def get_bs_abbreviation_homonyms():
abbreviation_bs = get_string_list_from_file('out/Аббревиатура. БС.txt')
word_forms_bases = read_src_bs('src_dict/БС 13.03.21.txt')
title_forms = [str(x.title_word_form) for x in word_forms_bases]
bs_abbreviation_homonyms = []
for abbreviation in abbreviation_bs:
abbreviation_name = get_bs_title_word_form(abbreviation).name
title_form_list = []
for title_form in title_forms:
form_name = get_bs_title_word_form(title_form).name.replace(
'*', '')
if form_name == abbreviation_name.lower():
print(title_form)
title_form_list.append(title_form)
if title_form_list:
bs_abbreviation_homonyms.append(abbreviation)
bs_abbreviation_homonyms += title_form_list
bs_abbreviation_homonyms.append('')
save_list_to_file(bs_abbreviation_homonyms[:-1],
'out/Аббревиатура. БС. Омонимы.txt')
def get_capital_letter_bs():
capital_letter_bs = list(
get_string_list_from_file('out/Большая буква. БС.txt'))
word_forms_bases = read_src_bs('src_dict/БС 13.03.21.txt')
title_forms = [str(x.title_word_form) for x in word_forms_bases]
capital_letter_bs_homonyms = []
for capital_word in capital_letter_bs:
capital_word_name = get_bs_title_word_form(capital_word).name
title_form_list = []
for title_form in title_forms:
form_name = get_bs_title_word_form(title_form).name.replace(
'*', '')
if form_name == capital_word_name.lower():
print(title_form)
title_form_list.append(title_form)
if title_form_list:
capital_letter_bs_homonyms.append(capital_word)
capital_letter_bs_homonyms += title_form_list
capital_letter_bs_homonyms.append('')
save_list_to_file(capital_letter_bs_homonyms[:-1],
'out/Большая буква. БС. Омонимы.txt')
def ordinary_words_bs():
word_forms_bases = read_src_bs('src_dict/БС 02.03.21.txt')
bs_word_forms = [str(x.title_word_form) for x in word_forms_bases]
exclusion_list = []
# Многокорневые слова БС
multi_root_bs_forms = get_string_list_from_file(
'src_dict/Многокорневые слова БС.txt')
exclusion_list += multi_root_bs_forms
# Омонимы БС
homonyms_bs = get_string_list_from_file('src_dict/Омонимы БС.txt')
exclusion_list += homonyms_bs
# Повторы ост. совпадает с БС
remaining_repetitions = get_string_list_from_file(
'src_dict/Повторы ост. совпадает с БС.txt')
remaining_repetitions = [str(get_bs_title_word_form(x))
for x in remaining_repetitions]
exclusion_list += remaining_repetitions
# Обычные слова БС
ordinary_words_bs_list = []
for bs_str_form in bs_word_forms:
if bs_str_form not in exclusion_list:
ordinary_words_bs_list.append(bs_str_form)
# print(bs_str_form)
save_list_to_file(ordinary_words_bs_list, 'out/Обычные слова БС.txt')
def compare_homonyms_spec_note():
homonyms_bg = get_string_list_from_file('src_dict/Омонимы БГ.txt')
homonyms_bg_str_form = []
for homonyms in homonyms_bg:
socket_form = get_socket_word_form(homonyms)
spec_note = socket_form.spec_note.replace('< ', '')
spec_note_socket_form = get_socket_word_form(spec_note)
spec_note = ' '.join(
filter(None, [
spec_note_socket_form.invisible,
spec_note_socket_form.name,
spec_note_socket_form.root_index,
spec_note_socket_form.idf,
' '.join(spec_note_socket_form.info),
spec_note_socket_form.note,
]))
string_form = ' '.join(
filter(None, [
socket_form.name,
socket_form.idf,
' '.join(socket_form.info),
spec_note,
]))
homonyms_bg_str_form.append(string_form)
homonyms_spec_note = get_string_list_from_file(
'src_dict/О-мы БС спец. прим. не совпадают с Повторами.txt')
