mirror/php-rss-bridge
1
0
Fork 0
mirror of https://github.com/RSS-Bridge/rss-bridge.git synced 2025-06-10 00:25:02 +02:00
Code Issues Releases Wiki Activity

[AllegroBridge]: fix, use JSON instead of HTML (#4536)

Browse Source 
Cookie is now obligatory, otherwise 403 is returned
This commit is contained in:
Dawid Wróbel 2025-05-09 05:06:23 +02:00 committed by GitHub
parent 083ba1e4f7
commit 72e1998e16
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 37 additions and 74 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
bridges/AllegroBridge.php
View File

@ -15,8 +15,8 @@ class AllegroBridge extends BridgeAbstract
],
'cookie' => [
'name' => 'The complete cookie value',
'title' => 'Paste the value of the cookie value from your browser if you want to prevent Allegro imposing rate limits',
'required' => false,
'title' => 'Paste the cookie value from your browser, otherwise 403 gets returned',
'required' => true,
],
'includeSponsoredOffers' => [
'type' => 'checkbox',
@ -65,93 +65,56 @@ class AllegroBridge extends BridgeAbstract
$url = preg_replace('/([?&])order=[^&]+(&|$)/', '$1', $this->getInput('url'));
$url .= (parse_url($url, PHP_URL_QUERY) ? '&' : '?') . 'order=n';
$opts = [];
$html = getContents($url, [], [CURLOPT_COOKIE => $this->getInput('cookie')]);
// If a cookie is provided
if ($cookie = $this->getInput('cookie')) {
$opts[CURLOPT_COOKIE] = $cookie;
$storeData = null;
if (preg_match('/<script[^>]*>\s*(\{\s*?"__listing_StoreState".*\})\s*<\/script>/i', $html, $match)) {
$data = json_decode($match[1], true);
$storeData = $data['__listing_StoreState'] ?? null;
}
$html = getSimpleHTMLDOM($url, [], $opts);
foreach ($storeData['items']['elements'] as $elements) {
if (!array_key_exists('offerId', $elements)) {
continue;
}
if (!$this->getInput('includeSponsoredOffers') && $elements['isSponsored']) {
continue;
}
if (!$this->getInput('includePromotedOffers') && $elements['promoted']) {
continue;
}
# if no results found
if ($html->find('.mzmg_6m.m9qz_yo._6a66d_-fJr5')) {
return;
}
$results = $html->find('article[data-analytics-view-custom-context="REGULAR"]');
if ($this->getInput('includeSponsoredOffers')) {
$results = array_merge($results, $html->find('article[data-analytics-view-custom-context="SPONSORED"]'));
}
if ($this->getInput('includePromotedOffers')) {
$results = array_merge($results, $html->find('article[data-analytics-view-custom-context="PROMOTED"]'));
}
foreach ($results as $post) {
$item = [];
$item['uid'] = $elements['offerId'];
$item['uri'] = $elements['url'];
$item['title'] = $elements['alt'];
$item['uid'] = $post->{'data-analytics-view-value'};
$item_link = $post->find('a[href*="' . $item['uid'] . '"], a[href*="allegrolokalnie"]', 0);
$item['uri'] = $item_link->href;
$item['title'] = $item_link->find('img', 0)->alt;
$image = $item_link->find('img', 0)->{'data-src'} ?: $item_link->find('img', 0)->src ?? false;
$image = $elements['photos'][0]['medium'];
if ($image) {
$item['enclosures'] = [$image . '#.image'];
}
$price = $post->{'data-analytics-view-json-custom-price'};
if ($price) {
$priceDecoded = json_decode(html_entity_decode($price));
$price = $priceDecoded->amount . ' ' . $priceDecoded->currency;
$price = $elements['price']['mainPrice']['amount'];
$currency = $elements['price']['mainPrice']['currency'];
$sellerType = $elements['seller']['title'];
$item['categories'] = [$sellerType];
$description = '';
foreach ($elements['parameters'] as $parameter) {
$item['categories'] = array_merge($item['categories'], $parameter['values']);
$description .= '<dt>' . $parameter['name'] . ': ' . implode(',', $parameter['values']) . '</dt>';
}
$descriptionPatterns = ['/<\s*dt[^>]*>\b/', '/<\/dt>/', '/<\s*dd[^>]*>\b/', '/<\/dd>/'];
$descriptionReplacements = ['<span>', ':</span> ', '<strong>', '&emsp;</strong> '];
$description = $post->find('.m7er_k4.mpof_5r.mpof_z0_s', 0)->innertext;
$descriptionPretty = preg_replace($descriptionPatterns, $descriptionReplacements, $description);
$pricingExtraInfo = array_filter($post->find('.mqu1_g3.mgn2_12'), function ($node) {
return empty($node->find('.mvrt_0'));
});
$pricingExtraInfo = $pricingExtraInfo[0]->plaintext ?? '';
$offerExtraInfo = array_map(function ($node) {
return str_contains($node->plaintext, 'zapłać później') ? '' : $node->outertext;
}, $post->find('div.mpof_ki.mwdn_1.mj7a_4.mgn2_12'));
$isSmart = $post->find('img[alt="Smart!"]', 0) ?? false;
if ($isSmart) {
$pricingExtraInfo .= $isSmart->outertext;
}
$item['categories'] = [];
$parameters = $post->find('dd');
foreach ($parameters as $parameter) {
if (in_array(strtolower($parameter->innertext), ['brak', 'nie'])) {
continue;
}
$item['categories'][] = $parameter->innertext;
}
$item['content'] = $descriptionPretty
. '<div><strong>'
. $price
. '</strong></div><div>'
. implode('</div><div>', $offerExtraInfo)
. '</div><dl>'
. $pricingExtraInfo
$item['content'] = '<div><strong>'
. $price . ' ' . $currency
. '</strong></div><dl><dt>'
. $sellerType . '</dt>'
. $description
. '</dl><hr>';
$this->items[] = $item;
}
}
}