Merge pull request #311 from LogMANOriginal/DeveloppezDotComBridge

Developpez dot com bridge
This commit is contained in:
Mitsu 2016-08-03 15:37:39 +02:00 committed by GitHub
commit 6ff73d47bb

View file

@ -2,24 +2,22 @@
class DeveloppezDotComBridge extends BridgeAbstract{ class DeveloppezDotComBridge extends BridgeAbstract{
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "polopollo"; $this->maintainer = "polopollo";
$this->name = "Developpez.com Actus (FR)"; $this->name = "Developpez.com Actus (FR)";
$this->uri = "http://www.developpez.com/"; $this->uri = "http://www.developpez.com/";
$this->description = "Returns the 15 newest posts from DeveloppezDotCom (full text)."; $this->description = "Returns the 15 newest posts from DeveloppezDotCom (full text).";
$this->update = "2014-07-14"; $this->update = "2016-08-03";
} }
public function collectData(array $param){
function DeveloppezDotComStripCDATA($string) { function DeveloppezDotComStripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string); $string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string); $string = str_replace(']]>', '', $string);
return $string; return $string;
} }
function convert_smart_quotes($string)//F***ing quotes from Microsoft Word badly encoded, here was the trick: http://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php // F***ing quotes from Microsoft Word badly encoded, here was the trick:
// http://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php
function convert_smart_quotes($string)
{ {
$search = array(chr(145), $search = array(chr(145),
chr(146), chr(146),
@ -38,27 +36,27 @@ class DeveloppezDotComBridge extends BridgeAbstract{
function DeveloppezDotComExtractContent($url) { function DeveloppezDotComExtractContent($url) {
$articleHTMLContent = $this->file_get_html($url); $articleHTMLContent = $this->file_get_html($url);
$text = convert_smart_quotes($articleHTMLContent->find('div.content', 0)->innertext); $text = $this->convert_smart_quotes($articleHTMLContent->find('div.content', 0)->innertext);
$text = utf8_encode($text); $text = utf8_encode($text);
return trim($text); return trim($text);
} }
public function collectData(array $param){
$rssFeed = $this->file_get_html('http://www.developpez.com/index/rss') or $this->returnError('Could not request http://www.developpez.com/index/rss', 404); $rssFeed = $this->file_get_html('http://www.developpez.com/index/rss') or $this->returnError('Could not request http://www.developpez.com/index/rss', 404);
$limit = 0; $limit = 0;
foreach($rssFeed->find('item') as $element) { foreach($rssFeed->find('item') as $element) {
if($limit < 10) { if($limit < 10) {
$item = new \Item(); $item = new \Item();
$item->title = DeveloppezDotComStripCDATA($element->find('title', 0)->innertext); $item->title = $this->DeveloppezDotComStripCDATA($element->find('title', 0)->innertext);
$item->uri = DeveloppezDotComStripCDATA($element->find('guid', 0)->plaintext); $item->uri = $this->DeveloppezDotComStripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext); $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$content = DeveloppezDotComExtractContent($item->uri); $content = $this->DeveloppezDotComExtractContent($item->uri);
$item->content = strlen($content) ? $content : $element->description; //In case of it is a tutorial, we just keep the original description $item->content = strlen($content) ? $content : $element->description; //In case of it is a tutorial, we just keep the original description
$this->items[] = $item; $this->items[] = $item;
$limit++; $limit++;
} }
} }
} }
public function getName(){ public function getName(){