From 9ed4c2643813f318d633a1f2596b147b53557e06 Mon Sep 17 00:00:00 2001
From: SansGuidon <sansguidon@noreply.gitea.zoemp.be>
Date: Thu, 21 Nov 2024 00:32:08 +0000
Subject: [PATCH] add(script) to generate indieblog rss feed

I'm hosting this on apache on cloudron and running it hourly
---
 generate_indieblog_rss.php | 63 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100644 generate_indieblog_rss.php

diff --git a/generate_indieblog_rss.php b/generate_indieblog_rss.php
new file mode 100644
index 0000000..3f7f2aa
--- /dev/null
+++ b/generate_indieblog_rss.php
@@ -0,0 +1,63 @@
+<?php
+
+// Configuration
+$jsonUrl = "https://indieblog.page/export"; // Source JSON URL
+$rssFile = "/app/data/public/indieblog_feed.xml"; // RSS output file
+$cacheFile = "/app/data/cache/indieblog_cache.txt"; // Cache for processed IDs
+$recentCutoff = time() - 86400; // Articles published within the last 24 hours
+
+// Ensure necessary directories exist
+@mkdir(dirname($rssFile), 0777, true);
+@mkdir(dirname($cacheFile), 0777, true);
+
+// Load JSON data
+$jsonData = @file_get_contents($jsonUrl);
+if (!$jsonData) {
+    die("Error: Unable to fetch JSON from $jsonUrl\n");
+}
+
+// Decode JSON into an array
+$data = json_decode($jsonData, true);
+if (!$data) {
+    die("Error: Invalid JSON data from $jsonUrl\n");
+}
+
+// Load cache of already processed IDs
+$includedIds = file_exists($cacheFile) ? file($cacheFile, FILE_IGNORE_NEW_LINES) : [];
+
+// Filter recent items
+$recentItems = array_filter($data, function ($item) use ($recentCutoff, $includedIds) {
+    return isset($item['published'], $item['itemid']) &&
+           $item['published'] >= $recentCutoff &&
+           !in_array($item['itemid'], $includedIds);
+});
+
+// Sort items by publication date (newest first)
+usort($recentItems, fn($a, $b) => $b['published'] <=> $a['published']);
+
+// Generate RSS feed
+$rss = new SimpleXMLElement('<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"></rss>');
+$channel = $rss->addChild('channel');
+$channel->addChild('title', 'IndieBlog Feed (Recent)');
+$channel->addChild('link', 'https://indieblog.page/');
+$channel->addChild('description', 'RSS feed of articles published in the last 24 hours');
+$channel->addChild('language', 'en');
+
+// Add recent items to RSS feed
+$newEntries = [];
+foreach ($recentItems as $item) {
+    $rssItem = $channel->addChild('item');
+    $rssItem->addChild('title', htmlspecialchars($item['itemtitle'] ?? 'No title'));
+    $rssItem->addChild('link', htmlspecialchars($item['itemurl'] ?? ''));
+    $rssItem->addChild('description', htmlspecialchars($item['feedtitle'] ?? ''));
+    $rssItem->addChild('pubDate', date(DATE_RSS, $item['published']));
+    $newEntries[] = $item['itemid'];
+}
+
+// Update cache with newly processed IDs
+if (!empty($newEntries)) {
+    file_put_contents($cacheFile, implode("\n", array_merge($includedIds, $newEntries)));
+}
+
+// Save RSS feed
+$rss->asXML($rssFile);