def get_bytes(name: str) -> bytearray:
""" Получить встроенный ресурс
Args:
name(str): имя, на который оканчивается ресурс
"""
names = Utils.getResourcesNames('pullenti.ner.org.properties',
'.png;.dat')
for n in names:
if (Utils.endsWithString(n, name, True)):
if (len(name) < len(n)):
if (n[len(n) - len(name) - 1] != '.'):
continue
try:
inf = Utils.getResourceInfo('pullenti.ner.org.properties',
n)
if (inf is None):
continue
with Utils.getResourceStream('pullenti.ner.org.properties',
n) as stream:
buf = Utils.newArrayOfBytes(stream.length, 0)
stream.read(buf, 0, len(buf))
return buf
except Exception as ex:
pass
return None
def getBytes(name: str) -> bytearray:
""" Получить встроенный ресурс
Args:
name(str): имя, на который оканчивается ресурс
"""
# ignored: assembly = EpNerBankInternalResourceHelper.
names = Utils.getResourcesNames('pullenti.ner.bank.properties',
'.png;.txt;.csv')
for n in names:
if (Utils.endsWithString(n, name, True)):
try:
inf = Utils.getResourceInfo('pullenti.ner.bank.properties',
n)
if (inf is None):
continue
with Utils.getResourceStream(
'pullenti.ner.bank.properties', n) as stream:
buf = Utils.newArrayOfBytes(Utils.getLengthIO(stream),
0)
Utils.readIO(stream, buf, 0, len(buf))
return buf
except Exception as ex:
pass
return None
def deserialize_string(stream : Stream) -> str:
len0_ = SerializerHelper.deserialize_int(stream)
if (len0_ < 0):
return None
if (len0_ == 0):
return ""
data = Utils.newArrayOfBytes(len0_, 0)
stream.read(data, 0, len(data))
return data.decode("UTF-8", 'ignore')
def _deserialize(self, stream: Stream, kit_: 'AnalysisKit',
vers: int) -> None:
from pullenti.ner.core.internal.SerializerHelper import SerializerHelper
super()._deserialize(stream, kit_, vers)
if (vers == 0):
buf = Utils.newArrayOfBytes(8, 0)
stream.read(buf, 0, 8)
lo = int.from_bytes(buf[0:0 + 8], byteorder="little")
self.value = str(lo)
else:
self.value = SerializerHelper.deserialize_string(stream)
self.typ = (Utils.valToEnum(SerializerHelper.deserialize_int(stream),
NumberSpellingType))
def get_bytes(name : str) -> bytearray:
names = Utils.getResourcesNames('pullenti.ner.core.properties', '.csv;.png')
for n in names:
if (Utils.endsWithString(n, name, True)):
if (len(name) < len(n)):
if (n[len(n) - len(name) - 1] != '.'):
continue
try:
inf = Utils.getResourceInfo('pullenti.ner.core.properties', n)
if (inf is None):
continue
with Utils.getResourceStream('pullenti.ner.core.properties', n) as stream:
buf = Utils.newArrayOfBytes(stream.length, 0)
stream.read(buf, 0, len(buf))
return buf
except Exception as ex:
pass
return None
def deflate_gzip(str0_: io.IOBase, res: io.IOBase) -> None:
with gzip.GzipFile(fileobj=str0_, mode='r') as deflate:
buf = Utils.newArrayOfBytes(100000, 0)
while True:
i = -1
try:
ii = 0
while ii < len(buf):
buf[ii] = (0)
ii += 1
i = Utils.readIO(deflate, buf, 0, len(buf))
except Exception as ex:
for i in range(len(buf) - 1, -1, -1):
if (buf[i] != (0)):
Utils.writeIO(res, buf, 0, i + 1)
break
else:
i = -1
break
if (i < 1):
break
Utils.writeIO(res, buf, 0, i)
def deflate_gzip(str0_: Stream, res: Stream) -> None:
with Stream(gzip.GzipFile(fileobj=str0_.getstream(),
mode='r')) as deflate:
buf = Utils.newArrayOfBytes(100000, 0)
len0_ = len(buf)
while True:
i = -1
try:
ii = 0
while ii < len0_:
buf[ii] = (0)
ii += 1
i = deflate.read(buf, 0, len0_)
except Exception as ex:
for i in range(len0_ - 1, -1, -1):
if (buf[i] != (0)):
res.write(buf, 0, i + 1)
break
else:
i = -1
break
if (i < 1):
break
res.write(buf, 0, i)
def deserialize_short(stream : Stream) -> int:
buf = Utils.newArrayOfBytes(2, 0)
stream.read(buf, 0, 2)
return int.from_bytes(buf[0:0+2], byteorder="little")
def create_subsents(sent: 'Sentence') -> typing.List['Subsent']:
if (len(sent.items) == 0):
return None
res = list()
begin = sent.items[0].begin_token.begin_char
end = sent.items[len(sent.items) - 1].end_token.end_char
map0_ = Utils.newArrayOfBytes((end + 1) - begin, 0)
if (sent.best_var is not None):
for seg in sent.best_var.segs:
if (seg is not None):
for li in seg.links:
