def createNumber(owner : 'FragToken', itok : 'InstrToken1') -> None:
""" Создать результирующий узел, представляющий номер
Args:
owner(FragToken):
itok(InstrToken1):
"""
from pullenti.ner.instrument.internal.FragToken import FragToken
if (itok.num_begin_token is None or itok.num_end_token is None):
return
num = FragToken._new1456(itok.num_begin_token, itok.num_end_token, InstrumentKind.NUMBER, True, itok)
owner.children.append(num)
if (itok.num_typ == NumberTypes.TWODIGITS):
owner.number = itok.first_number
owner.sub_number = itok.last_number
elif (itok.num_typ == NumberTypes.THREEDIGITS):
owner.number = itok.first_number
owner.sub_number = itok.middle_number
owner.sub_number2 = itok.last_number
elif (itok.num_typ == NumberTypes.FOURDIGITS and len(itok.numbers) == 4):
owner.number = itok.first_number
owner.sub_number = PartToken.getNumber(itok.numbers[1])
owner.sub_number2 = PartToken.getNumber(itok.numbers[2])
owner.sub_number3 = itok.last_number
else:
owner.number = itok.last_number
owner.min_number = itok.last_min_number
owner._itok = itok
def process(self, kit: 'AnalysisKit') -> None:
from pullenti.ner.instrument.internal.FragToken import FragToken
t = kit.first_token
t1 = t
if (t is None):
return
dfr = FragToken.createDocument(t, 0, InstrumentKind.UNDEFINED)
if (dfr is None):
return
ad = kit.getAnalyzerData(self)
res = dfr.createReferent(ad)
def create_number(owner : 'FragToken', itok : 'InstrToken1') -> None:
if (itok.num_begin_token is None or itok.num_end_token is None):
return
num = FragToken._new1562(itok.num_begin_token, itok.num_end_token, InstrumentKind.NUMBER, True, itok)
owner.children.append(num)
if (itok.num_typ == NumberTypes.TWODIGITS):
owner.number = itok.first_number
owner.sub_number = itok.last_number
elif (itok.num_typ == NumberTypes.THREEDIGITS):
owner.number = itok.first_number
owner.sub_number = itok.middle_number
owner.sub_number2 = itok.last_number
elif (itok.num_typ == NumberTypes.FOURDIGITS and len(itok.numbers) == 4):
owner.number = itok.first_number
owner.sub_number = PartToken.get_number(itok.numbers[1])
owner.sub_number2 = PartToken.get_number(itok.numbers[2])
owner.sub_number3 = itok.last_number
else:
owner.number = itok.last_number
owner.min_number = itok.last_min_number
owner._itok = itok
def __analize_list_items(chi : typing.List['FragToken'], ind : int) -> int:
if (ind >= len(chi)):
return -1
res = chi[ind]
ki = res.kind
if (((ki == InstrumentKind.CHAPTER or ki == InstrumentKind.CLAUSE or ki == InstrumentKind.CONTENT) or ki == InstrumentKind.ITEM or ki == InstrumentKind.SUBITEM) or ki == InstrumentKind.CLAUSEPART or ki == InstrumentKind.INDENTION):
pass
else:
return -1
if (res.has_changes and res.multiline_changes_value is not None):
ci = res.multiline_changes_value
cit = FragToken._new1340(ci.begin_token, ci.end_token, InstrumentKind.CITATION)
res.children.append(cit)
if (BracketHelper.is_bracket(cit.begin_token.previous, True)):
cit.begin_token = cit.begin_token.previous
if (BracketHelper.is_bracket(cit.end_token.next0_, True)):
cit.end_token = cit.end_token.next0_
if (cit.end_token.next0_ is not None and cit.end_token.next0_.is_char_of(";.")):
cit.end_token = cit.end_token.next0_
res.fill_by_content_children()
if (res.children[0].has_changes):
pass
