AnyNameHack / utils.py
Danil's picture
Upload utils.py
c82443e
raw
history blame
2.28 kB
import pymorphy2
morph = pymorphy2.MorphAnalyzer()
def counter(s: str) -> dict:
d = {}
for i in s:
if i not in d:
d[i] = 0
d[i] += 1
return d
def sweet_check(s1: str, s2: str):
STOP_PUNCT = list(',./!@#$%^&*()_+=-<>?\|{}[]`~/')
STOP = set(
["скидка", "скидкой", "скидки", "скидке", "скидкой", "скидке", "недорого", "дешево", "в", "на", "для", "о", "у",
"и", "с", "из"] + STOP_PUNCT)
s1 = s1.lower()
s2 = s2.lower()
set_s1 = set(s1.split(' ')) - STOP
set_s2 = set(s2.split(' ')) - STOP
if set_s1 == set_s2:
return False
diff_s1 = ' '.join(list(set_s1 - set_s2))
diff_s2 = ' '.join(list(set_s2 - set_s1))
if len(diff_s1) == 0:
# return diff_s2
return True
if len(diff_s2) == 0:
return False
return True
def check(s1: str, s2: str, debag=False, morph=morph) -> float:
STOP_PUNCT = list(',./!@#$%^&*()_+=-<>?\|{}[]`~/')
STOP = set(
["скидка", "скидкой", "скидки", "скидке", "скидкой", "скидке", "недорого", "дешево", "в", "на", "для", "о", "у",
"и", "с", "из"] + STOP_PUNCT)
s1 = s1.lower()
s2 = s2.lower()
s1 = [morph.parse(i)[0].normal_form for i in s1.split(' ')]
s2 = [morph.parse(i)[0].normal_form for i in s2.split(' ')]
set_s1 = set(s1) - STOP
set_s2 = set(s2) - STOP
if set_s1 == set_s2:
return False
diff_s1 = ' '.join(list(set_s1 - set_s2))
diff_s2 = ' '.join(list(set_s2 - set_s1))
if debag:
print(s1)
print(s2)
if len(diff_s1) == 0:
return True
if len(diff_s2) == 0:
return False
dt = {len(diff_s1): diff_s1, len(diff_s2): diff_s2}
c = 0
max_s, min_s = dt[max(len(diff_s1), len(diff_s2))], dt[min(len(diff_s1), len(diff_s2))]
c_s1 = counter(min_s)
c_s2 = counter(max_s)
for i in min_s:
if i in c_s2 and c_s2[i] > 0:
c += 1
c_s2[i] -= 1
else:
c -= 1
if len(diff_s2) == len(diff_s1):
c -= 1
if debag:
print(c / len(min_s))
if c / len(min_s) < 1.0:
return True
return False