rss-bridge/bridges/GlassdoorBridge.php

188 lines
5.1 KiB
PHP
Raw Normal View History

2018-10-01 19:00:50 +03:00
<?php
2018-10-01 19:00:50 +03:00
class GlassdoorBridge extends BridgeAbstract {
// Contexts
const CONTEXT_BLOG = 'Blogs';
const CONTEXT_REVIEW = 'Company Reviews';
const CONTEXT_GLOBAL = 'global';
// Global context parameters
const PARAM_LIMIT = 'limit';
// Blog context parameters
const PARAM_BLOG_TYPE = 'blog_type';
const PARAM_BLOG_FULL = 'full_article';
const BLOG_TYPE_HOME = 'Home';
const BLOG_TYPE_COMPANIES_HIRING = 'Companies Hiring';
const BLOG_TYPE_CAREER_ADVICE = 'Career Advice';
const BLOG_TYPE_INTERVIEWS = 'Interviews';
// Review context parameters
const PARAM_REVIEW_COMPANY = 'company';
const MAINTAINER = 'logmanoriginal';
const NAME = 'Glassdoor Bridge';
const URI = 'https://www.glassdoor.com/';
const DESCRIPTION = 'Returns feeds for blog posts and company reviews';
const CACHE_TIMEOUT = 86400; // 24 hours
const PARAMETERS = array(
self::CONTEXT_BLOG => array(
self::PARAM_BLOG_TYPE => array(
'name' => 'Blog type',
'type' => 'list',
'title' => 'Select the blog you want to follow',
'values' => array(
self::BLOG_TYPE_HOME => 'blog/',
self::BLOG_TYPE_COMPANIES_HIRING => 'blog/companies-hiring/',
self::BLOG_TYPE_CAREER_ADVICE => 'blog/career-advice/',
self::BLOG_TYPE_INTERVIEWS => 'blog/interviews/',
)
),
self::PARAM_BLOG_FULL => array(
'name' => 'Full article',
'type' => 'checkbox',
'title' => 'Enable to return the full article for each post'
),
),
self::CONTEXT_REVIEW => array(
self::PARAM_REVIEW_COMPANY => array(
'name' => 'Company URL',
'type' => 'text',
'required' => true,
'title' => 'Paste the company review page URL here!',
'exampleValue' => 'https://www.glassdoor.com/Reviews/GitHub-Reviews-E671945.htm'
)
),
self::CONTEXT_GLOBAL => array(
self::PARAM_LIMIT => array(
'name' => 'Limit',
'type' => 'number',
'defaultValue' => -1,
'title' => 'Specifies the maximum number of items to return (default: All)'
)
)
);
public function getURI() {
switch($this->queriedContext) {
case self::CONTEXT_BLOG:
return self::URI . $this->getInput(self::PARAM_BLOG_TYPE);
case self::CONTEXT_REVIEW:
return $this->filterCompanyURI($this->getInput(self::PARAM_REVIEW_COMPANY));
}
return parent::getURI();
}
public function collectData() {
$url = $this->getURI();
$html = getSimpleHTMLDOM($url);
$html = defaultLinkTo($html, $url);
2018-10-01 19:00:50 +03:00
$limit = $this->getInput(self::PARAM_LIMIT);
switch($this->queriedContext) {
case self::CONTEXT_BLOG:
$this->collectBlogData($html, $limit);
break;
case self::CONTEXT_REVIEW:
$this->collectReviewData($html, $limit);
break;
}
}
private function collectBlogData($html, $limit) {
$posts = $html->find('div.post')
2018-10-01 19:00:50 +03:00
or returnServerError('Unable to find blog posts!');
foreach($posts as $post) {
$item = [];
2018-10-01 19:00:50 +03:00
$item['uri'] = $post->find('a', 0)->href;
$item['title'] = $post->find('h3', 0)->plaintext;
$item['content'] = $post->find('p', 0)->plaintext;
$item['author'] = $post->find('p', -2)->plaintext;
$item['timestamp'] = strtotime($post->find('p', -1)->plaintext);
2018-10-01 19:00:50 +03:00
// TODO: fetch entire blog post content
2018-10-01 19:00:50 +03:00
$this->items[] = $item;
if ($limit > 0 && count($this->items) >= $limit) {
2018-10-01 19:00:50 +03:00
return;
}
2018-10-01 19:00:50 +03:00
}
}
private function collectReviewData($html, $limit) {
2019-06-09 20:32:22 +03:00
$reviews = $html->find('#ReviewsFeed li[id^="empReview]')
2018-10-01 19:00:50 +03:00
or returnServerError('Unable to find reviews!');
foreach($reviews as $review) {
$item = [];
2018-10-01 19:00:50 +03:00
$item['uri'] = $review->find('a.reviewLink', 0)->href;
// Not all reviews have a title
$item['title'] = $review->find('h2', 0)->plaintext ?? 'Glassdoor review';
2019-06-09 20:32:22 +03:00
[$date, $author] = explode('-', $review->find('span.authorInfo', 0)->plaintext);
2019-06-09 20:32:22 +03:00
$item['author'] = trim($author);
2019-06-09 20:32:22 +03:00
$createdAt = DateTimeImmutable::createFromFormat('F m, Y', trim($date));
if ($createdAt) {
$item['timestamp'] = $createdAt->getTimestamp();
2019-06-09 20:32:22 +03:00
}
$item['content'] = $review->find('.px-std', 2)->text();
2018-10-01 19:00:50 +03:00
$this->items[] = $item;
if($limit > 0 && count($this->items) >= $limit) {
2018-10-01 19:00:50 +03:00
return;
}
2018-10-01 19:00:50 +03:00
}
}
private function filterCompanyURI($uri) {
/* Make sure the URI is a valid review page. Unfortunately there is no
* simple way to determine if the URI is valid, because of automagic
* redirection and strange naming conventions.
*/
if(!filter_var($uri,
FILTER_VALIDATE_URL, FILTER_FLAG_PATH_REQUIRED)) {
2018-10-01 19:00:50 +03:00
returnClientError('The specified URL is invalid!');
}
$uri = filter_var($uri, FILTER_SANITIZE_URL);
$path = parse_url($uri, PHP_URL_PATH);
$parts = explode('/', $path);
$allowed_strings = array(
'de-DE' => 'Bewertungen',
'en-AU' => 'Reviews',
'nl-BE' => 'Reviews',
'fr-BE' => 'Avis',
'en-CA' => 'Reviews',
'fr-CA' => 'Avis',
'fr-FR' => 'Avis',
'en-IN' => 'Reviews',
'en-IE' => 'Reviews',
'nl-NL' => 'Reviews',
'de-AT' => 'Bewertungen',
'de-CH' => 'Bewertungen',
'fr-CH' => 'Avis',
'en-GB' => 'Reviews',
'en' => 'Reviews'
);
if(!in_array($parts[1], $allowed_strings)) {
returnClientError('Please specify a URL pointing to the companies review page!');
}
return $uri;
}
}