Derive feed from HTML (#10)

- Prevent duplicate feed subscriptions - Move FeedItem to its own file
2024-04-27 22:05:39 -04:00
parent 27f7a4c7d9
commit 1826ecd588
3 changed files with 177 additions and 108 deletions
--- a/src/lib/FeedItem.php
+++ b/src/lib/FeedItem.php
@@ -0,0 +1,80 @@
+<?php
+
+/**
+ * Information for a feed item
+ */
+class FeedItem {
+
+    /** @var string The title of the feed item */
+    public string $title = '';
+
+    /** @var string The unique ID for the feed item */
+    public string $guid = '';
+
+    /** @var string The link to the original content */
+    public string $link = '';
+
+    /** @var string When this item was published */
+    public string $publishedOn = '';
+
+    /** @var ?string When this item was last updated */
+    public ?string $updatedOn = null;
+
+    /** @var string The content for the item */
+    public string $content = '';
+
+    /**
+     * Construct a feed item from an Atom feed's `<entry>` tag
+     *
+     * @param DOMNode $node The XML node from which a feed item should be constructed
+     * @return FeedItem A feed item constructed from the given node
+     */
+    public static function fromAtom(DOMNode $node): FeedItem {
+        $guid = Feed::atomValue($node, 'id');
+        $link = '';
+        foreach ($node->getElementsByTagName('link') as $linkElt) {
+            if ($linkElt->hasAttributes()) {
+                $relAttr = $linkElt->attributes->getNamedItem('rel');
+                if ($relAttr && $relAttr->value == 'alternate') {
+                    $link = $linkElt->attributes->getNamedItem('href')->value;
+                    break;
+                }
+            }
+        }
+        if ($link == '' && str_starts_with($guid, 'http')) $link = $guid;
+
+        $item              = new FeedItem();
+        $item->guid        = $guid;
+        $item->title       = Feed::atomValue($node, 'title');
+        $item->link        = $link;
+        $item->publishedOn = Data::formatDate(Feed::atomValue($node, 'published'));
+        $item->updatedOn   = Data::formatDate(Feed::atomValue($node, 'updated'));
+        $item->content     = Feed::atomValue($node, 'content');
+
+        return $item;
+    }
+
+    /**
+     * Construct a feed item from an RSS feed's `<item>` tag
+     *
+     * @param DOMNode $node The XML node from which a feed item should be constructed
+     * @return FeedItem A feed item constructed from the given node
+     */
+    public static function fromRSS(DOMNode $node): FeedItem {
+        $itemGuid = Feed::rssValue($node, 'guid');
+        $updNodes = $node->getElementsByTagNameNS(Feed::ATOM_NS,    'updated');
+        $encNodes = $node->getElementsByTagNameNS(Feed::CONTENT_NS, 'encoded');
+
+        $item              = new FeedItem();
+        $item->guid        = $itemGuid == 'guid not found' ? Feed::rssValue($node, 'link') : $itemGuid;
+        $item->title       = Feed::rssValue($node, 'title');
+        $item->link        = Feed::rssValue($node, 'link');
+        $item->publishedOn = Data::formatDate(Feed::rssValue($node, 'pubDate'));
+        $item->updatedOn   = Data::formatDate($updNodes->length > 0 ? $updNodes->item(0)->textContent : null);
+        $item->content     = $encNodes->length > 0
+            ? $encNodes->item(0)->textContent
+            : Feed::rssValue($node, 'description');
+
+        return $item;
+    }
+}