),
modifiers.str_replace('||', 'П', target_field='title'),
modifiers.re_search_ban(r'[^-ЁёА-Яа-я]', target_field='title'),
modifiers.normalize_title(),
# Text OCR problems
# modifiers.translate(
# '?~[]{}',
# 'ё-()()',
# '|*o' # the o is latin
# ),
# modifiers.re_replace(r'знай\.', 'знач.'),
# modifiers.str_replace('3а', 'За'),
# modifiers.re_replace(r'(?<={alph})\d+(-\d+)?'.format(alph=modifiers.ALPH_RE), ''),
modifiers.str_replace(r' :', ':'),
modifiers.re_replace(r'([,:])(?=[^ 0-9])', r'\1 '),
# Text quality heuristics
modifiers.re_replace(r'\s+', ' '),
modifiers.re_replace(r' *[вк]о? *(\d|I)+( *, *(\d|I)+)*( *и *(\d|I)+)? *знач[,.]?', '', re.IGNORECASE),
modifiers.re_replace(r' *см\. *\S+((, ?| и )\S+)*', ''),
modifiers.re_replace(r'N((\d+)/)*(\d)+', ''),
modifiers.re_replace('<=', ''),
modifiers.re_replace('==', ''),
modifiers.re_replace('\|', ''),
modifiers.re_replace('Anti', 'противоположность'),
__author__ = 'moskupols'
import os
from hb_res.storage import get_storage, FileExplanationStorage
from preparation import modifiers
from preparation.resources.Resource import gen_resource, applied_modifiers
CUR_DIR = os.path.dirname(os.path.abspath(__file__))
INPUT_PATH = os.path.join(CUR_DIR, 'Selected.asset')
OUTPUT_PATH = os.path.join(CUR_DIR, 'SelectedAfterMissedModifiers.asset')
missed_modifiers = [
modifiers.str_replace('p', 'р'),
modifiers.re_replace(r'\s+', ' '),
modifiers.re_replace(r'([,:])(?=[^ ])', r'\1 '),
modifiers.str_replace(r' :', ':'),
modifiers.str_replace(r' ,', ','),
]
with FileExplanationStorage(INPUT_PATH) as inp:
PatchedResource = gen_resource('SelectedResource',
missed_modifiers)(inp.entries)
with FileExplanationStorage(OUTPUT_PATH) as outp:
outp.clear()
for e in applied_modifiers(PatchedResource()):
outp.add_entry(e)
import itertools
from preparation.resources.Resource import gen_resource
from preparation import modifiers
from hb_res.explanations.Explanation import Explanation
sample_modifiers = (
modifiers.str_replace('?', 'ё'),
modifiers.shadow_cognates(length_threshold=3, sep_re='(\\s|!|\\.)+'),
modifiers.re_replace('\\.', ' '),
modifiers.normalize_title(),
modifiers.calculate_key(),
)
@gen_resource('SampleResource', sample_modifiers)
def sample_parser():
raw_expls = itertools.starmap(Explanation, (
('ПОРА', 'Однажды.в.студ?ную.зимнюю.пору.я.из.лесу.вышел.'),
('ИВАН', 'Один день Ивана Денисовича'),
('унылая', 'Унылая пора! очей очарованье!')
))
return raw_expls
__author__ = 'Алексей'
# noinspection PyProtectedMember
from preparation.resources.synonyms import _raw_data
from preparation import modifiers
from preparation.resources.Resource import gen_resource
from hb_res.explanations import Explanation
from ._synonyms_quality import choose_best_synonyms
synonyms_mods = [
modifiers.normalize_title(0.01, True),
modifiers.re_replace('[^#]+? [^#]+?(#|$)', ''), # remove multi-word synonyms (containing spaces)
modifiers.re_fullmatch_ban(''),
modifiers.delete_cognates(4, '#'),
modifiers.choose_normal_words_in_explanation('#'),
choose_best_synonyms(5, '#'),
modifiers.calculate_prior_frequency_rate('#'),
modifiers.str_replace('#', ', '),
modifiers.calculate_key()
]
@gen_resource('SynonymsResource', synonyms_mods)
def read_data():
explanations = set()
with open(_raw_data, 'r', encoding='utf-8') as source:
for line in source:
__author__ = 'shkiper'
# noinspection PyProtectedMember
from preparation.resources.crosswords import _raw_data
from preparation.resources.Resource import gen_resource
from hb_res.explanations import Explanation
from preparation import modifiers
crosswords_mods = [
modifiers.re_replace('p', 'р'),
modifiers.strip(target_field='title'),
modifiers.strip(),
modifiers.normalize_title(),
modifiers.re_replace(r'\s+', ' '),
modifiers.re_replace(r'([,:])(?=[^ ])', r'\1 '),
modifiers.str_replace(r' :', ':'),
modifiers.str_replace(r' ,', ','),
modifiers.shadow_cognates(8, '\W+', with_pronoun=True),
modifiers.remove_to_much_gap_percentage(r'\W+', r'\*(\w+)[?]?\*', 0.5),
modifiers.calculate_key()
]
@gen_resource('CrosswordsResource', crosswords_mods)
def read_data():
with open(_raw_data, 'r', encoding='utf-8') as source:
for line in source:
tokens = line.split('$')
word_and_text = tokens[1], tokens[2]
yield Explanation(*word_and_text)
__author__ = 'shkiper'
# noinspection PyProtectedMember
from preparation.resources.phraseological import _raw_data
from preparation.resources.Resource import gen_resource
from hb_res.explanations import Explanation
from preparation import modifiers
import re
phraseological_mods = [
modifiers.check_contains_valid_parts(2, 0.1, '\W+'),
modifiers.shadow_title_with_question(),
modifiers.normalize_title(),
modifiers.shadow_cognates(5, '\W+'),
modifiers.delete_multiple_gaps(0),
modifiers.re_replace(' ([,!?])', r'\1'),
modifiers.strip(),
modifiers.calculate_key()
]
@gen_resource('PhraseologicalResource', phraseological_mods)
def read_data():
phrases = set()
with open(_raw_data, 'r', encoding='utf-8') as source:
for line in source:
phrases.add(line)
for line in sorted(phrases):
for word in sorted(set(re.split('\W+', line))):
if len(word) > 0:
yield Explanation(word, line.strip('\n'))
__author__ = 'ryad0m'
# noinspection PyProtectedMember
from preparation.resources.ngram import _raw_data
from preparation.resources.Resource import gen_resource
from hb_res.explanations import Explanation
from preparation import modifiers
ngram_mods = [
modifiers.ensure_russian_title(),
modifiers.re_replace('p', 'р'),
modifiers.check_contains_valid_parts(2, 0.1, '\W+'),
modifiers.shadow_cognates(5, '\W+', with_question=True),
modifiers.delete_multiple_gaps(0),
modifiers.calculate_key()
]
@gen_resource('NgramResource', ngram_mods)
def read_data():
with open(_raw_data, 'r', encoding='utf-8') as source:
explanations = dict()
for line in source:
word, expl, rate = line.strip('\n').split('\t')
if explanations.get((word, expl)) is None:
explanations[(word, expl)] = rate
else:
explanations[(word, expl)] += rate
for (word, expl), rate in sorted(explanations.items()):
yield Explanation(word, expl, prior_rating=int(rate) / 400000)
@modifiers.modifier_factory
def add_antonyms_common_text():
def apply(e: Explanation):
if e.text.find(',') == -1:
text = "антоним к слову " + e.text
else:
text = "антоним к словам " + e.text
ret = copy.copy(e)
ret.text = text
return ret
return apply
antonyms_mods = [
modifiers.normalize_title(0.01, True),
modifiers.re_replace('[^#]+ [^#]+(#|$)', ''), # remove multi-word antonyms (containing spaces)
modifiers.re_fullmatch_ban(''),
modifiers.delete_cognates(6, '#'),
modifiers.choose_normal_words_in_explanation('#'),
modifiers.calculate_prior_frequency_rate('#'),
modifiers.str_replace('#', ', ', target_field='text'),
add_antonyms_common_text(),
modifiers.calculate_key()
]
def add_antonyms_common_text():
def apply(e: Explanation):
if e.text.find(',') == -1:
text = "антоним к слову " + e.text
else:
text = "антоним к словам " + e.text
ret = copy.copy(e)
ret.text = text
return ret
return apply
antonyms_mods = [
modifiers.normalize_title(0.01, True),
modifiers.re_replace('[^#]+ [^#]+(#|$)',
''), # remove multi-word antonyms (containing spaces)
modifiers.re_fullmatch_ban(''),
modifiers.delete_cognates(6, '#'),
modifiers.choose_normal_words_in_explanation('#'),
modifiers.calculate_prior_frequency_rate('#'),
modifiers.str_replace('#', ', ', target_field='text'),
add_antonyms_common_text(),
modifiers.calculate_key()
]
@gen_resource('AntonymsResource', antonyms_mods)
def read_data():
explanations = set()
with open(_raw_data, 'r', encoding='utf-8') as source:
for line in source:
import itertools
from preparation.resources.Resource import gen_resource
from preparation import modifiers
from hb_res.explanations.Explanation import Explanation
sample_modifiers = (
modifiers.str_replace('?', 'ё'),
modifiers.shadow_cognates(length_threshold=3, sep_re='(\\s|!|\\.)+'),
modifiers.re_replace('\\.', ' '),
modifiers.normalize_title(),
modifiers.calculate_key(),
)
@gen_resource('SampleResource', sample_modifiers)
def sample_parser():
raw_expls = itertools.starmap(
Explanation,
(('ПОРА', 'Однажды.в.студ?ную.зимнюю.пору.я.из.лесу.вышел.'),
('ИВАН', 'Один день Ивана Денисовича'),
('унылая', 'Унылая пора! очей очарованье!')))
return raw_expls