rss-bridge/bridges/GoogleSearchBridge.php

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

107 lines
3.3 KiB
PHP
Raw Normal View History

2013-08-11 15:30:41 +04:00
<?php
class GoogleSearchBridge extends BridgeAbstract
{
bridges: Fix coding styles This commit is a squash of all commits that fix coding styles for the new coding style policy. [ABCTabsBridge] Fix coding style [AcrimedBridge] Fix coding style [AllocineFRBridge] Fix coding style [AnimeUltimeBridge] Fix coding style [Arte7Bridge] Fix coding style [AskfmBridge] Fix coding style [BandcampBridge] Fix coding style [BastaBridge] Fix coding style [BlaguesDeMerdeBridge] Fix coding style [BooruprojectBridge] Fix coding style [CADBridge] Fix coding style [CNETBridge] Fix coding style [CastorusBridge] Fix coding style [CollegeDeFranceBridge] Fix coding style [CommonDreamsBridge] Fix coding style [CopieDoubleBridge] Fix coding style [CourrierInternationalBridge] Fix coding style [CpasbienBridge] Fix coding style [CryptomeBridge] Fix coding style [DailymotionBridge] Fix coding style [DanbooruBridge] Fix coding style [DansTonChatBridge] Fix coding style [DauphineLibereBridge] Fix coding style [DeveloppezDotComBridge] Fix coding style [DemoBridge] Fix coding style [DilbertBridge] Fix coding style [DuckDuckGoBridge] Fix coding style [DollbooruBridge] Fix coding style [EliteDangerousGalnetBridge] Fix coding style [ElsevierBridge] Fix coding style [EstCeQuonMetEnProdBridge] Fix coding style [EZTVBridge] Fix coding style [FacebookBridge] Fix coding style [FeedExpanderExampleBridge] Fix coding style [FB2Bridge] Fix coding style [FierPandaBridge] Fix coding style [FlickrBridge] Fix coding style [FootitoBridge] Fix coding style [FourchanBridge] Fix coding style [FuturaSciencesBridge] Fix coding style [GBAtempBridge] Fix coding style [GelbooruBridge] Fix coding style [GiphyBridge] Fix coding style [GithubIssueBridge] Fix coding style [GizmodoBridge] Fix coding style [GoComicsBridge] Fix coding style [GooglePlusPostBridge] Fix coding style [GoogleSearchBridge] Fix coding style [HDWallpapersBridge] Fix coding style [HentaiHavenBridge] Fix coding style [IdenticaBridge] Fix coding style [InstagramBridge] Fix coding style [IsoHuntBridge] Fix coding style [JapanExpoBridge] Fix coding style [KonachanBridge] Fix coding style [KoreusBridge] Fix coding style [KununuBridge] Fix coding style [LeBonCoinBridge] Fix coding style [LegifranceJOBBridge] Fix coding style [LeMondeInformatiqueBridge] Fix coding style [LesJoiesDuCodeBridge] Fix coding style [LichessBridge] Fix coding style [LinkedInCompanyBridge] Fix coding style [LolibooruBridge] Fix coding style [LWNprevBridge] Fix coding style [MangareaderBridge] Fix coding style [MilbooruBridge] Fix coding style [MixCloudBridge] Fix coding style [MoebooruBridge] Fix coding style [MondeDiploBridge] Fix coding style [MsnMondeBridge] Fix coding style [MspabooruBridge] Fix coding style [NasaApodBridge] Fix coding style [NeuviemeArtBridge] Fix coding style [NextgovBridge] Fix coding style [NextInpactBridge] Fix coding style [NiceMatinBridge] Fix coding style [NovelUpdatesBridge] Fix coding style [OpenClassroomsBridge] Fix coding style [ParuVenduImmoBridge] Fix coding style [PickyWallpapersBridge] Fix coding style [PinterestBridge] Fix coding style [PlanetLibreBridge] Fix coding style [ReadComicsBridge] Fix coding style [Releases3DSBridge] Fix coding style [ReporterreBridge] Fix coding style [RTBFBridge] Fix coding style [Rue89Bridge] Fix coding style [Rule34Bridge] Fix coding style [Rule34pahealBridge] Fix coding style [SafebooruBridge] Fix coding style [SakugabooruBridge] Fix coding style [ScmbBridge] Fix coding style [ScoopItBridge] Fix coding style [SensCritiqueBridge] Fix coding style [SexactuBridge] Fix coding style [ShanaprojectBridge] Fix coding style [Shimmie2Bridge] Fix coding style [SoundcloudBridge] Fix coding style [StripeAPIChangeLogBridge] Fix coding style [SuperbWallpapersBridge] Fix coding style [T411Bridge] Fix coding style [TagBoardBridge] Fix coding style [TbibBridge] Fix coding style [TheCodingLoveBridge] Fix coding style [TheHackerNewsBridge] Fix coding style [ThePirateBayBridge] Fix coding style [TheTVDBBridge] Fix coding style [Torrent9Bridge] Fix coding style [TwitterBridge] Fix coding style [UnsplashBridge] Fix coding style [ViadeoCompanyBridge] Fix coding style [VineBridge] Fix coding style [VkBridge] Fix coding style [WallpaperStopBridge] Fix coding style [WebfailBridge] Fix coding style [WeLiveSecurityBridge] Fix coding style [WhydBridge] Fix coding style [WikipediaBridge] Fix coding style [WordPressBridge] Fix coding style [WorldOfTanksBridge] Fix coding style [XbooruBridge] Fix coding style [YandereBridge] Fix coding style [YoutubeBridge] Fix coding style [ZDNetBridge] Fix coding style
2017-02-11 18:16:56 +03:00
const MAINTAINER = 'sebsauvage';
const NAME = 'Google search';
const URI = 'https://www.google.com/';
const CACHE_TIMEOUT = 1800; // 30min
const DESCRIPTION = 'Returns max 100 results from the past year.';
2013-08-11 15:30:41 +04:00
const PARAMETERS = [[
'q' => [
'name' => 'keyword',
'required' => true,
'exampleValue' => 'rss-bridge',
],
'verbatim' => [
'name' => 'Verbatim',
'type' => 'checkbox',
'title' => 'Use literal keyword(s) without making improvements',
],
]];
public function collectData()
{
2022-10-26 01:47:45 +03:00
// todo: wrap this in try..catch because 429 too many requests happens a lot
$dom = getSimpleHTMLDOM($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 $b['timestamp'] <=> $a['timestamp'];
});
}
public function getURI()
{
if ($this->getInput('q')) {
$queryParameters = [
'q' => $this->getInput('q'),
'hl' => 'en',
'num' => '100', // get 100 results
'complete' => '0',
// in past year, sort by date, optionally verbatim
'tbs' => 'qdr:y,sbd:1' . ($this->getInput('verbatim') ? ',li:1' : ''),
];
return sprintf('https://www.google.com/search?%s', http_build_query($queryParameters));
}
return parent::getURI();
}
public function getName()
{
if (!is_null($this->getInput('q'))) {
return $this->getInput('q') . ' - Google search';
}
return parent::getName();
}
2014-05-21 21:15:52 +04:00
}