blog.hugopoi.net/v2/content/post/how-this-blog-is-made/index.md

143 lines
6.0 KiB
Markdown
Raw Normal View History

2022-11-25 17:23:17 +01:00
---
title: "How this blog is made, Archivarix, some PHP and Hugo"
date: 2022-11-06T15:07:16+01:00
draft: true
toc: true
tags: ["youpi"]
---
## The legacy of my blog, recover with Archivarix
For me this year was like a rollercoaster, and I forget my blog was
hosted on a very old Online.net Dedibox server, now called Scaleway.
This server was in a process to be decomission and I missed the 3 emails
annoncing the end of my services. Then Online.net decided that it was a
good idea to also delete the backups spaces attached to these machines.
To sumup I loose my blog and the recent backups. But I wanted to keep it
and a least serve the existing content that was linked on search
engines and other websites. I looked on the wayback machine and my blog
was in it. I found a cool all-in-one service to restore an entire
website from the Wayback machine called Archivarix, the cost was arround
10€.
I recovered a 300MB zip archive with a lot of content, some
images are missing but all the articles was there.
## Running the Archivarix Loader
Archivarix loader is a single php file using a sqlite database with all
your urls inside and the content is stored as files in `www/.content.EZtzwPjb/binary/`.
Each time a HTTP request is process, the script look in the database for
a matching url and serve the content linked to it. This mini cms is
license under GPL, and I put a copy [here](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/branch/master/www/index.php).
### With docker
You need PHP and SQLite extension, the PHP docker image already
contains that. I have done a small docker-compose for running
archivarix.
Simple as run `docker compose up`.
* [docker-compose.yml](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/branch/master/docker-compose.yml)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/branch/master/docker-compose.yml" data-range="1," class="line-numbers"></pre>
* [nginx.conf](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/branch/master/nginx.conf)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/branch/master/nginx.conf" data-range="1," class="line-numbers language-nginx"></pre>
{{< figureCupper
img="Screenshot 2022-11-20 at 18-54-58 HugoPoi Internet Hardware et Bidouille.png"
caption="First run of my old website"
command="Fill"
options="1024x500 Top" >}}
### With Yunohost
I'm mainly self-hosted with the Yunohost project, those next steps show
you how to easily add a small php inside your Yunohost instance.
1. Install the application `My Webapp` inside the yunohost admin panel
{{< figureCupper
img="Screenshot from 2022-11-21 18-40-13.png"
caption="TODO"
command="Fit"
options="1024x500" >}}
1. Fill the setup form
{{< figureCupper
img="Screenshot 2022-11-21 at 18-41-13 Install my_webapp _ Catalog YunoHost Admin.png"
caption="TODO"
command="Fill"
options="1024x500" >}}
1. You have an empty app inside `/var/www/my_webapp/www/`
{{< figureCupper
img="Screenshot from 2022-11-21 18-42-30.png"
caption="TODO"
command="Fit"
options="1024x500" >}}
1. You need to copy your files, I use rsync with the yunohost admin account
`rsync -rlgoD --checksum --verbose www/ admin@home.hugopoi.net:/var/www/my_webapp/www/`
1. Then you might need to `chmod 664 /var/www/my_webapp/www/.content.*/structure.*`, Archivarix required some write access on the sqlite files.
## Modding Archivarix Loader
### Fixing Wordpress version missing files
The homepage was looking good but some wordpress css and javascript assets were missing. Wordpress use a query params `?ver=`.
{{< figureCupper
img="Screenshot 2022-11-20 at 18-55-57 Linky opendata my ass HugoPoi.png"
caption="First run of my old website, some broken css looking wrong"
command="Fill"
options="1024x500 Top" >}}
{{< figureCupper
img="Screenshot-Firefix-debugger-404-ver-wordpress-archivarix.png"
caption="Missing files because of the Wordpress `?ver=` query params with Archivarix"
command="Fit"
options="1024x500" >}}
So I code a little function to load any version available for a
given url. And I take the most recent one.
* [`getOtherWordpressVersionUrls` function in www/index.php](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/commit/2a154a6eea510e08b2608fd55f6729056c363b25/www/index.php#L295-L305)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/commit/2a154a6eea510e08b2608fd55f6729056c363b25/www/index.php" data-range="295,305" class="line-numbers"></pre>
* [The call in www/index.php](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/commit/2a154a6eea510e08b2608fd55f6729056c363b25/www/index.php#L609-L614)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/commit/2a154a6eea510e08b2608fd55f6729056c363b25/www/index.php" data-range="608,620" class="line-numbers"></pre>
### Cleaning existing pages
After successfully running my backuped blog, I wanted to mod some
content.
* Replace the twitter widget
* Replace the hoster widget
* Add a legacy warning for visitor to redirect to the new blog
Archivarix has a `ARCHIVARIX_INCLUDE_CUSTOM` relying
on regular expression to replace content but I needed a more precise approach. I used the PHP XML extension which has a DOM parser buit in and can
parse HTML pages.
* [The easy config to add/replace/delete some html parts](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/src/commit/cd94d82c1a3dad22b026c9c26311b366f76dcd54/www/index.php#L97-L133)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/commit/cd94d82c1a3dad22b026c9c26311b366f76dcd54/www/index.php" data-range="115,134" class="line-numbers"></pre>
* [The clever mod](https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/commit/598d28551d071774007172782541e1d140b8a3c1#diff-eb630ac88267e24589fd94de0826721dff38beb4)
<pre data-src="https://home.hugopoi.net/gitea/hugopoi/blog.hugopoi.net/raw/commit/598d28551d071774007172782541e1d140b8a3c1/www/index.php" data-range="411,439" class="line-numbers"></pre>
## The new blog with Hugo
Go Hugo !