Migrating My Bookmarks From Raindrop to Linkding

I recently decided to migrate my bookmarks from Raindrop.io to Linkding, a self-hosted bookmark manager. I rather own my bookmarks' data instead of using a third-party service for this simple use-case.

While Linkding has a built-in HTML import feature, it doesn't preserve the folder structure from Raindrop so all bookmarks end up unorganized. It also doesn't help that Raindrop uses a folder structure, while Linkding uses a tag-based system (with a nifty feature called "Bundles" on top of that).

Let's ask AI to automate this for us!

I spun up a local intance of Linkding (See Linkding#Using Docker), mounted the data directory so I could have access to the SQLite database files, navigated into it and fired up Claude Code.

I had to give Claude Code all the necessary context so that it could successfully finish the task. I asked it to:

• Use a sqlite docker container to view the tables and structure of the Linkding database, in order to understand how the data and the relationship are stored.
• Read my Raindrop HTML export file containing all my bookmarks
• Look at how the bookmarks are organized in the Raindrop export file so it can understand the nesting of bookmarks inside collections (i.e. folders)
• Create a script to parse the Raindrop export file and generate a .sql file that will populate Linkding.

I also told it that all tags must be ASCII only, lowercase, without spaces (only dashes), and without special characters.

Claude generated the following script.

This script assumes that it's running inside Linkding's data directory, located at /data (Important!), and there is a filed called "Raindrop Export.HTML" in that directory.

Running this script creates a new file called organize_bookmarks.sql

... Profit ?

Now we can run the script:

docker run --rm -v $(pwd):/data python:3.11-slim python /data/parse_raindrop.py

Example output:

Found 500 bookmarks

Bookmark distribution by main folder:
  Software -> software: 100 bookmarks
  Misc -> misc: 50 bookmarks
  ...

After running the script, we now have a organize_bookmarks.sql file that we can run on Linkding's SQLite database to import all our data.

At this point, after all my local testing, I deployed Linkding on my actual server, and then copied over the .sql file to it for the real import.

First, stop the Linkding container:

docker stop linkding

Now, import the .sql file into the database:

docker run --rm -v $(pwd):/data \
  --user $(id -u www-data):$(id -g www-data) \
  keinos/sqlite3 sqlite3 /data/db.sqlite3 \
  '.read /data/organize_bookmarks.sql'

*Note: *It's important to run as the www-data user (which is the user owning the database files), otherwise you'll get the following error: Runtime error near line 2135: attempt to write a readonly database (8).

And that's it!

How Tags Are Organized

The script follows these rules:

1. Tag normalization: All tags are lowercase with dashes

   • "3D Printing" → 3d-printing
   • "Mechanical Keyboards" → mechanical-keyboards

2. Nested folders: Only the deepest folder becomes a tag

   • Bookmark in "Software > Rust" → gets tag rust

3. All tags are created: Even parent folders become separate tags for organization