from bokeh.plotting import figure, show
import datetime
import pandas as pd


def read(filename):
    with open('vin_data.txt') as f:
        s = f.read()

    return s


def parse_string(s):
    x = s.split(',')
    year = int(x[0][-4:])
    month = int(x[1].split(' - ')[0])
    day = int(x[2].split(')')[0])
    vin = int(x[3])
    date = datetime.date(year, month, day)

    return (date, vin)


def parse_data(s):
    ls = s[2:-3].split('], [')
    lx = [parse_string(s) for s in ls]

    return lx


raw_data = read('vin_data.txt')
list_of_tuples = parse_data(raw_data)

df_vin = pd.DataFrame(list_of_tuples, columns=['date', 'vin'])
df_vin = df_vin.groupby('vin').apply(min)
df_vin_unique = df_vin.groupby('date').count()
df_vin_unique['cumulative'] = df_vin_unique['vin'].cumsum()

df_production = pd.DataFrame(
    {'date': [datetime.date(2017, x, 1) for x in [7, 8, 9, 10, 11, 12]],
     'production': [0, 30, 75, 117, 145, 345]
    }
)

df_production['cumulative'] = df_production['production'].cumsum()

f = figure(width=800,
           height=600,
           x_axis_type='datetime',
           title='Model 3 Estimated Sales vs. Unique Sightings Over Time')
f.title.text_font_size = '18pt'

f.xaxis.axis_label='date'
f.yaxis.axis_label='cumulative number'

f.line(df_vin_unique.index.values, df_vin_unique['cumulative'].values,
       legend='Cumulative number of unique sightings (based on VIN) [source: blfire.lima-city.de]',
       line_width=2)

f.line(df_production['date'].values, df_production['cumulative'].values, color='red',
       legend = 'Cumulative number of estimated sales (interpolated) [source: insideevs.com]',
       line_width=2)

f.legend.location = 'top_left'

show(f)