diff --git a/bridges/DRKBlutspendeBridge.php b/bridges/DRKBlutspendeBridge.php
index 15075898..7c415d5d 100644
--- a/bridges/DRKBlutspendeBridge.php
+++ b/bridges/DRKBlutspendeBridge.php
@@ -37,6 +37,26 @@ class DRKBlutspendeBridge extends FeedExpander
]
];
+ const OFFER_LOW_PRIORITIES = [
+ 'Imbiss nach der Blutspende',
+ 'Registrierung als Stammzellspender',
+ 'Typisierung möglich!',
+ 'Allgemeine Informationen',
+ 'Krankenkassen belohnen Blutspender',
+ 'Wer benötigt eigentlich eine Blutspende?',
+ 'Win-Win-Situation für die Gesundheit!',
+ 'Terminreservierung',
+ 'Du möchtest das erste Mal Blut spenden?',
+ 'Spende-Check',
+ 'Sie haben Fragen vor Ihrer Blutspende?'
+ ];
+
+ const IMAGE_PRIORITIES = [
+ 'DRK',
+ 'Imbiss',
+ 'Obst',
+ ];
+
public function collectData()
{
$limitItems = intval($this->getInput('limit_items'));
@@ -45,37 +65,116 @@ class DRKBlutspendeBridge extends FeedExpander
protected function parseItem(array $item)
{
- $html = getSimpleHTMLDOM($item['uri']);
+ $html = getSimpleHTMLDOMCached($item['uri']);
$detailsElement = $html->find('.details', 0);
- $dateElement = $detailsElement->find('.datum', 0);
- $dateLines = self::explodeLines($dateElement->plaintext);
-
- $addressElement = $detailsElement->find('.adresse', 0);
- $addressLines = self::explodeLines($addressElement->plaintext);
+ $dateLines = self::explodeLines($detailsElement->find('.datum', 0)->plaintext);
+ $addressLines = self::explodeLines($detailsElement->find('.adresse', 0)->plaintext);
$infoElement = $detailsElement->find('.angebote > h4 + p', 0);
- $info = $infoElement ? $infoElement->innertext : '';
+ $info = $infoElement ? trim($infoElement->plaintext) : '';
- $imageElements = $detailsElement->find('.fotos img');
+ $offers = self::parseOffers($detailsElement->find('.angebote .item'));
- $item['title'] = $dateLines[0] . ' ' . $dateLines[1] . ' ' . $addressLines[0] . ' - ' . $addressLines[1];
+ $images = self::parseImages($detailsElement->find('.fotos', 0));
+ usort($images, function ($imageA, $imageB): int {
+ list($titleA) = $imageA;
+ list($titleB) = $imageB;
+ $prioA = 0;
+ $prioB = 0;
+ foreach (self::IMAGE_PRIORITIES as $prioIndex => $prioTitleNeedle) {
+ if (stripos($titleA, $prioTitleNeedle) !== false) {
+ $prioA = $prioIndex + 1;
+ }
+ if (stripos($titleB, $prioTitleNeedle) !== false) {
+ $prioB = $prioIndex + 1;
+ }
+ }
+ return $prioA - $prioB;
+ });
- $item['content'] = <<{$dateLines[0]} {$dateLines[1]}
- {$addressElement->innertext}
- {$info}
+ $itemContent = <<
+
+ {$dateLines[0]} {$dateLines[1]}
+ {$addressLines[3]}
+
+
+ {$addressLines[0]}
+ {$addressLines[1]}
+ {$addressLines[2]}
+
+
HTML;
- foreach ($imageElements as $imageElement) {
- $src = $imageElement->getAttribute('src');
- $item['content'] .= <<
+ if ($info) {
+ $itemContent .= <<
+ Infos
+ {$info}
+
HTML;
}
+ $majorOffers = array_filter($offers, fn($title): bool => !in_array($title, self::OFFER_LOW_PRIORITIES), ARRAY_FILTER_USE_KEY);
+ foreach ($majorOffers as $offerTitle => list($offerText, $offerImages)) {
+ $itemContent .= <<
+ {$offerTitle}
+ {$offerText}
+ HTML;
+ foreach ($offerImages as list($imageTitle, $imageUrl)) {
+ $itemContent .= <<
+ 
+ {$imageTitle}
+
+ HTML;
+ }
+ $itemContent .= <<
+ HTML;
+ }
+
+ if (count($images) > 0) {
+ $itemContent .= <<
+ Fotos
+ HTML;
+ foreach ($images as list($imageTitle, $imageUrl)) {
+ $itemContent .= <<
+
+ {$imageTitle}
+
+ HTML;
+ }
+ $itemContent .= <<
+ HTML;
+ }
+
+ $minorOffers = array_filter($offers, fn($title): bool => in_array($title, self::OFFER_LOW_PRIORITIES), ARRAY_FILTER_USE_KEY);
+ foreach ($minorOffers as $offerTitle => list($offerText)) {
+ $itemContent .= <<
+ {$offerTitle}
+ {$offerText}
+
+ HTML;
+ }
+
+ $item['title'] = $dateLines[0] . ' ' . $dateLines[1] . ' ' . $addressLines[0] . ' - ' . $addressLines[1];
+ $item['content'] = $itemContent;
$item['description'] = null;
+ $item['enclosures'] = array_map(
+ function ($image): string {
+ list($title, $url) = $image;
+ return $url . '#' . urlencode(str_replace(' ', '_', $title));
+ },
+ $images
+ );
return $item;
}
@@ -97,6 +196,67 @@ class DRKBlutspendeBridge extends FeedExpander
return self::BASE_URI . '/blutspendetermine/termine.rss?date_to=' . $dateTo . '&radius=' . $radius . '&term=' . $term;
}
+ private function parseImages($parentElement): array
+ {
+ $images = [];
+
+ if ($parentElement) {
+ $elements = $parentElement->find('a[data-lightbox]');
+ foreach ($elements as $i => $element) {
+ $url = trim($element->getAttribute('href'));
+ if (!$url) {
+ continue;
+ }
+
+ $title = trim($element->getAttribute('title'));
+ if (!$title) {
+ $number = $i + 1;
+ $title = "Foto {$number}";
+ }
+
+ $images[] = [$title, $url];
+ }
+ }
+
+ return $images;
+ }
+
+ private function parseOffers($offerElements): array
+ {
+ $offers = [];
+
+ foreach ($offerElements as $element) {
+ $title = self::getCleanPlainText($element->find(':is(h1,h2,h3,h4,h5,h6)', 0));
+ $text = trim(substr(self::getCleanPlainText($element), strlen($title)));
+ if (!$title || !$text) {
+ continue;
+ }
+
+ $linkElements = $element->find('a');
+ foreach ($linkElements as $linkElement) {
+ $linkText = trim($linkElement->plaintext);
+ $linkUrl = trim($linkElement->getAttribute('href'));
+ if (!$linkText || !$linkUrl) {
+ continue;
+ }
+
+ $linkHtml = <<{$linkText}
+ HTML;
+ $text = str_replace($linkText, $linkHtml, $text);
+ }
+
+ $offers[$title] = [$text, self::parseImages($element)];
+ }
+
+ return $offers;
+ }
+
+ private function getCleanPlainText($htmlElement): string
+ {
+ return $htmlElement ? trim(preg_replace('/\s+/', ' ', html_entity_decode($htmlElement->plaintext))) : '';
+ }
+
/**
* Returns an array of strings, each of which is a substring of string formed by splitting it on boundaries formed by line breaks.
*/