mirror of
https://github.com/RSS-Bridge/rss-bridge.git
synced 2025-01-16 13:50:01 +01:00
[ScientificAmericanBridge] Update bridge (#4074)
* [ScientificAmericanBridge] Update bridge * [ScientificAmericanBridge] Fix lint
This commit is contained in:
Korytov Pavel
GitHub
parent
58c254ad3b
commit
428c6c3c66
@ -25,7 +25,7 @@ class ScientificAmericanBridge extends FeedExpander
|
||||
];
|
||||
|
||||
const FEED = 'http://rss.sciam.com/ScientificAmerican-Global';
|
||||
const ISSUES = 'https://www.scientificamerican.com/archive/issues/2020s/';
|
||||
const ISSUES = 'https://www.scientificamerican.com/archive/issues/';
|
||||
|
||||
public function collectData()
|
||||
{
|
||||
@ -50,7 +50,7 @@ class ScientificAmericanBridge extends FeedExpander
|
||||
|
||||
if ($this->getInput('addContents') == 1) {
|
||||
usort($this->items, function ($item1, $item2) {
|
||||
return $item1['timestamp'] - $item2['timestamp'];
|
||||
return $item2['timestamp'] - $item1['timestamp'];
|
||||
});
|
||||
}
|
||||
}
|
||||
@ -66,8 +66,12 @@ class ScientificAmericanBridge extends FeedExpander
|
||||
private function collectIssues()
|
||||
{
|
||||
$html = getSimpleHTMLDOMCached(self::ISSUES);
|
||||
$content = $html->getElementById('app')->children(3);
|
||||
$issues = $content->children();
|
||||
$content = $html->getElementById('app');
|
||||
$issues_list = $content->find('div[class^="issue__list"]', 0);
|
||||
if ($issues_list == null) {
|
||||
return [];
|
||||
}
|
||||
$issues = $issues_list->find('div[class^="list__item"]');
|
||||
$issues_count = min(
|
||||
(int)$this->getInput('parseIssues'),
|
||||
count($issues)
|
||||
@ -87,36 +91,19 @@ class ScientificAmericanBridge extends FeedExpander
|
||||
$items = [];
|
||||
$html = getSimpleHTMLDOMCached($issue_link);
|
||||
|
||||
$features = $html->find('[class^=Detail_issue__article__previews__featured]', 0);
|
||||
if ($features != null) {
|
||||
$articles = $features->find('div', 0)->children();
|
||||
$blocks = $html->find('[class^="issueArchiveArticleListCompact"]');
|
||||
foreach ($blocks as $block) {
|
||||
$articles = $block->find('article[class*="article"]');
|
||||
foreach ($articles as $article) {
|
||||
$h4 = $article->find('h4', 0);
|
||||
$a = $h4->find('a', 0);
|
||||
$a = $article->find('a[class^="articleLink"]', 0);
|
||||
$link = 'https://scientificamerican.com' . $a->getAttribute('href');
|
||||
$title = $a->plaintext;
|
||||
$items[] = [
|
||||
$title = $a->find('h2[class^="articleTitle"]', 0);
|
||||
array_push($items, [
|
||||
'uri' => $link,
|
||||
'title' => $title,
|
||||
'title' => $title->plaintext,
|
||||
'uid' => $link,
|
||||
'content' => ''
|
||||
];
|
||||
}
|
||||
}
|
||||
|
||||
$departments = $html->find('[class^=Detail_issue__article__previews__departments]', 0);
|
||||
if ($departments != null) {
|
||||
$headers = $departments->find('[class*=Listing_article__listing__title]');
|
||||
foreach ($headers as $header) {
|
||||
$a = $header->find('a', 0);
|
||||
$link = 'https://scientificamerican.com' . $a->getAttribute('href');
|
||||
$title = $a->plaintext;
|
||||
$items[] = [
|
||||
'uri' => $link,
|
||||
'title' => $title,
|
||||
'uid' => $link,
|
||||
'content' => ''
|
||||
];
|
||||
]);
|
||||
}
|
||||
}
|
||||
|
||||
@ -125,65 +112,67 @@ class ScientificAmericanBridge extends FeedExpander
|
||||
|
||||
private function updateItem($item)
|
||||
{
|
||||
return $item;
|
||||
$html = getSimpleHTMLDOMCached($item['uri']);
|
||||
$article = $html->find('#sa_body', 0)->find('article', 0);
|
||||
$article = $html->find('#app', 0)->find('article', 0);
|
||||
|
||||
$time = $article->find('time[itemprop="datePublished"]', 0);
|
||||
if ($time == null) {
|
||||
