def process(self, kit: 'AnalysisKit') -> None:
""" Основная функция выделения телефонов
Args:
cnt:
stage:
"""
ad = kit.getAnalyzerData(self)
addunits = None
if (kit.ontology is not None):
addunits = TerminCollection()
for r in kit.ontology.items:
uu = Utils.asObjectOrNull(r.referent, UnitReferent)
if (uu is None):
continue
if (uu._m_unit is not None):
continue
for s in uu.slots:
if (s.type_name == UnitReferent.ATTR_NAME
or s.type_name == UnitReferent.ATTR_FULLNAME):
addunits.add(
Termin._new117(Utils.asObjectOrNull(s.value, str),
uu))
t = kit.first_token
first_pass3040 = True
while True:
if first_pass3040: first_pass3040 = False
else: t = t.next0_
if (not (t is not None)): break
mt = MeasureToken.tryParseMinimal(t, addunits, False)
if (mt is None):
mt = MeasureToken.tryParse(t, addunits, True, False)
if (mt is None):
continue
rts = mt.createRefenetsTokensWithRegister(ad, True)
if (rts is None):
continue
i = 0
while i < len(rts):
rt = rts[i]
t.kit.embedToken(rt)
t = (rt)
j = i + 1
while j < len(rts):
if (rts[j].begin_token == rt.begin_token):
rts[j].begin_token = t
if (rts[j].end_token == rt.end_token):
rts[j].end_token = t
j += 1
i += 1
if (kit.ontology is not None):
for e0_ in ad.referents:
u = Utils.asObjectOrNull(e0_, UnitReferent)
if (u is None):
continue
for r in kit.ontology.items:
uu = Utils.asObjectOrNull(r.referent, UnitReferent)
if (uu is None):
continue
ok = False
for s in uu.slots:
if (s.type_name == UnitReferent.ATTR_NAME
or s.type_name == UnitReferent.ATTR_FULLNAME):
if (u.findSlot(None, s.value, True) is not None):
ok = True
break
if (ok):
u.ontology_items = list()
u.ontology_items.append(r)
break
def process(self, kit: 'AnalysisKit') -> None:
ad = kit.getAnalyzerData(self)
models = TerminCollection()
objs_by_model = dict()
obj_by_names = TerminCollection()
t = kit.first_token
first_pass3158 = True
while True:
if first_pass3158: first_pass3158 = False
else: t = t.next0_
if (not (t is not None)): break
its = WeaponItemToken.tryParseList(t, 10)
if (its is None):
continue
rts = self.__tryAttach(its, False)
if (rts is not None):
for rt in rts:
rt.referent = ad.registerReferent(rt.referent)
kit.embedToken(rt)
t = (rt)
for s in rt.referent.slots:
if (s.type_name == WeaponReferent.ATTR_MODEL):
mod = str(s.value)
for k in range(2):
if (not str.isdigit(mod[0])):
li = []
wrapli2638 = RefOutArgWrapper(None)
inoutres2639 = Utils.tryGetValue(
objs_by_model, mod, wrapli2638)
li = wrapli2638.value
if (not inoutres2639):
li = list()
objs_by_model[mod] = li
if (not rt.referent in li):
li.append(rt.referent)
models.addStr(mod, li, None, False)
if (k > 0):
break
brand = rt.referent.getStringValue(
WeaponReferent.ATTR_BRAND)
if (brand is None):
break
mod = "{0} {1}".format(brand, mod)
elif (s.type_name == WeaponReferent.ATTR_NAME):
obj_by_names.add(
Termin._new117(str(s.value), rt.referent))
if (len(objs_by_model) == 0 and len(obj_by_names.termins) == 0):
return
t = kit.first_token
first_pass3159 = True
while True:
if first_pass3159: first_pass3159 = False
else: t = t.next0_
if (not (t is not None)): break
br = BracketHelper.tryParse(t, BracketParseAttr.NO, 10)
if (br is not None):
toks = obj_by_names.tryParse(t.next0_, TerminParseAttr.NO)
if (toks is not None
and toks.end_token.next0_ == br.end_token):
rt0 = ReferentToken(
Utils.asObjectOrNull(toks.termin.tag, Referent),
br.begin_token, br.end_token)
kit.embedToken(rt0)
t = (rt0)
continue
if (not ((isinstance(t, TextToken)))):
continue
if (not t.chars.is_letter):
continue
tok = models.tryParse(t, TerminParseAttr.NO)
if (tok is None):
if (not t.chars.is_all_lower):
tok = obj_by_names.tryParse(t, TerminParseAttr.NO)
if (tok is None):
continue
