from faker import Faker
import pandas as pd
import random
import matplotlib.pyplot as plt
import datetime

df = pd.read_csv('../datasets/nepali_first_name_gender.csv')

filename = '../datasets/lastnames.txt'
with open(filename) as f:
    content = f.readlines()
content = [x.strip()[:-1] for x in content] 

last_names = [x.capitalize() for x in content] 

first_name_male = [x.capitalize() for x in list(df[df.gender=='M'].first_name)] 
first_name_female = [x.capitalize() for x in list(df[df.gender=='F'].first_name)]

[random.choice(first_name_male) + " " + random.choice(last_names) for x in range(10)]

total_customer = 30000
male_percent = 58

male = {'name': [random.choice(first_name_male) + " " + random.choice(last_names) for x in range(int(total_customer*male_percent/100))]}
female = {'name': [random.choice(first_name_female) + " " + random.choice(last_names) for x in range(int(total_customer*(100-male_percent)/100))]}

df_male = pd.DataFrame(data=male)
df_male['gender'] = 'Male'
df_female = pd.DataFrame(data=female)
df_female['gender'] = 'Female'

df = df_male.append(df_female)

df.loc[df.sample(frac=.04).index,
    'gender'
] = 'Other'

df.loc[df.sample(frac=.02).index,
    'gender'
] = 'Prefer not to say'

df.gender.value_counts()

def email_gen(name):
    name = name.lower()
    return (name.split(' ')[0]+'.'+name.split(' ')[1]+'@fakeemail.com')

df['email'] = df.name.apply(email_gen)

df

df['phone'] = [random.randint(9841000000, 9842000000) for x in range(df.shape[0])]

df

df['channel'] = 'Word of Mouth'

df.loc[df.sample(frac=.5).index,
    'channel'
] = 'Facebook/Ads'

df.loc[df.sample(frac=.2).index,
    'channel'
] = 'Google Search'

df.loc[df.sample(frac=.1).index,
    'channel'
] = 'Other'

df

df.channel.value_counts()

df['first_contact'] = 'app'

df.loc[df.sample(frac=.676).index,
    'first_contact'
] = 'browser'

df

df['first_contact'].value_counts()

df = df.sample(frac=1)

df

from timeseries_generator import LinearTrend, Generator, WhiteNoise, RandomFeatureFactor, WeekdayFactor
import pandas as pd

start="09-01-2020"
end="08-31-2021"


# setting up a linear tren
lt = LinearTrend(coef=5.0, offset=1., col_name="my_linear_trend")

weekday_factor = WeekdayFactor(
    col_name="weekend_boost_factor",
    factor_values={4: 1.1, 5: 1.2, 6: 1.2}  # Here we assign a factor of 1.1 to Friday, and 1.2 to Sat/Sun 
)

# weekday_factor.plot(start_date=start, end_date=end)

g = Generator(factors={lt, weekday_factor}, features=None, date_range=pd.date_range(start=start, end=end))
g.generate()

# update by adding some white noise to the generator
wn = WhiteNoise(stdev_factor=0.06)
g.update_factor(wn)
g.generate()
g.plot()

data = g.generate()

data['rows'] = (10*data.value).astype(int)

data.rows.sum()

data

def nepali_holiday(row):
    date = datetime.datetime.strptime(str(row.date), "%Y-%m-%d %H:%M:%S")
    num = row.rows
    if date>datetime.datetime(2020, 10, 16) and date<datetime.datetime(2020, 10, 25):
        num = int (num*1.2)
    if date>datetime.datetime(2020, 12, 24) and date<datetime.datetime(2021, 1, 2):
        num = int (num*1.4)
    if date>datetime.datetime(2021, 1, 4) and date<datetime.datetime(2021, 1, 7):
        num = num - num
    if date>datetime.datetime(2021, 2, 13) and date<datetime.datetime(2021, 2, 15):
        num = int (num*1.4)
    if date>datetime.datetime(2021, 7, 1):
        num = int(num * 2)
    return num

data['rows'] = data.apply(nepali_holiday, axis=1)

# ["2020-09-01"]*100 + ["2020-09-02"] * 200
from datetime import datetime as dt
import datetime

def create_list(row):
    return [str(row.date)]*row.rows

final_dates = (data.apply(create_list, axis=1).sum())

clients = df.head(len(final_dates))
clients.reset_index(inplace = True, drop=True)

clients['created_at'] = final_dates

clients

clients.gender.value_counts()

clients.channel.value_counts()

clients['first_contact'].value_counts()

def random_times(input_time):
    random_hour = random.uniform(0, 23)
    return dt.strptime(input_time, "%Y-%m-%d %H:%M:%S") + datetime.timedelta(hours=random_hour)

clients['created_at'] = clients['created_at'].apply(random_times)

clients.sort_values('created_at').reset_index(drop=True, inplace=True)

clients['client_id'] = clients.index + 100000000

clients

def date_gen(start, end):
    start_date = datetime.date(start, 1, 1)
    end_date = datetime.date(end, 12, 31)
    time_between_dates = end_date - start_date
    days_between_dates = time_between_dates.days
    random_number_of_days = random.randrange(days_between_dates)
    random_date = start_date + datetime.timedelta(days=random_number_of_days)
    return str(random_date)

dob = [date_gen(1975, 1990) for x in range(clients.shape[0]//5)] + \
[date_gen(1991, 1995) for x in range(clients.shape[0]//5)] + \
[date_gen(1996, 2000) for x in range(clients.shape[0]//2)] + \
[date_gen(2001, 2005) for x in range(clients.shape[0]//2)] + \
[date_gen(2006, 2008) for x in range(clients.shape[0]//5)]

random.shuffle(dob)

clients['dob'] = dob[:clients.shape[0]]

clients

clients

location = pd.read_csv("../datasets/location.csv")

location = location[['lat', 'lon', 'name']]

location_list =  [location.sample().values.tolist() for x in range(clients.shape[0])]

location_df = pd.DataFrame(location_list)

location_df

location_df[['lat','lon', 'location_name']] = pd.DataFrame(location_df[0].tolist(), index= location_df.index)

location_df[['lat', 'lon', 'location_name']]

clients

clients = pd.concat([clients, location_df[['lat', 'lon', 'location_name']]], axis=1)

list(clients)

clients[
['client_id',
 'name',
 'gender',
 'dob',
 'email',
 'phone',
 'channel',
 'first_contact',
  'lat',
 'lon',
 'location_name',
 'created_at']
].to_csv('../datasets/clients.csv', index=False)