Cli

When viewing longer Markdown notes in Obsidian, tables of content (TOC) help a lot with navigation. There is a handful of community plugins to help with TOC generation, but I have two issues with them:

1. It creates a dependency on code whose developer may lose interest and eventually abandon the project. At least one dynamic TOC plugin has suffered this fate.
2. All of the TOC plugins have the same visual result. When you navigate to a note, Obsidian places the focus at the top of the note, beneath the frontmatter. That’s fine unless the content starts with a TOC markup block, in which case it’s not the TOC itself that is displayed, but the markup for the TOC plugin itself as depicted in the image below.

For me the solution was to write a script that scans the vault looking for this pair of markers:

Obsidian is pretty reckless with file creation dates. If you modify a note in Obsidian, it updates the file creation date. This renders Dataview queries that rely on it useless. For an introduction to this issue, see this lengthy thread on the Obsidian forums.

Workarounds

There are a several solutions to this problem.

1. YAML-based dates

One can include a cdate (or similar) field in the note’s front matter and just direct the Dataview query against that, e.g. LIST FROM "" WHERE startswith(cdate,"2023-05-29") SORT file.ctime asc. This works, but of course it requires you to always place that field ahead of your note content. Some people like that; others not so much.

I use Hazel extensively for automating file management tasks on my macOS systems. Recently I found that Hazel aggressively matches an invisible system file that appears whenever you use a custom file or folder icon. I’ll describe the problem and present a workaround.

In a handful of directories, I have a rule that prevents users (me) from adding certain file types. So the rule just matches any file that is not an image, for example, and deletes it. This is all well and good until to try to add a custom icon to this directory. Since the file Icon? that gets created as a result is not an image, the Hazel rule dutifully deletes it.

As its name implies, the AwesomeTTS Anki add-on is awesome. It’s nearly indispensable for language learners.

You can use it in one of two ways:

1. Subscribe on your own to the text-to-speech services that you plan to use and add those credentials to AwesomeTTS. (à la carte)
2. Subscribe to the AwesomeTTS+ service and gain access to these services. (prix fixe)

Because I had already subscribed to Google and Azure TTS before AwesomeTTS+ came on the scene, there was no reason for me to pay for the comprehensive prix fixe option. Furthermore, since I’ve never gone above the free tier on any of these services, it makes no sense for me to pay for something I’m already getting for free. For others, the convience of a one-stop-shopping experience probably makes the AwesomeTTS+ service worthwhile.

I make a lot of Anki cards, so I’m on a constant quest to make the process more efficient. Like a lot of language-learners, I use images on my cards where possible in order to make the word or sentence more memorable.

Process

When I find an image online that I want to use on the card, I download it to ~/Documents/ankibound. A Hazel rule then grabs the image file and converts it to a .webp file with relatively low quality and a maximum horizontal dimension of 200px. The size and low quality allow me to store lots of images without overwhelming storage capacity, or more importantly, resulting in long synchronization times.

A few years ago, I wrote about my problems with HTML in Anki fields. If you check out that previous post you’ll get the backstory about my objection.

The gist is this: If you copy something from the web, Anki tries to maintain the formatting. Basically it just pastes the HTML off the clipboard. Supposedly, Anki offers to strip the formatting with Shift-paste, but I’ve point out to the developer specific examples where this fails. Basically, I only want plain text. Ever. I will take care of any and all formatting needs via the card templates. Period.

I was using a REST API at https://textance.herokuapp.com/title but it seems awfully fragile. Sure enough this morning, the entire application is down. It’s also not open-source and I have no idea who actually runs this thing.

Here’s the solution:

#!/bin/bash

url=$(pbpaste)
curl $url -so - | pup 'meta[property=og:title] attr{content}'

It does require pup. On macOS, you can install via brew install pup.

There are other ways using regular expressions but no dependency on pup but parsing HTML with regex is not such a good idea.

I alluded to this nuance involving variable scope in my post on automating pdf processing, but I wanted to expand on it a bit.

Consider this little snippet:

i=0
printf "foo:bar:baz:quux" | grep -o '[^:]\+' | while read -r line ; do
   printf "Inner scope: %d - %s\n" $i $line
   ((i++))
   [ $i -eq 3 ] && break;
done
printf "====\nOuter scope\ni = %d\n" $i;

If you run this script - not in interactive mode in the shell - but as a script, what will i be in the outer scope? And why?