homonyms_spec_note_relevant = []
homonyms_spec_note_not_relevant = []
for homonym in homonyms_spec_note:
title_form = get_bs_title_word_form(homonym)
print(title_form.name, title_form.note)
string_form = ' '.join(
filter(None, [
title_form.name,
title_form.idf,
' '.join(title_form.info),
title_form.note.replace('.* < ', ''),
]))
if string_form in homonyms_bg_str_form:
homonyms_spec_note_relevant.append(homonym)
else:
homonyms_spec_note_not_relevant.append(homonym)
save_list_to_file(homonyms_spec_note_relevant,
'out/О-мы БС спец. прим. совпадают с О-мами БГ.txt')
save_list_to_file(homonyms_spec_note_not_relevant,
'out/О-мы БС спец. прим. не совпадают с О-мами БГ.txt')
def get_remaining_repetitions():
homonyms = []
for line in get_string_list_from_file(
'src_dict/О-мы БС спец. прим. совпадают с Повторами.txt'):
bs_form = get_bs_title_word_form(line)
homonyms.append(' '.join(filter(None, [
bs_form.name,
bs_form.idf,
' '.join(bs_form.info),
])))
for line in get_string_list_from_file(
'src_dict/О-мы БС совпадают с Повторами.txt'):
bs_form = get_bs_title_word_form(line)
homonyms.append(' '.join(filter(None, [
bs_form.name,
bs_form.idf,
' '.join(bs_form.info),
bs_form.note.replace('.*', '').strip()
])))
remaining_repetitions = []
replays_in_groups = get_string_list_from_file(
'out/Повторы в группах (без повторов).txt')
replays_in_groups = [
' '.join(filter(None, [
get_socket_word_form(x).name,
get_socket_word_form(x).idf,
' '.join(get_socket_word_form(x).info),
get_socket_word_form(x).note.replace('*', '').strip(),
]))
for x in replays_in_groups
]
for replay in replays_in_groups:
if replay not in homonyms:
remaining_repetitions.append(replay)
save_list_to_file(remaining_repetitions, 'out/Повторы ост.txt')
def get_capital_letter_bs():
capital_letter_bs = list(
get_string_list_from_file('src_dict/Большая буква. БС.txt'))
word_forms_bases = read_src_bs('src_dict/БС 09.03.21.txt')
title_forms = [str(x.title_word_form) for x in word_forms_bases]
capital_letter_bs_homonyms = []
for capital_word in capital_letter_bs:
capital_word_name = get_bs_title_word_form(capital_word).name
for title_form in title_forms:
form_name = get_bs_title_word_form(title_form).name.replace(
'*', '')
if form_name == capital_word_name.lower():
print(title_form)
capital_letter_bs_homonyms.append(title_form)
save_list_to_file(
sorted(capital_letter_bs_homonyms, key=lambda x: x.replace('*', '')),
'out/Большая буква. БС. Омонимы.txt')
def get_bs_abbreviation_homonyms():
abbreviation_bs = get_string_list_from_file(
'src_dict/Аббревиатура. БС.txt')
word_forms_bases = read_src_bs('src_dict/БС 09.03.21.txt')
title_forms = [str(x.title_word_form) for x in word_forms_bases]
bs_abbreviation_homonyms = []
for abbreviation in abbreviation_bs:
abbreviation_name = get_bs_title_word_form(abbreviation).name
for title_form in title_forms:
form_name = get_bs_title_word_form(title_form).name.replace(
'*', '')
if form_name == abbreviation_name.lower():
print(title_form)
bs_abbreviation_homonyms.append(title_form)
save_list_to_file(
sorted(bs_abbreviation_homonyms, key=lambda x: x.replace('*', '')),
'out/Аббревиатура. БС. Омонимы.txt')
def check_g58():
endings = ('греть', 'мять', 'оть', 'ыть', 'пеленать')
g58_list = get_string_list_from_file('src_dict/Г58 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g58_list]
out_list = []
