spaCy/spacy/lang/ar/tokenizer_exceptions.py

47 lines
1.4 KiB
Python
Raw Normal View History

from ...symbols import ORTH, LEMMA
_exc = {}
# Time
for exc_data in [
{LEMMA: "قبل الميلاد", ORTH: "ق.م"},
{LEMMA: "بعد الميلاد", ORTH: "ب. م"},
{LEMMA: "ميلادي", ORTH: ""},
{LEMMA: "هجري", ORTH: ".هـ"},
{LEMMA: "توفي", ORTH: ""},
]:
_exc[exc_data[ORTH]] = [exc_data]
# Scientific abv.
for exc_data in [
{LEMMA: "صلى الله عليه وسلم", ORTH: "صلعم"},
{LEMMA: "الشارح", ORTH: "الشـ"},
{LEMMA: "الظاهر", ORTH: "الظـ"},
{LEMMA: "أيضًا", ORTH: "أيضـ"},
{LEMMA: "إلى آخره", ORTH: "إلخ"},
{LEMMA: "انتهى", ORTH: "اهـ"},
{LEMMA: "حدّثنا", ORTH: "ثنا"},
{LEMMA: "حدثني", ORTH: "ثنى"},
{LEMMA: "أنبأنا", ORTH: "أنا"},
{LEMMA: "أخبرنا", ORTH: "نا"},
{LEMMA: "مصدر سابق", ORTH: "م. س"},
{LEMMA: "مصدر نفسه", ORTH: "م. ن"},
]:
_exc[exc_data[ORTH]] = [exc_data]
# Other abv.
for exc_data in [
{LEMMA: "دكتور", ORTH: "د."},
{LEMMA: "أستاذ دكتور", ORTH: "أ.د"},
{LEMMA: "أستاذ", ORTH: "أ."},
{LEMMA: "بروفيسور", ORTH: "ب."},
]:
_exc[exc_data[ORTH]] = [exc_data]
for exc_data in [{LEMMA: "تلفون", ORTH: "ت."}, {LEMMA: "صندوق بريد", ORTH: "ص.ب"}]:
_exc[exc_data[ORTH]] = [exc_data]
TOKENIZER_EXCEPTIONS = _exc