Introduction

Hatsu is a self-hosted bridge that interacts with Fediverse on behalf of your static site.

Normally it can do all the:

  • When a Fediverse user searches for a user of your site (@catch-all@example.com), redirects to the corresponding user of the Hatsu instance.
  • When a Fediverse user searches for your site URL (https://example.com/hello-world), redirects to the corresponding post on the Hatsu instance.
  • Accepts follow requests and pushes new posts to the follower's homepage as they become available.
  • Receive replies from Fediverse users and backfeed to your static site.

Best of all, these are fully automated! Just set it up once and you won't need to do anything else.

Comparison

Hatsu is still a Work-In-Progress. It is similar to Bridgy Fed but different:

  • Hatsu uses Feed (JSON / Atom / RSS) as a data source instead of HTML pages with microformats2.
  • Hatsu doesn't require you to automatically or manually send Webmention reminders for create and update, it's all fully automated.
  • Hatsu is ActivityPub only, which means it doesn't handle Nostr, AT Protocol (Bluesky) or other protocols.

If you don't want to self-host, you may still want to use Bridgy Fed or Bridgy in some cases:

Bridgy Fed

  • You don't mind compatibility with platforms other than Mastodon.
  • Your site has good microformats2 markup.

Bridgy

  • You already have a Fediverse account ready to be used for this purpose.
  • Your site has good microformats2 markup.

Getting Started

Setting up your static site

Once you register in Hatsu instance, it will be updated fully automatically.

However, your site will need to make some changes accordingly to take advantage of the ActivityPub features that Hatsu brings to the table.

Choose instance

After Hatsu supports public instances, there may be a list of instances here.

Until then, you'll need to self-host the instance or find an person running a Hatsu instance and have them create an account.

Feed

For Hatsu to work, your site needs to have one of the valid JSON / Atom / RSS feeds.

These feeds should be auto-discoverable on the homepage:

<!-- https://example.com -->
<!DOCTYPE html>
<html>
    <head>
        ...
        <link rel="alternate" type="application/feed+json" href="https://example.com/feed.json" />
        <link rel="alternate" type="application/atom+xml" href="https://example.com/atom.xml" />
        <link rel="alternate" type="application/rss+xml" href="https://example.com/rss.xml" />
    </head>
    <body>
        ...
    </body>
</html>

Hatsu detects all available feeds and prioritizes them in order of JSON > Atom > RSS.

JSON Feed

Hatsu uses serde to parse JSON Feed directly, so you can expect it to have first-class support.

Please make sure your feed is valid in JSON Feed Validator first.

JSON Feed Items

Hatsu infers object id from item.url and item.id.

It will use the item.url first, and if it doesn't exist, it will try to convert the item.id to an absolute url.

https://example.com/foo/bar => https://example.com/foo/bar
/foo/bar => https://example.com/foo/bar 
foo/bar => https://example.com/foo/bar

Ideally, your item.id and item.url should be consistent absolute links:

{
	"id": "https://example.com/foo/bar",
    "url": "https://example.com/foo/bar",
	"title": "...",
	"content_html": "...",
	"date_published": "..."
}

JSON Feed Extension

If you can customize your site's JSON Feed, you might also want to check out the Hatsu JSON Feed Extension.

Atom / RSS

Hatsu uses feed-rs to parse XML feeds and convert them manually.

Please make sure your feed is valid in W3C Feed Validation Service first.

This section is currently lacking testing, so feel free to report bugs.

Redirecting

There are two types of redirects required by Hatsu:

  1. Well Known files, redirecting them to make your username searchable.

    • before: https://example.com/.well-known/webfinger?resource=acct:carol@example.com
    • after: https://hatsu.local/.well-known/webfinger?resource=acct:carol@example.com
  2. Requests accept of type application/activity+json, redirecting them to make your page searchable.

    • before: https://example.com/foo/bar
    • after: https://hatsu.local/posts/https://example.com/foo/bar

There are many ways to redirect them and you can pick one you like:

with Static files and Markup

This should apply to most hosting services and SSG.

with Redirects file

Works with Netlify and Cloudflare Pages.

with Platform-Specific Configuration

Works with Netlify and Vercel.

with Aoba (Lume & Hono)

SSG plugin for Lume and Server Middleware for Deno Deploy and Netlify.

Redirecting with Static files and Markup

This should apply to most hosting services and SSG.

