From de4952d31fd9a2d8339eb675e3c79ad167cb2392 Mon Sep 17 00:00:00 2001 From: Frans de Jonge Date: Sun, 1 May 2016 12:39:48 +0200 Subject: [PATCH] Fix up CollegeDeFranceBridge The site changed. --- bridges/CollegeDeFranceBridge.php | 83 +++++++++++++++++++------------ 1 file changed, 52 insertions(+), 31 deletions(-) diff --git a/bridges/CollegeDeFranceBridge.php b/bridges/CollegeDeFranceBridge.php index e9e0dbc0..f7a086e6 100644 --- a/bridges/CollegeDeFranceBridge.php +++ b/bridges/CollegeDeFranceBridge.php @@ -2,41 +2,62 @@ class CollegeDeFranceBridge extends BridgeAbstract{ public function loadMetadatas() { - $this->maintainer = "pit-fgfjiudghdf"; $this->name = "CollegeDeFrance"; $this->uri = "http://www.college-de-france.fr/"; - $this->description = "Returns the 10 newest posts from CollegeDeFrance"; - $this->update = "2014-05-26"; - + $this->description = "Returns the latest audio and video from CollegeDeFrance"; + $this->update = "2016-05-01"; } - public function collectData(array $param){ -$find = array('janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'novembre', 'décembre'); -$replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December'); - $html = file_get_html('http://www.college-de-france.fr/site/audio-video/_audiovideos.jsp?index=0&prompt=&fulltextdefault=mots-cles...&fulltext=mots-cles...&fields=TYPE2_ACTIVITY&fieldsdefault=0_0&TYPE2=0') or $this->returnError('Could not request CollegeDeFrance.', 404); - $limit = 0; - foreach($html->find('li.audio') as $element) { - if($limit < 10) { - $item = new \Item(); - $item->title = $element->find('span.title', 0)->plaintext; - $item->timestamp = strtotime(str_replace($find, $replace, $element->find('span.date', 0)->plaintext)); - $item->content = $element->find('span.lecturer', 0)->innertext . ' - ' . $element->find('span.title', 0)->innertext; - $item->uri = $element->find('a', 0)->href; - $this->items[] = $item; - $limit++; - } - } + public function collectData(array $param) { + $months = array( + '01' => 'janv.', + '02' => 'févr.', + '03' => 'mars', + '04' => 'avr.', + '05' => 'mai', + '06' => 'juin', + '07' => 'juil.', + '08' => 'août', + '09' => 'sept.', + '10' => 'oct.', + '11' => 'nov.', + '12' => 'déc.' + ); + // The "API" used by the site returns a list of partial HTML in this form + /*
  • + * + * 15 avr. 2016 + * Christopher Hays + * Imagery of Divine Suckling in the Hebrew Bible and the Ancient Near East + * + *
  • + */ + $html = file_get_html('http://www.college-de-france.fr/components/search-audiovideo.jsp?fulltext=&siteid=1156951719600&lang=FR&type=all') or $this->returnError('Could not request CollegeDeFrance.', 404); + foreach($html->find('a[data-target]') as $element) { + $item = new \Item(); + $item->title = $element->find('.title', 0)->plaintext; + // Most relative URLs contains an hour in addition to the date, so let's use it + // + // But unfortunately some don't + // + $d = DateTime::createFromFormat('!Y-m-d-H\hi', substr($element->href, -20, -4)) ?: DateTime::createFromFormat('!H m Y', str_replace(array_values($months), array_keys($months), $element->find('.date', 0)->innertext)); + $item->timestamp = $d->format('U'); + $item->content = $element->find('.lecturer', 0)->innertext . ' - ' . $element->find('.title', 0)->innertext; + $item->uri = 'http://www.college-de-france.fr' . $element->href; + $this->items[] = $item; + } + } - } - public function getName(){ - return 'CollegeDeFrance'; - } - public function getURI(){ - return 'http://www.college-de-france.fr/'; - } - public function getCacheDuration(){ - return 3600*3; // 3 hour - } + public function getName(){ + return 'CollegeDeFrance'; + } + + public function getURI(){ + return 'http://www.college-de-france.fr/'; + } + + public function getCacheDuration(){ + return 3600*3; // 3 hour + } } -