cit_kind = InstrumentKind.UNDEFINED
if (isinstance(ci.tag, DecreeChangeReferent)):
dcr = Utils.asObjectOrNull(ci.tag, DecreeChangeReferent)
if (dcr.value is not None and len(dcr.value.new_items) > 0):
mnem = dcr.value.new_items[0]
i = 0
i = mnem.find(' ')
if (((i)) > 0):
mnem = mnem[0:0+i]
cit_kind = PartToken._get_instr_kind_by_typ(PartToken._get_type_by_attr_name(mnem))
elif (len(dcr.owners) > 0 and (isinstance(dcr.owners[0], DecreePartReferent)) and dcr.kind == DecreeChangeKind.NEW):
pat = Utils.asObjectOrNull(dcr.owners[0], DecreePartReferent)
min0_ = 0
for s in pat.slots:
ty = PartToken._get_type_by_attr_name(s.type_name)
if (ty == PartToken.ItemType.UNDEFINED):
continue
l_ = PartToken._get_rank(ty)
if (l_ == 0):
continue
if (l_ > min0_ or min0_ == 0):
min0_ = l_
cit_kind = PartToken._get_instr_kind_by_typ(ty)
sub = None
if (cit_kind != InstrumentKind.UNDEFINED and cit_kind != InstrumentKind.APPENDIX):
sub = FragToken(ci.begin_token, ci.end_token)
wr = ContentAnalyzeWhapper()
wr.analyze(sub, None, True, cit_kind)
sub.kind = InstrumentKind.CONTENT
else:
sub = FragToken.create_document(ci.begin_token, ci.end_char, cit_kind)
if (sub is None or len(sub.children) == 0):
pass
elif ((sub.kind == InstrumentKind.CONTENT and len(sub.children) > 0 and sub.children[0].begin_token == sub.begin_token) and sub.children[len(sub.children) - 1].end_token == sub.end_token):
cit.children.extend(sub.children)
else:
cit.children.append(sub)
return 1
end_char = res.end_char
if (res._itok is None):
res._itok = InstrToken1.parse(res.begin_token, True, None, 0, None, False, res.end_char, False, False)
lines = ListHelper.LineToken.parse_list(res.begin_token, end_char, None)
if (lines is None or (len(lines) < 1)):
return -1
ret = 1
if (res.kind == InstrumentKind.CONTENT):
j = ind + 1
while j < len(chi):
if (chi[j].kind == InstrumentKind.CONTENT):
lines2 = ListHelper.LineToken.parse_list(chi[j].begin_token, chi[j].end_char, lines[len(lines) - 1])
if (lines2 is None or (len(lines2) < 1)):
break
if (not lines2[0].is_list_item):
if ((len(lines2) > 1 and lines2[1].is_list_item and lines2[0].end_token.is_char_of(":")) and not lines2[0].begin_token.chars.is_capital_upper):
lines2[0].is_list_item = True
else:
break
lines.extend(lines2)
ret = ((j - ind) + 1)
elif (chi[j].kind != InstrumentKind.EDITIONS and chi[j].kind != InstrumentKind.COMMENT):
break
j += 1
if (len(lines) < 2):
return -1
if ((len(lines) > 1 and lines[0].is_list_item and lines[1].is_list_item) and lines[0].number != 1):
if (len(lines) == 2 or not lines[2].is_list_item):
lines[1].is_list_item = False
lines[0].is_list_item = lines[1].is_list_item
i = 0
first_pass3276 = True
while True:
if first_pass3276: first_pass3276 = False
else: i += 1
if (not (i < len(lines))): break
if (lines[i].is_list_item):
if (i > 0 and lines[i - 1].is_list_item):
continue
if (((i + 1) < len(lines)) and lines[i + 1].is_list_item):
pass
else:
lines[i].is_list_item = False
continue
j = 0
new_line = False
j = (i + 1)
while j < len(lines):
if (not lines[j].is_list_item):
break
elif (lines[j].is_newline_before):
new_line = True
j += 1
if (new_line):
continue
if (i > 0 and lines[i - 1].end_token.is_char(':')):
continue
j = i
while j < len(lines):
if (not lines[j].is_list_item):
break
else:
