import sqlite3 as sql
from scipy.stats import logistic
import math
import numpy as np
import numpy.random as rnd
import matplotlib.pyplot as plt
from sklearn.neural_network import MLPRegressor

sdata =[]
sql1= "select ticker, date, open, high, low, close, vol \
    from Hist_1m where ticker_id=1 order by Date;"
con=sql.connect('C:/Users/ubase/Documents/StockDB/StockDB21.sqlite')
cur=con.cursor()
cur.execute(sql1)
sdata=cur.fetchall()
con.commit()
con.close()

Ldata = len(sdata)
N = 8000 # Количество сделок
ld = 5 #Продолжительность сделки
NNinterval = 20 # Количество входов NN

# Генерация случайных чисел
rng = rnd.default_rng()
rm=rng.integers(0, Ldata, N )

class Candle:
    tr = 0
    dt = 1
    o = 2
    h = 3
    l = 4
    c = 5
    v = 6
    
cl = Candle
DataC =[sdata[i][cl.c] for i in range(0,Ldata)]

# sigmoid линейность до 0.5
def sigmoidnorm(x, alfa = 0.9, xmin = -1.3, xmax = 1.3):
    return (xmax - xmin)*((1 / (1 + math.exp(-x*2.0*alfa))) - 1.0) + xmax

x = [0.002 * i - 3 for i in range(0,3000)]
y = [sigmoidnorm(x[i]) for i in range(len(x))]


plt.plot(x,y)
plt.grid()
plt.show()

# формируем сделки.
def DealsGenL(rm,ld):
   #Lm = len(rm)
   ix = []
   x = []
   pr = []
   
   for i in range(0,N):
        if rm[i] + ld < Ldata and rm[i] - NNinterval - 1 > 0:
            delta = (sdata[rm[i]+ld][cl.c] - sdata[rm[i]][cl.c])/sdata[rm[i]+ld][cl.c]*100
            x0 = [sigmoidnorm((sdata[rm[i] - j][cl.c] - sdata[rm[i]][cl.c])/sdata[rm[i]][cl.c]*100) \
                 for j in range(0, NNinterval)]
            ix.append(rm[i])
            x.append(x0)
            pr.append(delta)
   return ix, x, pr


Ix, X, Pr = DealsGenL(rm,ld)



Ib = 0
Ie = 100

plt.plot(X)
plt.legend()
plt.grid()
plt.show()


plt.plot(Pr, label = 'Prof')
plt.legend()
plt.grid()
plt.show()


regr = MLPRegressor(hidden_layer_sizes = [30,20,15,10,5], \
                    max_iter=500, activation = 'tanh')

regr.fit(X, Pr)
Out = regr.predict(X)

plt.plot(Pr, Out, '.')
plt.grid()
plt.show()