Jpdict

jpdict

BundleMode

Bases: str, Enum

How aggressively the stitch function groups UniDic short-unit tokens into words

The modes trade granularity for a learner's needs, from whole dictionary words down to raw morphemes

words

• The coarsest grouping

• Merges compound nouns and the whole inflected tail of a predicate, including the connecting て, bound auxiliary verbs, and politeness, into single dictionary words (e.g. 図書館, 食べてみたかった)

• Best for looking words up

grammar (default)

• The learning view

• Keeps compound nouns and a predicate's inflectional auxiliaries together, but breaks off the pieces a learner parses separately, like the connecting て, bound auxiliary verbs (みる/いる/出す), and the politeness stems (ます/です) (e.g. 食べ | て | みたかった, 読み | ました)

morphemes

• The finest grouping

• No stitching at all, one word per UniDic short unit (e.g. 図書 | 館, 読み | まし | た)

EnrichedJapaneseWord

Bases: BaseModel

A JapaneseWord paired with its dictionary data

Returned by the enrichment endpoint (one kotobase lookup per stitched word), as opposed to the fast tokenize endpoint which returns bare JapaneseWord models

Parameters:

Name	Type	Description	Default
word	JapaneseWord	The stitched word	required
kotobase_data	KotobaseData	The dictionary data for the word's lemma	required

JMEntry

Bases: BaseModel

Represents a single word entry in the Japanase-Multilingual Dictionary

rank

Kotobase calculates the rank attribute based on JMDict's <pri> tags. The following are the possible values and their meanings

• 0 → High-frequency words found across standard textbooks (ichi1) and newspapers (news1)

• 1-48 → The specific 500-word corpus interval the word belongs to (e.g., tier 5 means the word is within the top 2001-2500 most common words)

• 99 → Low-priority or niche words containing auxiliary tags

Parameters:

Name	Type	Description	Default
rank	int	A categorized numerical value mapping the word's real-world popularity	required
kana	list[str] | None	List of kana readings	required
kanji	list[str] | None	List of kanji readings	required
senses	list[WordSense] | None	List of JMWordSense models	required

JMNEntry

Bases: BaseModel

Represents a single name entry in the Japanese Multi-Lingual Dictionary

Parameters:

Name	Type	Description	Default
kana	list[str] | None	List of kana readings	required
kanji	list[str] | None	List of kanji readings	required
translation_type	str | None	Type of name	required
gloss	list[str] | None	list of translation strings	required

JMWordSense

Bases: BaseModel

Represents a single sense (distinct meanings, translations, or nuances of a Japanese word) for a word within the Japanese-Multilingual Dictionary

order

• Represents the sequential arrangement of word meanings based on lexicographical hierarchy

• Senses progress logically from primary, literal definitions to secondary, figurative, or technical nuances

• This order is editorially curated and does not reflect mathematical usage frequency

Parameters:

Name	Type	Description	Default
order	int	editorial priority order	required
pos	str	Grammatical classifications like verb (v5u), noun (n), or adjective (adj-no) that apply to this specific meaning	required
gloss	str	The English equivalent of the word	required

JapaneseWord

Bases: BaseModel

Represents a single useful word stitched from one or more UniDic short-unit tokens

Stitching

• UniDic segments at the short-unit level, which is often too granular to be useful (e.g. 図書館 -> 図書 + 館, or a verb split from its auxiliaries)

• A JapaneseWord re-bundles those short units into the word a learner actually wants to click

• The original short-unit Token models are kept in tokens so no morphological detail is lost

the word built from 読み + まし + た

• surface = "読みました" (the pieces joined as written)
• reading = "ヨミマシタ" (their katakana readings joined)
• lemma = "読む" (the dictionary form, for look-ups)
• pos = "動詞" (verb -- the head piece's part of speech)
• tokens = [読み, まし, た] (the three original short units)

Parameters:

Name	Type	Description	Default
surface	str	The bundle's combined surface form	required
reading	str	The combined katakana reading of the component tokens	required
lemma	str	Dictionary-lookup form (the head token's UniDic's orthBase for inflected words, or the combined surface for noun compounds)	required
pos	str	The head token's top-level part of speech	required
tokens	list[Token]	The component short-unit tokens, in order	required

KanjiInfo

Bases: BaseModel

Represents a single Kanji entry in KANJIDIC2

Parameters:

Name	Type	Description	Default
literal	str	Kanji literal	required
grade	int | None	Optional Japanese grade in which Kanji is learned	required
stroke_count	int | None	Number of strokes in handwriting	required
meanings	list[str] | None	List of known meanings	required
onyomi	list[str] | None	List of on readings	required
kunyomi	list[str] | None	List of kun readings	required
jlpt_kanjidic	int | None	Optional JLPT level present in KANJIDIC2	required
jlpt_tanos	int | None	Optional JLPT level in Tanos list	required

KotobaseData

Bases: BaseModel

Represents all information extracted from kotobase for a single query (either a single Japanese word, or a wildcard pattern matching multiple words)

meanings

• Exposes the gloss attributes (English equivalent of the word) of all JMWordSense models contained inside the first Japanase-Multilingual Dictionary entry for the query

• If the query has only JMNEntry entries, the first entry's gloss attribute is used

Parameters:

Name	Type	Description	Default
query	str	query literal (either a single Japanese word or a wildcard pattern)	required
jmentries	list[JMEntry]	All Japanese-Multilingual Dictionary entries for the query	required
jmnentries	list[JMNEntry]	All Japanese-Multilingual Dictionary name entries for the query	required
kanji	list[KanjiInfo]	KANJIDIC2 entries for all Kanji present in the query	required
meanings	list[str]	All English equivalents contained in the first JMEntry, or JMNEntry	required
jlpt	str	JLPT vocabulary level for the word extracted from the Tanos list. Defaults to Unknown when it's a wildcard query or the word is not in the list	required
examples	list[str]	List of example sentences containing the single word or any words matched by the wildcard query	required

Token

Bases: BaseModel

Represents morphological data extracted for a single Japanese token

Maps all core token features and deep UniDic morphological data produced by Fugashi. Converts internal dictionary symbols (like asterisks) into clean pythonic types.

Attributes:

Name	Type	Description
surface	str	The raw string exactly as it appears in the text.
lemma	str	The dictionary base form (語彙素) of the word.
reading	str	The standard reading of the token in Katakana.
pos	str	The broad, top-level part of speech (品詞).
pos2	str	Sub-category level 2 part of speech.
pos3	str	Sub-category level 3 part of speech.
pos4	str	Sub-category level 4 part of speech.
c_type	str	Conjugation type (活用型) if applicable.
c_form	str	Conjugation form (活用形) if applicable.
l_form	str	Lemma reading in Katakana.
orth	str	Orthographic surface representation.
pron	str	Actual pronunciation including long vowels.
orth_base	str	Base form using current orthography.
pron_base	str	Pronunciation of the base form.
goshu	str	Word origin type (語種) e.g., Native, Sino-Japanese.
i_type	str	Word-initial transformation type.
i_form	str	Word-initial transformation form.
f_type	str	Word-final transformation type.
f_form	str	Word-final transformation form.

clear_asterisks(data) classmethod

Cleans incoming dictionary fields by converting UniDic's "*" sentinel and any missing (None) feature to an empty string, since unknown / out-of-vocabulary tokens leave some features unset

Parameters:

Name	Type	Description	Default
data	dict	Raw dictionary data containing morphological fields	required

Returns:

Name	Type	Description
dict	dict[str, Any]	The modified dictionary with "*"/None values replaced by ""