mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-11 12:18:04 +03:00
24 lines
855 B
Python
24 lines
855 B
Python
# coding: utf8
|
|
from __future__ import unicode_literals
|
|
|
|
from ..char_classes import ALPHA, ALPHA_LOWER, ALPHA_UPPER, CONCAT_QUOTES, HYPHENS
|
|
from ..char_classes import LIST_ELLIPSES, LIST_ICONS
|
|
|
|
_hyphens_no_dash = HYPHENS.replace("-", "").strip("|").replace("||", "")
|
|
_infixes = (
|
|
LIST_ELLIPSES
|
|
+ LIST_ICONS
|
|
+ [
|
|
r"(?<=[{al}])\.(?=[{au}])".format(al=ALPHA_LOWER, au=ALPHA_UPPER),
|
|
r"(?<=[{a}])[,!?/()]+(?=[{a}])".format(a=ALPHA),
|
|
r"(?<=[{a}{q}])[:<>=](?=[{a}])".format(a=ALPHA, q=CONCAT_QUOTES),
|
|
r"(?<=[{a}])--(?=[{a}])".format(a=ALPHA),
|
|
r"(?<=[{a}]),(?=[{a}])".format(a=ALPHA),
|
|
r"(?<=[{a}])([{q}\)\]\(\[])(?=[\-{a}])".format(a=ALPHA, q=CONCAT_QUOTES),
|
|
r"(?<=[{a}])(?:{h})(?=[{a}])".format(a=ALPHA, h=_hyphens_no_dash),
|
|
r"(?<=[0-9])-(?=[0-9])",
|
|
]
|
|
)
|
|
|
|
TOKENIZER_INFIXES = _infixes
|