import pandas as pd
import numpy as np
import random
import time

start_time = time.time()
data = pd.read_csv("podstawa.csv", sep=',', engine='python')
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"+str(time.time() - start_time))
print(data.info())
print(data.values[0])
print(data.values[15])
uniquelist = data['session_id'].unique().tolist()
d = int(len(uniquelist)*3/4)
trainlist = uniquelist[:d]
testlist = uniquelist[d:]
#df = data
print(len(data['session_id']))
print(len(uniquelist))
print(len(trainlist))
print(len(testlist))
print(testlist[0])
print(testlist[1])
ll = []
#data = data[data['session_id'].isin(ll)]
#len(uniquelist)*0,25
ii = len(trainlist)/5
print("!!!!!!!!!!")
print(ii)
for i in range(int(ii)):
    x = random.choice(trainlist)
    ll.append(x)
    trainlist.remove(x)
print(len(testlist))
print(len(ll))
dt = data.copy()
trainset = dt[dt['session_id'].isin(ll)]
print("@@@@@@@@@@@@@@@@@@@@@@@@@@@")
print(trainset.info())
#tainset ^^^
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"+str(time.time() - start_time))
ll = []
ii = len(testlist)/5
print("!!!!!!!!!!")
print(ii)
for i in range(int(ii)):
    x = random.choice(testlist)
    ll.append(x)
    testlist.remove(x)
testset = data[data['session_id'].isin(ll)]
gtset = testset.copy()
print(testset.info())
print(len(testset['session_id']))
print(testset['session_id'])
print(testset['session_id'].value_counts())
print("010")
w = testset['session_id'].value_counts().to_frame()
print(w)
for i in range(len(testset['session_id'])):
    print(testset.values[i][5])
print("!!!!")
for i in range(len(w)):
    testset.loc[((testset['session_id'] == w.index.values[i]) & (testset['step'] == w['session_id'].values[i])), 'reference'] = ''
for i in range(len(testset['session_id'])):
    print(testset.values[i][5])

trainset.to_csv('C:\\Users\\Nico\\Desktop\\SR\\1\\1\\data\\train.csv')
gtset.to_csv('C:\\Users\\Nico\\Desktop\\SR\\1\\1\\data\\GroundTruth.csv')
testset.to_csv('C:\\Users\\Nico\\Desktop\\SR\\1\\1\\data\\test.csv')
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"+str(time.time() - start_time))