mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-27 18:36:36 +03:00
20 lines
688 B
Python
20 lines
688 B
Python
from ..tokenizer_exceptions import BASE_EXCEPTIONS
|
|
from ...symbols import ORTH, NORM
|
|
from ...util import update_exc
|
|
|
|
|
|
_exc = {
|
|
"tayo'y": [{ORTH: "tayo"}, {ORTH: "'y", NORM: "ay"}],
|
|
"isa'y": [{ORTH: "isa"}, {ORTH: "'y", NORM: "ay"}],
|
|
"baya'y": [{ORTH: "baya"}, {ORTH: "'y", NORM: "ay"}],
|
|
"sa'yo": [{ORTH: "sa"}, {ORTH: "'yo", NORM: "iyo"}],
|
|
"ano'ng": [{ORTH: "ano"}, {ORTH: "'ng", NORM: "ang"}],
|
|
"siya'y": [{ORTH: "siya"}, {ORTH: "'y", NORM: "ay"}],
|
|
"nawa'y": [{ORTH: "nawa"}, {ORTH: "'y", NORM: "ay"}],
|
|
"papa'no": [{ORTH: "papa'no", NORM: "papaano"}],
|
|
"'di": [{ORTH: "'di", NORM: "hindi"}],
|
|
}
|
|
|
|
|
|
TOKENIZER_EXCEPTIONS = update_exc(BASE_EXCEPTIONS, _exc)
|