Well Known

For the .well-known/* files, you need to get the corresponding contents from the hatsu instance and output them as a static file.

Replace hatsu.local with your Hatsu instance and example.com with your site.

Open your Hatsu instance home page in a browser and F12 -> Console to run:

// .well-known/webfinger
await fetch('https://hatsu.local/.well-known/webfinger?resource=acct:example.com@hatsu.local').then(res => res.text())
// .well-known/nodeinfo
await fetch('https://hatsu.local/.well-known/nodeinfo').then(res => res.text())
// .well-known/host-meta
await fetch('https://hatsu.local/.well-known/host-meta').then(res => res.text()).then(text => text.split('\n').map(v => v.trim()).join(''))
// .well-known/host-meta.json
await fetch('https://hatsu.local/.well-known/host-meta.json').then(res => res.text())

This will fetch their text contents, which you need to save to the SSG equivalent of the static files directory and make sure they are output to the .well-known folder.

AS2 Alternate

Only Mastodon and Misskey (and their forks) is known to support auto-discovery, other software requires redirection to search correctly. w3c/activitypub#310

Make your posts searchable on Fediverse by setting up auto-discovery.

Since Hatsu's object URLs are predictable, you just need to make sure:

  • The page you want to set up for auto-discovery is in the Feed.
  • The actual URL of the page is the same as in the Feed. (see ./feed)

That's it! For https://example.com/foo/bar, just add the following tag to the document.head:

Replace hatsu.local with your Hatsu instance.

<link rel="alternate" type="application/activity+json" href="https://hatsu.local/posts/https://example.com/foo/bar" />

Redirecting with Redirects file

Works with Netlify and Cloudflare Pages.

Well Known

Create a _redirects file in the SSG static files directory containing the following:

Replace hatsu.local with your Hatsu instance.

/.well-known/host-meta* https://hatsu.local/.well-known/host-meta:splat 307
/.well-known/nodeinfo* https://hatsu.local/.well-known/nodeinfo 307
/.well-known/webfinger* https://hatsu.local/.well-known/webfinger 307

AS2

Redirects file only applies to .well-known. for AS2 redirects, you need to use AS2 Alternate.

Redirecting with Platform-Specific Configuration

Works with Netlify and Vercel.

Well Known

Netlify (netlify.toml)

Create a netlify.toml file in the root directory containing the following:

Replace hatsu.local with your Hatsu instance.

[[redirects]]
  from = "/.well-known/host-meta*"
  to = "https://hatsu.local/.well-known/host-meta:splat"
  status = 307
[[redirects]]
  from = "/.well-known/nodeinfo*"
  to = "https://hatsu.local/.well-known/nodeinfo"
  status = 307
[[redirects]]
  from = "/.well-known/webfinger*"
  to = "https://hatsu.local/.well-known/webfinger"
  status = 307

Vercel (vercel.json)

Create a vercel.json file in the root directory containing the following:

Replace hatsu.local with your Hatsu instance.

{
  "redirects": [
    {
      "source": "/.well-known/host-meta",
      "destination": "https://hatsu.local/.well-known/host-meta"
    },
    {
      "source": "/.well-known/host-meta.json",
      "destination": "https://hatsu.local/.well-known/host-meta.json"
    },
    {
      "source": "/.well-known/nodeinfo",
      "destination": "https://hatsu.local/.well-known/nodeinfo"
    },
    {
      "source": "/.well-known/webfinger",
      "destination": "https://hatsu.local/.well-known/webfinger"
    }
  ]
}

AS2

Redirects file only applies to .well-known. for AS2 redirects, you need to use AS2 Alternate.

Redirecting with Aoba (Lume & Hono)

SSG plugin for Lume and Server Middleware for Deno Deploy and Netlify.

Aoba provides some plugins and server middleware for Lume and Hono, including Hatsu integration.

Lume

The Lume plugin will do what you did in Redirecting with Static files and Markup for you.

Replace hatsu.local with your Hatsu instance and example.com with your site.

import lume from 'lume/mod.ts'
import { hatsuPlugin } from 'aoba/lume/plugins/hatsu.ts'

export default lume({ location: new URL('https://example.com') })
    .use(hatsuPlugin({
        // Hatsu instance
        instance: new URL('https://hatsu.local'),
        // match /posts/*
        match: [/^\/posts\/(.+)$/],
    }))

Lume Server

On top of that, the Lume server middleware can redirect .well-known/* and AS2 request.

