JapaNote

Anki add-on to quickly make notes for Japanese words.

Features

• Type a word in kanji, kana or romaji, and press Enter.
• Type a conjugated word to get its dictionary form (e.g. 食べました gives 食べる).
• Check the note in the editor of Anki, and change any field or tag, before adding it.
• Pick the word you want when several match.
• Look up proper nouns, and technical words on the Japanese Wikipedia.
• If nothing matches, you still get automatic furigana placement.
• Paste a list of words, or open a text or CSV file, and add them all at once.
• You can have each card of a new note go into a specific deck.
• Tag new notes with their JLPT level (e.g. JLPT::N3), and common for common words.

Available fields, for example for 食べる:

• kanji: 食べる
• kana: たべる
• furigana: 食[た]べる
• definition: 1. to eat 2. to live on (e.g. a salary); to live off; to subsist on
• JMdict ID: EntL1358280
• pitch accent: たべ＼る (a drop after べ)
• audio: taberu.wav (by Fukujinzuke, from Lingua Libre, CC BY 4.0)
• example sentence: もっと果物を食べるべきです。 You should eat more fruit.
• kanji details: 食 eat, food (ショク, ジキ; く.う, く.らう, た.べる, は.む)

Screenshots

Dictionaries

• The add-on does not ship any dictionary.
• On first start, it downloads the main dictionary (about 10 MB).
• It downloads the names dictionary (about 5 MB) the first time it needs it.
• It updates both in the background, at most once a day, with small patches.
• It stores them in its user_files folder.

A workflow builds them every week with tools/build_dictionaries.py. It publishes them in the dictionaries release (see japanote/package.py for the format).

Development

# build the dictionaries locally (about a minute)
python3 tools/build_dictionaries.py download sources
python3 tools/build_dictionaries.py build sources dist http://127.0.0.1:8000
python3 tools/build_dictionaries.py install dist db

# run the tests
JAPANOTE_DB_DIR=db uv run pytest

# use local dictionaries in Anki
(cd dist && python3 -m http.server 8000) &
JAPANOTE_MANIFEST_URL=http://127.0.0.1:8000/manifest-v1.json anki

# package the add-on
make -C japanote package

• The add-on must stay compatible with Python 3.9, which Anki 24.06 to 25.02 ship.
• The package can be installed with "Install from file…" in Anki. It replaces the version from AnkiWeb.
• AnkiWeb ignores manifest.json: set the oldest version of Anki (24.06) when uploading.
• tools/e2e tests the add-on in a real Anki, and takes the screenshots.

Credits

• JMdict, JMnedict and KANJIDIC2 are the property of the Electronic Dictionary Research and Development Group, and are used in conformance with the Group's licence (CC BY-SA 4.0)
• furigana from JmdictFurigana (CC BY-SA 4.0)
• JLPT levels from Jonathan Waller's lists (CC BY), matched to JMdict by yomitan-jlpt-vocab (CC BY-SA 4.0)
• the pitch accent notation was provided by Uros O. through his free database Kanjium (CC BY-SA 4.0)
• example sentences from Tatoeba (CC BY 2.0 FR), chosen with the index of the words they contain (originally from the Tanaka Corpus)
• readings and translations of technical words from Wikipedia (CC BY-SA 4.0)
• recordings from Lingua Libre, hosted on Wikimedia Commons (mostly CC BY-SA 4.0; the name of the speaker is kept in the name of the file)
• sentences are analyzed with the IPA dictionary (© Nara Institute of Science and Technology), as compiled by Janome (Apache 2.0), whose algorithm is reproduced in japanote/analyzer.py (see NOTICE-IPADIC.txt)
• romaji conversion by python-romkan, a port of Ruby/Romkan by Satoru Takabayashi (Ruby's licence)