From 4f8518b1a98f28d371eafbd4e24fcb24ff6c093d Mon Sep 17 00:00:00 2001
From: SansGuidon <sansguidon@noreply.gitea.zoemp.be>
Date: Tue, 3 Dec 2024 00:16:48 +0000
Subject: [PATCH] add(cron) daily indieblog link list

---
 cron/generate_indieblog_daily_rss.php | 78 +++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)
 create mode 100644 cron/generate_indieblog_daily_rss.php

diff --git a/cron/generate_indieblog_daily_rss.php b/cron/generate_indieblog_daily_rss.php
new file mode 100644
index 0000000..11edd58
--- /dev/null
+++ b/cron/generate_indieblog_daily_rss.php
@@ -0,0 +1,78 @@
+<?php
+
+// Configuration
+$jsonUrl = "https://indieblog.page/export";
+$rssFile = "/app/data/public/indieblogv2_feed.xml";
+$cacheFile = "/app/data/cache/indieblogv2_cache.txt";
+$cutoffDate = strtotime('-10 days'); // Seulement les 10 derniers jours
+
+@mkdir(dirname($rssFile), 0777, true);
+@mkdir(dirname($cacheFile), 0777, true);
+
+$jsonData = @file_get_contents($jsonUrl);
+if (!$jsonData) die("Error: Unable to fetch JSON from $jsonUrl\n");
+
+$data = json_decode($jsonData, true);
+if (!$data) die("Error: Invalid JSON data from $jsonUrl\n");
+
+$includedIds = file_exists($cacheFile) ? file($cacheFile, FILE_IGNORE_NEW_LINES) : [];
+
+// Filtrer les éléments récents et non déjà inclus
+$newItems = array_filter($data, function ($item) use ($includedIds, $cutoffDate) {
+    return isset($item['published'], $item['itemid']) &&
+        !in_array($item['itemid'], $includedIds) &&
+        $item['published'] >= $cutoffDate; // Publie dans les 10 derniers jours
+});
+
+usort($newItems, fn($a, $b) => $b['published'] <=> $a['published']);
+
+$groupedByDay = [];
+foreach ($newItems as $item) {
+    $day = isset($item['published']) && is_numeric($item['published'])
+        ? date('Y-m-d', $item['published'])
+        : date('Y-m-d');
+    $groupedByDay[$day][] = $item;
+}
+
+$rss = new SimpleXMLElement('<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"></rss>');
+$channel = $rss->addChild('channel');
+$channel->addChild('title', 'IndieBlog Feed (Last 10 Days)');
+$channel->addChild('link', 'https://indieblog.page/');
+$channel->addChild('description', 'RSS feed of articles from the last 10 days');
+$channel->addChild('language', 'en');
+
+$newEntries = [];
+foreach ($groupedByDay as $day => $items) {
+    $rssItem = $channel->addChild('item');
+    $rssItem->addChild('title', "Links for $day");
+    $rssItem->addChild('link', "https://indieblog.page/$day#" . md5(json_encode($items)));
+    $rssItem->addChild('guid', "https://indieblog.page/$day#" . md5(json_encode($items)));
+    $rssItem->addChild('pubDate', date(DATE_RSS, strtotime($day)));
+
+    $description = "<ul>";
+    foreach ($items as $item) {
+        $postTitle = htmlspecialchars($item['itemtitle'] ?? 'No title', ENT_XML1);
+        $postUrl = htmlspecialchars($item['itemurl'] ?? '#', ENT_XML1);
+        $blogTitle = htmlspecialchars($item['feedtitle'] ?? 'Unknown Blog', ENT_XML1);
+        $blogUrl = htmlspecialchars($item['feedurl'] ?? '#', ENT_XML1);
+
+        $description .= "<li><a href=\"$postUrl\">$postTitle</a> (<a href=\"$blogUrl\">$blogTitle</a>)</li>";
+        $newEntries[] = $item['itemid'];
+    }
+    $description .= "</ul>";
+
+    addCData($rssItem->addChild('description'), $description);
+}
+
+if (!empty($newEntries)) {
+    file_put_contents($cacheFile, implode("\n", array_merge($includedIds, $newEntries)));
+}
+
+$rss->asXML($rssFile);
+
+function addCData(SimpleXMLElement $node, $content)
+{
+    $domNode = dom_import_simplexml($node);
+    $domOwner = $domNode->ownerDocument;
+    $domNode->appendChild($domOwner->createCDATASection($content));
+}