Maison", "value" : "" },
  55. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Maison", "value" : "maison/actualites" },
  56. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières astuces de Futura-Maison", "value" : "maison/questions-reponses" },
  57. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Maison", "value" : "maison/definitions" },
  58. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Maison", "value" : "maison/dossiers" },
  59. { "name" : "", "value" : "" },
  60. { "name" : "Les flux Nature", "value" : "" },
  61. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Nature", "value" : "nature/actualites" },
  62. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Nature", "value" : "nature/definitions" },
  63. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions-réponses de Futura-Nature", "value" : "nature/questions-reponses" },
  64. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Nature", "value" : "nature/dossiers" },
  65. { "name" : "", "value" : "" },
  66. { "name" : "Les flux Terre", "value" : "" },
  67. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Terre", "value" : "terre/actualites" },
  68. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Terre", "value" : "terre/definitions" },
  69. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions-réponses de Futura-Terre", "value" : "terre/questions-reponses" },
  70. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Terre", "value" : "terre/dossiers" },
  71. { "name" : "", "value" : "" },
  72. { "name" : "Les flux Matière", "value" : "" },
  73. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Matière", "value" : "matiere/actualites" },
  74. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières définitions de Futura-Matière", "value" : "matiere/definitions" },
  75. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières questions-réponses de Futura-Matière", "value" : "matiere/questions-reponses" },
  76. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Matière", "value" : "matiere/dossiers" },
  77. { "name" : "", "value" : "" },
  78. { "name" : "Les flux Mathématiques", "value" : "" },
  79. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les dernières actualités de Futura-Mathématiques", "value" : "mathematiques/actualites" },
  80. { "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Les derniers dossiers de Futura-Mathématiques", "value" : "mathematiques/dossiers" }
  81. ]
  82. }
  83. ]';
  84. }
  85. public function collectData(array $param) {
  86. function StripCDATA($string) {
  87. $string = str_replace('<![CDATA[', '', $string);
  88. $string = str_replace(']]>', '', $string);
  89. return $string;
  90. }
  91. function StripWithDelimiters($string, $start, $end) {
  92. while (strpos($string, $start) !== false) {
  93. $section_to_remove = substr($string, strpos($string, $start));
  94. $section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
  95. $string = str_replace($section_to_remove, '', $string);
  96. } return $string;
  97. }
  98. function StripRecursiveHTMLSection($string, $tag_name, $tag_start) {
  99. $open_tag = '<'.$tag_name;
  100. $close_tag = '</'.$tag_name.'>';
  101. $close_tag_length = strlen($close_tag);
  102. if (strpos($tag_start, $open_tag) === 0) {
  103. while (strpos($string, $tag_start) !== false) {
  104. $max_recursion = 100;
  105. $section_to_remove = null;
  106. $section_start = strpos($string, $tag_start);
  107. $search_offset = $section_start;
  108. do {
  109. $max_recursion--;
  110. $section_end = strpos($string, $close_tag, $search_offset);
  111. $search_offset = $section_end + $close_tag_length;
  112. $section_to_remove = substr($string, $section_start, $section_end - $section_start + $close_tag_length);
  113. $open_tag_count = substr_count($section_to_remove, $open_tag);
  114. $close_tag_count = substr_count($section_to_remove, $close_tag);
  115. } while ($open_tag_count > $close_tag_count && $max_recursion > 0);
  116. $string = str_replace($section_to_remove, '', $string);
  117. }
  118. }
  119. return $string;
  120. }
  121. if (empty($param['feed']))
  122. $this->returnError('Please select a feed to display.'.$url, 400);
  123. if ($param['feed'] !== preg_replace('/[^a-zA-Z-\/]+/', '', $param['feed']) || substr_count($param['feed'], '/') > 1 || strlen($param['feed'] > 64))
  124. $this->returnError('Invalid "feed" parameter.'.$url, 400);
  125. $url = $this->getURI().'rss/'.$param['feed'].'.xml';
  126. $html = $this->file_get_html($url) or $this->returnError('Could not request Futura-Sciences: '.$url, 500);
  127. $limit = 0;
  128. foreach($html->find('item') as $element) {
  129. if ($limit < 10) {
  130. $article_url = str_replace('#xtor=RSS-8', '', StripCDATA($element->find('guid', 0)->plaintext));
  131. $article = $this->file_get_html($article_url) or $this->returnError('Could not request Futura-Sciences: '.$article_url, 500);
  132. $contents = $article->find('div.content', 0)->innertext;
  133. $author = trim(str_replace(', Futura-Sciences', '', $article->find('span.author', 0)->plaintext));
  134. if (empty($author))
  135. $author = StripCDATA($element->find('author', 0)->plaintext);
  136. foreach (array(
  137. '<div class="clear',
  138. '<div class="sharebar2',
  139. '<div class="diaporamafullscreen"',
  140. '<div style="margin-bottom:10px;" class="noprint"',
  141. '<div class="ficheprevnext',
  142. '<div class="bar noprint',
  143. '<div class="toolbar noprint',
  144. '<div class="addthis_toolbox',
  145. '<div class="noprint',
  146. '<div class="bg bglight border border-full noprint',
  147. '<div class="httplogbar-wrapper noprint',
  148. '<div id="forumcomments'
  149. ) as $div_start) {
  150. $contents = StripRecursiveHTMLSection($contents , 'div', $div_start);
  151. }
  152. $contents = StripWithDelimiters($contents, '<hr ', '/>');
  153. $contents = StripWithDelimiters($contents, '<p class="content-date', '</p>');
  154. $contents = StripWithDelimiters($contents, '<h1 class="content-title', '</h1>');
  155. $contents = StripWithDelimiters($contents, 'fs:definition="', '"');
  156. $contents = StripWithDelimiters($contents, 'fs:xt:clicktype="', '"');
  157. $contents = StripWithDelimiters($contents, 'fs:xt:clickname="', '"');
  158. $item = new \Item();
  159. $item->author = $author;
  160. $item->uri = $article_url;
  161. $item->title = StripCDATA($element->find('title', 0)->innertext);
  162. $item->thumbnailUri = StripCDATA($element->find('enclosure', 0)->url);
  163. $item->timestamp = strtotime(StripCDATA($element->find('pubDate', 0)->plaintext));
  164. $item->content = trim($contents);
  165. $this->items[] = $item;
  166. $limit++;
  167. }
  168. }
  169. }
  170. public function getName() {
  171. return 'Futura-Sciences Bridge';
  172. }
  173. public function getURI() {
  174. return 'http://www.futura-sciences.com/';
  175. }
  176. public function getCacheDuration() {
  177. return 3600;
  178. }
  179. }