Sunday  9 June 2024

Migrated links data to recutils
===============================

Yesterday I wrote that I am going to shutdown my Linkding bookmark
manager [1].

I use Linkding not only to manage my bookmarks, but also to collect
web pages for my links page [2]. In Linkding, I use a special tag to
identify the web pages that will be included on the links page.

Linkding API
------------
Up until now, the links page is build by the same home-grown SBCL
program that builds my web site.

To build the links page, it uses the Linkding API to get the data for
it. It queries for the specific tag to get the right bookmarks.

Export to GNU recutils recfile
------------------------------
Today, I added a temporary function to the SBCL program that exports
the data for the links page. It creates a GNU recutils recfile,
containing the category, URL and the title for each web page.

The GNU recutils tools are specially created for human-editable,
text-based databases [3].

When a bookmark is added to Linkding, Linkding fetches the title of
the page. Because of this, some links in the link page got a label
like "Home | <website>", or "Blog | <website>".

I used the opportunity to go manually through the recfile and made
some edits.

Small shell script
------------------
I wrote a recfmt file and a small shell script. For each category, the
shell script queries the recfile, assemble the links with recfmt, and
sort the links. The result is a file with the contents of the links
page.

I also modified the SBCL program. To build the links HTML-file, it no
longer calls the Linkding API, but simply reads the content from the
file, that is created by the new shell script.

Maintenance of the links .
--------------------------
From now on, the recfile is used for the maintenance of the links.

To add a web page to the links page, I will manually add it to the
recfile. Or remove a web page from the links page by removing it from
the recfile.

Before running the SBCL program to update my website, the shell script
is run again. This will query the recutils database and recreate the
contents file for the links page. The SBCL program rebuilds all the
files of the website, including the links page.

Final step
----------
Now the creation of the links page has been migrated to a GNU recutils
database, nothing stands in the way of shutting down the bookmark
manager. Which still is scary :)


[1]: gopher://box.matto.nl/0/bookmark-manager-the-siren-call-of-web-browsing.txt
[2]: http://box.matto.nl/links.html
[3]: https://www.gnu.org/software/recutils/manual/html_node/Purpose.html


Last edited: $Date: 2024/06/09 15:55:52 $