mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-11 12:18:04 +03:00
951825532c
* Improved Dutch language resources and Dutch lemmatization * Fix conftest * Update punctuation.py * Auto-format * Format and fix tests * Remove unused test file * Re-add deleted test * removed redundant infix regex pattern for ','; note: brackets + simple hyphen remains * Cleaner lemmatization files
41 lines
1.0 KiB
Python
41 lines
1.0 KiB
Python
# coding: utf8
|
|
from __future__ import unicode_literals
|
|
|
|
from ._verbs_irreg import VERBS_IRREG
|
|
from ._nouns_irreg import NOUNS_IRREG
|
|
from ._adjectives_irreg import ADJECTIVES_IRREG
|
|
from ._adverbs_irreg import ADVERBS_IRREG
|
|
|
|
from ._adpositions_irreg import ADPOSITIONS_IRREG
|
|
from ._determiners_irreg import DETERMINERS_IRREG
|
|
from ._pronouns_irreg import PRONOUNS_IRREG
|
|
|
|
from ._verbs import VERBS
|
|
from ._nouns import NOUNS
|
|
from ._adjectives import ADJECTIVES
|
|
|
|
from ._adpositions import ADPOSITIONS
|
|
from ._determiners import DETERMINERS
|
|
|
|
from .lookup import LOOKUP
|
|
|
|
from ._lemma_rules import RULES
|
|
|
|
from .lemmatizer import DutchLemmatizer
|
|
|
|
|
|
LEMMA_INDEX = {"adj": ADJECTIVES,
|
|
"noun": NOUNS,
|
|
"verb": VERBS,
|
|
"adp": ADPOSITIONS,
|
|
"det": DETERMINERS}
|
|
|
|
LEMMA_EXC = {"adj": ADJECTIVES_IRREG,
|
|
"adv": ADVERBS_IRREG,
|
|
"adp": ADPOSITIONS_IRREG,
|
|
"noun": NOUNS_IRREG,
|
|
"verb": VERBS_IRREG,
|
|
"det": DETERMINERS_IRREG,
|
|
"pron": PRONOUNS_IRREG}
|
|
|