Anki Decks with Orgmode

Setting up unicode math and orgmode for painless Anki deck building

Background

A recent Hacker News post reminded me of Anki, and that brought back memories of my Anki orgmode setup. I thought I’d re-create and immortalize it.

The standard way of working with Anki, is with a pretty awkward GUI. There are changes to be made here, which make life a little easier, including the setup of custom cards, but the inherent concerns of the WYSIWYG editor are basically insurmountable.

The goal is to get this a better workflow than manual editing of Anki decks. orgmode is perfect for making cards, especially in the larger context of using it for storing images and rich pdfs.

Figure 2: A pleasant way to make anki decks

Methodology

To accomplish this, we basically need to have the following:

anki-editor: This emacs plugin will facilitate the conversion from our orgmode files to the Anki markup
anki-connect: We need a server of sorts set up to allow us to push pull and get errors from the running Anki server, this is an Anki plugin
LaTeX process editor: It wouldn’t be much better than manually making cards in Anki if we couldn’t leverage unicode characters, so we need to modify the internal Anki build process for TeX

Anki Editor

As with all emacs related setup snippets on this site, these should be modified and adapted as needed, especially for those not using doom-emacs.

(use-package anki-editor
  :after org-noter
  :config
  ; I like making decks
  (setq anki-editor-create-decks 't))

Also, my full configuration has additional non-essential quality of life keybindings amongst other things.

Anki Connect

CTRL+Shift+A will bring up the addon settings, and Anki has to be restarted after installing the addons. Anki Connect itself does not need any further configuration, though the readme is very comprehensive.

TeX Setup

The LaTeX process editor can be set in two stages, wherein we will first ensure that we can use xelatex and that we can generate an svg.

{
    "svgCommands": [
        [
            "xelatex",
            "--no-pdf",
            "-interaction=nonstopmode",
            "tmp.tex"
        ],
        [
            "dvisvgm",
            "--no-fonts",
            "-Z",
            "2",
            "tmp.xdv",
            "-o",
            "tmp.svg"
        ]
    ]
}

The png settings can be modified in a similar manner if required, but it is better to generate svg files, which will set up in the cosmetics section. Note that we pass --no-pdf to get the xdv file which has replaced dvi files for xelatex.

Cosmetics

The final aspect of this is to be configured with the GUI. The easiest option is to clone the Basic card type and customize that. CTRL+Shift+N should bring up the card editor. The relevant styles are¹ (from the Cards option):

.card {
 font-family: Literata;
 font-size: 26px;
 text-align: center;
 color: black;
 background-color: white;
}
img {
max-height:1000px;
height: auto;
width: auto;
}
img[src*="latex"] {
  vertical-align: middle;
}

Now we need setup our TeX headers as well, and enable the Create scalable images with dvisvgm option. The header needs to have (minimally):

\documentclass[12pt]{article}
\special{papersize=3in,5in}
\usepackage{geometry}
\usepackage{unicode-math}
\usepackage{mathtools}
\pagestyle{empty}
\setlength{\parindent}{0in}
\begin{document}

While the footer is simply \end{document}. With this, we have achieved pretty formatting.

Font Locking

Inspired by this post, we will also use orgcss to obtain some orgmode font-locking. We will add the following styles:

:not(pre) > code {
  padding: 2px 5px;
  margin: auto 1px;
  border: 1px solid #ddd;
  border-radius: 3px;
  background-clip: padding-box;
  color: #333;
  font-size: $code-size;
}

.org-src-container {
  border: 1px solid #ccc;
  box-shadow: 3px 3px 3px #eee;
  font-family: $monospace;
  font-size: $code-size;
  margin: 1em auto;
  padding: 0.1em 0.5em;
  position: relative;
}

.org-src-container > pre {
  overflow: auto;
}

.org-src-container > pre:before {
  display: block;
  position: absolute;
  background-color: #b3b3b3;
  top: 0;
  right: 0;
  padding: 0 0.5em;
  border-bottom-left-radius: 8px;
  border: 0;
  color: white;
  font-size: $code-size;
}

