From abf8b1d21bf2c355c588e990decf043011750376 Mon Sep 17 00:00:00 2001
From: aleteoryx <alyx@aleteoryx.me>
Date: Tue, 12 Nov 2024 15:13:06 -0500
Subject: mostly feature-complete

---
 rss.php | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 210 insertions(+)
 create mode 100644 rss.php

(limited to 'rss.php')

diff --git a/rss.php b/rss.php
new file mode 100644
index 0000000..012bf4e
--- /dev/null
+++ b/rss.php
@@ -0,0 +1,210 @@
+<?php define("VERSION", "0.0.1");
+/* --- FEEDS - EDIT AS NEEDED --- */
+
+
+$feeds["OTW News"]       ["url"] = "https://www.transformativeworks.org/category/announcement/feed/";
+$feeds["Dreamwidth News"]["url"] = "https://dw-news.dreamwidth.org/data/rss";
+
+/* --- CONFIG - EDIT AS NEEDED --- */
+
+
+/// Directory to store RSS cache.
+///
+/// Multiple instances can share one dir.
+$config["cache_dir"] = "/tmp/rss_dot_php";
+
+
+/// Custom CSS
+$config["custom_css"] = <<<'EOC'
+
+/* custom CSS goes here! */
+
+EOC;
+
+
+/// Document Language
+$config["lang"] = "en";
+
+
+/// Date Format
+///
+/// Displayed under every article, see
+/// <https://www.php.net/manual/en/datetime.format.php>
+/// for documentation.
+$config["date_fmt"] = "l, M jS, Y, H:i T";
+
+
+/// Timezone
+///
+/// A value of type DateTimeZone, see
+/// <https://www.php.net/manual/en/class.datetimezone.php>
+/// for documentation.
+$config["timezone"] = new DateTimeZone('UTC');
+
+
+/* --- CODE - DO NOT TOUCH --- */
+
+function load_rss(string $uri, string $linkrel = "alternate"): array {
+  global $config;
+
+  $xml = file_get_contents($uri);
+
+  // if the file doesn't contain an encoding, attempt to read it from http headers and re-encode
+  if (!preg_match("/^[^>]+encoding/", $xml) && str_starts_with($uri, "http")) {
+    foreach ($http_response_header as $header) {
+      if (!str_starts_with(strtolower($header), "content-type")) continue;
+      if (preg_match("/(?<=charset=)[a-z0-9_-]+/i", $header, $matches)) {
+        $xml = iconv($matches[0], "UTF-8", $xml);
+        $doc = new DOMDocument(encoding: "UTF-8");
+      }
+      break;
+    }
+  }
+
+  $doc ??= new DOMDocument();
+  $doc->loadXML($xml);
+
+  if ($doc->documentElement->nodeName == "rss") {
+    // TODO: better rss / atom sniffing
+    foreach ($doc->getElementsByTagName("item") as $node) {
+      $data["title"] = $node->getElementsByTagName("title")
+                            ?->item(0)?->textContent;
+      $data["link"] ??= $node->getElementsByTagName("link")
+                             ?->item(0)?->textContent;
+
+      $data["date"] = new DateTime($node->getElementsByTagName("pubDate")
+                                        ->item(0)->textContent);
+      $data["date"]->setTimezone($config["timezone"]);
+
+      $parsed[] = $data;
+    }
+  } else {
+    // assume atom
+    foreach ($doc->getElementsByTagName("entry") as $node) {
+      $data["title"] = $node->getElementsByTagName("title")
+                          ?->item(0)?->textContent;
+      $data["links"] = [];
+      foreach ($node->getElementsByTagName("link")->getIterator() as $link) {
+        $date["links"][] = ["rel" => $link->getAttribute("rel"),
+                          "href" => $link->getAttribute("href")];
+        if ($link->getAttribute("rel") === $linkrel) {
+          $data["link"] ??= $link->getAttribute("href");
+        }
+      }
+      $data["link"] ??= @$data["links"][0];
+
+      $data["date"] = $node->getElementsByTagName("published")
+                           ?->item(0)?->textContent;
+      $data["date"] ??= $node->getElementsByTagName("updated")
+                             ?->item(0)?->textContent;
+      $data["date"] = new DateTime($data["date"]);
+      $data["date"]->setTimezone($config["timezone"]);
+
+      $parsed[] = $data;
+    }
+  }
+
+  return $parsed??[];
+}
+
+function load_cached(int $ttl, string $uri, string $linkrel = "alternate"): array {
+  global $config;
+  $path = $config["cache_dir"]."/".md5($uri);
+//  echo $path."\n";
+  if ((@filemtime($path) ?? 0) + $ttl < time()) {
+//    echo "cache miss, loading over network\n";
+    $data = load_rss($uri, $linkrel);
+    file_put_contents($path, serialize($data));
+    return $data;
+  } else {
+//    echo "cache hit, loading from file\n";
+    return unserialize(file_get_contents($path));
+  }
+}
+
+@mkdir($config["cache_dir"], recursive: true);
+
+foreach ($_GET["disabled"]??[] as $idx => $feed) {
+  if (!array_key_exists($feed, $feeds)) {
+    unset($_GET["disabled"][$idx]);
+    continue;
+  }
+  $off_feeds[$feed] = @$feeds[$feed];
+  unset($feeds[$feed]);
+}
+
+$combined = [];
+// Real Feed Processing Happens Here
+foreach ($feeds as $name => $data) {
+  if (!isset($data["url"])) {
+    error_log("Feed \"$name\" missing url. Ignoring.");
+    continue;
+  }
+  if (!isset($data["ttl"])) $data["ttl"] = 3600;
+  if (!isset($data["linkrel"])) $data["linkrel"] = "alternate";
+
+  foreach(load_cached($data["ttl"], $data["url"], $data["linkrel"]) as $entry) {
+    $entry["source"] = $name;
+    $combined[] = $entry;
+  }
+}
+
+// reverse-chronological by default
+usort($combined, fn($a, $b) => $b["date"]->getTimestamp() <=> $a["date"]->getTimestamp());
+
+if (isset($_GET["reverse"]))
+  $combined = array_reverse($combined);
+
+$base = parse_url($_SERVER["REQUEST_URI"], PHP_URL_PATH);
+
+?>
+<!doctype html>
+<html lang="<?= $config['lang'] ?>">
+<head>
+  <meta charset="utf-8">
+  <style><?= $config['custom_css'] ?></style>
+</head>
+<body>
+  <nav>
+    <div>
+      <b>Toggle Feeds</b>:
+<?php foreach ($feeds??[] as $name => $data):
+  $query = $_GET;
+  $query["disabled"][] = $name;
+  $uri = $base."?".http_build_query($query);
+?>
+        <span class="source" data-source="<?= htmlentities($name) ?>">
+          <a href="<?= htmlentities($uri) ?>"><?= htmlentities($name) ?></a>
+        </span>
+<?php endforeach; ?>
+<?php foreach ($off_feeds??[] as $name => $data):
+  $query = $_GET;
+  $query["disabled"] = array_filter($query["disabled"], fn($x) => $x !== $name);
+  $uri = $base."?".http_build_query($query);
+?>
+        <span class="source disabled" data-source="<?= htmlentities($name) ?>">
+          <a href="<?= htmlentities($uri) ?>"><?= htmlentities($name) ?></a>
+        </span>
+<?php endforeach; ?>
+    </div>
+  </nav>
+  <main>
+<?php if (!count($combined) && isset($_GET['disabled'])): ?>
+<h1>Looks like you filtered out everything...</h1>
+<p>Try unfiltering some feeds!</p>
+<?php endif;
+      foreach ($combined as $entry): ?>
+    <article>
+      <h1><a href="<?= htmlentities($entry['link']) ?>"><?= htmlentities($entry['title'] ?? "[[[No Title]]]") ?></a></h1>
+      <span class="source" data-source="<?= htmlentities($entry['source']) ?>"><?= htmlentities($entry['source']) ?></span>
+      &bullet;
+      <time datetime="<?= htmlentities($entry['date']->format(DateTime::ISO8601)) ?>">
+        <?= htmlentities($entry['date']->format($config['date_fmt'])) ?>
+      </time>
+    </article>
+<?php endforeach; ?>
+  </main>
+  <!-- generated by rss_dot_php <?= VERSION ?>
+       https://git.aleteoryx.me/cgit/rss_dot_php -->
+</body>
+</html>
-- 
cgit v1.2.3-54-g00ecf