Replace hatsu.local with your Hatsu instance.

import Server from 'lume/core/server.ts'
import site from './_config.ts'
import { hatsuMiddleware } from 'aoba/lume/middlewares/hatsu.ts'

const server = new Server()

server.use(hatsuMiddleware({
    // Hatsu instance
    instance: new URL('https://hatsu.local'),
    // site location
    location: site.options.location,
}))

server.start()

Hono

It's not published to npm, so feel free to copy and paste it if you need to use it in a Node.js.

Replace hatsu.local with your Hatsu instance.

import { Hono } from 'hono'
import { hatsuWellKnown, hatsuObject } from 'aoba/hono/middlewares/hatsu.ts'

const app = new Hono()
const instance = new URL('https://hatsu.local')

// https://example.com/.well-known/* => https://hatsu.local/.well-known/*
app.use('/.well-known/*', hatsuWellKnown({ instance }))
// https://example.com/posts/foo => https://hatsu.local/posts/https://example.com/posts/foo
app.use('/posts/*', hatsuObject({ instance }))

Redirecting with FEP-612d

There doesn't seem to be software currently implements FEP-612d, but that won't stop us from setting it up.

just add the following TXT record:

Replace hatsu.local with your Hatsu instance and example.com with your site.

_apobjid.example.com https://hatsu.local/users/example.com

That's it!

Backfeed

Display mentions received by Hatsu on your site.

based on KKna

based on Mastodon Comments

based on Webmention (TODO)

Backfeed based on KKna

Written by the same authors as Hatsu, KKna provides the simplest integration for Hatsu.

Examples

Replace hatsu.local with your Hatsu instance.

<script type="module">
  import { defineConfig } from 'https://esm.sh/@kkna/context'
  import { hatsu } from 'https://esm.sh/@kkna/preset-hatsu'

  defineConfig({
    presets: [hatsu({ instance: 'https://hatsu.local' })],
  })
</script>
<script type="module" src="https://esm.sh/@kkna/component-material"></script>
<kkna-material></kkna-material>

You can use it with other presets or write your own components, see the KKna Documentation for details.

Backfeed based on Mastodon Comments

Examples

"Mastodon Comments" refers to the @oom/mastodon-components library.

<script type="module">
  import Comments from 'https://esm.run/@oom/mastodon-comments'
  customElements.define('oom-comments', Comments)
</script>
<link
  rel="stylesheet"
  href="https://cdn.jsdelivr.net/npm/@oom/mastodon-comments/src/styles.css"
/>
<oom-comments src="https://mastodon.gal/@misteroom/110810445656343599">
  No comments yet
</oom-comments>

The basic example should look something like the above, where https://mastodon.gal/@misteroom/110810445656343599 is the link to the post in Fediverse.

Hatsu uses predictable URLs, you just need to change the src:

// trim url
// input:
// https://example.com/foo/bar#baz
// https://example.com/foo/bar?baz=qux
// output:
// https://example.com/foo/bar
const { origin, pathname } = new URL(window.location.href)
const url = new URL(pathname, origin).href

// get id (base64url encode)
// aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy
const id = btoa(url).replaceAll('+', '-').replaceAll('/', '_')

// oom-comments src
// https://hatsu.local/notice/aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy
const src = new URL(`/notice/${id}`, 'https://hatsu.local').href

So eventually it will look like this:

<script type="module">
  import Comments from 'https://esm.run/@oom/mastodon-comments'
  customElements.define('oom-comments', Comments)
</script>
<link
  rel="stylesheet"
  href="https://cdn.jsdelivr.net/npm/@oom/mastodon-comments/src/styles.css"
/>
<oom-comments
  src="https://hatsu.local/notice/aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy"
>
  No comments yet
</oom-comments>

It's a real pain in the ass, but you can try to automate it.

Lume

If you're using Lume and Theme Simple Blog, it will read data.comments.src.

So you can do this:

// _config.ts
import lume from 'lume/mod.ts'
import blog from 'https://deno.land/x/lume_theme_simple_blog@v0.14.0/mod.ts'

const site = lume()

site.use(blog())

// add this:
site.preprocess(['.md'], (pages) =>
  pages
    .filter((page) => page.type === 'post')
    .forEach((page) => {
      page.data.comments = {
        src: new URL(
          `/notice/${btoa(site.url(page.data.url, true))
            .replaceAll('+', '-')
            .replaceAll('/', '_')}`,
          'https://hatsu.local' // your hatsu instance
        ).href,
      }
    })
)

export default site

How it works?