for word_form in word_forms[:]:
if not word_form.name.endswith(endings):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г58 ещё изм.txt')
def check_socket_bs():
multi_root_words = get_dicts_from_csv_file('out/Многокорневые слова.csv')
multi_root_bs_forms = get_string_list_from_file(
'out/Многокорневые слова БС.txt')
multi_root_bs_forms = [
get_bs_title_word_form(x).name for x in multi_root_bs_forms
]
for multi_root_word in multi_root_words:
for root_index_key in list(multi_root_word)[1:]:
if multi_root_word[root_index_key]:
socket_form = get_socket_word_form(
multi_root_word[root_index_key])
if socket_form.name not in multi_root_bs_forms:
print(socket_form)
def check_g21():
endings = ('верещать', 'верещаться', 'дышать', 'дышаться', 'кишеть',
'кишеться', 'пищать', 'пищаться', 'слышать', 'слышаться',
'трещать', 'трещаться', 'ршать', 'ршаться', 'жить', 'житься',
'чить', 'читься', 'шить', 'шиться', 'щить', 'щиться')
g21_list = get_string_list_from_file('src_dict/Г21 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g21_list]
out_list = []
for word_form in word_forms[:]:
if not word_form.name.endswith(endings):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г21 ещё изм.txt')
def check_g15():
endings = ('бормотать', 'бормотаться', 'лепетать', 'лепетаться', 'плакать',
'плакаться', 'прятать', 'прятаться', 'скакать', 'скакаться',
'топтать', 'топтаться', 'шептать', 'шептаться', 'щебетать',
'щебетаться', 'готать', 'готаться', 'котать', 'котаться',
'потать', 'потаться', 'хотать', 'хотаться', 'хтать', 'хтаться')
g15_list = get_string_list_from_file('src_dict/Г15 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g15_list]
out_list = []
for word_form in word_forms[:]:
if not word_form.name.endswith(endings):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г15 ещё изм.txt')
def check_g6():
endings = ('нуть', 'нуться', 'зиждить', 'зиждиться', 'врать', 'враться',
'жрать', 'жраться', 'рвать', 'рваться', 'реветь', 'реветься',
'ржать', 'ржаться', 'сосать', 'сосаться', 'срать', 'сраться',
'стонать', 'стонаться', 'ткать', 'ткаться', 'шибить',
'шибиться')
g6_list = get_string_list_from_file('src_dict/Г6 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g6_list]
out_list = []
for word_form in word_forms[:]:
if not word_form.name.endswith(endings):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г6 ещё изм.txt')
def compare_homonyms():
homonyms_bg = get_string_list_from_file(
'src_dict/Омонимы БГ без этим. примечаний.txt')
homonyms_bg_str_form = []
for homonyms in homonyms_bg:
socket_form = get_socket_word_form(homonyms)
string_form = ' '.join(filter(
None, [
socket_form.name,
socket_form.idf,
' '.join(socket_form.info),
socket_form.note
]
))
homonyms_bg_str_form.append(string_form)
homonyms_bs = get_string_list_from_file(
'src_dict/О-мы БС не совпадают с Повторами.txt')
homonyms_bs_relevant = []
homonyms_bs_not_relevant = []
for homonym in homonyms_bs:
title_form = get_bs_title_word_form(homonym)
string_form = ' '.join(filter(
None,
[
title_form.name,
title_form.idf,
' '.join(title_form.info),
title_form.note.replace('.* ', '* '),
]
))
if string_form in homonyms_bg_str_form:
homonyms_bs_relevant.append(homonym)
else:
homonyms_bs_not_relevant.append(homonym)
save_list_to_file(homonyms_bs_relevant,
'out/О-мы БС совпадают с О-мами БГ.txt')
save_list_to_file(homonyms_bs_not_relevant,
