Rss-Bridge/bridges/LWNprevBridge.php

146 lines
3.6 KiB
PHP
Raw Normal View History

<?php
class LWNprevBridge 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 16:16:56 +01:00
const MAINTAINER = 'Pierre Mazière';
const NAME = 'LWN Free Weekly Edition';
const URI = 'https://lwn.net/';
const CACHE_TIMEOUT = 604800; // 1 week
const DESCRIPTION = 'LWN Free Weekly Edition available one week late';
function getURI(){
return self::URI . 'free/bigpage';
}
private function jumpToNextTag(&$node){
while($node && $node->nodeType === XML_TEXT_NODE){
$nextNode = $node->nextSibling;
if(!$nextNode){
break;
}
$node = $nextNode;
}
}
private function jumpToPreviousTag(&$node){
while($node && $node->nodeType === XML_TEXT_NODE){
$previousNode = $node->previousSibling;
if(!$previousNode){
break;
}
$node = $previousNode;
}
}
public function collectData(){
// Because the LWN page is written in loose HTML and not XHTML,
// Simple HTML Dom is not accurate enough for the job
$content = getContents($this->getURI())
or returnServerError('No results for LWNprev');
libxml_use_internal_errors(true);
$html = new DOMDocument();
$html->loadHTML($content);
libxml_clear_errors();
$cat1 = '';
$cat2 = '';
foreach($html->getElementsByTagName('a') as $a){
if($a->textContent === 'Multi-page format'){
break;
}
}
$realURI = self::URI . $a->getAttribute('href');
$URICounter = 0;
$edition = $html->getElementsByTagName('h1')->item(0)->textContent;
$editionTimeStamp = strtotime(
substr($edition, strpos($edition, 'for ') + strlen('for '))
);
foreach($html->getElementsByTagName('h2') as $h2){
if($h2->getAttribute('class') !== 'SummaryHL'){
continue;
}
$item = array();
$h2NextSibling = $h2->nextSibling;
$this->jumpToNextTag($h2NextSibling);
switch($h2NextSibling->getAttribute('class')){
case 'FeatureByline':
$item['author'] = $h2NextSibling->getElementsByTagName('b')->item(0)->textContent;
break;
case 'GAByline':
$text = $h2NextSibling->textContent;
$item['author'] = substr($text,strpos($text, 'by '));
break;
default:
$item['author'] = 'LWN';
break;
};
$h2FirstChild = $h2->firstChild;
$this->jumpToNextTag($h2FirstChild);
if($h2FirstChild->nodeName === 'a'){
$item['uri'] = self::URI . $h2FirstChild->getAttribute('href');
}else{
$item['uri'] = $realURI . '#' . $URICounter;
}
$URICounter++;
$item['timestamp'] = $editionTimeStamp + $URICounter;
$h2PrevSibling = $h2->previousSibling;
$this->jumpToPreviousTag($h2PrevSibling);
switch($h2PrevSibling->getAttribute('class')){
case 'Cat2HL':
$cat2 = $h2PrevSibling->textContent;
$h2PrevSibling = $h2PrevSibling->previousSibling;
$this->jumpToPreviousTag($h2PrevSibling);
if($h2PrevSibling->getAttribute('class') !== 'Cat1HL'){
break;
}
$cat1 = $h2PrevSibling->textContent;
break;
case 'Cat1HL':
$cat1 = $h2PrevSibling->textContent;
$cat2 = '';
break;
default:
break;
}
$h2PrevSibling = null;
$item['title'] = '';
if(!empty($cat1)){
$item['title'] .= '[' . $cat1 . ($cat2 ? '/' . $cat2 : '') . '] ';
}
$item['title'] .= $h2->textContent;
$node = $h2;
$content = '';
$contentEnd = false;
while(!$contentEnd){
$node = $node->nextSibling;
if(!$node || (
$node->nodeType !== XML_TEXT_NODE && (
$node->nodeName === 'h2' || (
!is_null($node->attributes) &&
!is_null($class = $node->attributes->getNamedItem('class')) &&
in_array($class->nodeValue, array('Cat1HL', 'Cat2HL'))
)
)
)
){
$contentEnd = true;
}else{
$content .= $node->C14N();
}
}
$item['content'] = $content;
$this->items[] = $item;
}
}
}