maintainer = "pit-fgfjiudghdf"; $this->name = "CollegeDeFrance"; $this->uri = "http://www.college-de-france.fr/"; $this->description = "Returns the latest audio and video from CollegeDeFrance"; $this->update = "2016-08-09"; } 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 = $this->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 // // // Sometimes there's an __1, perhaps it signifies an update "/site/patrick-boucheron/seminar-2016-05-03-18h00__1.htm" // // But unfortunately some don't have any hours info // $timezone = new DateTimeZone('Europe/Paris'); // strpos($element->href, '201') will break in 2020 but it'll probably break prior to then due to site changes anyway $d = DateTime::createFromFormat('!Y-m-d-H\hi', substr($element->href, strpos($element->href, '201'), 16), $timezone) ?: DateTime::createFromFormat('!d m Y', trim(str_replace(array_values($months), array_keys($months), $element->find('.date', 0)->plaintext)), $timezone); $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 getCacheDuration(){ return 3600*3; // 3 hours } }