/* from http://demo.thi.ng/org-spec/ */

.org-src-container > pre.src-sh:before {
  content: "sh";
}
.org-src-container > pre.src-bash:before {
  content: "bash";
}
.org-src-container > pre.src-emacs-lisp:before {
  content: "Emacs Lisp";
}
.org-src-container > pre.src-R:before {
  content: "R";
}
.org-src-container > pre.src-org:before {
  content: "Org";
}
.org-src-container > pre.src-cpp:before {
  content: "C++";
}
.org-src-container > pre.src-c:before {
  content: "C";
}
.org-src-container > pre.src-html:before {
  content: "HTML";
}
.org-src-container > pre.src-js:before {
  content: "Javascript";
}
.org-src-container > pre.src-javascript:before {
  content: "Javascript";
}

// More languages from http://orgmode.org/worg/org-contrib/babel/languages.html

.org-src-container > pre.src-abc:before {
  content: "ABC";
}
.org-src-container > pre.src-asymptote:before {
  content: "Asymptote";
}
.org-src-container > pre.src-awk:before {
  content: "Awk";
}
.org-src-container > pre.src-C:before {
  content: "C";
}
.org-src-container > pre.src-calc:before {
  content: "Calc";
}
.org-src-container > pre.src-clojure:before {
  content: "Clojure";
}
.org-src-container > pre.src-comint:before {
  content: "comint";
}
.org-src-container > pre.src-css:before {
  content: "CSS";
}
.org-src-container > pre.src-D:before {
  content: "D";
}
.org-src-container > pre.src-ditaa:before {
  content: "Ditaa";
}
.org-src-container > pre.src-dot:before {
  content: "Dot";
}
.org-src-container > pre.src-ebnf:before {
  content: "ebnf";
}
.org-src-container > pre.src-forth:before {
  content: "Forth";
}
.org-src-container > pre.src-F90:before {
  content: "Fortran";
}
.org-src-container > pre.src-gnuplot:before {
  content: "Gnuplot";
}
.org-src-container > pre.src-haskell:before {
  content: "Haskell";
}
.org-src-container > pre.src-io:before {
  content: "Io";
}
.org-src-container > pre.src-java:before {
  content: "Java";
}
.org-src-container > pre.src-latex:before {
  content: "LaTeX";
}
.org-src-container > pre.src-ledger:before {
  content: "Ledger";
}
.org-src-container > pre.src-ly:before {
  content: "Lilypond";
}
.org-src-container > pre.src-lisp:before {
  content: "Lisp";
}
.org-src-container > pre.src-makefile:before {
  content: "Make";
}
.org-src-container > pre.src-matlab:before {
  content: "Matlab";
}
.org-src-container > pre.src-max:before {
  content: "Maxima";
}
.org-src-container > pre.src-mscgen:before {
  content: "Mscgen";
}
.org-src-container > pre.src-Caml:before {
  content: "Objective";
}
.org-src-container > pre.src-octave:before {
  content: "Octave";
}
.org-src-container > pre.src-org:before {
  content: "Org";
}
.org-src-container > pre.src-perl:before {
  content: "Perl";
}
.org-src-container > pre.src-picolisp:before {
  content: "Picolisp";
}
.org-src-container > pre.src-plantuml:before {
  content: "PlantUML";
}
.org-src-container > pre.src-python:before {
  content: "Python";
}
.org-src-container > pre.src-ruby:before {
  content: "Ruby";
}
.org-src-container > pre.src-sass:before {
  content: "Sass";
}
.org-src-container > pre.src-scala:before {
  content: "Scala";
}
.org-src-container > pre.src-scheme:before {
  content: "Scheme";
}
.org-src-container > pre.src-screen:before {
  content: "Screen";
}
.org-src-container > pre.src-sed:before {
  content: "Sed";
}
.org-src-container > pre.src-shell:before {
  content: "shell";
}
.org-src-container > pre.src-shen:before {
  content: "Shen";
}
.org-src-container > pre.src-sql:before {
  content: "SQL";
}
.org-src-container > pre.src-sqlite:before {
  content: "SQLite";
}
.org-src-container > pre.src-stan:before {
  content: "Stan";
}
.org-src-container > pre.src-vala:before {
  content: "Vala";
}
.org-src-container > pre.src-axiom:before {
  content: "Axiom";
}
.org-src-container > pre.src-browser:before {
  content: "HTML";
}
.org-src-container > pre.src-cypher:before {
  content: "Neo4j";
}
.org-src-container > pre.src-elixir:before {
  content: "Elixir";
}
.org-src-container > pre.src-request:before {
  content: "http";
}
.org-src-container > pre.src-ipython:before {
  content: "iPython";
}
.org-src-container > pre.src-kotlin:before {
  content: "Kotlin";
}
.org-src-container > pre.src-Flavored Erlang lfe:before {
  content: "Lisp";
}
.org-src-container > pre.src-mongo:before {
  content: "MongoDB";
}
.org-src-container > pre.src-prolog:before {
  content: "Prolog";
}
.org-src-container > pre.src-rec:before {
  content: "rec";
}
.org-src-container > pre.src-ML sml:before {
  content: "Standard";
}
.org-src-container > pre.src-Translate translate:before {
  content: "Google";
}
.org-src-container > pre.src-typescript:before {
  content: "Typescript";
}
.org-src-container > pre.src-rust:before {
  content: "Rust";
}