$time = $html->find('span[itemprop="datePublished"]', 0);
|
||||
}
|
||||
$time = $article->find('p[class^="article_pub_date"]', 0);
|
||||
if ($time) {
|
||||
$datetime = DateTime::createFromFormat('F j, Y', $time->plaintext);
|
||||
$datetime->setTime(0, 0, 0, 0);
|
||||
$item['timestamp'] = $datetime->format('U');
|
||||
}
|
||||
$main = $article->find('section.article-grid__main', 0);
|
||||
|
||||
if ($main == null) {
|
||||
$main = $article->find('div.article-text', 0);
|
||||
$authors = $article->find('a[class^="article_authors__link"]');
|
||||
if ($authors) {
|
||||
$author = implode('; ', array_map(fn($a) => $a->plaintext, $authors));
|
||||
$item['author'] = $author;
|
||||
}
|
||||
|
||||
if ($main == null) {
|
||||
return $item;
|
||||
$res = '';
|
||||
$desc = $article->find('div[class^="article_dek"]', 0);
|
||||
if ($desc) {
|
||||
$res .= $desc->innertext;
|
||||
}
|
||||
|
||||
foreach ($main->find('img') as $img) {
|
||||
$img->removeAttribute('width');
|
||||
$img->removeAttribute('height');
|
||||
$img->setAttribute('style', 'height: auto; width: auto; max-height: 768px');
|
||||
$lead_figure = $article->find('figure[class^="lead_image"]', 0);
|
||||
if ($lead_figure) {
|
||||
$res .= $lead_figure->outertext;
|
||||
}
|
||||
|
||||
$rights_link = $main->find('div.article-rightslink', 0);
|
||||
if ($rights_link != null) {
|
||||
$rights_link->parent->removeChild($rights_link);
|
||||
}
|
||||
$reprints_link = $main->find('div.article-reprintsLink', 0);
|
||||
if ($reprints_link != null) {
|
||||
$reprints_link->parent->removeChild($reprints_link);
|
||||
}
|
||||
$about_section = $main->find('section.article-author-container', 0);
|
||||
if ($about_section != null) {
|
||||
$about_section->parent->removeChild($about_section);
|
||||
}
|
||||
$read_next = $main->find('#read-next', 0);
|
||||
if ($read_next != null) {
|
||||
$read_next->parent->removeChild($read_next);
|
||||
$content = $article->find('div[class^="article__content"]', 0);
|
||||
if ($content) {
|
||||
foreach ($content->children() as $block) {
|
||||
if (str_contains($block->innertext, 'On supporting science journalism')) {
|
||||
continue;
|
||||
}
|
||||
if (
|
||||
($block->tag == 'p' && $block->getAttribute('data-block') == 'sciam/paragraph')
|
||||
|| ($block->tag == 'figure' && str_starts_with($block->class, 'article__image'))
|
||||
) {
|
||||
$iframe = $block->find('iframe', 0);
|
||||
if ($iframe) {
|
||||
$res .= "<a href=\"{$iframe->src}\">{$iframe->src}</a>";
|
||||
} else {
|
||||
$res .= $block->outertext;
|
||||
}
|
||||
} else if ($block->tag == 'h2') {
|
||||
$res .= '<h3>' . $block->innertext . '</h3>';
|
||||
} else if ($block->tag == 'blockquote') {
|
||||
$res .= $block->outertext;
|
||||
} else if ($block->tag == 'hr' && $block->getAttribute('data-block') == 'sciam/raw_html') {
|
||||
$res .= '<hr />';
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
foreach ($main->find('iframe') as $iframe) {
|
||||
$a = $html->createElement('a');
|
||||
$a->href = $iframe->src;
|
||||
$a->innertext = $iframe->src;
|
||||
$iframe->parent->appendChild($a);
|
||||
$iframe->parent->removeChild($iframe);
|
||||
$footer = $article->find('footer[class*="footer"]', 0);
|
||||
if ($footer) {
|
||||
$bios = $footer->find('div[class^=bio]');
|
||||
$bio = implode('', array_map(fn($b) => $b->innertext, $bios));
|
||||
$res .= $bio;
|
||||
}
|
||||
|
||||
$authors = $main->find('span[itemprop="author"]', 0);
|
||||
if ($authors != null) {
|
||||
$item['author'] = $authors->plaintext;
|
||||
}
|
||||
|
||||
$item['content'] = $main->innertext;
|
||||
$item['content'] = $res;
|
||||
return $item;
|
||||
}
|
||||
}
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user