Rss-Bridge/bridges/KununuBridge.php

237 lines
6.4 KiB
PHP
Raw Normal View History

2016-08-10 23:18:35 +02:00
<?php
class KununuBridge extends BridgeAbstract {
const MAINTAINER = "logmanoriginal";
const NAME = "Kununu Bridge";
const URI = "https://www.kununu.com/";
const CACHE_TIMEOUT = 86400; // 24h
const DESCRIPTION = "Returns the latest reviews for a company and site of your choice.";
2016-08-10 23:18:35 +02:00
2016-12-17 17:11:58 +01:00
const PARAMETERS = array(
'global' => array(
'site'=>array(
'name'=>'Site',
'type'=>'list',
'required'=>true,
'title'=>'Select your site',
'values'=>array(
'Austria'=>'at',
'Germany'=>'de',
'Switzerland'=>'ch',
'United States'=>'us'
)
),
'full'=>array(
'name'=>'Load full article',
'type'=>'checkbox',
'required'=>false,
'exampleValue'=>'checked',
'title'=>'Activate to load full article'
)
),
2016-12-17 17:11:58 +01:00
array(
'company'=>array(
'name'=>'Company',
'required'=>true,
'exampleValue'=>'kununu-us',
'title'=>'Insert company name (i.e. Kununu US) or URI path (i.e. kununu-us)'
)
)
);
2016-08-10 23:18:35 +02:00
2016-12-17 17:11:58 +01:00
private $companyName='';
2016-12-17 17:11:58 +01:00
public function getURI(){
$company = $this->encode_umlauts(strtolower(str_replace(' ', '-', trim($this->getInput('company')))));
$site=$this->getInput('site');
$section = '';
switch($site){
case 'at':
case 'de':
case 'ch':
$section = 'kommentare';
break;
case 'us':
$section = 'reviews';
break;
}
2016-12-17 17:11:58 +01:00
return self::URI.$site.'/'.$company.'/'.$section;
}
2016-12-17 17:11:58 +01:00
function getName(){
$company = $this->encode_umlauts(strtolower(str_replace(' ', '-', trim($this->getInput('company')))));
return ($this->companyName?:$company).' - '.self::NAME;
}
2016-08-10 23:18:35 +02:00
public function collectData(){
2016-12-17 17:11:58 +01:00
$full = $this->getInput('full');
2016-08-10 23:18:35 +02:00
// Load page
$html = getSimpleHTMLDOM($this->getURI());
if(!$html)
returnServerError('Unable to receive data from ' . $this->getURI() . '!');
2016-08-10 23:18:35 +02:00
// Update name for this request
$this->companyName = $this->extract_company_name($html);
2016-08-10 23:18:35 +02:00
// Find the section with all the panels (reviews)
$section = $html->find('section.kununu-scroll-element', 0);
if($section === false)
returnServerError('Unable to find panel section!');
2016-08-10 23:18:35 +02:00
// Find all articles (within the panels)
$articles = $section->find('article');
if($articles === false || empty($articles))
returnServerError('Unable to find articles!');
2016-08-10 23:18:35 +02:00
// Go through all articles
foreach($articles as $article){
$item = array();
2016-08-10 23:18:35 +02:00
$item['author'] = $this->extract_article_author_position($article);
$item['timestamp'] = $this->extract_article_date($article);
$item['title'] = $this->extract_article_rating($article) . ' : ' . $this->extract_article_summary($article);
$item['uri'] = $this->extract_article_uri($article);
2016-08-10 23:18:35 +02:00
if($full)
$item['content'] = $this->extract_full_description($item['uri']);
2016-08-10 23:18:35 +02:00
else
$item['content'] = $this->extract_article_description($article);
2016-08-10 23:18:35 +02:00
$this->items[] = $item;
}
}
/**
* Fixes relative URLs in the given text
*/
private function fix_url($text){
return preg_replace('/href=(\'|\")\//i', 'href="'.self::URI, $text);
2016-08-10 23:18:35 +02:00
}
/**
* Encodes unmlauts in the given text
*/
private function encode_umlauts($text){
$umlauts = Array("/ä/","/ö/","/ü/","/Ä/","/Ö/","/Ü/","/ß/");
$replace = Array("ae","oe","ue","Ae","Oe","Ue","ss");
return preg_replace($umlauts, $replace, $text);
}
/**
* Returns the company name from the review html
*/
private function extract_company_name($html){
$panel = $html->find('div.panel', 0);
if(is_null($panel))
returnServerError('Cannot find panel for company name!');
2016-08-10 23:18:35 +02:00
$company_name = $panel->find('h1', 0);
if(is_null($company_name))
returnServerError('Cannot find company name!');
2016-08-10 23:18:35 +02:00
return $company_name->plaintext;
}
/**
* Returns the date from a given article
*/
private function extract_article_date($article){
// They conviniently provide a time attribute for us :)
$date = $article->find('time[itemprop=dtreviewed]', 0);
if(is_null($date))
returnServerError('Cannot find article date!');
2016-08-10 23:18:35 +02:00
return strtotime($date->datetime);
}
/**
* Returns the rating from a given article
*/
private function extract_article_rating($article){
$rating = $article->find('span.rating', 0);
if(is_null($rating))
returnServerError('Cannot find article rating!');
2016-08-10 23:18:35 +02:00
return $rating->getAttribute('aria-label');
}
/**
* Returns the summary from a given article
*/
private function extract_article_summary($article){
$summary = $article->find('[itemprop=summary]', 0);
if(is_null($summary))
returnServerError('Cannot find article summary!');
2016-08-10 23:18:35 +02:00
return strip_tags($summary->innertext);
}
/**
* Returns the URI from a given article
*/
private function extract_article_uri($article){
// Notice: This first part is the same as in extract_article_summary!
$summary = $article->find('[itemprop=summary]', 0);
if(is_null($summary))
returnServerError('Cannot find article summary!');
2016-08-10 23:18:35 +02:00
$anchor = $summary->find('a', 0);
if(is_null($anchor))
returnServerError('Cannot find article URI!');
return self::URI . $anchor->href;
2016-08-10 23:18:35 +02:00
}
/**
* Returns the position of the author from a given article
*/
private function extract_article_author_position($article){
// We need to parse the aside manually
$aside = $article->find('aside', 0);
if(is_null($aside))
returnServerError('Cannot find article author information!');
2016-08-10 23:18:35 +02:00
// Go through all h2 elements to find index of required span (I know... it's stupid)
$author_position = 'Unknown';
foreach($aside->find('h2') as $subject){
if(stristr(strtolower($subject->plaintext), 'position')){ /* This works for at, ch, de, us */
$author_position = $subject->next_sibling()->plaintext;
break;
}
}
2016-08-10 23:18:35 +02:00
return $author_position;
}
/**
* Returns the description from a given article
*/
private function extract_article_description($article){
$description = $article->find('div[itemprop=description]', 0);
if(is_null($description))
returnServerError('Cannot find article description!');
2016-08-10 23:18:35 +02:00
return $this->fix_url($description->innertext);
}
/**
* Returns the full description from a given uri
*/
private function extract_full_description($uri){
// Load full article
$html = getSimpleHTMLDOMCached($uri);
2016-08-10 23:18:35 +02:00
if($html === false)
returnServerError('Could not load full description!');
2016-08-10 23:18:35 +02:00
// Find the article
$article = $html->find('article', 0);
if(is_null($article))
returnServerError('Cannot find article!');
2016-08-10 23:18:35 +02:00
// Luckily they use the same layout for the review overview and full article pages :)
return $this->extract_article_description($article);
}
}