2013-08-11 15:30:41 +04:00
|
|
|
<?php
|
|
|
|
/**
|
|
|
|
* 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)
|
|
|
|
*/
|
2016-12-07 00:45:52 +03:00
|
|
|
class GoogleSearchBridge extends BridgeAbstract {
|
2013-08-11 15:30:41 +04:00
|
|
|
|
2017-02-11 18:16:56 +03:00
|
|
|
const MAINTAINER = 'sebsauvage';
|
|
|
|
const NAME = 'Google search';
|
|
|
|
const URI = 'https://www.google.com/';
|
2016-09-25 18:04:28 +03:00
|
|
|
const CACHE_TIMEOUT = 1800; // 30min
|
2017-02-11 18:16:56 +03:00
|
|
|
const DESCRIPTION = 'Returns most recent results from Google search.';
|
2015-11-05 18:50:18 +03:00
|
|
|
|
2016-12-07 00:45:52 +03:00
|
|
|
const PARAMETERS = array(array(
|
|
|
|
'q' => array(
|
2018-06-30 00:55:33 +03:00
|
|
|
'name' => 'keyword',
|
2016-12-07 00:45:52 +03:00
|
|
|
'required' => true
|
|
|
|
)
|
|
|
|
));
|
|
|
|
|
2017-02-11 18:16:56 +03:00
|
|
|
public function collectData(){
|
2016-12-07 00:45:52 +03:00
|
|
|
$html = '';
|
|
|
|
|
2017-02-11 18:16:56 +03:00
|
|
|
$html = getSimpleHTMLDOM(self::URI
|
|
|
|
. 'search?q='
|
|
|
|
. urlencode($this->getInput('q'))
|
2018-11-05 14:55:58 +03:00
|
|
|
. '&num=100&complete=0&tbs=qdr:y,sbd:1')
|
2017-02-11 18:16:56 +03:00
|
|
|
or returnServerError('No results for this query.');
|
2016-12-07 00:45:52 +03:00
|
|
|
|
|
|
|
$emIsRes = $html->find('div[id=ires]', 0);
|
|
|
|
|
2017-07-29 20:28:00 +03:00
|
|
|
if(!is_null($emIsRes)) {
|
|
|
|
foreach($emIsRes->find('div[class=g]') as $element) {
|
2016-12-07 00:45:52 +03:00
|
|
|
|
|
|
|
$item = array();
|
|
|
|
|
|
|
|
// Extract direct URL from google href (eg. /url?q=...)
|
|
|
|
$t = $element->find('a[href]', 0)->href;
|
2017-02-11 18:16:56 +03:00
|
|
|
$item['uri'] = '' . $t;
|
2016-12-07 00:45:52 +03:00
|
|
|
parse_str(parse_url($t, PHP_URL_QUERY), $parameters);
|
2017-07-29 20:28:00 +03:00
|
|
|
if(isset($parameters['q'])) {
|
2017-02-11 18:16:56 +03:00
|
|
|
$item['uri'] = $parameters['q'];
|
|
|
|
}
|
2016-12-07 00:45:52 +03:00
|
|
|
|
|
|
|
$item['title'] = $element->find('h3', 0)->plaintext;
|
|
|
|
$item['content'] = $element->find('span[class=st]', 0)->plaintext;
|
|
|
|
|
|
|
|
$this->items[] = $item;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public function getName(){
|
2017-07-29 20:28:00 +03:00
|
|
|
if(!is_null($this->getInput('q'))) {
|
2017-02-15 00:20:55 +03:00
|
|
|
return $this->getInput('q') . ' - Google search';
|
|
|
|
}
|
|
|
|
|
|
|
|
return parent::getName();
|
2016-12-07 00:45:52 +03:00
|
|
|
}
|
2014-05-21 21:15:52 +04:00
|
|
|
}
|