Distribution

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import openpyxl as pxl
import seaborn as sns
import statistics
from collections import OrderedDict
from dateutil.relativedelta import relativedelta
from datetime import datetime
import textwrap


def get_val(df):
    x = 0
    for i in df:
        x = i
    return x




def graph_id_rating(df):
    df = pd.DataFrame(df.groupby(['рейтинг', 'флаг']).count()).reset_index().set_index('рейтинг')
    df.drop('report_date', inplace=True, axis=1)
    df = df.pivot(columns='флаг')
    return df

#function for create df year per id
def graph_year_id(df):
    df.drop('флаг', inplace=True, axis=1)
    df.drop('рейтинг', inplace=True, axis=1)
    df = pd.DataFrame(df.groupby(['report_date']).count())
    df = df.reset_index()
    df['report_date'] = pd.DatetimeIndex(df['report_date']).year
    df = df.set_index('report_date')
    return df


# 1y
# 2y
# 3m

# {'scale': 'month', 'scale_factor': 4}
# {'scale': 'year', 'scale_factor': 2}
# {'scale': 'quartal', 'scale_factor': 1 or 2}




#function for different check
#TODO check solution using udf functions, not iterate over dicts
def step_graph(df, step):
    year_dict = {}
    temp_2 = {}
    temp = {}
    for date, i in df.iterrows():
        if i[1] not in year_dict:
            year_dict[i[1]] = []
            temp_2[i[1]] = []
        year_dict[i[1]].append(i[0].strftime('%Y-%m'))
    value_get = ''
    for key, values in year_dict.items():
        value_get = year_dict[key][0]
        while value_get < year_dict[key][len(year_dict[key]) - 1]:
            # TODO CHECK WHILE OR FOR TO ITERATE OVER THE ALL DATES
            if step['scale'] == 'month':
                value_get = datetime.strptime(value_get, '%Y-%m')
                value_get = value_get + (relativedelta(months=step['scale_factor']))
                value_get = value_get.strftime('%Y-%m')
            elif step['scale'] == 'year':
                value_get = datetime.strptime(value_get, '%Y-%m')
                value_get = value_get + (relativedelta(years=step['scale_factor']))
                value_get = value_get.strftime('%Y-%m')
            elif step['scale'] == "quartal":
                value_get = datetime.strptime(value_get, '%Y-%m')
                value_get = value_get + (relativedelta(months=step['scale_factor'] * 3))    #todo redact this step
                value_get = value_get.strftime('%Y-%m')
            if value_get in values:
                temp_2[key].append(value_get)
    check = []
    df_1 = pd.DataFrame(columns=df.columns)
    df_1 = df_1.dropna()
    for valeu in temp_2.values():
        check.append(len(valeu))
    for x in check:
        if x not in temp:
            temp[x] = []
        temp[x].append(check.count(x))
    for key, value in temp.items():
        df_1.at[key] = len(value)
        print(key, len(value))
    df_1.drop('report_date', inplace=True, axis=1)
    df_1 = df_1.sort_index()
    print(df_1)
    return df_1


#reading excel file
step = {'scale': 'year', 'scale_factor': 2}
df = pd.read_excel("cases_sample_rebuilt.xlsx", sheet_name="Лист3")

year_dict = {}  #dict for create df
temp = []   #mass for count number of id/years/etc
val = []    #mass for median, mean, etc
for date, i in df.iterrows():
    if i[1] not in year_dict:
        year_dict[i[1]] = []
    temp.append(i[0].strftime('%Y'))
    year_dict[i[1]].append(i[0].strftime('%Y'))

df_1 = pd.DataFrame(columns=df.columns) #creating df
df_1 = df_1.dropna()


for key, value in year_dict.items():
    df_1.at[key] = len(set(temp)) - len(value)  #key --- (value = count of id)
    val.append(len(set(temp)) - len(value))
print(year_dict)

med = get_val(df_1.median())    #median
mean = get_val(df_1.mean())     #mean
mod = (statistics.mode(val))    #mod
sted = get_val(df_1.std())      #std

types = ['density', 'count', 'percent', 'frequency', 'dependence id-year', 'dependence year-id', 'step']      #todo add step checker
for i in types:
    fig = plt.figure(figsize=(10, 10))
    if i == 'dependence id-year':
        df_1 = graph_id_rating(df)
        df_1.plot.bar(stacked=True)
        plt.xlabel('Dates')
        plt.ylabel('number of unique objects')
    elif i == 'dependence year-id':
        df_1 = graph_year_id(df)
        df_1.plot.bar()
        plt.ylabel('number of unique objects at date')
        plt.xlabel('Number of dates')
    elif i == 'step' :
        df_1 = step_graph(df, step)
        df_1.plot.bar()
        plt.ylabel('number of unique objects')
    else:
        sns.histplot(data=df_1, kde=True,bins = list(range(0, len(set(temp)), 1)), stat=i, legend=False)
    plt.xlabel(
        f'median:{med}, mod:{mod}, mean:{round(mean, 2)}, sted:{round(sted, 2)}'
        , labelpad=20, fontsize=5, loc='center')
    plt.savefig(f'picture{i}', dpi=350, bbox_inches='tight')
    plt.close(fig)