def resolveGender(self, name, country):
# Check if name is written in Cyrillic or Greek script, and transliterate
if only_cyrillic_chars(name) or only_greek_chars(name):
name = unidecode(name)
firstName = extractFirstName(name, 'direct')
gender = self.tryCrossCountry(firstName)
if gender is not None:
return gender
gender = self.tryUnidecoded(name)
if gender is not None:
return gender
gender = self.tryRemovingFirstAndLastLetters(name)
if gender is not None:
return gender
name = re.sub("\d+", "", name)
gender = self.tryRemovingFirstAndLastLetters(name)
if gender is not None:
return gender
return None
def resolveGender(self, name, country):
# Check if name is written in Cyrillic or Greek script, and transliterate
if only_cyrillic_chars(name) or only_greek_chars(name):
name = unidecode(name)
firstName = extractFirstName(name, 'direct')
gender = self.tryCrossCountry(firstName)
if gender is not None:
return gender
gender = self.tryUnidecoded(name)
if gender is not None:
return gender
gender = self.tryRemovingFirstAndLastLetters(name)
if gender is not None:
return gender
name = re.sub("\d+", "", name)
gender = self.tryRemovingFirstAndLastLetters(name)
if gender is not None:
return gender
return None
def resolveGender(self, name, country):
'''Check if name is written in Cyrillic or Greek script, and transliterate'''
if only_cyrillic_chars(name) or only_greek_chars(name):
name = unidecode(name)
'''Check if empty string'''
if len(name.strip()) < 1:
return None
'''Initial check for gender-specific words at the beginning of the name'''
f = name.split()[0]
if f in self.maleWords:
return 'male'
elif f in self.femaleWords:
return 'female'
'''Extract first name from name string'''
firstName = extractFirstName(name, 'direct')
if country is not None:
'''Start with suffixes
Works well for Russians (can determine gender based on surname suffix)'''
if country in self.suffixes.keys():
gender = self.suffixLookup(name, country)
if gender is not None:
return gender
'''If still no luck, extract first name and try to resolve'''
gender = self.resolveFirstName(firstName, country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try to inverse if no luck
Hungarians use reversed first/last names order'''
if country in self.invOrder:
gender = self.suffixLookup(inverseNameParts(name), country)
if gender is not None:
return gender
gender = self.resolveFirstName(
extractFirstName(name, 'inverse'), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Starting to get desperate by now. Assume name is in fact username,
and try different tricks:'''
if len(name.split()) == 1:
'''- Try the Dutch tricks'''
if country in ['Belgium', 'The Netherlands', 'South Africa']:
positions = [m.start() for m in re.finditer('v', name)]
bestMatch = []
if len(positions):
for pos in positions:
gender = self.resolveFirstName(
name[:pos], country, True)
if gender is not None:
if gender != 'blacklist':
bestMatch.append(gender)
gender = next((g for g in bestMatch if g != 'unisex'),
None)
if gender is not None:
return gender
if 'unisex' in bestMatch:
return 'unisex'
'''- Try to guess first name from: bogdanv, vbogdan'''
# bogdanv
gender = self.resolveFirstName(name[:-1].lower(), country,
True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
# vbogdan
gender = self.resolveFirstName(name[1:].lower(), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''I can't believe I'm trying leet'''
nameL = leet2eng(name)
gender = self.resolveFirstName(extractFirstName(nameL, 'direct'),
country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try also the unidecoded version'''
dname = unidecode(name)
gender = self.resolveFirstName(extractFirstName(dname, 'direct'),
country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''If everything failed, try cross-country'''
gender = self.resolveFirstNameOverall(firstName, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try also unidecoded version'''
dname = unidecode(name)
gender = self.resolveFirstNameOverall(
extractFirstName(dname, 'direct'), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
if len(name.split()) == 1:
'''- Try to guess first name from: bogdanv, vbogdan'''
# bogdanv
gender = self.resolveFirstNameOverall(name[:-1].lower(), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
# vbogdan
gender = self.resolveFirstNameOverall(name[1:].lower(), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
return None
def resolveGenderByCountry(self, name, country, script='Latin'):
(gender, conf) = self.initialCheckName(name)
if gender is not None:
return (gender, conf)
'''Extract first name from name string'''
firstName = extractFirstName(name, 'direct')
if country is not None:
'''Start with suffixes
Works well for Russians (can determine gender based on surname suffix)'''
if country in self.suffixes.keys():
(gender, conf) = self.suffixLookup(name, country)
if gender is not None:
return (gender, conf)
'''If still no luck, extract first name and try to resolve'''
(gender, conf) = self.resolveFirstName(firstName, country, True,
script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Try to inverse if no luck
Hungarians use reversed first/last names order'''
if country in self.invOrder:
(gender, conf) = self.suffixLookup(inverseNameParts(name),
country)
if gender is not None:
return (gender, conf)
(gender, conf) = self.resolveFirstName(
