Posts

For one-off projects that target Anki collections, I often use Python in a standalone application rather than an Anki add-on. Since I’m not going to distribute these little creations that are specific to my own needs, there’s no reason to create an add-on. These are just a few notes - nothing comprehensive - on the process.

One thing to be aware of is that there must be a perfect match between the Anki major and minor version numbers for the Python anki module to work. If you are running Anki 2.1.48 on your desktop application but have the Python module built for 2.1.49, it will not work. This is a huge irritation and there’s no backwards compatibility; the versions must match precisely.

This may be obvious to some, but visually-recognizing character encoding at a glance is not always obvious.

For example, pronunciation files downloaded form Forvo have the following appearance:

pronunciation_ru_оÑ‚бывание.mp3

How can we extact the actual word from this gibberish? Optimally, the filename should reflect that actual word uttered in the pronunciation file, after all.

Step 1 - Extracting the interesting bits

The gibberish begins after the pronunciation_ru_ and ends before the file extension. Any regex tool can tease that out.

I’ve written a lot about applying and removing syllabic stress marks in Russian text because I use it a lot when making Anki cards.

This iteration is a command line tool for applying the stress mark at a particular character index. The advantage of these little shell tools is that they can be composable, integrating into different tools as the need arises.

#!/usr/local/bin/zsh

while getopts i:w: flag
do
    case "${flag}" in
        i) index=${OPTARG};;
        w) word=${OPTARG};;
    esac
done

if [ $word ]; then
    temp=$word
else
    read temp
fi

outword=""
for (( i=0; i<${#temp}; i++ )); do
    thischar="${temp:$i:1}"
    if [ $i -eq $index ]; then
        thischar=$(echo $thischar | perl -C -pe 's/(.)/\1\x{301}/g;')
    fi
    outword="$outword$thischar"
done

echo $outword

We can use it in a couple different ways. For example, we can provide all of the arguments in a declarative way:

Introducing sterilize-ng [GitHub link] - a URL sterilizer made to work flexibily on the command line.

Background

The surveillance capitalist economy is built on the relentless tracking of users. Imagine going about town running errands but everywhere you go, someone is quietly following you. When you pop into the grocery, they examine your receipt. They look into the bags to see what you bought. Then they hop in the car with you and keep careful records of where you go, how fast you drive, whom you talk with on the phone. This is surveillance capitalism - the relentless “digital exhaust” left by our actions online.

One of the things that I love about Keyboard Maestro is the ability to chain together disparate technologies to achieve some automation goal on macOS.

In most of my previous posts about Keyboard Maestro macros, I’ve used Python or shell scripts, but I decided to draw on some decades-old experience with Perl to do a little text processing for a specific need.

Background

I want this text from Wiktionary:

to look like this:

Russian text intended for learners sometimes contains marks that indicate the syllabic stress. It is usually rendered as a vowel + a combining diacritical mark, typically the combining acute accent \u301. Here are a couple ways of stripping these marks on the command line:

First is a version using Perl

#!/bin/bash

f='покупа́ешья́';
echo $f | perl -C -pe 's/\x{301}//g;'

And then another using the sd tool:

#!/bin/bash

f='покупа́ешья́';
echo $f | sd "\u0301" ""

Both rely on finding the combining diacritical mark and removing it with regex.

It seems like the command line is one of those places where you can accomplish crazy efficient things with one-liners.

Here’s a perfect use case for a CLI one-liner:

In Anki, I often add lists of synonyms and antonyms to my vocabulary cards, but I like them formatted as a bulleted list. My usual route to that involves Markdown. But how to convert this:

известный, точный, определённый, достоверный

to

- `известный`
- `точный`
- `определённый`
- `достоверный`

After trying to come up with a single text replacement strategy to make this work, the best I could do was this:

Often when I import a pronunciation file into Anki, from Forvo for example, the volume isn’t quite right or there’s a lot of background noise; and I want to edit the sound file. How?

The solution for me, as it often the case is a Keyboard Maestro macro.

Prerequisites

• Keyboard Maestro - if you are a macOS power user and don’t have KM, then your missing on a lot.
• Audacity - the multi-platform FOSS audio editor

Outline of the approach

Since Keyboard Maestro won’t know the path to our file in Anki’s collection.media directory, we have to find it. But the first task is to extract the filename. In the Anki note field, it’s going to have this format:

When the Anki application is open on the desktop, it places a lock on the sqlite3 database such that it can’t be queried by another process. One workaround is to try to open the database and if it fails, then make a temporary copy and query that. Of course, this only works with read-only queries. Here’s the basic strategy:

#!/usr/local/bin/python3
# -*- coding: utf-8 -*-

# requires python >= 3.8 to run because of anki module

from anki import Collection, errors

if __name__ == "__main__":
    try:
        col = Collection(path_to_anki_db)
    except (errors.DBError:
        # anki is open, copy to temp file
        import tempfile
        import shutil
        import os

        with tempfile.TemporaryDirectory() as tmpdir:
            dst = os.path.join(tmpdir, 'collectiontemp.anki2')
            shutil.copy(COLLECTION_PATH, dst)
            col = Collection(dst)
            # do something with Anki db

Note that the tempfile context manager will discard the database, if there are actions on the collection that are common to the Anki-is-open and Anki-is-not-open paths then those should be abstracted to separate function.

The Russian letters ё and e have a complex and troubled relationship. The two letters are pronounced differently, but usually appear the same in written text. This presents complications for Russian learners and for text-to-speech systems. In several recent projects, I have needed to normalize the spelling of Russian words. For examples, if I have the written word определенно , is the word actually определенно ? Or is it определённо ?

This was a larger challenge than I imagined. Apart from udar¹, I failed to find any off-the-shelf solutions to what I call normalizing the spelling of words that should be spelled with ё . It turns out that the Russian language Wiktionary respects URLs whether spelled with ё or e . Therefore, one way of normalizing the spelling is to query Wiktionary and grab the headword from the page. Normally I don’t like creating this sort of dependency; but it’s the only solution that presented itself so far. Here’s the approach I took: