def build_tokenizer(
token_type: str,
bpemodel: Union[Path, str, Iterable[str]] = None,
non_linguistic_symbols: Union[Path, str, Iterable[str]] = None,
remove_non_linguistic_symbols: bool = False,
space_symbol: str = "<space>",
delimiter: str = None,
g2p_type: str = None,
) -> AbsTokenizer:
"""A helper function to instantiate Tokenizer"""
assert check_argument_types()
if token_type == "bpe":
if bpemodel is None:
raise ValueError('bpemodel is required if token_type = "bpe"')
if remove_non_linguistic_symbols:
raise RuntimeError(
"remove_non_linguistic_symbols is not implemented for token_type=bpe"
)
return SentencepiecesTokenizer(bpemodel)
elif token_type == "word":
if remove_non_linguistic_symbols and non_linguistic_symbols is not None:
return WordTokenizer(
delimiter=delimiter,
non_linguistic_symbols=non_linguistic_symbols,
remove_non_linguistic_symbols=True,
)
else:
return WordTokenizer(delimiter=delimiter)
elif token_type == "char":
return CharTokenizer(
non_linguistic_symbols=non_linguistic_symbols,
space_symbol=space_symbol,
remove_non_linguistic_symbols=remove_non_linguistic_symbols,
)
elif token_type == "phn":
if g2p_type is None:
raise ValueError("g2p_type is required if token_type=phn")
return PhonemeTokenizer(
g2p_type=g2p_type,
non_linguistic_symbols=non_linguistic_symbols,
space_symbol=space_symbol,
remove_non_linguistic_symbols=remove_non_linguistic_symbols,
)
else:
raise ValueError(f"token_mode must be one of bpe, word, char or phn: "
f"{token_type}")
def word_tokenizer(request):
return WordTokenizer(delimiter=request.param)
def test_tokens2text(word_tokenizer: WordTokenizer):
assert word_tokenizer.tokens2text(
"Hello World!!".split()) == "Hello World!!"
def test_text2tokens(word_tokenizer: WordTokenizer):
assert word_tokenizer.text2tokens("Hello World!! Ummm") == [
"Hello",
"World!!",
"Ummm",
]
def test_Text2Words_tokens2text(word_converter: WordTokenizer):
assert word_converter.tokens2text("Hello World!!".split()) == "Hello World!!"