[bridges] extract nested functions

Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
Pierre Mazière 2016-08-21 11:46:23 +02:00
parent fd5c8ea658
commit dd1474f154
2 changed files with 65 additions and 66 deletions

View file

@ -3,66 +3,68 @@
#error_reporting(E_ALL); #error_reporting(E_ALL);
class ArstechnicaBridge extends BridgeAbstract { class ArstechnicaBridge extends BridgeAbstract {
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "prysme"; $this->maintainer = "prysme";
$this->name = "ArstechnicaBridge"; $this->name = "ArstechnicaBridge";
$this->uri = "http://arstechnica.com"; $this->uri = "http://arstechnica.com";
$this->description = "The PC enthusiast's resource. Power users and the tools they love, without computing religion"; $this->description = "The PC enthusiast's resource. Power users and the tools they love, without computing religion";
} }
public function collectData(array $param) { function StripWithDelimiters($string, $start, $end) {
function StripWithDelimiters($string, $start, $end) { while (strpos($string, $start) !== false) {
while (strpos($string, $start) !== false) { $section_to_remove = substr($string, strpos($string, $start));
$section_to_remove = substr($string, strpos($string, $start)); $section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
$section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end)); $string = str_replace($section_to_remove, '', $string);
$string = str_replace($section_to_remove, '', $string); } return $string;
} return $string; }
}
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ExtractContent($url) { function StripCDATA($string) {
#echo $url; $string = str_replace('<![CDATA[', '', $string);
$html2 = getSimpleHTMLDOM($url); $string = str_replace(']]>', '', $string);
return $string;
}
$text = $html2->find("section[id='article-guts']", 0); function ExtractContent($url) {
#echo $url;
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find("section[id='article-guts']", 0);
/*foreach ($text->find('<aside id="social-left">') as $node) /*foreach ($text->find('<aside id="social-left">') as $node)
{ $node = NULL; }*/ { $node = NULL; }*/
$text = StripWithDelimiters($text->innertext,'<aside id="social-left">','</aside>'); $text = $this->StripWithDelimiters($text->innertext,'<aside id="social-left">','</aside>');
$text = StripWithDelimiters($text,'<figcaption class="caption">','</figcaption>'); $text = $this->StripWithDelimiters($text,'<figcaption class="caption">','</figcaption>');
$text = StripWithDelimiters($text,'<div class="gallery shortcode-gallery">','</div>'); $text = $this->StripWithDelimiters($text,'<div class="gallery shortcode-gallery">','</div>');
//error_log("ICI", 0); //error_log("ICI", 0);
//error_log($text, 0); //error_log($text, 0);
return $text; return $text;
} }
$html = $this->getSimpleHTMLDOM('http://feeds.arstechnica.com/arstechnica/index') or $this->returnServerError('Could not request NextInpact.'); public function collectData(array $param) {
$limit = 0;
foreach($html->find('item') as $element) { $html = $this->getSimpleHTMLDOM('http://feeds.arstechnica.com/arstechnica/index') or $this->returnServerError('Could not request NextInpact.');
if($limit < 5) { $limit = 0;
$item = new \Item();
$item->title = StripCDATA($element->find('title', 0)->innertext); foreach($html->find('item') as $element) {
$item->uri = StripCDATA($element->find('guid', 0)->plaintext); if($limit < 5) {
$item->author = StripCDATA($element->find('author', 0)->innertext); $item = new \Item();
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext); $item->title = $this->StripCDATA($element->find('title', 0)->innertext);
$item->content = ExtractContent($item->uri); $item->uri = $this->StripCDATA($element->find('guid', 0)->plaintext);
//$item->content = $item->uri; $item->author = $this->StripCDATA($element->find('author', 0)->innertext);
$this->items[] = $item; $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$limit++; $item->content = $this->ExtractContent($item->uri);
} //$item->content = $item->uri;
} $this->items[] = $item;
$limit++;
} }
}
public function getCacheDuration() {
return 7200; // 2h }
}
public function getCacheDuration() {
return 7200; // 2h
}
} }

View file

@ -41,21 +41,18 @@ class DailymotionBridge extends BridgeAbstract{
]'; ]';
} }
function getMetadata($id) {
$metadata=array();
$html2 = $this->getSimpleHTMLDOM('http://www.dailymotion.com/video/'.$id) or $this->returnServerError('Could not request Dailymotion.');
$metadata['title'] = $html2->find('meta[property=og:title]', 0)->getAttribute('content');
$metadata['timestamp'] = strtotime($html2->find('meta[property=video:release_date]', 0)->getAttribute('content') );
$metadata['thumbnailUri'] = $html2->find('meta[property=og:image]', 0)->getAttribute('content');
$metadata['uri'] = $html2->find('meta[property=og:url]', 0)->getAttribute('content');
return $metadata;
}
public function collectData(array $param){ public function collectData(array $param){
function getMetadata($id) {
$metadata=array();
$html2 = $this->getSimpleHTMLDOM('http://www.dailymotion.com/video/'.$id) or $this->returnServerError('Could not request Dailymotion.');
$metadata['title'] = $html2->find('meta[property=og:title]', 0)->getAttribute('content');
$metadata['timestamp'] = strtotime($html2->find('meta[property=video:release_date]', 0)->getAttribute('content') );
$metadata['thumbnailUri'] = $html2->find('meta[property=og:image]', 0)->getAttribute('content');
$metadata['uri'] = $html2->find('meta[property=og:url]', 0)->getAttribute('content');
return $metadata;
}
$html = ''; $html = '';
$limit = 5; $limit = 5;
$count = 0; $count = 0;
@ -80,7 +77,7 @@ class DailymotionBridge extends BridgeAbstract{
if($count < $limit) { if($count < $limit) {
$item = new \Item(); $item = new \Item();
$item->id = str_replace('/video/', '', strtok($element->href, '_')); $item->id = str_replace('/video/', '', strtok($element->href, '_'));
$metadata = getMetadata($item->id); $metadata = $this->getMetadata($item->id);
$item->uri = $metadata['uri']; $item->uri = $metadata['uri'];
$item->title = $metadata['title']; $item->title = $metadata['title'];
$item->timestamp = $metadata['timestamp']; $item->timestamp = $metadata['timestamp'];