I semi-regularly collect my reading into blog posts (see here). The tracking of what I read is via Things, where I store article/book/talk title (as task title) and URL in the notes section (for articles and talks usually). For books, I usually add the book title only and look for the book URL later. The process to create these posts has evolved:
- Delete uninteresting, manually copy the interesting title+link into a Markdown document, add any commentary
- AppleScript that fetches done tasks in the
Articlesproject and auto-formats title+link. Add any commentary - Add the commentary just after reading the article (or before running the above script), modify script to add the notes as well
- Delete uninteresting articles instead of marking as done to reduce overhead, run the script
This has reduced the time needed tremendously, but there was a problematic step left: since I usually didn't have URLs for books, the AppleScript script would silently fail for book records. Also, adding functionality to Applescripts makes for a not-fun experience.
So late this weekend I wrote this small Haskell utility to remove AppleScript from my life. It has the added bonus of
- Not failing for tasks without URL (it marks them with
¿which is a glyph I never use so I can search for it) - Adding the Hugo header and the footer I use
- Being easily extensible (I could add a "search in amazon" for books directly)
Note 1: that this could be a stack script, but I find having a binary I can easily stack install to be very convenient.
Note 2: yes, I'm prefixing all my Haskell repos with h.
Note 3: I tried using Selda for the SQL, but couldn't figure out how to go from Result to a record. sqlite-simple is easy enough especially for such a simple use case. I shouldn't have bothered!
Note 4: There are no tests on purpose, this is small enoguh that I can just run and confirm it works as expected or will fail with enough information to fix.