cuqmbr/rss-bridge
1
0
Fork 0
mirror of https://github.com/RSS-Bridge/rss-bridge.git synced 2025-04-05 09:09:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

[GoogleSearch] fix: improve bridge (#2848)

Browse Source 
Sort properly by date.
Fix php errors.
Improve the date parsing logic.
Improve the content parsing by also including those with · as separator
This commit is contained in:
Dag 2022-06-22 18:32:54 +02:00 committed by GitHub
parent e7aebb223d
commit 9d5f59e2db
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 60 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

100
bridges/GoogleSearchBridge.php
View File

@ -1,19 +1,12 @@
<?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)
*/
class GoogleSearchBridge extends BridgeAbstract {
const MAINTAINER = 'sebsauvage';
const NAME = 'Google search';
const URI = 'https://www.google.com/';
const CACHE_TIMEOUT = 1800; // 30min
const DESCRIPTION = 'Returns most recent results from Google search.';
const DESCRIPTION = 'Returns max 100 results from the past year.';
const PARAMETERS = array(array(
'q' => array(
@ -24,43 +17,70 @@ class GoogleSearchBridge extends BridgeAbstract {
));
public function collectData(){
$header = array('Accept-language: en-US');
$html = getSimpleHTMLDOM($this->getURI(), $header)
or returnServerError('No results for this query.');
$emIsRes = $html->find('div[id=res]', 0);
if(!is_null($emIsRes)) {
foreach($emIsRes->find('div[class~=g]') as $element) {
$item = array();
$t = $element->find('a[href]', 0)->href;
$item['uri'] = htmlspecialchars_decode($t);
$item['title'] = $element->find('h3', 0)->plaintext;
$resultComponents = explode(' — ', $element->find('div[data-content-feature=1]', 0)->plaintext);
$item['content'] = $resultComponents[1];
if(strpos($resultComponents[0], 'day') === true) {
$daysago = explode(' ', $resultComponents[0])[0];
$item['timestamp'] = date('d M Y', strtotime('-' . $daysago . ' days'));
} else {
$item['timestamp'] = $resultComponents[0];
}
$this->items[] = $item;
}
$dom = getSimpleHTMLDOM($this->getURI(), ['Accept-language: en-US']);
if (!$dom) {
returnServerError('No results for this query.');
}
$result = $dom->find('div[id=res]', 0);
if(!$result) {
return;
}
foreach ($result->find('div[class~=g]') as $element) {
$item = [];
$url = $element->find('a[href]', 0)->href;
$item['uri'] = htmlspecialchars_decode($url);
$item['title'] = $element->find('h3', 0)->plaintext;
$resultDom = $element->find('div[data-content-feature=1]', 0);
if ($resultDom) {
// Split by — or ·
$resultParts = preg_split('/( — | · )/', $resultDom->plaintext);
$resultDate = trim($resultParts[0]);
$resultContent = trim($resultParts[1] ?? '');
} else {
// Some search results don't have this particular dom identifier
$resultDate = null;
$resultContent = null;
}
if ($resultDate) {
try {
$createdAt = new \DateTime($resultDate);
// Set to midnight for consistent datetime
$createdAt->setTime(0, 0);
$item['timestamp'] = $createdAt->format('U');
} catch (\Exception $e) {
$item['timestamp'] = 0;
}
} else {
$item['timestamp'] = 0;
}
if ($resultContent) {
$item['content'] = $resultContent;
}
$this->items[] = $item;
}
// Sort by descending date
usort($this->items, function($a, $b) {
return $a['timestamp'] < $b['timestamp'];
return $b['timestamp'] <=> $a['timestamp'];
});
}
public function getURI() {
if (!is_null($this->getInput('q'))) {
return self::URI
. 'search?q='
. urlencode($this->getInput('q'))
. '&hl=en&num=100&complete=0&tbs=qdr:y,sbd:1';
if ($this->getInput('q')) {
$queryParameters = [
'q' => $this->getInput('q'),
'hl' => 'en',
'num' => '100', // get 100 results
'complete' => '0',
'tbs' => 'qdr:y,sbd:1' // in past year, sort by date
];
return sprintf('https://www.google.com/search?%s', http_build_query($queryParameters));
}
return parent::getURI();