LesLibres.org
/
Rss-Bridge
2
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[NextINpact] Upgrade for NextINpact v7 (#1708)

This commit is contained in:
ORelio 2020-08-20 14:49:26 +02:00 committed by GitHub
parent 859053ef7a
commit 07c71b3b36
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 100 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

127
bridges/NextInpactBridge.php
View File

@ -1,9 +1,10 @@
<?php
class NextInpactBridge extends FeedExpander {
const MAINTAINER = 'qwertygc';
const MAINTAINER = 'qwertygc and ORelio';
const NAME = 'NextInpact Bridge';
const URI = 'https://www.nextinpact.com/';
const URI_HARDWARE = 'https://www.inpact-hardware.com/';
const DESCRIPTION = 'Returns the newest articles.';
const PARAMETERS = array( array(
@ -11,10 +12,30 @@ class NextInpactBridge extends FeedExpander {
'name' => 'Feed',
'type' => 'list',
'values' => array(
'Tous nos articles' => 'news',
'Nos contenus en accès libre' => 'acces-libre',
'Blog' => 'blog',
'Bons plans' => 'bonsplans'
'Nos actualités' => array(
'Toutes nos publications' => 'news',
'Toutes nos publications sauf #LeBrief' => 'nobrief',
'Toutes nos publications sauf INpact Hardware' => 'noih',
'Seulement les publications INpact Hardware' => 'hardware:news',
'Seulement les publications Next INpact' => 'nobrief-noih',
'Seulement les publications #LeBrief' => 'lebrief',
),
'Flux spécifiques' => array(
'Le blog' => 'blog',
'Les bons plans' => 'bonsplans',
'Publications INpact Hardware en accès libre' => 'hardware:acces-libre',
'Publications Next INpact en accès libre' => 'acces-libre',
),
'Flux thématiques' => array(
'Tech' => 'category:1',
'Logiciel' => 'category:2',
'Internet' => 'category:3',
'Mobilité' => 'category:4',
'Droit' => 'category:5',
'Économie' => 'category:6',
'Culture numérique' => 'category:7',
'Next INpact' => 'category:8',
)
)
),
'filter_premium' => array(
@ -39,9 +60,27 @@ class NextInpactBridge extends FeedExpander {
public function collectData(){
$feed = $this->getInput('feed');
if (empty($feed))
$base_uri = self::URI;
$args = '';
if (empty($feed)) {
// Default to All articles
$feed = 'news';
$this->collectExpandableDatas(self::URI . 'rss/' . $feed . '.xml');
}
if (strpos($feed, 'hardware:') === 0) {
// Feed hosted on Hardware domain
$base_uri = self::URI_HARDWARE;
$feed = str_replace('hardware:', '', $feed);
}
if (strpos($feed, 'category:') === 0) {
// Feed with specific category parameter
$args = '?CategoryIds=' . str_replace('category:', '', $feed);
$feed = 'params';
}
$this->collectExpandableDatas($base_uri . 'rss/' . $feed . '.xml' . $args);
}
protected function parseItem($newsItem){
@ -57,9 +96,10 @@ class NextInpactBridge extends FeedExpander {
if (!is_object($html))
return 'Failed to request NextInpact: ' . $url;
// Filter premium and brief articles?
foreach(array(
'filter_premium' => 'h2.title_reserve_article',
'filter_brief' => 'div.brief-inner-content'
'filter_premium' => 'p.red-msg',
'filter_brief' => 'div.brief-container'
) as $param_name => $selector) {
$param_val = intval($this->getInput($param_name));
if ($param_val != 0) {
@ -71,38 +111,71 @@ class NextInpactBridge extends FeedExpander {
}
}
if (is_object($html->find('div[itemprop=articleBody], div.brief-inner-content', 0))) {
$article_content = $html->find('div.article-content', 0);
if (!is_object($article_content)) {
$article_content = $html->find('div.content', 0);
}
if (is_object($article_content)) {
$subtitle = trim($html->find('span.sub_title, div.brief-head', 0));
if(is_object($subtitle) && $subtitle->plaintext !== $item['title']) {
$subtitle = '<p><em>' . $subtitle->plaintext . '</em></p>';
// Subtitle
$subtitle = $html->find('small.subtitle', 0);
if(!is_object($subtitle)) {
$subtitle = $html->find('small', 0);
}
if(!is_object($subtitle)) {
$content_wrapper = $html->find('div.content-wrapper', 0);
if (is_object($content_wrapper)) {
$subtitle = $content_wrapper->find('h2.title', 0);
}
}
if(is_object($subtitle) && (!isset($item['title']) || $subtitle->plaintext != $item['title'])) {
$subtitle = '<p><em>' . trim($subtitle->plaintext) . '</em></p>';
} else {
$subtitle = '';
}
$postimg = $html->find(
'div.container_main_image_article, div.image-brief-container, div.image-brief-side-container', 0
);
// Image
$postimg = $html->find('div.article-image, div.image-container', 0);
if(is_object($postimg)) {
$postimg = '<p><img src="'
. $postimg->find('img.dedicated', 0)->src
. '" alt="-" /></p>';
$postimg = $postimg->find('img', 0);
if (!empty($postimg->src)) {
$postimg = $postimg->src;
} else {
$postimg = $postimg->srcset; //"url 355w, url 1003w, url 748w"
$postimg = explode(', ', $postimg); //split by ', ' to get each url separately
$postimg = end($postimg); //Get last item: "url 748w" which is of largest size
$postimg = explode(' ', $postimg); //split by ' ' to separate url from res
$postimg = array_reverse($postimg); //reverse array content to have url last
$postimg = end($postimg); //Get last item of array: "url"
}
$postimg = '<p><img src="' . $postimg . '" alt="-" /></p>';
} else {
$postimg = '';
}
// Paywall
$paywall = $html->find('div.paywall-restriction', 0);
if (is_object($paywall) && is_object($paywall->find('p.red-msg', 0))) {
$paywall = '<p><em>' . $paywall->find('span.head-mention', 0)->innertext . '</em></p>';
} else {
$paywall = '';
}
// Content
$article_content = $article_content->outertext;
$article_content = str_replace('>Signaler une erreur</span>', '></span>', $article_content);
// Result
$text = $subtitle
. $postimg
. $html->find('div[itemprop=articleBody], div.brief-inner-content', 0)->outertext;
. $article_content
. $paywall;
} else {
$text = $item['content']
. '<p><em>Failed retrieve full article content</em></p>';
}
$premium_article = $html->find('h2.title_reserve_article', 0);
if (is_object($premium_article)) {
$text .= '<p><em>' . $premium_article->innertext . '</em></p>';
$text = '<p><em>Failed to retrieve full article content</em></p>';
if (isset($item['content'])) {
$text = $item['content'] . $text;
}
}
return $text;