extractFirstName(name, 'inverse'), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Starting to get desperate by now. Assume name is in fact username,
and try different tricks:'''
if len(name.split()) == 1:
'''- Try the Dutch tricks'''
if country in ['Belgium', 'The Netherlands', 'South Africa']:
positions = [m.start() for m in re.finditer('v', name)]
bestMatch = []
if len(positions):
for pos in positions:
(gender, conf) = self.resolveFirstName(
name[:pos], country, True, script)
if gender is not None:
if gender != 'blacklist':
bestMatch.append(gender)
gender = next((g for g in bestMatch if g != 'unisex'),
None)
if gender is not None:
return (gender, conf)
if 'unisex' in bestMatch:
return ('unisex', 0)
'''- Try to guess first name from: bogdanv, vbogdan'''
#bogdan
(gender,
conf) = self.resolveFirstNameOverall(name.lower(), True,
script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
# bogdanv
(gender, conf) = self.resolveFirstName(name[:-1].lower(),
country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
# vbogdan
(gender, conf) = self.resolveFirstName(name[1:].lower(),
country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''I can't believe I'm trying leet'''
nameL = leet2eng(name)
(gender,
conf) = self.resolveFirstName(extractFirstName(nameL, 'direct'),
country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Try also the unidecoded version'''
dname = unidecode(name)
(gender,
conf) = self.resolveFirstName(extractFirstName(dname, 'direct'),
country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
return (None, 0)
def resolveGenderOverall(self, name, script='Latin'):
(gender, conf) = self.initialCheckName(name)
if gender is not None:
return (gender, conf)
'''Extract first name from name string'''
firstName = extractFirstName(name, 'direct')
'''If everything failed, try cross-country'''
(gender, conf) = self.resolveFirstNameOverall(firstName, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Try also unidecoded version'''
dname = unidecode(name)
(gender, conf) = self.resolveFirstNameOverall(
extractFirstName(dname, 'direct'), True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Try also inverse names: (i.e. family name)'''
(gender, conf) = self.resolveFirstNameOverall(
extractFirstName(name, 'inverse'), True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Try also middle names: (e.g. J. Marcos Nieto)'''
if len(name.split()) == 3:
mid = name.split()[1]
(gender, conf) = self.resolveFirstNameOverall(mid, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
if len(name.split()) == 1:
#bogdan
(gender,
conf) = self.resolveFirstNameOverall(name.lower(), True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''- Try to guess first name from: bogdanv, vbogdan'''
# bogdanv
(gender,
conf) = self.resolveFirstNameOverall(name[:-1].lower(), True,
script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
#vbogdan
(gender,
conf) = self.resolveFirstNameOverall(name[1:].lower(), True,
script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
return (gender, conf)
def resolveGender(self, name, country):
'''Check if name is written in Cyrillic or Greek script, and transliterate'''
if only_cyrillic_chars(name) or only_greek_chars(name):
name = unidecode(name)
'''Initial check for gender-specific words at the beginning of the name'''
f = name.split()[0]
if f in self.maleWords:
return 'male'
elif f in self.femaleWords:
return 'female'
'''Extract first name from name string'''
firstName = extractFirstName(name, 'direct')
if country is not None:
'''Start with suffixes
Works well for Russians (can determine gender based on surname suffix)'''
if country in self.suffixes.keys():
gender = self.suffixLookup(name, country)
if gender is not None:
return gender
'''If still no luck, extract first name and try to resolve'''
gender = self.resolveFirstName(firstName, country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try to inverse if no luck
Hungarians use reversed first/last names order'''
if country in self.invOrder:
gender = self.suffixLookup(inverseNameParts(name), country)
if gender is not None:
return gender
gender = self.resolveFirstName(extractFirstName(name, 'inverse'), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Starting to get desperate by now. Assume name is in fact username,
and try different tricks:'''
if len(name.split()) == 1:
'''- Try the Dutch tricks'''
if country in ['Belgium', 'The Netherlands', 'South Africa']:
positions = [m.start() for m in re.finditer('v', name)]
bestMatch = []
if len(positions):
for pos in positions:
gender = self.resolveFirstName(name[:pos], country, True)
if gender is not None:
if gender != 'blacklist':
bestMatch.append(gender)
gender = next((g for g in bestMatch if g != 'unisex'), None)
if gender is not None:
return gender
if 'unisex' in bestMatch:
return 'unisex'
'''- Try to guess first name from: bogdanv, vbogdan'''
# bogdanv
gender = self.resolveFirstName(name[:-1].lower(), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
# vbogdan
gender = self.resolveFirstName(name[1:].lower(), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''I can't believe I'm trying leet'''
nameL = leet2eng(name)
gender = self.resolveFirstName(extractFirstName(nameL, 'direct'), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try also the unidecoded version'''
