import datetime
def get_df_sellers(sellers, items, products, orders, translation):
    my_sellers = sellers[['seller_id', 'seller_state']].drop_duplicates()
    merged_df = my_sellers.merge(items[['order_id', 'product_id', 'seller_id']], on = 'seller_id')
    df = products.merge(merged_df, on = 'product_id')
    df = df[['seller_state', 'product_category_name', 'order_id']].merge(orders[['order_status', 'order_id', 'order_delivered_customer_date']], on = 'order_id')
    df = df.merge(translation, on = "product_category_name")
    df.drop(columns = ['product_category_name'], inplace = True)
    df.rename(columns = {'product_category_name_english': 'category', 'seller_state': 'state', 'order_status': 'status', 'order_delivered_customer_date' : 'date'}, inplace = True)
    df = df.dropna()
    df['date'] = pd.to_datetime(df['date']).apply(lambda x: x.strftime("%Y-%m"))
    df = df.groupby(['date', 'state' , 'status','category']).agg({'order_id' : 'count'}).reset_index().rename(columns = {"order_id" : "count"})
    return df
def get_df_customers(customers, items, products, orders, translation):
    my_customers = customers[['customer_id', 'customer_state']].drop_duplicates()
    merged_df = my_customers.merge(orders, on = 'customer_id')
    df = items.merge(merged_df, on = 'order_id')
    df = df.merge(products, on = 'product_id')
    df = df.merge(translation, on = "product_category_name")
    df.drop(columns = ['product_category_name'], inplace = True)
    df.rename(columns = {'product_category_name_english': 'category', 'customer_state': 'state', 'order_status': 'status', 'order_delivered_customer_date' : 'date'}, inplace = True)
    df = df.dropna()
    df['date'] = pd.to_datetime(df['date']).apply(lambda x: x.strftime("%Y-%m"))
    df = df.groupby(['date', 'state' , 'status','category']).agg({'order_id' : 'count'}).reset_index().rename(columns = {"order_id" : "count"})
    return df
df = get_df_sellers(sellers, items, products, orders, translation)

all_dates = df
all_dates['date'] = pd.to_datetime(all_dates.date).apply(lambda x: x.date())
all_dates = all_dates.sort_values(by='date')

marks = {i: str(sorted(all_dates['date'])[i]) for i in range(0, all_dates.shape[0], 7850)}
all_states = ['All'] + sorted(list(set(list(sellers.seller_state.unique()) + list(customers.customer_state.unique()))))
all_statuses = orders.order_status.unique()
    from dash import Dash, dcc, html, Input, Output
    app = Dash(__name__)

    app.layout = html.Div([
        html.Label('Order statuses to display:'),
        dcc.Checklist(
            all_statuses,
            [df['status'][0]],
            id = 'status-filter'
        ),
        html.Label('State:', style={'font-size': '120%'}),
        html.Div(dcc.Dropdown(
            all_states,
            'All',
            id = 'state-dropdown'
        )),
        dcc.Graph(id='sellers-graph'),
        dcc.Graph(id='customers-graph'),
        dcc.RangeSlider(
            min=0,
            max=all_dates.shape[0] - 1,
            step=1,
            dots=False,
            value=[0, all_dates.shape[0] - 1],
            marks = marks,
            id='date-slider',
        ),
    ])



    @app.callback(
        Output('sellers-graph', 'figure'),
        [Input('date-slider', 'value'),
         Input('state-dropdown', 'value'),
         Input('status-filter', 'value')])

    def update_figure(time_gap, state_to_display, statuses_filter):
        cur_df = get_df_sellers(sellers, items, products, orders, translation)
        cur_df = cur_df[cur_df['status'].isin(statuses_filter)]
        if state_to_display == 'All':
            cur_df['state'] = 'All'
        else:
            cur_df = cur_df[cur_df.state == state_to_display]
        cur_df = cur_df[pd.to_datetime(sorted(all_dates['date'])[time_gap[0]]) <= pd.to_datetime(cur_df.date)]
        cur_df = cur_df[pd.to_datetime(cur_df.date) <= pd.to_datetime(sorted(all_dates['date'])[time_gap[1]])]
        cur_df = cur_df.groupby(['category']).agg({'count' : 'count', 'state':'first'}).reset_index()
        all_count = np.einsum('i->', cur_df['count'])
        cur_df['count'] = cur_df['count'] * 100/all_count
        cur_df.rename(columns = {"count":"propotion"})
        fig = px.bar(cur_df, y="state", x="count", color="category",
                     title="Distribution by sales categories", orientation='h', height=300)
        fig.update_layout(transition_duration=500)
        return fig

    @app.callback(
        Output('customers-graph', 'figure'),
        [Input('date-slider', 'value'),
         Input('state-dropdown', 'value'),
         Input('status-filter', 'value')])
    def update_figure(time_gap, state_to_display, statuses_filter):
        cur_df = get_df_customers(customers, items, products, orders, translation)
        cur_df = cur_df[cur_df['status'].isin(statuses_filter)]
        if state_to_display == 'All':
            cur_df['state'] = 'All'
        else:
            cur_df = cur_df[cur_df.state == state_to_display]
        cur_df = cur_df[pd.to_datetime(sorted(all_dates['date'])[time_gap[0]]) <= pd.to_datetime(cur_df.date)]
        cur_df = cur_df[pd.to_datetime(cur_df.date) <= pd.to_datetime(sorted(all_dates['date'])[time_gap[1]])]
        cur_df = cur_df.groupby(['category']).agg({'count' : 'count', 'state':'first'}).reset_index()
        all_count = np.einsum('i->', cur_df['count'])
        cur_df['count'] = cur_df['count'] * 100/all_count
        cur_df.rename(columns = {"count":"propotion"})
        fig = px.bar(cur_df, y="state", x="count", color="category",
                     title="Distribution by customers categories", orientation='h', height=300)
        fig.update_layout(transition_duration=500)
        return fig
    if __name__ == '__main__':
        app.run_server(debug=True)