def __deserialize_item(self, stream: Stream) -> 'MorphBaseInfo':
from pullenti.ner.core.internal.SerializerHelper import SerializerHelper
ty = stream.readbyte()
res = (MorphBaseInfo() if ty == 0 else MorphWordForm())
res.class0_ = MorphClass._new53(
SerializerHelper.deserialize_short(stream))
res.case_ = MorphCase._new29(
SerializerHelper.deserialize_short(stream))
res.gender = Utils.valToEnum(
SerializerHelper.deserialize_short(stream), MorphGender)
res.number = Utils.valToEnum(
SerializerHelper.deserialize_short(stream), MorphNumber)
res.language = MorphLang._new56(
SerializerHelper.deserialize_short(stream))
if (ty == 0):
return res
wf = Utils.asObjectOrNull(res, MorphWordForm)
wf.normal_case = SerializerHelper.deserialize_string(stream)
wf.normal_full = SerializerHelper.deserialize_string(stream)
wf.undef_coef = SerializerHelper.deserialize_short(stream)
cou = SerializerHelper.deserialize_int(stream)
i = 0
while i < cou:
if (wf.misc is None):
wf.misc = MorphMiscInfo()
wf.misc.attrs.append(SerializerHelper.deserialize_string(stream))
i += 1
return res
def __init__(self, source: 'MorphCollection' = None) -> None:
super().__init__()
self.__m_class = MorphClass()
self.__m_gender = MorphGender.UNDEFINED
self.__m_number = MorphNumber.UNDEFINED
self.__m_case = MorphCase()
self.__m_language = MorphLang()
self.__m_voice = MorphVoice.UNDEFINED
self.__m_need_recalc = True
self.__m_items = None
if (source is None):
return
for it in source.items:
mi = None
if (isinstance(it, MorphWordForm)):
wf = MorphWordForm()
wf.copy_from_word_form(Utils.asObjectOrNull(it, MorphWordForm))
mi = (wf)
else:
mi = MorphBaseInfo()
mi.copy_from(it)
if (self.__m_items is None):
self.__m_items = list()
self.__m_items.append(mi)
self.__m_class = MorphClass._new53(source.__m_class.value)
self.__m_gender = source.__m_gender
self.__m_case = MorphCase._new29(source.__m_case.value)
self.__m_number = source.__m_number
self.__m_language = MorphLang._new56(source.__m_language.value)
self.__m_voice = source.__m_voice
self.__m_need_recalc = False
def clone(self) -> 'MorphCollection':
""" Создать копию
"""
res = MorphCollection()
if (self.__m_items is not None):
res.__m_items = list()
try:
res.__m_items.extend(self.__m_items)
except Exception as ex:
pass
if (not self.__m_need_recalc):
res.__m_class = MorphClass._new53(self.__m_class.value)
res.__m_gender = self.__m_gender
res.__m_case = MorphCase._new29(self.__m_case.value)
res.__m_number = self.__m_number
res.__m_language = MorphLang._new56(self.__m_language.value)
res.__m_need_recalc = False
res.__m_voice = self.__m_voice
return res
def _deserialize(self, stream: Stream) -> None:
from pullenti.ner.core.internal.SerializerHelper import SerializerHelper
self.__m_class = MorphClass._new53(
SerializerHelper.deserialize_short(stream))
self.__m_case = MorphCase._new29(
SerializerHelper.deserialize_short(stream))
self.__m_gender = (Utils.valToEnum(
SerializerHelper.deserialize_short(stream), MorphGender))
self.__m_number = (Utils.valToEnum(
SerializerHelper.deserialize_short(stream), MorphNumber))
self.__m_voice = (Utils.valToEnum(
SerializerHelper.deserialize_short(stream), MorphVoice))
self.__m_language = MorphLang._new56(
SerializerHelper.deserialize_short(stream))
cou = SerializerHelper.deserialize_int(stream)
self.__m_items = list()
i = 0
while i < cou:
it = self.__deserialize_item(stream)
if (it is not None):
self.__m_items.append(it)
i += 1
self.__m_need_recalc = False
def get_normal_case_text(self,
mc: 'MorphClass' = None,
num: 'MorphNumber' = MorphNumber.UNDEFINED,
gender: 'MorphGender' = MorphGender.UNDEFINED,
keep_chars: bool = False) -> str:
if ((isinstance(self.begin_token, ReferentToken))
and self.begin_token == self.end_token):
return self.begin_token.get_normal_case_text(
mc, num, gender, keep_chars)
res = None
max_coef = 0
def_coef = -1
for it in self.morph.items:
v = Utils.asObjectOrNull(it, NounPhraseItemTextVar)
if (v is None):
continue
if (v.undef_coef > 0
and (((v.undef_coef < max_coef) or def_coef >= 0))):
continue
if (num == MorphNumber.SINGULAR
and v.single_number_value is not None):
if (mc is not None and ((gender == MorphGender.NEUTER
or gender == MorphGender.FEMINIE))
and mc.is_adjective):
bi = MorphBaseInfo._new401(MorphClass._new53(mc.value),
gender, MorphNumber.SINGULAR,
