spaCy/spacy/lang/da/punctuation.py

25 lines
882 B
Python
Raw Normal View History

# coding: utf8
from __future__ import unicode_literals
from ..char_classes import LIST_ELLIPSES, LIST_ICONS
from ..char_classes import QUOTES, ALPHA, ALPHA_LOWER, ALPHA_UPPER
from ..punctuation import TOKENIZER_SUFFIXES
_quotes = QUOTES.replace("'", '')
_infixes = (LIST_ELLIPSES + LIST_ICONS +
[r'(?<=[{}])\.(?=[{}])'.format(ALPHA_LOWER, ALPHA_UPPER),
r'(?<=[{a}])[,!?](?=[{a}])'.format(a=ALPHA),
r'(?<=[{a}"])[:<>=](?=[{a}])'.format(a=ALPHA),
r'(?<=[{a}]),(?=[{a}])'.format(a=ALPHA),
r'(?<=[{a}])([{q}\)\]\(\[])(?=[\{a}])'.format(a=ALPHA, q=_quotes),
r'(?<=[{a}])--(?=[{a}])'.format(a=ALPHA)])
_suffixes = [suffix for suffix in TOKENIZER_SUFFIXES if suffix not in ["'s", "'S", "s", "S", r"\'"]]
_suffixes += [r"(?<=[^sSxXzZ])\'"]
TOKENIZER_INFIXES = _infixes
TOKENIZER_SUFFIXES = _suffixes