'out/О-мы БС не совпадают с О-мами БГ.txt')
def remove_strings():
relevant_repetitions = get_string_list_from_file(
'src_dict/Повторы ост. совпадает с БС.txt')
multi_root_bs = list(get_string_list_from_file(
'src_dict/Многокорневые слова БС.txt'))
multi_root_bs = [get_bs_title_word_form(x) for x in multi_root_bs]
multi_root_bs = [
' '.join(filter(None, [
x.name,
x.idf,
' '.join(x.info),
x.note,
]))
for x in multi_root_bs
]
save_list_to_file(sorted(list(set(
relevant_repetitions) - set(multi_root_bs))),
'out/Повторы ост. совпадает с БС.txt')
def check_presence():
homonyms = [get_bs_title_word_form(x).name
for x in get_string_list_from_file('src_dict/Омонимы БС.txt')]
word_forms_bases = list(read_src_bs('src_dict/БС 16.03.21 изм.txt'))
bs_names = [
x.title_word_form.name.replace('*', '').strip()
for x in word_forms_bases
]
for file_path in Path('out/lst').glob('*'):
file_stem = file_path.stem
presence = [] # 28а. Если слово имеется в док-те Омонимы БС.txt
nouns = [] # 29а. если слово находится
absent = [] # 29б. если слово НЕ находится
words = get_string_list_from_file(file_path, encoding='cp1251')
for word in words:
print(word)
if word in homonyms:
presence.append(word)
else:
if word in bs_names:
for group in word_forms_bases:
if word == group.title_word_form.name:
print(group.title_word_form)
nouns.append(str(group.title_word_form))
else:
absent.append(word)
if presence:
save_list_to_file(
sorted(presence), f'out/homonyms/{file_stem} омонимы.txt')
if nouns:
save_list_to_file(sorted(nouns), f'out/C/С{file_stem}.txt')
if absent:
save_list_to_file(
sorted(absent), f'out/absent/{file_stem} отсутствует.txt')
def compare_replays_in_groups():
homonyms_bs = get_string_list_from_file('src_dict/Омонимы БС.txt')
bs_str_forms = []
for homonym in homonyms_bs:
title_form = get_bs_title_word_form(homonym)
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(title_form.info),
title_form.note.replace('.*', '').strip()
]))
bs_str_forms.append(string_form)
relevant = []
not_relevant = []
with open('src_dict/Повторы в группах.txt', encoding='utf-8') as f_in:
groups = (x.strip() for x in f_in.read().split('\n\n'))
for group in groups:
for line in group.split('\n')[1:]:
if not line.startswith('!'):
word_form = get_socket_word_form(line)
line_form = ' '.join(
filter(None, [
word_form.name,
word_form.idf,
' '.join(word_form.info),
word_form.note.replace('*', '').strip(),
]))
if line_form in bs_str_forms:
relevant.append(line)
else:
not_relevant.append(line)
save_list_to_file(
sorted(relevant, key=lambda x: x.replace('*', '').strip().lower()),
'out/19.1 Совпадающие.txt')
save_list_to_file(
sorted(not_relevant, key=lambda x: x.replace('*', '').strip().lower()),
'out/19.1 Не совпадающие.txt')
def check_socket_bs():
multi_root_words = get_dicts_from_csv_file('out/Многокорневые слова.csv')
multi_root_bs_forms = get_string_list_from_file(
'out/Многокорневые слова БС.txt')
multi_root_bs_forms = [
get_bs_title_word_form(x).name for x in multi_root_bs_forms
]
not_found_in_bs = []
for multi_root_word in multi_root_words:
for root_index_key in list(multi_root_word)[1:]:
if multi_root_word[root_index_key]:
socket_form = get_socket_word_form(
multi_root_word[root_index_key])
if socket_form.name not in multi_root_bs_forms:
print(socket_form)
not_found_in_bs.append(str(socket_form))
save_list_to_file(not_found_in_bs, 'out/Не найденные в БС.txt')
