boyska/rss-bridge
1
0
Fork 0
forked from blallo/rss-bridge
Code Releases Activity

Merge pull request #248 from Frenzie/master

Browse source 
Fix up CollegeDeFranceBridge
This commit is contained in:
Mitsu 2016-05-01 20:58:41 +02:00
commit bfaade37ab
1 changed files with 52 additions and 31 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

55
bridges/CollegeDeFranceBridge.php
View file

@ -2,41 +2,62 @@
class CollegeDeFranceBridge extends BridgeAbstract{ class CollegeDeFranceBridge extends BridgeAbstract{
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "pit-fgfjiudghdf"; $this->maintainer = "pit-fgfjiudghdf";
$this->name = "CollegeDeFrance"; $this->name = "CollegeDeFrance";
$this->uri = "http://www.college-de-france.fr/"; $this->uri = "http://www.college-de-france.fr/";
$this->description = "Returns the 10 newest posts from CollegeDeFrance"; $this->description = "Returns the latest audio and video from CollegeDeFrance";
$this->update = "2014-05-26"; $this->update = "2016-05-01";
} }
public function collectData(array $param) { public function collectData(array $param) {
$find = array('janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'novembre', 'décembre'); $months = array(
$replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December'); '01' => 'janv.',
$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); '02' => 'févr.',
$limit = 0; '03' => 'mars',
foreach($html->find('li.audio') as $element) { '04' => 'avr.',
if($limit < 10) { '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
/* <li>
* <a href="/site/thomas-romer/guestlecturer-2016-04-15-14h30.htm" data-target="after">
* <span class="date"><span class="list-icon list-icon-video"></span><span class="list-icon list-icon-audio"></span>15 avr. 2016</span>
* <span class="lecturer">Christopher Hays</span>
* <span class='title'>Imagery of Divine Suckling in the Hebrew Bible and the Ancient Near East</span>
* </a>
* </li>
*/
$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 = new \Item();
$item->title = $element->find('span.title', 0)->plaintext; $item->title = $element->find('.title', 0)->plaintext;
$item->timestamp = strtotime(str_replace($find, $replace, $element->find('span.date', 0)->plaintext)); // Most relative URLs contains an hour in addition to the date, so let's use it
$item->content = $element->find('span.lecturer', 0)->innertext . ' - ' . $element->find('span.title', 0)->innertext; // <a href="/site/yann-lecun/course-2016-04-08-11h00.htm" data-target="after">
$item->uri = $element->find('a', 0)->href; // But unfortunately some don't
// <a href="/site/institut-physique/The-Mysteries-of-Decoherence-Sebastien-Gleyzes-[Video-3-35].htm" data-target="after">
$d = DateTime::createFromFormat('!Y-m-d-H\hi', substr($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; $this->items[] = $item;
$limit++;
} }
} }
}
public function getName(){ public function getName(){
return 'CollegeDeFrance'; return 'CollegeDeFrance';
} }
public function getURI(){ public function getURI(){
return 'http://www.college-de-france.fr/'; return 'http://www.college-de-france.fr/';
} }
public function getCacheDuration(){ public function getCacheDuration(){
return 3600*3; // 3 hour return 3600*3; // 3 hour
} }
} }