if (not tok.is_whitespace_after):
if (tok.end_token.next0_ is None
or not tok.end_token.next0_.isCharOf(",.)")):
if (not BracketHelper.isBracket(tok.end_token.next0_,
False)):
continue
tr = None
li = Utils.asObjectOrNull(tok.termin.tag, list)
if (li is not None and len(li) == 1):
tr = li[0]
else:
tr = (Utils.asObjectOrNull(tok.termin.tag, Referent))
if (tr is not None):
tit = WeaponItemToken.tryParse(tok.begin_token.previous, None,
False, True)
if (tit is not None and tit.typ == WeaponItemToken.Typs.BRAND):
tr.addSlot(WeaponReferent.ATTR_BRAND, tit.value, False, 0)
tok.begin_token = tit.begin_token
rt0 = ReferentToken(tr, tok.begin_token, tok.end_token)
kit.embedToken(rt0)
t = (rt0)
continue
def process(self, kit : 'AnalysisKit') -> None:
ad = kit.get_analyzer_data(self)
models = TerminCollection()
objs_by_model = dict()
obj_by_names = TerminCollection()
t = kit.first_token
first_pass3406 = True
while True:
if first_pass3406: first_pass3406 = False
else: t = t.next0_
if (not (t is not None)): break
its = TransItemToken.try_parse_list(t, 10)
if (its is None):
continue
rts = self.__try_attach(its, False)
if (rts is not None):
for rt in rts:
cou = 0
tt = t.previous
first_pass3407 = True
while True:
if first_pass3407: first_pass3407 = False
else: tt = tt.previous; cou += 1
if (not (tt is not None and (cou < 1000))): break
tr = Utils.asObjectOrNull(tt.get_referent(), TransportReferent)
if (tr is None):
continue
ok = True
for s in rt.referent.slots:
if (tr.find_slot(s.type_name, s.value, True) is None):
ok = False
break
if (ok):
rt.referent = (tr)
break
rt.referent = ad.register_referent(rt.referent)
kit.embed_token(rt)
t = (rt)
for s in rt.referent.slots:
if (s.type_name == TransportReferent.ATTR_MODEL):
mod = str(s.value)
for k in range(2):
if (not str.isdigit(mod[0])):
li = [ ]
wrapli2702 = RefOutArgWrapper(None)
inoutres2703 = Utils.tryGetValue(objs_by_model, mod, wrapli2702)
li = wrapli2702.value
if (not inoutres2703):
li = list()
objs_by_model[mod] = li
if (not rt.referent in li):
li.append(rt.referent)
models.add_string(mod, li, None, False)
if (k > 0):
break
brand = rt.referent.get_string_value(TransportReferent.ATTR_BRAND)
if (brand is None):
break
mod = "{0} {1}".format(brand, mod)
elif (s.type_name == TransportReferent.ATTR_NAME):
obj_by_names.add(Termin._new100(str(s.value), rt.referent))
if (len(objs_by_model) == 0 and len(obj_by_names.termins) == 0):
return
t = kit.first_token
first_pass3408 = True
while True:
if first_pass3408: first_pass3408 = False
else: t = t.next0_
if (not (t is not None)): break
br = BracketHelper.try_parse(t, BracketParseAttr.NO, 10)
if (br is not None):
toks = obj_by_names.try_parse(t.next0_, TerminParseAttr.NO)
if (toks is not None and toks.end_token.next0_ == br.end_token):
rt0 = ReferentToken(Utils.asObjectOrNull(toks.termin.tag, Referent), br.begin_token, br.end_token)
kit.embed_token(rt0)
t = (rt0)
continue
if (not (isinstance(t, TextToken))):
continue
if (not t.chars.is_letter):
continue
tok = models.try_parse(t, TerminParseAttr.NO)
if (tok is None):
if (not t.chars.is_all_lower):
tok = obj_by_names.try_parse(t, TerminParseAttr.NO)
if (tok is None):
continue
if (not tok.is_whitespace_after):
if (tok.end_token.next0_ is None or not tok.end_token.next0_.is_char_of(",.)")):
if (not BracketHelper.is_bracket(tok.end_token.next0_, False)):
continue
tr = None
li = Utils.asObjectOrNull(tok.termin.tag, list)
if (li is not None and len(li) == 1):
tr = li[0]
else:
tr = (Utils.asObjectOrNull(tok.termin.tag, Referent))
if (tr is not None):
tit = TransItemToken.try_parse(tok.begin_token.previous, None, False, True)
if (tit is not None and tit.typ == TransItemToken.Typs.BRAND):
