mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-28 02:46:35 +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
32 lines
789 B
Python
32 lines
789 B
Python
# coding: utf8
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
NUMBERS_IRREG = {
|
|
'achten': ('acht',),
|
|
'biljoenen': ('biljoen',),
|
|
'drieën': ('drie',),
|
|
'duizenden': ('duizend',),
|
|
'eentjes': ('één',),
|
|
'elven': ('elf',),
|
|
'miljoenen': ('miljoen',),
|
|
'negenen': ('negen',),
|
|
'negentiger': ('negentig',),
|
|
'tienduizenden': ('tienduizend',),
|
|
'tienen': ('tien',),
|
|
'tientjes': ('tien',),
|
|
'twaalven': ('twaalf',),
|
|
'tweeën': ('twee',),
|
|
'twintiger': ('twintig',),
|
|
'twintigsten': ('twintig',),
|
|
'vieren': ('vier',),
|
|
'vijftiger': ('vijftig',),
|
|
'vijven': ('vijf',),
|
|
'zessen': ('zes',),
|
|
'zestiger': ('zestig',),
|
|
'zevenen': ('zeven',),
|
|
'zeventiger': ('zeventig',),
|
|
'zovele': ('zoveel',),
|
|
'zovelen': ('zoveel',)
|
|
}
|