Weekend Project: Firehose RSS Feed

8 November 2026

This project started on a whim where I mentioned on Mastodon that I was thinking of putting together a firehose feed of my social media presence on a variety of different sites. This was because I realized that I use different sites for different things, and it would be convenient if there was one place I could collect everything. It's also based loosely on the wording of Bluesky's "firehose" - a feed of all updates without any discrimination as to relevance.

The project turned out to be very rudimentary, as it should be, considering RSS is just XML, and everything can parse XML. I went with PHP for a lot of the back-end stuff because I'm not a snob and it Just Works™.

The ingestion

I exist across a selection of sites that were mostly open. We can basically split the ingestion stage into three parts, describing what worked out-of-the-box, what needed some custom wiring, and what was simply impossible.

Hall of Fame

The following services had RSS ready to go and I didn't need to do anything:

Homework

I had to do a little extra work to get a couple things working. I'm going to outline them there.

compend

Compend is very baroquely written on the JavaScript side, but it turns out that eventually if you prod enough you can tell exactly how its probing user reviews. This made it extremely easy for me to write a PHP script to take that backend query and turn it into an RSS feed. I don't want to fully publish it because I feel like this is a little gray-hat, but here's a relevant portion of example:

$reviews = json_decode($response, true);
foreach ($reviews as $_ => $review) { ?>
	<item>
		<title><?= $review["items"]["title"] ?> - <?= str_repeat("★", $review["rating"] / 1) . (fmod($review["rating"], 1.0) == 0.5 ? "⯪" : "") . str_repeat("☆", 5 - $review["rating"]) ?></title>
		<link>https://compend.org/u/hellgnoll/<?= $review["id"] ?></link>
		<description><![CDATA[<?= $review["review"] ?>]]></description>
		<author>gwyn@shadenexus.ca</author>
		<category>Compend</category>
		<pubDate><?= $review["logged_at"] ?></pubDate>
    <media:content url="<?= htmlspecialchars($review["items"]["image_url"], ENT_XML1, 'UTF-8'); ?>" medium="image"  />
	</item>
<?php } ?>

Hardcover

Hardcover on the other hand has a fully open API that uses GraphQL as its system of interaction, and was equally very easy to interact with. I found Emma Goto's blog a really helpful guide on how to probe what I needed out of the API. Most notably the magic query I tweaked:

query {
  me {
    user_books(where: {has_review: {_eq: true}}, order_by: [{reviewed_at: desc}]) {
      review
      rating
      book {
        title
        slug
      }
      id
      reviewed_at
    }
  }
}

Hall of Failure

The following platforms are impossible to bridge or use RSS from because of the provided very stupid reasons:

How it's Wired

The feeds are wired together by a PHP script called PHP Merger that requests all of them and sorts the posts based on date. This is then wired through my own script which adjusts header information to declare itself the firehose!

$channel = $mergedRSS->getElementsByTagName("channel")->item(0);
$channel->getElementsByTagName("title")->item(0)->textContent = "Gwyndolyn Marchant's RSS Firehose";
$channel->getElementsByTagName("link")->item(0)->textContent = "https://shadenexus.ca";
$channel->getElementsByTagName("description")->item(0)->textContent = "Every feed for every social media and updateable site that Gwyn has an account on. NO sorting, NO cleanliness, just the raw firehose of all RSS all at once.";

This script is run every 13 minutes by cron, so that end users can read a cached version that isn't too far behind reality.

*/13 	* 	* 	* 	* 	/usr/local/bin/php /home/████████████/firehose.php > /home/████████/public_html/d/rss/firehose.xml

The result

It's all wired together and ready to grab here! Now you can be annoyed by me no matter where I am.