However, in the interests of sanity, we will leverage the Syntax Highlighting Anki plugin for managing the actual style-sheets instead of manual edits to each card type.

Figure 4: A screencast from the plugin readme

At this stage, we have a card which can gracefully handle both XeLaTeX and code in an elegant manner. An example is presented in the next section.

Usage

For the sample card² shown, the markup is dead simple.

* Basis Vectors :math:quantum:linear:
:properties:
:anki_deck: CompChem
:anki_note_type: LaTeX
:ANKI_NOTE_ID: 1603755931922
🔚
** Front
For a three dimensional vector with components $aᵢ,i=1,2,3$ what are the basis vectors?
** Back
This is defined as follows:
$$
\mathbf{a}=\mathbf{e}₁a₁+\mathbf{e}₂a₂+\mathbf{e}₃a₃=∑ᵢ\mathbf{e}ᵢaᵢ
$$

Essentially:

Enable and load anki-editor
- Add local variable section to ensure we load anki-editor. This is essentially via eval: (anki-editor-mode) in the Local variables block
Fire up Anki
Export at will, and continue adding more cards or non-card details to the orgmode file

The Anki editor examples file is excellent and the issue tracker also has a ton of information.

Code

* Test Code :code:python:
:properties:
:anki_deck: CodeWiki
:anki_note_type: myTex
:ANKI_NOTE_ID: 1603891864091
🔚
** Front
What is the definition of an inner product? What are some examples of a code block in Python and R?
** Back
This is essentially a *norm* with more structure. The first two properties, positive definiteness and symmetry (conjugate) defines a *norm*.
$$
\mathbf{a}=\mathbf{e}₁a₁+\mathbf{e}₂a₂+\mathbf{e}₃a₃=∑ᵢ\mathbf{e}ᵢaᵢ
$$

#+begin_src python
def test():
    x = [1,2,3]
    for i in x:
        print(i)
#+end_src

#+begin_src R
library("dplyr")
x = 1
#+end_src

Conclusions

Some final comments:

Screenshots and other images linked are automatically synced
The TeX is best rendered on the PC first, so run through these at-least once

A missing link in this setup is the ability to use a touch screen and stylus to write proofs or skip the TeX setup altogether, but that would require another post altogether. Additionally, all the standard bells and whistles of having an orgmode document can be applied, including, crucially, the ability to have long-form notes as well, a coherent approach to this can also be covered later.

The alignment trick is from this post ↩︎
It is a gag card, no judgement here ↩︎
org-drill doesn’t support any kind of mobile synchronization ↩︎