mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-10 19:57:17 +03:00
49 lines
1.5 KiB
Python
49 lines
1.5 KiB
Python
from ..tokenizer_exceptions import BASE_EXCEPTIONS
|
|
from ...symbols import ORTH, NORM
|
|
from ...util import update_exc
|
|
|
|
|
|
_exc = {}
|
|
|
|
|
|
# Time
|
|
for exc_data in [
|
|
{NORM: "قبل الميلاد", ORTH: "ق.م"},
|
|
{NORM: "بعد الميلاد", ORTH: "ب. م"},
|
|
{NORM: "ميلادي", ORTH: ".م"},
|
|
{NORM: "هجري", ORTH: ".هـ"},
|
|
{NORM: "توفي", ORTH: ".ت"},
|
|
]:
|
|
_exc[exc_data[ORTH]] = [exc_data]
|
|
|
|
# Scientific abv.
|
|
for exc_data in [
|
|
{NORM: "صلى الله عليه وسلم", ORTH: "صلعم"},
|
|
{NORM: "الشارح", ORTH: "الشـ"},
|
|
{NORM: "الظاهر", ORTH: "الظـ"},
|
|
{NORM: "أيضًا", ORTH: "أيضـ"},
|
|
{NORM: "إلى آخره", ORTH: "إلخ"},
|
|
{NORM: "انتهى", ORTH: "اهـ"},
|
|
{NORM: "حدّثنا", ORTH: "ثنا"},
|
|
{NORM: "حدثني", ORTH: "ثنى"},
|
|
{NORM: "أنبأنا", ORTH: "أنا"},
|
|
{NORM: "أخبرنا", ORTH: "نا"},
|
|
{NORM: "مصدر سابق", ORTH: "م. س"},
|
|
{NORM: "مصدر نفسه", ORTH: "م. ن"},
|
|
]:
|
|
_exc[exc_data[ORTH]] = [exc_data]
|
|
|
|
# Other abv.
|
|
for exc_data in [
|
|
{NORM: "دكتور", ORTH: "د."},
|
|
{NORM: "أستاذ دكتور", ORTH: "أ.د"},
|
|
{NORM: "أستاذ", ORTH: "أ."},
|
|
{NORM: "بروفيسور", ORTH: "ب."},
|
|
]:
|
|
_exc[exc_data[ORTH]] = [exc_data]
|
|
|
|
for exc_data in [{NORM: "تلفون", ORTH: "ت."}, {NORM: "صندوق بريد", ORTH: "ص.ب"}]:
|
|
_exc[exc_data[ORTH]] = [exc_data]
|
|
|
|
TOKENIZER_EXCEPTIONS = update_exc(BASE_EXCEPTIONS, _exc)
|