import psycopg2
import struct
import array
import matplotlib.pyplot as plt
import numpy as np
from sklearn.svm import SVR
from sklearn.decomposition import PCA

def normalize(v):
    norm = np.linalg.norm(v)
    if norm == 0:
       return v
    return v / norm

def getMuestra(n):
    conn = psycopg2.connect(host="127.0.0.1",database="TPAS_FIX", user="postgres", password="pass")
    cur  = conn.cursor()
    cur.execute("SELECT * FROM muestravraw WHERE idmuestratiposuelo = " + str(n)) #120
    print("The number of raw sample: ", cur.rowcount)
    if (cur.rowcount<1):
        return 0
    row = cur.fetchone()
    m = {}
    for i in range(8):
    #m[i] = np.divide(row[i+2],10000.0)
        var = row[i+2]
        var = var[510:]
        #m[i] = normalize(row[i+2])
        m[i] = normalize(var)

    return m

def getAllMuestras():
    conn = psycopg2.connect(host="127.0.0.1",database="TPAS_FIX", user="postgres", password="pass")
    cur  = conn.cursor()
    cur.execute("SELECT * FROM muestravraw") #120
    print("The number of raw sample: ", cur.rowcount)
    row = cur.rowcount
    bd = {}
    while(True):
        row = cur.fetchone()
        m = {}
        if (row == None):
            break
        for j in range(8):
            var = row[j+2]
            var = var[510:]
            var1 = var[35:70]
            var2 = var[110:170]
            var1.extend(var2)
            #var = [var1[33], var1[39], var1[40], var1[46], var1[61], var1[62], var1[70], var1[71], var1[82]]
            #var = [sum(var[:4]), sum(var[4:-1])]
            var = var1
            m[j] = normalize(var)
        bd[str(row[0])] = m
        #print (row[0])
    return bd
    #m = {}
    #for i in range(8):
    ##m[i] = np.divide(row[i+2],10000.0)
    #    var = row[i+2]
    #    var = var[510:]
    #    #m[i] = normalize(row[i+2])
    #    m[i] = normalize(var)

    #return m

def getValues(muestra):
    conn = psycopg2.connect(host="127.0.0.1",database="TPAS_FIX", user="postgres", password="pass")
    cur  = conn.cursor()
    cur.execute("SELECT * FROM analisisquimico WHERE idmuestratiposuelo = " + str(muestra)) #120
    row = cur.fetchone()
    idQ = row[0]
    cur.execute("SELECT * FROM analisisquimicoelemento WHERE idanalisisquimico = " + str(idQ) + "and idelemento = 21") #120
    while(True):
        row = cur.fetchone()
        if (row == None):
            break
        return (row[2])



dic = getAllMuestras()
l = []
X = []
y = []
for key in dic:
    X.append(np.multiply(dic[key][0], 100000))
    l.append(key)
    y.append(getValues(key))

plt.plot(X[30])
plt.plot(X[10])
plt.plot(X[100])
plt.show()

print (len(X))
print (len(y))
#getValues(l[0])
X.pop(46)
y.pop(46)
pca = PCA(n_components = 5)
X = pca.fit(X).transform(X)

k = 80
y = np.multiply(y, 1)
X_train = X[:k]
y_train = y[:k]
X_test = X[k:110]
y_test = y[k:110]


X1 = np.asarray(X_train)
#X1 = X1.transpose()
y1 = np.asarray(y_train)
y_test = np.asarray(y_test)
t = np.linspace(1.0, k, num=k)
#svr_rbf = SVR(kernel='rbf', C=1e9, gamma=0.00001)
svr_rbf = SVR(kernel='poly', C=1e3, degree=5)
#svr_rbf = SVR(kernel='linear', C=1e3)

y_rbf = svr_rbf.fit(X1, y1).predict(X1)
plt.scatter(y1, t,color='darkorange', label='data')
plt.scatter(y_rbf, t, color='navy', label ='sim')
plt.show()

X_total = np.asarray(X)
y_total = np.asarray(y)
y_sim = svr_rbf.predict(X_test)
y_ori = y_test
t_comp = np.linspace(1, len(y_ori), len(y_ori))
plt.scatter(y_sim, t_comp, label="DATA SIMULADA")
plt.scatter(y_ori, t_comp, label="DATA")
plt.show()



t2 = np.linspace(1, len(y_test), len(y_test))

y_sim = svr_rbf.predict(X_test)
plt.scatter(y_sim, t2,color='darkorange', label='data')
plt.scatter(y_test, t2, color='navy', label ='sim')
plt.show()

dif = []
for i in range(len(y_test)):
    #print (str(y1[i]) + " " + str(y_rbf[i])+ " " + str(y1[i] - y_rbf[i] ) )
    d = abs(y_test[i]- y_train[i])
    res = d*100.0/y_test[i]
    dif.append(res)
    #print (i)#

plt.scatter(t2, dif)
plt.show()


#plt.scatter(err, t, label='error porc')
#plt.show()