def get_singles_bases():
src_group_word_form_list = get_string_list_from_file(
'src_dict/Добавить одиночки в БС.txt')
for src_title_word_form in src_group_word_form_list:
title_word_form = TitleWordForm(src_title_word_form, '', [], '')
group_word_form = GroupWordForm(title_word_form, [])
yield group_word_form
def get_group_word_form(src_dict: dict) -> GroupWordForm:
name = src_dict['name']
info = [src_dict['Inf_0']]
if src_dict['Inf_1']:
info.append(''.join(list(filter(None, [
src_dict['Inf_1'],
src_dict['Inf_2'],
src_dict['Inf_3']]))))
if src_dict['Inf_4']:
info.append(''.join(list(filter(None, [
src_dict['Inf_4'],
src_dict['Inf_5'],
src_dict['Inf_6']]))))
word_forms = []
if src_dict['Inf_1']:
singl_word_forms = get_singl_word_forms(src_dict)
word_forms += singl_word_forms
if src_dict['Inf_4']:
plurl_word_forms = get_plurl_word_forms(
src_dict, singl_word_forms)
word_forms += plurl_word_forms
elif src_dict['Inf_4']:
plurl_word_forms = get_plurl_word_forms(src_dict, [])
word_forms += plurl_word_forms
pprint(word_forms)
title_word_form = TitleWordForm(name, word_forms[0].idf, info, '')
return GroupWordForm(title_word_form, word_forms[1:])
def add_groups_to_bs():
"""
2 (продолжение).
Соблюдая алфавитный порядок ЗС в док-те БС 15.01.21.txt,
добавить в док-т БС 15.01.21.txt
одиночки из док-та Добавить одиночки в БС 2.txt .
4. Соблюдая алфавитный порядок ЗС в БС, добавить в БС группы
из док-тов Добавить группы в БС. Сущ-ные изм.txt
и Добавить группы в БС. Сущ-ные 2.txt .
"""
word_forms_bases = list(read_src_bs('src_dict/БС 15.01.21.txt'))
loners = get_string_list_from_file(
'src_dict/Добавить одиночки в БС 2.txt')
loners_list = []
for loner in loners:
title_word_form = TitleWordForm(loner, '', [], '')
group_word_form = GroupWordForm(title_word_form, [])
loners_list.append(group_word_form)
modified_nouns = list(read_src_bs(
'src_dict/Добавить группы в БС. Сущ-ные изм.txt'))
nouns_2 = list(read_src_bs('src_dict/Добавить группы в БС. Сущ-ные 2.txt',
encoding='utf-8'))
word_forms_bases += loners_list
word_forms_bases += modified_nouns
word_forms_bases += nouns_2
save_bs_dicts_to_txt(sorted(word_forms_bases), 'out/БС 15.01.21.txt')
def get_socket_group_word_form(src_dict: dict) -> GroupWordForm:
name = src_dict['name']
info = [
src_dict['Inf_0'], ' '.join(
list(
filter(None, [
src_dict['Inf_1'],
src_dict['Inf_2'],
src_dict['Inf_3'],
src_dict['Inf_4'],
src_dict['Inf_5'],
])))
]
word_forms = []
adjectives_full_forms = get_full_forms(src_dict)
word_forms += adjectives_full_forms
if src_dict['Inf_1']:
adjectives_short_forms = get_short_forms(src_dict)
word_forms += adjectives_short_forms
if src_dict['Inf_2']:
adjectives_comparative_forms = get_comparative_forms(src_dict)
word_forms += adjectives_comparative_forms
if src_dict['Inf_3']:
adjectives_superlative_forms = get_superlative_forms(src_dict)
word_forms += adjectives_superlative_forms
title_word_form = TitleWordForm(name, word_forms[0].idf, info, '')
return GroupWordForm(title_word_form, word_forms[1:])
def get_bs_title_word_form(src_title_word_form):
if '.*' in src_title_word_form:
src_title_word_form_w_note, src_note = [
x.strip() for x in src_title_word_form.split(' .* ')]
note = ' '.join(['.*', src_note])
else:
src_title_word_form_w_note = src_title_word_form
note = ''
name, *idf_info, = src_title_word_form_w_note.split()
if idf_info:
idf, *info = idf_info
else:
idf, *info = '', ''
return TitleWordForm(name, idf, info, note)
def get_group_word_form(src_dict: dict) -> GroupWordForm:
name = src_dict['name']
info = [src_dict['Inf_0'], src_dict['Inf_1']]
info += list(
filter(None, [
src_dict['Inf_2'],
src_dict['Inf_3'],
src_dict['Inf_4'],
src_dict['Inf_5'],
src_dict['Inf_6'],
src_dict['Inf_7'],
src_dict['Inf_8'],
src_dict['Inf_9'],
src_dict['Inf_10'],
src_dict['Inf_11'],
src_dict['Inf_12'],
]))
title_word_form = TitleWordForm(name, '.ГИ', info, '')
word_forms = []
if src_dict['Inf_3']:
present_future_forms = get_present_future_forms(src_dict)
word_forms += present_future_forms
if src_dict['Inf_4']:
past_tense_forms = get_past_tense_forms(src_dict)
word_forms += past_tense_forms
if src_dict['Inf_5']:
imperative_mood_forms = get_imperative_mood_forms(src_dict)
word_forms += imperative_mood_forms
if src_dict['Inf_6']:
joint_action_forms = get_joint_action_forms(src_dict)
word_forms += joint_action_forms
if src_dict['Inf_7']:
present_participle_is_valid_forms = get_present_participle_is_valid(
src_dict)
word_forms += present_participle_is_valid_forms
if src_dict['Inf_8']:
passive_present_participle_forms = get_passive_present_participle(
src_dict)
word_forms += passive_present_participle_forms
if src_dict['Inf_9']:
past_participle_is_valid_forms = get_past_participle_is_valid(src_dict)
word_forms += past_participle_is_valid_forms
if src_dict['Inf_10']:
passive_past_participle_forms = get_passive_past_participle(src_dict)
word_forms += passive_past_participle_forms
if src_dict['Inf_11']:
present_participle_forms = get_present_participle(src_dict)
word_forms += present_participle_forms
if src_dict['Inf_12']:
past_participle_forms = get_past_participle(src_dict)
word_forms += past_participle_forms
return GroupWordForm(title_word_form, word_forms)