def check_g50_51():
endings = ('выкнуть', 'выкнуться', 'гибнуть', 'гибнуться', 'глохнуть',
'глохнуться', 'грязнуть', 'грязнуться', 'дрыхнуть',
'дрыхнуться', 'дрябнуть', 'дрябнуться', 'зябнуть', 'зябнуться',
'киснуть', 'киснуться', 'крепнуть', 'крепнуться', 'липнуть',
'липнуться', 'мерзнуть', 'мерзнуться', 'мокнуть', 'мокнуться',
'мякнуть', 'мякнуться', 'пухнуть', 'пухнуться', 'сохнуть',
'сохнуться', 'чахнуть', 'чахнуться')
g50_51_list = get_string_list_from_file('src_dict/Г50, Г51 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g50_51_list]
out_list = []
for word_form in word_forms[:]:
if not word_form.name.endswith(endings):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г50, Г51 ещё изм.txt')
def check_g53_56():
endings = ('шел', 'авать', 'ивать', 'увать', 'ывать', 'вевать', 'мевать',
'певать', 'ревать', 'севать', 'тевать', 'щевать', 'длевать',
'тлевать', 'одолевать', 'одолевать', 'разевать')
nes_endings = ('водить', 'возить', 'носить', 'ходить')
g53_56_list = get_string_list_from_file(
'src_dict/Г53, Г54, Г55, Г56 ещё.txt')
word_forms = [get_bs_title_word_form(x) for x in g53_56_list]
out_list = []
for word_form in word_forms[:]:
if not ('неп' in word_form.info or 'б' in word_form.info
or word_form.name.endswith(endings) or
('нес' in word_form.info
and word_form.name.endswith(nes_endings))):
print(word_form)
out_list.append(str(word_form))
save_list_to_file(sorted(out_list), 'out/Г53, Г54, Г55, Г56 ещё изм.txt')
def compare_dicts():
ordinary_words_bg_list = list(
get_string_list_from_file('out/Обычные слова БГ.txt'))
ordinary_words_bs_list = list(
get_string_list_from_file('out/Обычные слова БС.txt'))
bg_compare_forms = [
get_socket_word_form(x) for x in ordinary_words_bg_list
]
bg_compare_forms = [
' '.join(filter(None, [
x.name,
x.idf,
' '.join(x.info),
x.note[2:],
])) for x in bg_compare_forms
]
bs_compare_forms = [
get_bs_title_word_form(x) for x in ordinary_words_bs_list
]
bs_compare_forms = [
' '.join(filter(None, [
x.name,
x.idf,
' '.join(x.info),
x.note[3:],
])) for x in bs_compare_forms
]
intersection = list(set(bg_compare_forms) & set(bs_compare_forms))
matching_lines = []
bg_unique = []
bs_unique = []
for bg_string in ordinary_words_bg_list:
bg_form = get_socket_word_form(bg_string)
compare_form = ' '.join(
filter(None, [
bg_form.name,
bg_form.idf,
' '.join(bg_form.info),
bg_form.note[2:],
]))
if compare_form in intersection:
matching_lines.append(bg_string)
else:
bg_unique.append(bg_string)
for bs_string in ordinary_words_bs_list:
bs_form = get_bs_title_word_form(bs_string)
compare_form = ' '.join(
filter(None, [
bs_form.name, bs_form.idf, ' '.join(bs_form.info),
bs_form.note[3:]
]))
if compare_form not in intersection:
matching_lines.append(bs_unique)
save_list_to_file(matching_lines, 'out/Строки совпадают.txt')
save_list_to_file(bg_unique, 'out/Уникальные строки БГ.txt')
save_list_to_file(bs_unique, 'out/Уникальные строки БС.txt')
def get_diff_lists():
ordinary_words_bs_list = get_string_list_from_file(
'out/Обычные слова БС.txt')
ordinary_words_bs_list = [
get_bs_title_word_form(x) for x in ordinary_words_bs_list
]
ordinary_words_bs_list = [
' '.join(
filter(None, [
x.name,
x.idf,
' '.join(x.info),
x.note.replace('.*', '').strip(),
])) for x in ordinary_words_bs_list
]
ordinary_words_bg_list = get_string_list_from_file(
'out/Обычные слова БГ.txt')
ordinary_words_bg_list = [
