[Dilbert] Improve metadata extraction

This commit is contained in:
ORelio 2016-02-14 13:43:58 +01:00
parent 54812d626e
commit dd0be3f4db

View file

@ -1,45 +1,58 @@
<?php <?php
class DilbertBridge extends BridgeAbstract{ class DilbertBridge extends BridgeAbstract {
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "kranack"; $this->maintainer = "kranack";
$this->name = "Dilbert Daily Strip"; $this->name = "Dilbert Daily Strip";
$this->uri = "http://dilbert.com/strips/"; $this->uri = "http://dilbert.com/";
$this->description = "The Unofficial Dilbert Daily Comic Strip"; $this->description = "The Unofficial Dilbert Daily Comic Strip";
$this->update = "30/01/2015"; $this->update = "14/02/2016";
} }
public function collectData(array $param){ public function collectData(array $param) {
$html = file_get_html('http://dilbert.com/strips/') or $this->returnError('Could not request Dilbert.', 404);
foreach($html->find('section.comic-item') as $element) { $html = file_get_html('http://dilbert.com/') or $this->returnError('Could not request Dilbert.', 500);
$comic = $element->find('img', 0);
$item = new Item(); foreach ($html->find('section.comic-item') as $element) {
$item->uri = $element->find('a',0)->href;
$item->content = '<img src="'. $comic->src . '" alt="' . $comic->alt . '" />'; $img = $element->find('img', 0);
$item->title = $comic->alt; $comic = $img->src;
$item->timestamp = strtotime($element->find('h3', 0)->plaintext); $title = $img->alt;
$url = $element->find('a', 0)->href;
$author = trim(substr($title, strpos($title, ' - Dilbert by ') + 14));
$title = trim(substr($title, 0, strpos($title, ' - ')));
$date = substr($url, 25);
if (empty($title))
$title = "Dilbert Comic Strip on ".$date;
$date = strtotime($date);
$item = new \Item();
$item->uri = $url;
$item->thumbnailUri = $comic;
$item->title = $title;
$item->author = $author;
$item->timestamp = $date;
$item->content = '<img src="'.$comic.'" alt="'.$title.'" />';
$this->items[] = $item; $this->items[] = $item;
} }
} }
public function getName(){ public function getName() {
return 'Dilbert'; return 'Dilbert Bridge';
} }
public function getURI(){ public function getURI() {
return 'http://dilbert.com'; return 'http://dilbert.com';
} }
public function getDescription(){ public function getDescription() {
return 'Dilbert via rss-bridge'; return 'Dilbert Daily Strip Bridge';
} }
public function getCacheDuration(){ public function getCacheDuration() {
return 14400; // 4 hours return 21600; // 6 hours
} }
} }
?> ?>