tr.add_slot(TransportReferent.ATTR_BRAND, tit.value, False, 0)
tok.begin_token = tit.begin_token
rt0 = ReferentToken(tr, tok.begin_token, tok.end_token)
kit.embed_token(rt0)
t = (rt0)
continue
class IntOntologyCollection:
# Внутренний онтологический словарь. По сути, некоторая надстройка над TerminCollection.
# Не помню уже, зачем был введён, но для чего-то нужен.
class OntologyTermin(Termin):
def __init__(self) -> None:
super().__init__(None, None, False)
self.owner = None;
@staticmethod
def _new489(_arg1 : 'IntOntologyItem', _arg2 : object) -> 'OntologyTermin':
res = IntOntologyCollection.OntologyTermin()
res.owner = _arg1
res.tag = _arg2
return res
def __init__(self) -> None:
self.is_ext_ontology = False
self.__m_items = list()
self.__m_termins = TerminCollection()
@property
def items(self) -> typing.List['IntOntologyItem']:
return self.__m_items
def add_item(self, di : 'IntOntologyItem') -> None:
self.__m_items.append(di)
di.owner = self
i = 0
while i < len(di.termins):
if (isinstance(di.termins[i], IntOntologyCollection.OntologyTermin)):
di.termins[i].owner = di
self.__m_termins.add(di.termins[i])
else:
nt = IntOntologyCollection.OntologyTermin._new489(di, di.termins[i].tag)
di.termins[i].copy_to(nt)
self.__m_termins.add(nt)
di.termins[i] = (nt)
i += 1
def add_referent(self, referent : 'Referent') -> bool:
if (referent is None):
return False
oi = None
if (referent._int_ontology_item is not None and referent._int_ontology_item.owner == self):
oi1 = referent.create_ontology_item()
if (oi1 is None or len(oi1.termins) == len(referent._int_ontology_item.termins)):
return True
for t in referent._int_ontology_item.termins:
self.__m_termins.remove(t)
i = Utils.indexOfList(self.__m_items, referent._int_ontology_item, 0)
if (i >= 0):
del self.__m_items[i]
oi = oi1
else:
oi = referent.create_ontology_item()
if (oi is None):
return False
oi.referent = referent
referent._int_ontology_item = oi
self.add_item(oi)
return True
def add_termin(self, di : 'IntOntologyItem', t : 'Termin') -> None:
nt = IntOntologyCollection.OntologyTermin._new489(di, t.tag)
t.copy_to(nt)
self.__m_termins.add(nt)
def add(self, t : 'Termin') -> None:
self.__m_termins.add(t)
def find_termin_by_canonic_text(self, text : str) -> typing.List['Termin']:
return self.__m_termins.find_termins_by_canonic_text(text)
def try_attach(self, t : 'Token', referent_type_name : str=None, can_be_geo_object : bool=False) -> typing.List['IntOntologyToken']:
tts = self.__m_termins.try_parse_all(t, (TerminParseAttr.CANBEGEOOBJECT if can_be_geo_object else TerminParseAttr.NO))
if (tts is None):
return None
res = list()
dis = list()
for tt in tts:
di = None
if (isinstance(tt.termin, IntOntologyCollection.OntologyTermin)):
di = tt.termin.owner
if (di is not None):
if (di.referent is not None and referent_type_name is not None):
if (di.referent.type_name != referent_type_name):
continue
if (di in dis):
continue
dis.append(di)
res.append(IntOntologyToken._new491(tt.begin_token, tt.end_token, di, tt.termin, tt.morph))
return (None if len(res) == 0 else res)
def try_attach_by_item(self, item : 'IntOntologyItem') -> typing.List['IntOntologyItem']:
if (item is None):
return None
res = None
for t in item.termins:
li = self.__m_termins.find_termins_by_termin(t)
if (li is not None):
for tt in li:
if (isinstance(tt, IntOntologyCollection.OntologyTermin)):
oi = tt.owner
if (res is None):
res = list()
if (not oi in res):
res.append(oi)
return res
def try_attach_by_referent(self, referent : 'Referent', item : 'IntOntologyItem'=None, must_be_single : bool=False) -> typing.List['Referent']:
if (referent is None):
return None
if (item is None):
item = referent.create_ontology_item()