Hatsu mimics Pleroma's URL format.

@oom/mastodon-components will extract the ID from the URL and query the corresponding API for the data.

// oom-comments src
const src = 'https://hatsu.local/notice/aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy'
// origin: 'https://hatsu.local
// pathname: '/notice/aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy'
const { origin, pathname } = new URL(src)
// id: 'aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy'
const [, id] = pathname.match(/^\/notice\/([^\/?#]+)/)
// api url: https://hatsu.local/api/v1/statuses/aHR0cHM6Ly9leGFtcGxlLmNvbS9mb28vYmFy/context
const url = new URL(`/api/v1/statuses/${id}/context`, origin)

Upon receiving a request, Hatsu's corresponding API will attempt to decode the base64url ID and return the data.

If you're interested in the code, you can also take a look at routes/statuses/status_context.rs and entities/context.rs.

Backfeed based on Webmention

This section is not yet implemented in Hatsu.

TODO

Install

Docker Installation

Binary Installation

Docker Installation

Hatsu uses the x86-64-v3 target architecture for optimal performance.

If you are using an older processor, you currently need to build locally and change the corresponding values in .cargo/config.toml.

You can find images on GitHub: https://github.com/importantimport/hatsu/pkgs/container/hatsu

Hatsu uses three primary tags: latest (stable), beta and nightly, literally.

docker run

Replace {{version}} with the version you want to use.

docker run -d \
  --name hatsu \
  --restart unless-stopped \
  -p 3939:3939 \
  -v /opt/hatsu/hatsu.sqlite3:/app/hatsu.sqlite3 \
  -e HATSU_DATABASE_URL=sqlite://hatsu.sqlite3 \
  -e HATSU_DOMAIN={{hatsu-instance-domain}} \
  -e HATSU_LISTEN_HOST=0.0.0.0 \
  -e HATSU_PRIMARY_ACCOUNT={{your-static-site}} \
  -e HATSU_ACCESS_TOKEN=123e4567-e89b-12d3-a456-426614174000 \
  ghcr.io/importantimport/hatsu:{{version}}

You need to specify all environment variables at once. For more information, see Environments.

docker compose

The examples folder contains some sample docker compose configurations,

You can make your own modifications based on them.

Binary Installation

Hatsu uses the x86-64-v3 target architecture for optimal performance.

If you are using an older processor, you currently need to build locally and change the corresponding values in .cargo/config.toml.

Releases

You can download both stable and beta versions of Hatsu from the Releases page.

Artifacts

You can find the latest artifacts on the Workflow runs page.

GitHub has a document that tells you how to download artifact: https://docs.github.com/en/actions/managing-workflow-runs/downloading-workflow-artifacts

Environments

Hatsu supports dotenv, and you can set environment variables via the .env file.

It is required unless it has a suffix (optional).

However, it may exist as a built-in preset (in the source code) or an example preset (in .env.example)

HATSU_ENV_FILE

  • default: /etc/hatsu/.env

Hatsu will first try to find the dotenv file in the current directory, and if unsuccessful will try to use the path indicated by HATSU_ENV_FILE.

HATSU_DATABASE_URL

  • default: sqlite::memory:
  • example: sqlite://hatsu.sqlite3

Should be a valid sqlite: or postgres: URL. see sea-ql.org

sqlite::memory: is used by default to ensure that hatsu does not report errors as a result.

If you're not using a Postgres database, I recommend keeping the sqlite://hatsu.sqlite3.

HATSU_DOMAIN

  • default: None
  • example: hatsu.local

The domain name you assigned to this Hatsu instance.

HATSU_LISTEN_HOST

  • default: 127.0.0.1
  • example: 0.0.0.0

The hostname on which the Hatsu server listens.

HATSU_LISTEN_PORT

  • default: 3939
  • example: 3939

The port on which the Hatsu server listens.

HATSU_PRIMARY_ACCOUNT

  • default: None
  • example: None

The primary account for this Hatsu instance, which cannot be removed and is used as a signed_fetch_actor.

HATSU_ACCESS_TOKEN (optional)

  • default: None
  • example: None

For accessing Admin API.

If this value is not set, the Hatsu Admin API will not be available.