lines[j].is_list_item = False
j += 1
if (len(lines) > 2):
last = lines[len(lines) - 1]
last2 = lines[len(lines) - 2]
if ((not last.is_list_item and last.end_token.is_char('.') and last2.is_list_item) and last2.end_token.is_char(';')):
if ((last.length_char < (last2.length_char * 2)) or last.begin_token.chars.is_all_lower):
last.is_list_item = True
i = 0
while i < (len(lines) - 1):
if (not lines[i].is_list_item and not lines[i + 1].is_list_item):
if (((i + 2) < len(lines)) and lines[i + 2].is_list_item and lines[i + 1].end_token.is_char(':')):
pass
else:
lines[i].end_token = lines[i + 1].end_token
del lines[i + 1]
i -= 1
i += 1
i = 0
while i < (len(lines) - 1):
if (lines[i].is_list_item):
if (lines[i].number == 1):
ok = True
num = 1
nonum = 0
j = i + 1
while j < len(lines):
if (not lines[j].is_list_item):
ok = False
break
elif (lines[j].number > 0):
num += 1
if (lines[j].number != num):
ok = False
break
else:
nonum += 1
j += 1
if (not ok or nonum == 0 or (num < 2)):
break
lt = lines[i]
j = i + 1
while j < len(lines):
if (lines[j].number > 0):
lt = lines[j]
else:
chli = Utils.asObjectOrNull(lt.tag, list)
if (chli is None):
chli = list()
lt.tag = (chli)
lt.end_token = lines[j].end_token
chli.append(lines[j])
del lines[j]
j -= 1
j += 1
i += 1
cou = 0
for li in lines:
if (li.is_list_item):
cou += 1
if (cou < 2):
return -1
i = 0
first_pass3277 = True
while True:
if first_pass3277: first_pass3277 = False
else: i += 1
if (not (i < len(lines))): break
if (lines[i].is_list_item):
i0 = i
ok = True
cou = 1
while i < len(lines):
if (not lines[i].is_list_item):
break
elif (lines[i].number != cou):
ok = False
i += 1; cou += 1
if (not ok):
i = i0
while i < len(lines):
if (not lines[i].is_list_item):
break
else:
lines[i].number = 0
i += 1
if (cou > 3 and lines[i0].begin_token.get_source_text() != lines[i0 + 1].begin_token.get_source_text() and lines[i0 + 1].begin_token.get_source_text() == lines[i0 + 2].begin_token.get_source_text()):
pref = lines[i0 + 1].begin_token.get_source_text()
ok = True
j = i0 + 2
while j < i:
if (pref != lines[j].begin_token.get_source_text()):
ok = False
break
j += 1
if (not ok):
continue
tt = None
ok = False
tt = lines[i0].end_token.previous
while tt is not None and tt != lines[i0].begin_token:
if (tt.get_source_text() == pref):
ok = True
break
tt = tt.previous
if (ok):
li0 = ListHelper.LineToken(lines[i0].begin_token, tt.previous)
lines[i0].begin_token = tt
lines.insert(i0, li0)
i += 1
for li in lines:
li.correct_begin_token()
ch = FragToken._new1357(li.begin_token, li.end_token, (InstrumentKind.LISTITEM if li.is_list_item else InstrumentKind.CONTENT), li.number)
if (ch.kind == InstrumentKind.CONTENT and ch.end_token.is_char(':')):
ch.kind = InstrumentKind.LISTHEAD
res.children.append(ch)
chli = Utils.asObjectOrNull(li.tag, list)
if (chli is not None):
for lt in chli:
ch.children.append(FragToken._new1340(lt.begin_token, lt.end_token, InstrumentKind.LISTITEM))
if (ch.begin_char < ch.children[0].begin_char):
ch.children.insert(0, FragToken._new1340(ch.begin_token, ch.children[0].begin_token.previous, InstrumentKind.CONTENT))
return ret
def analyze(res : 'FragToken') -> None:
if (res.number == 4):
pass
if (len(res.children) == 0):
ki = res.kind
