mirror of
https://github.com/wezm/wezm.net.git
synced 2024-12-21 03:39:54 +00:00
61 lines
2.6 KiB
Markdown
61 lines
2.6 KiB
Markdown
|
+++
|
||
|
|
title = "Generating a Static Website From a Pleroma Archive"
|
||
|
|
date = 2024-11-25T20:21:44+10:00
|
||
|
|
|
||
|
|
#[extra]
|
||
|
|
#updated = 2024-06-06T08:24:45+10:00
|
||
|
|
+++
|
||
|
|
|
||
|
|
Almost two years ago, in Jan 2023 I migrated from my Fediverse presence from my
|
||
|
|
self-hosted [Pleroma] instance to a single user Mastodon instance hosted by
|
||
|
|
[masto.host]. Since then I've wanted to retire the Pleroma instance, but I
|
||
|
|
didn't want to just take it offline. I wanted to preserve my posts and
|
||
|
|
links to them. That became a priority over the weekend so I built a
|
||
|
|
tool, `pleroma-archive` to do it.
|
||
|
|
|
||
|
|
<!-- more -->
|
||
|
|
|
||
|
|
A few months after the switch to Mastodon I tried pulling my posts via RSS hit
|
||
|
|
a runtime error. [I reported it to the project][runtime-error] but nothing came
|
||
|
|
of it.
|
||
|
|
|
||
|
|
I ignored it for another 18 months until this weekend when I tried to
|
||
|
|
upgrade my PostgreSQL server from version 12 to 16 and migrate it to a new
|
||
|
|
host. I chose the dump and load method of doing this but when restoring the Pleroma
|
||
|
|
database it appeared to get stuck building one of the indexes. I eventually
|
||
|
|
concluded that it was going to take something like 30 hours to complete. [I'm
|
||
|
|
not the first one to hit this problem either][pg-restore].
|
||
|
|
|
||
|
|
Retiring Pleroma had now become a priority. I discovered that there was now an
|
||
|
|
account backup option in the import/export section of the settings. I
|
||
|
|
downloaded my archive and set about building a tool that could generate a
|
||
|
|
website from it.
|
||
|
|
|
||
|
|
As usual I built the tool in Rust, my scripting language of choice. It's
|
||
|
|
imaginatively called `pleroma-archive`. It generates an index page of all posts
|
||
|
|
as well as a page for each individual post. The public URLs that Pleroma uses
|
||
|
|
are not part of the archive, so for each post the tool does a `HEAD` request
|
||
|
|
with the post id to determine the public URL of the post. The results of this
|
||
|
|
are cached so it only needs to do it once for each post.
|
||
|
|
|
||
|
|
With a little bit of help from [Nginx try\_files][try_files] a page like
|
||
|
|
`notice/ARQGKLTJNiP8Lu2gT2.html` can be served at `/notice/ARQGKLTJNiP8Lu2gT2`,
|
||
|
|
matching the URL it had when served by Pleroma:
|
||
|
|
|
||
|
|
```nginx
|
||
|
|
location / {
|
||
|
|
try_files $uri $uri/index.html $uri.html =404;
|
||
|
|
}
|
||
|
|
```
|
||
|
|
|
||
|
|
The end result is at <https://decentralised.social/>.
|
||
|
|
|
||
|
|
Source code and instructions for using `pleroma-archive` is available at
|
||
|
|
<https://forge.wezm.net/wezm/pleroma-archive>.
|
||
|
|
|
||
|
|
[Pleroma]: https://pleroma.social/
|
||
|
|
[masto.host]: https://masto.host/
|
||
|
|
[runtime-error]: https://git.pleroma.social/pleroma/pleroma/-/issues/3149
|
||
|
|
[pg-restore]: https://git.pleroma.social/pleroma/pleroma/-/issues/3031
|
||
|
|
[try_files]: https://nginx.org/en/docs/http/ngx_http_core_module.html#try_files
|