if (item is None):
return None
li = self.try_attach_by_item(item)
if (li is None):
return None
res = None
for oi in li:
r = Utils.ifNotNull(oi.referent, Utils.asObjectOrNull(oi.tag, Referent))
if (r is not None):
if (referent.can_be_equals(r, ReferentsEqualType.WITHINONETEXT)):
if (res is None):
res = list()
if (not r in res):
res.append(r)
if (must_be_single):
if (res is not None and len(res) > 1):
i = 0
while i < (len(res) - 1):
j = i + 1
while j < len(res):
if (not res[i].can_be_equals(res[j], ReferentsEqualType.FORMERGING)):
return None
j += 1
i += 1
return res
def remove(self, r : 'Referent') -> None:
i = 0
i = 0
while i < len(self.__m_items):
if (self.__m_items[i].referent == r):
oi = self.__m_items[i]
oi.referent = (None)
r._int_ontology_item = (None)
del self.__m_items[i]
for t in oi.termins:
self.__m_termins.remove(t)
break
i += 1
@staticmethod
def _new2815(_arg1 : bool) -> 'IntOntologyCollection':
res = IntOntologyCollection()
res.is_ext_ontology = _arg1
return res
class IntOntologyCollection:
""" Онтологический словарь """
class OntologyTermin(Termin):
def __init__(self) -> None:
super().__init__(None, None, False)
self.owner = None;
@staticmethod
def _new546(_arg1 : 'IntOntologyItem', _arg2 : object) -> 'OntologyTermin':
res = IntOntologyCollection.OntologyTermin()
res.owner = _arg1
res.tag = _arg2
return res
def __init__(self) -> None:
self.is_ext_ontology = False
self.__m_items = list()
self.__m_termins = TerminCollection()
@property
def items(self) -> typing.List['IntOntologyItem']:
""" Список элементов онтологии """
return self.__m_items
def addItem(self, di : 'IntOntologyItem') -> None:
""" Добавить элемент (внимание, после добавления нельзя менять термины у элемента)
Args:
di(IntOntologyItem):
"""
self.__m_items.append(di)
di.owner = self
i = 0
while i < len(di.termins):
if (isinstance(di.termins[i], IntOntologyCollection.OntologyTermin)):
(di.termins[i]).owner = di
self.__m_termins.add(di.termins[i])
else:
nt = IntOntologyCollection.OntologyTermin._new546(di, di.termins[i].tag)
di.termins[i].copyTo(nt)
self.__m_termins.add(nt)
di.termins[i] = (nt)
i += 1
def addReferent(self, referent : 'Referent') -> bool:
""" Добавить в онтологию сущность
Args:
referent(Referent):
"""
if (referent is None):
return False
oi = None
if (referent._int_ontology_item is not None and referent._int_ontology_item.owner == self):
oi1 = referent.createOntologyItem()
if (oi1 is None or len(oi1.termins) == len(referent._int_ontology_item.termins)):
return True
for t in referent._int_ontology_item.termins:
self.__m_termins.remove(t)
i = Utils.indexOfList(self.__m_items, referent._int_ontology_item, 0)
if (i >= 0):
del self.__m_items[i]
oi = oi1
else:
oi = referent.createOntologyItem()
if (oi is None):
return False
oi.referent = referent
referent._int_ontology_item = oi
self.addItem(oi)
return True
def addTermin(self, di : 'IntOntologyItem', t : 'Termin') -> None:
""" Добавить термин в существующий элемент
Args:
di(IntOntologyItem):
t(Termin):
"""
nt = IntOntologyCollection.OntologyTermin._new546(di, t.tag)
t.copyTo(nt)
self.__m_termins.add(nt)
def add(self, t : 'Termin') -> None:
""" Добавить отдельный термин (после добавления нельзя изменять свойства термина)
Args:
t(Termin):
"""
self.__m_termins.add(t)
def findTerminByCanonicText(self, text : str) -> typing.List['Termin']:
return self.__m_termins.findTerminByCanonicText(text)
def tryAttach(self, t : 'Token', referent_type_name : str=None, can_be_geo_object : bool=False) -> typing.List['IntOntologyToken']:
""" Привязать с указанной позиции
Args:
t(Token):
can_be_geo_object(bool): при True внутри может быть географический объект (Министерство РФ по делам ...)
