rss-bridge/bridges/GoogleSearchBridge.php

63 lines
2.1 KiB
PHP
Raw Normal View History

2013-08-11 13:30:41 +02:00
<?php
/**
* RssBridgeGoogleMostRecent
* Search Google for most recent pages regarding a specific topic.
* Returns the 100 most recent links in results in past year, sorting by date (most recent first).
* Example:
* http://www.google.com/search?q=sebsauvage&num=100&complete=0&tbs=qdr:y,sbd:1
* complete=0&num=100 : get 100 results
* qdr:y : in past year
* sbd:1 : sort by date (will only work if qdr: is specified)
2014-05-25 23:27:14 +02:00
* 2014-05-25
2013-08-11 13:30:41 +02:00
*
* @name Google search
2014-05-25 23:27:14 +02:00
* @homepage https://www.google.com/
2013-08-11 13:30:41 +02:00
* @description Returns most recent results from Google search.
2014-05-21 19:44:14 +02:00
* @maintainer sebsauvage
2013-08-11 22:34:14 +02:00
* @use1(q="keyword")
2013-08-11 13:30:41 +02:00
*/
class GoogleSearchBridge extends BridgeAbstract{
private $request;
2013-08-11 13:30:41 +02:00
public function collectData(array $param){
$html = '';
if (isset($param['q'])) { /* keyword search mode */
$this->request = $param['q'];
2015-01-20 19:06:51 +01:00
$html = file_get_html('https://www.google.com/search?q=' . urlencode($this->request) . '&num=100&complete=0&tbs=qdr:y,sbd:1') or $this->returnError('No results for this query.', 404);
2013-08-11 13:30:41 +02:00
}
else{
$this->returnError('You must specify a keyword (?q=...).', 400);
}
$emIsRes = $html->find('div[id=ires]',0);
if( !is_null($emIsRes) ){
foreach($emIsRes->find('li[class=g]') as $element) {
$item = new Item();
// Extract direct URL from google href (eg. /url?q=...)
$t = $element->find('a[href]',0)->href;
2015-01-20 19:06:51 +01:00
$item->uri = ''.$t;
parse_str(parse_url($t, PHP_URL_QUERY),$parameters);
if (isset($parameters['q'])) { $item->uri = $parameters['q']; }
2013-08-11 13:30:41 +02:00
$item->title = $element->find('h3',0)->plaintext;
$item->content = $element->find('span[class=st]',0)->plaintext;
$this->items[] = $item;
}
}
}
public function getName(){
return (!empty($this->request) ? $this->request .' - ' : '') .'Google search';
2013-08-11 13:30:41 +02:00
}
public function getURI(){
return 'http://google.com';
}
public function getCacheDuration(){
return 1800; // 30 minutes
}
2014-05-21 19:15:52 +02:00
}