MorphCase.NOMINATIVE,
self.morph.language)
str0_ = MorphologyService.get_wordform(
v.single_number_value, bi)
if (str0_ is not None):
res = str0_
else:
res = v.single_number_value
if (v.undef_coef == 0):
break
max_coef = v.undef_coef
continue
if (Utils.isNullOrEmpty(v.normal_value)):
continue
if (str.isdigit(v.normal_value[0]) and mc is not None
and mc.is_adjective):
val = 0
wrapval402 = RefOutArgWrapper(0)
inoutres403 = Utils.tryParseInt(v.normal_value, wrapval402)
val = wrapval402.value
if (inoutres403):
str0_ = NumberHelper.get_number_adjective(
val, gender,
(MorphNumber.SINGULAR if num == MorphNumber.SINGULAR
or val == 1 else MorphNumber.PLURAL))
if (str0_ is not None):
res = str0_
if (v.undef_coef == 0):
break
max_coef = v.undef_coef
continue
res1 = it.normal_value
if (num == MorphNumber.SINGULAR):
if (res1 == "ДЕТИ"):
res1 = "РЕБЕНОК"
elif (res1 == "ЛЮДИ"):
res1 = "ЧЕЛОВЕК"
max_coef = v.undef_coef
if (v.undef_coef > 0):
res = res1
continue
def_co = 0
if (mc is not None and mc.is_adjective and v.undef_coef == 0):
pass
elif (
((isinstance(self.begin_token, TextToken))
and res1 == self.begin_token.term and it.case_.is_nominative)
and it.number == MorphNumber.SINGULAR):
def_co = 1
if (num == MorphNumber.PLURAL and
((v.number) & (MorphNumber.PLURAL)) == (MorphNumber.PLURAL)):
def_co += 3
if (res is None or def_co > def_coef):
res = res1
def_coef = def_co
if (def_co > 0):
break
if (res is not None):
return self.__corr_chars(res, keep_chars)
if (res is None and self.begin_token == self.end_token):
res = self.begin_token.get_normal_case_text(
mc, num, gender, keep_chars)
elif (res is None):
res = self.begin_token.get_normal_case_text(
mc, num, gender, keep_chars)
if (res is None):
res = MiscHelper.get_text_value_of_meta_token(
self, (GetTextAttr.KEEPREGISTER
if keep_chars else GetTextAttr.NO))
else:
res = "{0} {1}".format(
res,
MiscHelper.get_text_value(
self.begin_token.next0_, self.end_token,
(GetTextAttr.KEEPREGISTER
if keep_chars else GetTextAttr.NO)))
return Utils.ifNotNull(res, "?")
def get_normal_case_text(self,
mc: 'MorphClass' = None,
num: 'MorphNumber' = MorphNumber.UNDEFINED,
gender: 'MorphGender' = MorphGender.UNDEFINED,
keep_chars: bool = False) -> str:
from pullenti.ner.core.MiscHelper import MiscHelper
empty = True
if (mc is not None and mc.is_preposition):
return LanguageHelper.normalize_preposition(self.term)
for it in self.morph.items:
if (mc is not None and not mc.is_undefined):
cc = (it.class0_) & mc
if (cc.is_undefined):
continue
if (cc.is_misc and not cc.is_proper and mc != it.class0_):
continue
wf = Utils.asObjectOrNull(it, MorphWordForm)
normal_full = False
if (gender != MorphGender.UNDEFINED):
if (((it.gender) & (gender)) == (MorphGender.UNDEFINED)):
if ((gender == MorphGender.MASCULINE and
((it.gender != MorphGender.UNDEFINED or it.number
== MorphNumber.PLURAL)) and wf is not None)
and wf.normal_full is not None):
normal_full = True
elif (gender == MorphGender.MASCULINE
and it.class0_.is_personal_pronoun):
pass
else:
continue
if (not it.case_.is_undefined):
empty = False
if (wf is not None):
res = None
if (num == MorphNumber.SINGULAR
and it.number == MorphNumber.PLURAL
and wf.normal_full is not None):
le = len(wf.normal_case)
if ((le == (len(wf.normal_full) + 2) and le > 4
and wf.normal_case[le - 2] == 'С')
and wf.normal_case[le - 1] == 'Я'):
res = wf.normal_case
else:
res = (wf.normal_full
if normal_full else wf.normal_full)
else:
res = (wf.normal_full if normal_full else
(Utils.ifNotNull(wf.normal_case, self.term)))
if (num == MorphNumber.SINGULAR and mc is not None
and mc == MorphClass.NOUN):
if (res == "ДЕТИ"):
res = "РЕБЕНОК"
if (keep_chars):
if (self.chars.is_all_lower):
res = res.lower()
elif (self.chars.is_capital_upper):
res = MiscHelper.convert_first_char_upper_and_other_lower(
res)
return res
if (not empty):
return None
te = None
if (num == MorphNumber.SINGULAR and mc is not None):
bi = MorphBaseInfo._new492(MorphClass._new53(mc.value), gender,
MorphNumber.SINGULAR,
self.morph.language)
vars0_ = MorphologyService.get_wordform(self.term, bi)
if (vars0_ is not None):
te = vars0_
if (te is None):
te = self.term
if (keep_chars):
if (self.chars.is_all_lower):
return te.lower()
elif (self.chars.is_capital_upper):
return MiscHelper.convert_first_char_upper_and_other_lower(te)
return te