"""
tts = self.__m_termins.tryParseAll(t, (TerminParseAttr.CANBEGEOOBJECT if can_be_geo_object else TerminParseAttr.NO))
if (tts is None):
return None
res = list()
dis = list()
for tt in tts:
di = None
if (isinstance(tt.termin, IntOntologyCollection.OntologyTermin)):
di = (tt.termin).owner
if (di is not None):
if (di.referent is not None and referent_type_name is not None):
if (di.referent.type_name != referent_type_name):
continue
if (di in dis):
continue
dis.append(di)
res.append(IntOntologyToken._new548(tt.begin_token, tt.end_token, di, tt.termin, tt.morph))
return (None if len(res) == 0 else res)
def tryAttachByItem(self, item : 'IntOntologyItem') -> typing.List['IntOntologyItem']:
""" Найти похожие онтологические объекты
Args:
item(IntOntologyItem):
"""
if (item is None):
return None
res = None
for t in item.termins:
li = self.__m_termins.tryAttach(t)
if (li is not None):
for tt in li:
if (isinstance(tt, IntOntologyCollection.OntologyTermin)):
oi = (tt).owner
if (res is None):
res = list()
if (not oi in res):
res.append(oi)
return res
def tryAttachByReferent(self, referent : 'Referent', item : 'IntOntologyItem'=None, must_be_single : bool=False) -> typing.List['Referent']:
""" Найти эквивалентные сущности через онтологические объекты
Args:
item(IntOntologyItem):
referent(Referent):
"""
if (referent is None):
return None
if (item is None):
item = referent.createOntologyItem()
if (item is None):
return None
li = self.tryAttachByItem(item)
if (li is None):
return None
res = None
for oi in li:
r = Utils.ifNotNull(oi.referent, (Utils.asObjectOrNull(oi.tag, Referent)))
if (r is not None):
if (referent.canBeEquals(r, Referent.EqualType.WITHINONETEXT)):
if (res is None):
res = list()
if (not r in res):
res.append(r)
if (must_be_single):
if (res is not None and len(res) > 1):
i = 0
while i < (len(res) - 1):
j = i + 1
while j < len(res):
if (not res[i].canBeEquals(res[j], Referent.EqualType.FORMERGING)):
return None
j += 1
i += 1
return res
def remove(self, r : 'Referent') -> None:
""" Произвести привязку, если элемент найдётся, то установить ссылку на OntologyElement
Args:
referent:
mergeSlots:
Удалить всё, что связано с сущностью
r(Referent):
"""
i = 0
while i < len(self.__m_items):
if (self.__m_items[i].referent == r):
oi = self.__m_items[i]
oi.referent = (None)
r._int_ontology_item = (None)
del self.__m_items[i]
for t in oi.termins:
self.__m_termins.remove(t)
break
i += 1
@staticmethod
def _new2649(_arg1 : bool) -> 'IntOntologyCollection':
res = IntOntologyCollection()
res.is_ext_ontology = _arg1
return res