dname = unidecode(name)
gender = self.resolveFirstName(extractFirstName(dname, 'direct'), country, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''If everything failed, try cross-country'''
gender = self.resolveFirstNameOverall(firstName, True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
'''Try also unidecoded version'''
dname = unidecode(name)
gender = self.resolveFirstNameOverall(extractFirstName(dname, 'direct'), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
if len(name.split()) == 1:
'''- Try to guess first name from: bogdanv, vbogdan'''
# bogdanv
gender = self.resolveFirstNameOverall(name[:-1].lower(), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
# vbogdan
gender = self.resolveFirstNameOverall(name[1:].lower(), True)
if gender is not None:
if gender == 'blacklist':
return None
return gender
return None
def tryUnidecoded(self, name): dname = unidecode(name) firstName = extractFirstName(dname, 'direct') return self.tryCrossCountry(firstName)
def resolveGenderByCountry(self, name, country, script = 'Latin'):
(gender,conf) = self.initialCheckName(name)
if gender is not None:
return (gender,conf)
'''Extract first name from name string'''
firstName = extractFirstName(name, 'direct')
if country is not None:
'''Start with suffixes
Works well for Russians (can determine gender based on surname suffix)'''
if country in self.suffixes.keys():
(gender,conf) = self.suffixLookup(name, country)
if gender is not None:
return (gender,conf)
'''If still no luck, extract first name and try to resolve'''
(gender,conf) = self.resolveFirstName(firstName, country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None,0)
return (gender, conf)
'''Try to inverse if no luck
Hungarians use reversed first/last names order'''
if country in self.invOrder:
(gender, conf) = self.suffixLookup(inverseNameParts(name), country)
if gender is not None:
return (gender, conf)
(gender, conf) = self.resolveFirstName(extractFirstName(name, 'inverse'), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''Starting to get desperate by now. Assume name is in fact username,
and try different tricks:'''
if len(name.split()) == 1:
'''- Try the Dutch tricks'''
if country in ['Belgium', 'The Netherlands', 'South Africa']:
positions = [m.start() for m in re.finditer('v', name)]
bestMatch = []
if len(positions):
for pos in positions:
(gender, conf) = self.resolveFirstName(name[:pos], country, True, script)
if gender is not None:
if gender != 'blacklist':
bestMatch.append(gender)
gender = next((g for g in bestMatch if g != 'unisex'), None)
if gender is not None:
return (gender, conf)
if 'unisex' in bestMatch:
return ('unisex',0)
'''- Try to guess first name from: bogdanv, vbogdan'''
#bogdan
(gender,conf) = self.resolveFirstNameOverall(name.lower(), True, script)
if gender is not None:
if gender == 'blacklist':
return (None,0)
return (gender,conf)
# bogdanv
(gender, conf) = self.resolveFirstName(name[:-1].lower(), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None,0)
return (gender, conf)
# vbogdan
(gender, conf) = self.resolveFirstName(name[1:].lower(), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
'''I can't believe I'm trying leet'''
nameL = leet2eng(name)
(gender, conf) = self.resolveFirstName(extractFirstName(nameL, 'direct'), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None,0)
return (gender, conf)
'''Try also the unidecoded version'''
dname = unidecode(name)
(gender, conf) = self.resolveFirstName(extractFirstName(dname, 'direct'), country, True, script)
if gender is not None:
if gender == 'blacklist':
return (None, 0)
return (gender, conf)
return (None, 0)
def resolveGenderOverall(self, name, script= 'Latin'): (gender,conf) = self.initialCheckName(name) if gender is not None: return (gender,conf) '''Extract first name from name string''' firstName = extractFirstName(name, 'direct') '''If everything failed, try cross-country''' (gender,conf) = self.resolveFirstNameOverall(firstName, True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) '''Try also unidecoded version''' dname = unidecode(name) (gender,conf) = self.resolveFirstNameOverall(extractFirstName(dname, 'direct'), True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) '''Try also inverse names: (i.e. family name)''' (gender,conf) = self.resolveFirstNameOverall(extractFirstName(name, 'inverse'), True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) '''Try also middle names: (e.g. J. Marcos Nieto)''' if len(name.split()) == 3: mid = name.split()[1] (gender,conf) = self.resolveFirstNameOverall(mid, True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) if len(name.split()) == 1: #bogdan (gender,conf) = self.resolveFirstNameOverall(name.lower(), True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) '''- Try to guess first name from: bogdanv, vbogdan''' # bogdanv (gender,conf) = self.resolveFirstNameOverall(name[:-1].lower(), True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) #vbogdan (gender,conf) = self.resolveFirstNameOverall(name[1:].lower(), True, script) if gender is not None: if gender == 'blacklist': return (None,0) return (gender,conf) return (gender,conf)
def tryUnidecoded(self, name):
dname = unidecode(name)
firstName = extractFirstName(dname, 'direct')
return self.tryCrossCountry(firstName)