def get_author_gender(name):
"""
"""
gender = getGenders(name.split()[0])
return gender
def save_name_genders():
name_pool = get_names()
result = {}
bad_names = []
chunk_size = 50
total_chunks = len(name_pool)/chunk_size
rest = len(name_pool)%chunk_size
for j in range(0, total_chunks):
print '*******Chunk', j, '/', total_chunks
names = []
if j == total_chunks - 1:
names = name_pool[j * chunk_size:total_chunks*chunk_size+rest]
else:
names = name_pool[j * chunk_size:(j+1)*chunk_size]
gender_list = gender.getGenders(names) #gender, prob, count
for i, name in enumerate(names):
infered_gender = gender_list[i][0]
prob = float(gender_list[i][1])
print name, infered_gender, prob
if infered_gender == 'None' or prob < 0.6:
bad_names.append(name)
result[name] = infered_gender
with open( "genders.p", "wb" ) as outfile:
pickle.dump(result, outfile)
with open( "badnames.p", "wb" ) as outfile:
pickle.dump(bad_names, outfile)
return bad_names
def get_name_genders_genderize():
name_pool = get_names()
chunk_size = 50
total_chunks = len(name_pool)/chunk_size
rest = len(name_pool)%chunk_size
genders = {}
for j in range(0, total_chunks):
print '*******Chunk', j, '/', total_chunks
names = []
if j == total_chunks - 1:
names = name_pool[j * chunk_size:total_chunks*chunk_size+rest]
else:
names = name_pool[j * chunk_size:(j+1)*chunk_size]
gender_list = gender.getGenders(names)
# [
# {"name":"peter","gender":"male","probability":"1.00","count":796},
# {"name":"lois","gender":"female","probability":"0.94","count":70},
# {"name":"stevie","gender":"male","probability":"0.63","count":39}
# ]
for g in gender_list:
genders[g[0]] = g[1]
json.dump(genders, open('genders_genderize.json', 'w'))
def get_genderize(name):
"""
Uses gender.py library to assign gender to first name using genderize.io API
Must only use first name --> "Mark"
name (string)
"""
info = getGenders(name)
return info
def get_gender_from_cache_or_genderize(artist, cached_genders):
if(cached_genders.get(artist) is not None):
return cached_genders.get(artist)
else:
#not in cache -> use https://github.com/block8437/gender.py (https://genderize.io)
gender, _, _ = getGenders([artist])[0]
cached_genders[artist] = gender
update_cached_genders(cached_genders)
return gender
def _get_genders(df):
genTuples = []
for fullname in df["Name"].values:
firstname = get_firstname(fullname)
genTuple = getGenders(firstname)[0]
genTuples.append(genTuple)
genders = [gen[0] if gen is not None else "n" for gen, _, _ in genTuples]
gender_probas = [genpro for _, genpro, _ in genTuples]
return genders, gender_probas
def gender_voter(name):
votes = {"male": 0, "female": 0, "unknown": 0}
# fist gender detector (male, female, None)
found_gender, precision, n_docs = getGenders(name)[0]
# second gender detector (male, female, mostly_male, mostly_female, andy, unknown)
d = gndr.Detector(case_sensitive=False)
found_gender_2 = d.get_gender(name)
# spanish better detector (Male, Female)
guesser = genderator.Parser()
answer = guesser.guess_gender(name)
if answer:
spanish_gender = answer['gender']
else:
spanish_gender = 'not found'
# VOTING
# add votes from first generator
if found_gender == 'male':
votes['male'] += 1
if found_gender == 'female':
votes['female'] += 1
# add votes from second generator
if found_gender_2 == 'male':
votes['male'] += 1
if found_gender_2 == 'female':
votes['female'] += 1
if found_gender_2 == 'mostly_male':
votes['male'] += 1
if found_gender_2 == 'mostly_female':
votes['female'] += 1
if found_gender_2 == 'andy':
votes['male'] += 1
votes['female'] += 1
# add votes from third spanish gen
if spanish_gender == 'Male':
votes['male'] += 1
if spanish_gender == 'Female':
votes['female'] += 1
return most_votes(votes)
def update_name_genders():
name_pool = get_names()
bad_names = []
config = load_config()
cnx = mysql.connector.connect(**config)
cursor = cnx.cursor()
chunk_size = 50
total_chunks = len(name_pool)/chunk_size
rest = len(name_pool)%chunk_size
for j in range(0, total_chunks):
print '*******Chunk', j, '/', total_chunks
names = []
if j == total_chunks - 1:
names = name_pool[j * chunk_size:total_chunks*chunk_size+rest]
else:
names = name_pool[j * chunk_size:(j+1)*chunk_size]
gender_list = gender.getGenders(names) #gender, prob, count
for i, name in enumerate(names):
infered_gender = gender_list[i][0]
prob = float(gender_list[i][1])
#print name, infered_gender, prob
if infered_gender == 'None' or prob < 0.6:
bad_names.append(name)
query = "UPDATE person SET gender = '%s' WHERE first_name = '%s'" % (infered_gender, name)
cursor.execute(query)
cursor.close()
cnx.close()
import pandas as pd
from gender import getGenders
import time
df = pd.read_csv("user_names/name_sorted", header=None)
print(df)
nameList = df[0].values.tolist()
nameList = [str(x) for x in nameList]
index = 994
while index < len(nameList):
try:
name = nameList[index]
print("name = ", name)
gender = getGenders(name)
print("gender = ", gender)
# genderList.append(gender[0])
with open("genderList.csv", "a") as op:
op.write(name + "," + str(gender[0]) + "\n")
index += 1
except Exception as e:
print("exept: " + str(e))
time.sleep(3600)
continue
print("Done")
except:
print("oops")
print(row_index)
print(out)
'''
out = 'org'
# coding for checking the left out record
for row_index, row in data_all.iterrows():
if row["user_fname_count"] == 0 and row[
"user_predicted_fname_genderO"] != 'Org':
usr_fname = row["user_fname"]
time.sleep(0.1)
#result = getGenders(usr_fname)
try:
print(row_index)
out = getGenders(usr_fname)
print(out)
if (out[0][0] == 'male'):
data_all.loc[row_index,
"user_predicted_fname_genderM"] = out[0][1]
data_all.loc[row_index, "user_fname_count"] = out[0][2]
if (out[0][0] == 'female'):
data_all.loc[row_index,
"user_predicted_fname_genderF"] = out[0][1]
data_all.loc[row_index, "user_fname_count"] = out[0][2]
if (out[0][0] == 'None'):
data_all.loc[row_index, "user_predicted_fname_genderO"] = 'Org'
data_all.loc[row_index, "user_fname_count"] = out[0][2]
except:
import glob
import sys
from gender import getGenders
indir = sys.argv[1]
infile = sys.argv[2]
cur_names = set()
with open(infile) as f:
for line in f:
cur_names.add(line.split()[0].strip())
dirlist = glob.glob('%s/*' % indir)
namelist = [s.split('/')[-1].split('_')[0] for s in dirlist]
names = list(set(namelist) - cur_names)
gender_dict = dict.fromkeys(names, None)
for i in range(0, len(names), 10):
g = getGenders(names[i:i + 10])
for name, gender in zip(names[i:i + 10], g):
gender_dict[name] = gender
print name, gender