...

ebook2audiobook.egg-info/requires.txt

@@ -8,8 +8,7 @@ ebooklib
 fastapi
 hf_xet
 beautifulsoup4
-sudachipy
-sudachidict-core
+nagisa
 pymupdf
 pymupdf-layout
 pytesseract

lib/core.py

@@ -1039,10 +1039,19 @@ def get_sentences(text:str, id:str)->list|None:
                         jieba.dt.cache_file = os.path.join(models_dir, 'jieba.cache')
                         result.extend([t for t in jieba.cut(segment) if t.strip()])
                     elif lang == 'jpn':
+                        """
                         from sudachipy import dictionary, tokenizer
                         sudachi = dictionary.Dictionary().create()
                         mode = tokenizer.Tokenizer.SplitMode.C
                         result.extend([m.surface() for m in sudachi.tokenize(segment, mode) if m.surface().strip()])
+                        """
+                        import nagisa
+                        tokens = nagisa.tagging(segment).words
+                        result.extend([
+                            token
+                            for token in tokens
+                            if token.strip()
+                        ])
                     elif lang == 'kor':
                         from soynlp.tokenizer import LTokenizer
                         ltokenizer = LTokenizer()

pyproject.toml

@@ -27,8 +27,7 @@ dependencies = [
 	"fastapi",
 	"hf_xet",
 	"beautifulsoup4",
-	"sudachipy",
-	"sudachidict-core",
+	"nagisa",
 	"pymupdf",
 	"pymupdf-layout",
 	"pytesseract",

requirements.txt

@@ -8,8 +8,7 @@ ebooklib
 fastapi
 hf_xet
 beautifulsoup4
-sudachipy
-sudachidict-core
+nagisa
 pymupdf
 pymupdf-layout
 pytesseract