get_socket_word_form(x) for x in ordinary_words_bg_list
]
ordinary_words_bg_list = [
' '.join(filter(None, [
x.name,
x.idf,
' '.join(x.info),
x.note[2:],
])) for x in ordinary_words_bg_list
]
matching_src = list(
set(ordinary_words_bs_list)
& set(ordinary_words_bg_list))
unique_src_bs = list(
set(ordinary_words_bs_list) - set(ordinary_words_bg_list))
unique_src_bg = list(
set(ordinary_words_bg_list) - set(ordinary_words_bs_list))
matching_forms = []
unique_bs_forms = []
unique_bg_forms = []
for word in ordinary_words_bs_list:
title_form = get_bs_title_word_form(word)
str_form = ' '.join(
filter(None, [
title_form.name,
title_form.idf,
' '.join(title_form.info),
title_form.note.replace('.*', '').strip(),
]))
if str_form in matching_src:
matching_forms.append(title_form)
elif str_form in unique_src_bs:
unique_bs_forms.append(title_form)
for word in ordinary_words_bg_list:
socket_form = get_socket_word_form(word)
str_form = ' '.join(
filter(None, [
socket_form.name,
socket_form.idf,
' '.join(socket_form.info),
socket_form.note[2:],
]))
if str_form in unique_src_bg:
unique_bg_forms.append(socket_form)
save_list_to_file(matching_forms, 'out/Строки совпадают.txt')
save_list_to_file(unique_bs_forms, 'out/Уникальные строки БС.txt')
save_list_to_file(unique_bg_forms, 'out/Уникальные строки БГ.txt')
def check_presence():
homonyms = [
get_bs_title_word_form(x)
for x in get_string_list_from_file('src_dict/Омонимы БС.txt')
]
homonyms = [
' '.join(
filter(None,
[x.name, ' '.join([y
for y in x.info if y in IDENTIFIERS])]))
for x in homonyms
]
word_forms_bases = list(read_src_bs('src_dict/БС 27.03.21.txt'))
bs_names = [
' '.join(
filter(None, [
x.title_word_form.name, ' '.join(
[y for y in x.title_word_form.info if y in IDENTIFIERS])
])) for x in word_forms_bases
]
for file_path in Path('out/lst').glob('*'):
file_stem = file_path.stem
presence = [] # 28а. Если слово имеется в док-те Омонимы БС.txt
nouns = [] # 29а. если слово находится
absent = [] # 29б. если слово НЕ находится
words = get_string_list_from_file(file_path)
for word in words:
clear_word = word.split('.*')[0]
clear_word = ' '.join([x for x in clear_word.split() if x])
print(word)
if clear_word in homonyms:
presence.append(word)
else:
if clear_word in bs_names:
for group in word_forms_bases:
if clear_word == ' '.join(
filter(None, [
group.title_word_form.name, ' '.join([
y for y in group.title_word_form.info
if y in IDENTIFIERS
])
])):
nouns.append(str(group.title_word_form))
else:
absent.append(word)
if presence:
save_list_to_file(
sorted(presence,
key=lambda x: x.replace('*', '').strip().lower()),
f'out/homonyms/{file_stem} омонимы.txt')
if nouns:
save_list_to_file(
sorted(nouns,
key=lambda x: x.replace('*', '').strip().lower()),
f'out/verbs/Г{file_stem}.txt')
if absent:
save_list_to_file(
sorted(absent,
key=lambda x: x.replace('*', '').strip().lower()),
f'out/absent/{file_stem} отсутствует.txt')
def get_homonyms_bs():
homonyms_bs = get_string_list_from_file('src_dict/Омонимы БС.txt')
replays_in_groups_spec_note = [] # (".* <")
replays_in_groups = [] # (".*")
with open('src_dict/Повторы в группах.txt', encoding='utf-8') as f_in:
groups = (x.strip() for x in f_in.read().split('\n\n'))
for group in groups:
for line in group.split('\n')[1:]:
if line.startswith('!'):
title_word_sub_group = line
title_name_sub_group = get_socket_word_form(