This can be any string, but I recommend generating a random uuid v4.

echo "\nHATSU_ACCESS_TOKEN = \"$(cat /proc/sys/kernel/random/uuid)\"" >> .env

HATSU_NODE_NAME (optional)

  • default: None
  • example: None

Used for NodeInfo metadata.nodeName.

HATSU_NODE_DESCRIPTION (optional)

  • default: None
  • example: None

Used for NodeInfo metadata.nodeDescription.

Create Account

Ensure you set HATSU_ACCESS_TOKEN correctly in the previous section first, otherwise you will not be able to use the Hatsu Admin API.

just

The easiest way to create an account is the just command line tool:

just account create example.com

If you are using docker, you need to exec to the container first.

docker exec -it hatsu /bin/bash

curl

You can also access the API via curl, as Justfile does.

curl -X POST "http://localhost:$(echo $HATSU_LISTEN_PORT)/api/v0/admin/create-account?token=$(echo $HATSU_ACCESS_TOKEN)" \
  -H "Content-Type: application/json" \
  -d "{\"name\": \"example.com\"}"

Prepare

Clone Repository

It will create a hatsu subfolder in the current path.

git clone https://github.com/importantimport/hatsu.git && cd hatsu

Contributing

Go to the hatsu folder and you can see these:

Local Development

You'll need to complete prepare before you do this.

Dependencies

To develop Hatsu, you should first install Rust and some dependencies.

# Arch-based distro
sudo pacman -S git cargo openssl

# Debian-based distro
sudo apt install git cargo libssl-dev

Running

First copy the variables,

Set HATSU_DOMAIN to your prepared domain (e.g. hatsu.example.com without https://)

and HATSU_PRIMARY_ACCOUNT to your desired user domain (e.g. blog.example.com without https://)

# copy env example
cp .env.example .env
# edit env
nano .env

Then create the database file and run:

# create database
touch hatsu.sqlite3
# run hatsu
cargo run

Hatsu now listen on localhost:3939, and in order for it to connect to Fediverse, you'll also need to set up a reverse proxy.

Docker Development

You'll need to complete prepare before you do this.

Dependencies

To use Docker, you only need to install Docker and Docker Compose.

# Arch-based distro
sudo pacman -S docker docker-compose

# Debian-based distro
sudo apt install docker.io docker-compose

Running

First copy the variables,

Set HATSU_DOMAIN to your prepared domain (e.g. hatsu.example.com without https://)

and HATSU_PRIMARY_ACCOUNT to your desired user domain (e.g. blog.example.com without https://)

# copy env example
cp .env.example .env
# edit env
nano .env

Then create the database file and run:

# create database
touch hatsu.sqlite3
# run hatsu
docker-compose up -d

If there is no build image, it will be built automatically at execution time. Hatsu uses cargo-chef in the Dockerfile, which caches dependencies to avoid duplicate build dependencies.

If you need to rebuild, add the --build flag:

docker-compose up -d --build

Compatibility Chart

Hatsu is primarily geared towards the micro-blogging platform in the Fediverse.

Currently I've created a chart for all the platforms I expect to be compatible with, and hopefully it will be filled in later:

Send

Create (Note)Accept (Follow)
Mastodon
GoToSocial
Misskey
Pleroma

Receive

Akkoma, Sharkey, etc. forks should be compatible with upstream, so they are not listed separately.

Federation in Hatsu

Supported federation protocols and standards

Supported FEPs

ActivityPub

The following activities and object types are supported:

Send

  • Accept(Follow)
  • Create(Note), Update(Note)

Receive

  • Follow(Actor), Undo(Follow)
  • Create(Note)
  • Like(Note), Undo(Like)
  • Announce(Note), Undo(Announce)

Activities are implemented in way that is compatible with Mastodon and other popular ActivityPub servers.

Notable differences

  • No shared inbox.

Additional documentation

Hatsu JSON Feed Extension

To allow you to customize your postings, Hatsu defines a JSON Feed extension that uses the _hatsu key.

All extension keys for the Hatsu JSON Feed Extension are optional.

Note: everything here is experimental. It is always subject to breaking changes and does not follow semver.

Top-level

The following applies to the Top-level JSON Feed.

  • about (optional but strongly recommended, string) is the URL used to introduce this extension to humans. should be https://github.com/importantimport/hatsu/issues/1 .
  • aliases (optional, string) is the customized username used for FEP-4adb and FEP-2c59.
  • banner_image (optional, string) is the URL of the banner image for the website in hatsu.

Items

The following applies to the JSON Feed Item.