mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-14 21:57:15 +03:00
14 lines
456 B
Python
14 lines
456 B
Python
from ..punctuation import TOKENIZER_INFIXES as BASE_TOKENIZER_INFIXES
|
|
from ..punctuation import TOKENIZER_PREFIXES as BASE_TOKENIZER_PREFIXES
|
|
from ..punctuation import TOKENIZER_SUFFIXES as BASE_TOKENIZER_SUFFIXES
|
|
|
|
_prefixes = [r"\w{1,3}\$"] + BASE_TOKENIZER_PREFIXES
|
|
|
|
_suffixes = BASE_TOKENIZER_SUFFIXES
|
|
|
|
_infixes = [r"(\w+-\w+(-\w+)*)"] + BASE_TOKENIZER_INFIXES
|
|
|
|
TOKENIZER_PREFIXES = _prefixes
|
|
TOKENIZER_SUFFIXES = _suffixes
|
|
TOKENIZER_INFIXES = _infixes
|