title_word_sub_group.replace('!', '').strip()).name
else:
word_form = get_socket_word_form(line)
replays_in_groups_spec_note.append(' '.join(
filter(None, [
word_form.name,
word_form.idf,
' '.join(word_form.info),
title_name_sub_group,
])))
replays_in_groups.append(' '.join(
filter(None, [
word_form.name,
word_form.idf,
' '.join(word_form.info),
word_form.note.replace('*', '').strip(),
])))
homonyms_spec_note_relevant = []
homonyms_spec_note_not_relevant = []
homonyms_relevant = []
homonyms_not_relevant = []
for homonym in homonyms_bs:
title_form = get_bs_title_word_form(homonym)
# имеющие специальное примечание (".* <")
if title_form.note.startswith('.* <'):
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(title_form.info),
title_form.note.replace('.* <', '').strip()
]))
if string_form in replays_in_groups_spec_note:
homonyms_spec_note_relevant.append(homonym)
else:
homonyms_spec_note_not_relevant.append(homonym)
# НЕ имеющие специальное примечание (".* <")
else:
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(title_form.info),
title_form.note.replace('.*', '').strip()
]))
if string_form in replays_in_groups:
homonyms_relevant.append(homonym)
else:
homonyms_not_relevant.append(homonym)
save_list_to_file(homonyms_spec_note_relevant,
'out/О-мы БС спец. прим. совпадают с Повторами.txt')
save_list_to_file(homonyms_spec_note_not_relevant,
'out/О-мы БС спец. прим. не совпадают с Повторами.txt')
save_list_to_file(homonyms_relevant,
'out/О-мы БС совпадают с Повторами.txt')
save_list_to_file(homonyms_not_relevant,
'out/О-мы БС не совпадают с Повторами.txt')
def get_homonyms_bs():
homonymous_repetitions = list(
get_string_list_from_file(
'src_dict/Слова, омонимичные повторам в группе.txt'))
homonyms_bs = get_string_list_from_file('src_dict/Омонимы БС.txt')
homonyms_spec_note_relevant = []
homonyms_spec_note_not_relevant = []
homonyms_relevant = []
homonyms_not_relevant = []
for homonym in homonyms_bs:
title_form = get_bs_title_word_form(homonym)
# без примечаний
if not title_form.note:
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(title_form.info),
title_form.note.replace('.*', '').strip()
]))
if string_form in homonymous_repetitions:
homonyms_relevant.append(homonym)
else:
homonyms_not_relevant.append(homonym)
else:
# имеющие специальное примечание (".* <")
if title_form.note.startswith('.* <'):
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(
title_form.info),
title_form.note.replace('.* <', '<').strip()
]))
if string_form in homonymous_repetitions:
homonyms_spec_note_relevant.append(homonym)
else:
homonyms_spec_note_not_relevant.append(homonym)
# НЕ имеющие специальное примечание (".* <")
else:
string_form = ' '.join(
filter(None, [
title_form.name, title_form.idf, ' '.join(
title_form.info),
title_form.note.replace('.*', '*').strip()
]))
if string_form in homonymous_repetitions:
homonyms_relevant.append(homonym)
else:
homonyms_not_relevant.append(homonym)
save_list_to_file(
homonyms_spec_note_relevant,
'out/О-мы БС спец. прим. совпадают с Слова, омонимичные повторам.txt')
save_list_to_file(
homonyms_spec_note_not_relevant,
'out/О-мы БС спец. прим. не совпадают с Слова, омонимичные повторам.txt'
)
save_list_to_file(
homonyms_relevant,
'out/О-мы БС совпадают с Слова, омонимичные повторам.txt')
save_list_to_file(
homonyms_not_relevant,
'out/О-мы БС не совпадают с Слова, омонимичные повторам.txt')