if (((ki == InstrumentKind.CHAPTER or ki == InstrumentKind.CLAUSE or ki == InstrumentKind.CONTENT) or ki == InstrumentKind.ITEM or ki == InstrumentKind.SUBITEM) or ki == InstrumentKind.CLAUSEPART or ki == InstrumentKind.INDENTION):
tmp = list()
tmp.append(res)
ListHelper.__analize_list_items(tmp, 0)
return
if (res.kind == InstrumentKind.CLAUSE and res.number == 12):
pass
i = 0
first_pass3273 = True
while True:
if first_pass3273: first_pass3273 = False
else: i += 1
if (not (i < len(res.children))): break
if (res.children[i].kind == InstrumentKind.INDENTION and ((res.children[i].end_token.is_char_of(":;") or ((((i + 1) < len(res.children)) and res.children[i + 1].kind == InstrumentKind.EDITIONS and res.children[i + 1].end_token.is_char_of(":;")))))):
j = 0
cou = 1
list_bullet = chr(0)
j = (i + 1)
first_pass3274 = True
while True:
if first_pass3274: first_pass3274 = False
else: j += 1
if (not (j < len(res.children))): break
ch = res.children[j]
if (ch.kind == InstrumentKind.COMMENT or ch.kind == InstrumentKind.EDITIONS):
continue
if (ch.kind != InstrumentKind.INDENTION):
break
if (ch.end_token.is_char_of(";") or ((((j + 1) < len(res.children)) and res.children[j + 1].kind == InstrumentKind.EDITIONS and res.children[j + 1].end_token.is_char(';')))):
cou += 1
if ((isinstance(ch.begin_token, TextToken)) and not ch.chars.is_letter):
list_bullet = ch.kit.get_text_character(ch.begin_char)
continue
if (ch.end_token.is_char_of(".")):
cou += 1
j += 1
break
if (ch.end_token.is_char_of(":")):
if ((ord(list_bullet)) != 0 and ch.begin_token.is_char(list_bullet)):
tt = ch.begin_token.next0_
while tt is not None and (tt.end_char < ch.end_char):
if (tt.previous.is_char('.') and MiscHelper.can_be_start_of_sentence(tt)):
ch2 = FragToken._new1357(tt, ch.end_token, InstrumentKind.INDENTION, ch.number)
ch.end_token = tt.previous
res.children.insert(j + 1, ch2)
k = j + 1
while k < len(res.children):
if (res.children[k].kind == InstrumentKind.INDENTION):
res.children[k].number += 1
k += 1
cou += 1
j += 1
break
tt = tt.next0_
break
cou += 1
j += 1
break
if (cou < 3):
i = j
continue
if ((i > 0 and not res.children[i].end_token.is_char(':') and res.children[i - 1].kind2 == InstrumentKind.UNDEFINED) and res.children[i - 1].end_token.is_char(':')):
res.children[i - 1].kind2 = InstrumentKind.LISTHEAD
first_pass3275 = True
while True:
if first_pass3275: first_pass3275 = False
else: i += 1
if (not (i < j)): break
ch = res.children[i]
if (ch.kind != InstrumentKind.INDENTION):
continue
if (ch.end_token.is_char(':')):
ch.kind2 = InstrumentKind.LISTHEAD
elif (((i + 1) < j) and res.children[i + 1].kind == InstrumentKind.EDITIONS and res.children[i + 1].end_token.is_char(':')):
ch.kind2 = InstrumentKind.LISTHEAD
else:
ch.kind2 = InstrumentKind.LISTITEM
changed = list()
i = 0
while i < len(res.children):
if (res.number == 7):
pass
if (len(res.children[i].children) > 0):
ListHelper.analyze(res.children[i])
else:
co = ListHelper.__analize_list_items(res.children, i)
if (co > 0):
changed.append(res.children[i])
if (co > 1):
del res.children[i + 1:i + 1+co - 1]
i += (co - 1)
i += 1
for i in range(len(changed) - 1, -1, -1):
if (changed[i].kind == InstrumentKind.CONTENT):
j = Utils.indexOfList(res.children, changed[i], 0)
if (j < 0):
continue
del res.children[j]
res.children[j:j] = changed[i].children