fasiha / ebieki

All the vocab Wanikani except using open-source JMdict definitions

Ebi-eki 🦐🚉

Wanikani. It's superb. Go pay them for it. They gave us a discount when my whole family signed up for it—small business FTW.

But I just wanted to review the vocabulary (the purple words) with the definitions from JMdict, the open-source Japanese dictionary everyone uses.

So this repo contains a big table of Wanikani vocabulary† using JMdict definitions. For each line, the first word is the kanji, the next word or words are acceptable readings (per JMdict), followed by the JMdict gloss in parentheses.

† I left out twenty-five vocabulary (out of 6369) because they didn't have entries for JMdict. For twenty-six other vocabulary, JMdict didn't have an entry but I made one.

If you just want to see that big table, click here.

Dev

If you want to run this code repository to generate the results yourself—

1. Get a Wanikani v2 API key.
2. Install Git and Node.js.
3. Run in your terminal:

git checkout https://github.com/fasiha/ebieki
cd ebieki
npm i

5. Download and uncompress jmdict-eng-3.0.1.json from the JMdict-Simplified project's downloads page. (If you download a newer version, change the filename in wanikani-parse.ts.) Put this file in the ebieki directory you just created (via git clone).
6. Create a .env file in the ebieki directory and put your Wanikani v2 API token in the following format:

WANIKANI_TOKEN=YOUR_TOKEN_GOES_HERE

7. Run

npm run download
npm run table

The npm run download invokation will hit the Wanikani API server to download all 6000+ vocabulary. This takes a few seconds (since the server gives us a thousand at a time, and we wait a bit before going back for more to avoid overloading them).

Then the npm run table runs some custom code that looks up all the vocab in JMdict. It skips twenty-five vocab that aren't in JMdict and that I felt were ok to skip (e.g., 二万, 二台). For another twenty-six that aren't in JMdict, I created translations; most were obvious (新宿→Shinjuku, 福島→Fukushima) but a couple I just used Google Translate. It takes less than ten seconds to run on my old laptop, and should print out a summary like this:

Statistics:
- 6369 vocabulary from Wanikani
- 25 skipped
- 26 custom definitions used
- 0 unable to find JMdict defintion ✅
- 0 found multiple JMdict definitions  ✅

It outputs table.txt, which contains no Wanikani-copyrighted material, as well as table-with-wanikani.txt which contains the translations from Wanikani.

Notes

Here's an example couple of lines from table.txt:

一 いち (§1.44. one; best; first, foremost; beginning, start; a (single), one (of many); ace (playing card); bottom string (on a shamisen, etc.). #1160790, common!)
一つ ひとつ (§1.45. one; for one thing; only; (not) even; just (e.g. "just try it"); some kind of, one type of. #1160820, common!)
七 なな しち (§1.46. seven; hepta-. #1319210, common!)

As you see, the format is

<kanji> <kana, possibly multiple> (§<Wanikani lesson number (skips because we're only looking at vocabulary)> <JMdict definition> #<JMdict entry ID> <indicator if this word is common>)

The "JMdict entry ID" may need some explication. All JMdict entries have an ID associated with them, even though most dictionaries don't expose it. You can look up an ID using JMdict advanced search by typing it into the "Search by Id or Seq number" section. Doing so can lead you to the JMdict entry for #1319210, 七なな.

Go pay Wanikani money