genanki VS anki-connect

I'm working with ChatGPT to create python scripts that generate Anki decks with Genanki.

Use genanki to create a note for each street, and then two cards (name, direction -> gif, and gif -> name, direction) for each note.

Anki has many types of cards, the main ones being the basic (memory trigger + answer) and cloze (one or more memory triggers embedded in an outline). This example uses the basic and reversed card, because both the 'question' and 'answer' can be a memory trigger for each other; if you want a simple, basic card you should consult the example at Genanki's README. I hope this script serves as at least a starting point for your own applications of programmable Anki, it is probably my main study method nowadays, and it got a whole more powerful after ChatGPT. Have fun with your coding.

If you want to do something really custom you may want to look at something like https://github.com/kerrickstaley/genanki (I haven't tried it).

Many people consider the prep time to be valuable itself as it usually involves you gaining understanding. That said, I don't really buy that argument and the vast majority of my cards these days are auto-generated using https://github.com/kerrickstaley/genanki.

It makes use of these frequency lists (have to be manually downloaded). It will make a request to the tatoeba api to grab a sentence (or multiple) including this word and if you would like it will also request an audio file from google translate (while the code is running a temporary recordings folder is created). It then packages it up into an anki deck using the genanki python module.

In the end I feed the data together with audio to another script that generate the Anki deck using genanki.

Yes, it's actually pretty easy to automate the process with genanki.

Since it looks like the export to Anki is still in progress, I can recommend using https://github.com/kerrickstaley/genanki -- it's been very stable over the 2 years I've been using it. I think it's a much better solution than csv or AnkiConnect, since it requires no actual work from the user except for opening an apkg.

If you're comfortable with Python, https://github.com/kerrickstaley/genanki is amazing. You can pull your content from any source, use templates for the question and answer format, and even add media to each card (image, audio).

Creating a Node.js Scraper: Initially, I created a web scraper for my Anki collection using Anki Connect and Node.js. I chose Node.js because I believed that using Rust would require defining every field for the response object(later found out I was wrong). Encountered issues: My program kept crashing due to not specifying the query to be of the same type(as I am writing this I realize I could have just did if(note.field == undefined) continue;) or the TypeScript ?. Resorting to single requests: To resolve the issue, I decided to send one request per card, resulting in processing 50,000 cards, which took 15-30 minutes.

Has anyone used the insertReviews() function before to simulate answering cards programmatically? I’m not able to find much information online regarding it (perhaps because it is relatively new). I am not confident enough in my knowledge of Anki’s review algorithm to attempt replicating its behavior on all 9 of the required inputs for each review (see the 'Manual Analysis' subsection here); wanted to reach out here to see if anyone had experience with the function they would like to share.

I do however create all my decks in Emacs' Org mode using louietan/anki-editor and export to Anki via the plugin FooSoft/anki-connect. This way I never worry about my decks getting corrupted. I actually just have one large deck but anki-editor allows me to separate my deck into separate org files which is convenient.

Another tool I've used if you have the Anki app is the Anki-Connect plugin https://github.com/FooSoft/anki-connect

It supports curl, python, javascript, etc to add cards and supports an incredible amount of actions to interact with Anki.

Example:

Just to add a footnote to the above: I wasn't aware of https://foosoft.net/projects/anki-connect/ so what I'm thinking about is very doable right now, and probably being done.

I've made a small script to make cards with audio. Following the manual at https://foosoft.net/projects/anki-connect/, I have an audio field of:

I dont think there's a public/documented Ankiweb api. You could use AnkiConnect to interface with your locally running instance of Anki: https://foosoft.net/projects/anki-connect/

Read the AnkiConnect documentation. You can interact with AnkiConnect in any language environment you wish. I've mostly - but not exclusively - used Python. It would probably be a very forgiving way to begin.

I write my cards in plaintext (Emacs Org mode via anki-editor) and push them to Anki via anki-connect. The advantages for me are:

create Anki cards with Emac's anki-editor and push them to Anki with anki-connect. There is also a markdown equivalent of anki-editor if you prefer that format plus packages to perform your SRS within Emacs itself such as org-drill, org-fc and pamparam.