if (li is not None and li.typ == NGLinkType.LIST):
i = (li.to_verb.begin_char if li.to is None else
li.to.source.begin_token.begin_char)
while i <= li.from0_.source.end_token.end_char:
po = i - begin
if (po >= 0 and (po < len(map0_))):
map0_[po] = (1)
i += 1
ss = Subsent()
has_verb = False
i = 0
first_pass3469 = True
while True:
if first_pass3469: first_pass3469 = False
else: i += 1
if (not (i < len(sent.items))): break
it = sent.items[i]
delim = False
if (it.typ == SentItemType.DELIM):
delim = True
elif (it.typ == SentItemType.CONJ
and map0_[it.begin_token.begin_char - begin] == (0)):
delim = True
if (it.source.typ == ConjunctionType.COMMA):
if (not has_verb):
delim = False
if (not delim):
if (it.typ == SentItemType.VERB):
has_verb = True
ss.items.append(it)
continue
if (len(ss.items) == 0):
ss.delims.append(it.source)
continue
if (len(ss.items) > 0):
res.append(ss)
ss = Subsent()
has_verb = False
ss.delims.append(it.source)
if (len(ss.items) > 0):
res.append(ss)
i = 0
first_pass3470 = True
while True:
if first_pass3470: first_pass3470 = False
else: i += 1
if (not (i < len(res))): break
r = res[i]
j = 0
if (r.check(DelimType.IF)):
has_then = False
has_else = False
j = (i + 1)
while j < len(res):
if (res[j].check(DelimType.THEN)):
if (has_then):
break
res[j].owner = r
res[j].question = "если"
res[j].typ = SemFraglinkType.IFTHEN
has_then = True
r.is_then_else_root = True
elif (res[j].check(DelimType.ELSE)):
if (has_else):
break
res[j].owner = r
res[j].question = "иначе"
res[j].typ = SemFraglinkType.IFELSE
has_else = True
r.is_then_else_root = True
elif (res[j].check(DelimType.IF)):
if (res[j].check(DelimType.AND)):
res[j].owner = r
else:
break
j += 1
if (not has_then and i > 0):
if (res[0].owner is None and res[0].only_conj()):
res[0].owner = r
res[0].question = "если"
r.is_then_else_root = True
res[0].typ = SemFraglinkType.IFTHEN
elif (res[0].owner is not None):
r.owner = res[0]
r.question = "если"
r.typ = SemFraglinkType.IFTHEN
continue
if (r.check(DelimType.BECAUSE)):
has_then = False
j = (i + 1)
while j < len(res):
if (res[j].check(DelimType.THEN)):
if (has_then):
break
res[j].owner = r
res[j].question = "по причине"
res[j].typ = SemFraglinkType.BECAUSE
has_then = True
r.is_then_else_root = True
j += 1
if (not has_then and i > 0):
if (res[0].owner is None and res[0].only_conj()):
res[0].owner = r
res[0].question = "по причине"
r.is_then_else_root = True
res[0].typ = SemFraglinkType.BECAUSE
continue
if (not has_then and ((i + 1) < len(res))):
if (res[i + 1].owner is None and res[i + 1].only_conj()):
res[i + 1].owner = r
res[i + 1].question = "по причине"
r.is_then_else_root = True
res[i + 1].typ = SemFraglinkType.BECAUSE
continue
continue
if (r.check(DelimType.BUT)):
if (i > 0):
if (res[i - 1].owner is None and res[i - 1].only_conj()):
res[i - 1].owner = r
res[i - 1].question = "но"
r.is_then_else_root = True
res[i - 1].typ = SemFraglinkType.BUT
continue
if (r.check(DelimType.WHAT)):
if (i > 0):
if (res[i - 1].owner is None and res[i - 1].only_conj()):
res[i - 1].owner = r
res[i - 1].question = "что"
r.is_then_else_root = True
res[i - 1].typ = SemFraglinkType.WHAT
continue
if (r.check(DelimType.FOR)):
if ((i + 1) < len(res)):
if (res[i + 1].owner is None and res[i + 1].only_conj()):
res[i + 1].owner = r
res[i + 1].question = "чтобы"
r.is_then_else_root = True
res[i + 1].typ = SemFraglinkType.FOR
continue
if (i > 0):
if (res[i - 1].owner is None and res[i - 1].only_conj()):
res[i - 1].owner = r
res[i - 1].question = "чтобы"
r.is_then_else_root = True
res[i - 1].typ = SemFraglinkType.FOR
continue
i = 1
first_pass3471 = True
while True:
if first_pass3471: first_pass3471 = False
else: i += 1
if (not (i < len(res))): break
r = res[i]
if (not r.check(DelimType.AND) or r.owner is not None):
continue
for j in range(i - 1, -1, -1):
rr = res[j]
if (rr.can_be_next_in_list(r)
and ((rr.owner is None or
((rr.owner_root is not None
and rr.owner_root.can_be_next_in_list(r)))))):
if (r.check_or()):
rr.is_or = True
rr.items.extend(r.items)
del res[i]
i -= 1
break
return res
