Dict

dict

This module defines the dict_router of the Mirumoji API

Attributes:

Name	Type	Description
LOGGER	Logger	Module's logging object
dict_router	APIRouter	The FastAPI router object

AUDIO_MEDIA_TYPES = {'mp3': 'audio/mpeg', 'ogg': 'audio/ogg'} module-attribute

Maps a pronunciation clip's container format to its HTTP media type

analyze(sentence, mode=BundleMode.grammar) async

Tokenizes a sentence and enriches every stitched word with dictionary data

Slower than /dict/tokenize (one dictionary lookup per word). Intended for on-demand analysis rather than bulk rendering

Parameters:

Name	Type	Description	Default
sentence	str	The Japanese sentence to analyze	required
mode	BundleMode	How aggressively to group tokens into words	grammar

Returns:

Type	Description
list[EnrichedJapaneseWord]	A list of EnrichedJapaneseWord models, one per stitched word

Raises:

Type	Description
FugashiError	If tokenization fails
KotobaseError	If a dictionary lookup fails

by_radicals(radicals, mode='all') async

Finds Kanji that contain one, or every one of the given radicals

Parameters:

Name	Type	Description	Default
radicals	str	The radical glyphs, concatenated or comma-separated (e.g. 言口 or 言,口)	required
mode	Literal	When all, every one of radicals must be present in the kanji. When any, kanjis that contain at least one of radicals are matched	'all'

Returns:

Type	Description
list[KanjiInfo]	The matching KanjiInfo models

Raises:

Type	Description
HTTPException	400 when no radical glyphs are provided
KotobaseError	If the search fails

kanji(literal) async

Fetches the full profile of a single Kanji

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal	required

Returns:

Type	Description
KanjiInfo	The KanjiInfo profile

Raises:

Type	Description
HTTPException	404 when the Kanji is unknown
KotobaseError	If the lookup fails

kanji_audio(literal) async

Lists the pronunciation clips available for a Kanji (Kanji Alive)

The clips list is empty when the audio pack is not installed or the Kanji has no recorded clips

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal	required

Returns:

Type	Description
KanjiAudio	The KanjiAudio clip listing

Raises:

Type	Description
KotobaseError	If the lookup fails

kanji_audio_clip(literal, clip) async

Streams the raw bytes of one pronunciation clip

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal the clip belongs to	required
clip	str	The clip id from the audio listing endpoint	required

Returns:

Type	Description
Response	The audio bytes with the clip's media type

Raises:

Type	Description
HTTPException	404 when no clip matches
KotobaseError	If the lookup fails

kanji_sentences(literal, limit=10) async

Lists example sentences that contain a Kanji

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal	required
limit	int	Maximum number of sentences to return	10

Returns:

Type	Description
list[Example]	The matching Example sentences with translations

Raises:

Type	Description
KotobaseError	If the lookup fails

kanji_strokes(literal) async

Fetches a Kanji's stroke-order diagram as an SVG document (KanjiVG)

The SVG is self-contained, so the frontend can inline it and animate the stroke paths

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal	required

Returns:

Type	Description
Response	The SVG document as image/svg+xml

Raises:

Type	Description
HTTPException	404 when no stroke-order data exists for the Kanji
KotobaseError	If the lookup fails

kanji_words(literal, limit=50) async

Lists dictionary entries whose written form uses a Kanji

Parameters:

Name	Type	Description	Default
literal	str	The Kanji literal	required
limit	int	Maximum number of entries to return	50

Returns:

Type	Description
list[JMEntry]	The matching JMEntry models

Raises:

Type	Description
KotobaseError	If the lookup fails

query(word, wildcard=False, reading=None, pos=None) async

Looks up dictionary data for a single word or a wildcard pattern

Token Hints

• reading and pos are optional hints from tokenization

• When provided, entries whose kana form matches the reading and whose part of speech matches the token rank first

• Unknown written forms fall back to a lookup by reading

Parameters:

Name	Type	Description	Default
word	str	Word or wildcard pattern to look up	required
wildcard	bool	When True, treat word as a wildcard pattern matching multiple words	False
reading	str | None	The token's reading (any kana), when known	None
pos	str | None	The token's UniDic top-level part of speech, when known	None

Returns:

Type	Description
KotobaseData	The KotobaseData for the query

Raises:

Type	Description
KotobaseError	If the lookup fails

radicals() async

Lists every search radical with its stroke count (RADKFILE)

Returns:

Type	Description
list[RadicalInfo]	All search radicals

Raises:

Type	Description
KotobaseError	If the lookup fails

resolve(ref) async

Resolves a sense cross-reference or antonym code into its entries

Parameters:

Name	Type	Description	Default
ref	str	The reference code from a sense's xrefs or antonyms	required

Returns:

Type	Description
list[JMEntry]	The referenced JMEntry models

Raises:

Type	Description
KotobaseError	If the resolution fails

search(q, limit=50) async

Searches dictionary entries by English meaning (reverse lookup)

Japanese word and wildcard searches go through /dict/query instead

Parameters:

Name	Type	Description	Default
q	str	The English search text	required
limit	int	Maximum number of entries to return	50

Returns:

Type	Description
list[JMEntry]	The matching JMEntry models

Raises:

Type	Description
KotobaseError	If the search fails

tokenize(sentence, mode=BundleMode.grammar) async

Tokenizes a sentence into useful, stitched words (no dictionary lookups)

This is the fast path for rendering clickable text. Call /dict/analyze or /dict/query to fetch dictionary data for a word

Parameters:

Name	Type	Description	Default
sentence	str	The Japanese sentence to tokenize	required
mode	BundleMode	How aggressively to group tokens into words	grammar

Returns:

Type	Description
list[JapaneseWord]	A list of JapaneseWord models, one per stitched word

Raises:

Type	Description
FugashiError	If tokenization fails

tokenize_batch(req) async

Tokenizes many sentences in one request (stitched words, no dict data)

Lets a client tokenize a whole subtitle file up front in a single call, so playback never tokenizes per-cue

Parameters:

Name	Type	Description	Default
req	TokenizeBatchRequest	The sentences to tokenize	required

Returns:

Type	Description
list[list[JapaneseWord]]	One JapaneseWord list per input sentence, in the same order

Raises:

Type	Description
FugashiError	If tokenization fails