# Clustering and other steps :)
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import style
style.use("ggplot")
from sklearn.cluster import KMeans
from scipy.stats import norm
import argparse

argparser = argparse.ArgumentParser()

argparser.add_argument("--filename", help="input_file", required=False)

argparser.add_argument("--measurements", help="is 10-measurements provided", default = "yes", required=False)

argparser.add_argument("--clusters", help="number of clusters", default = 2, required=False)

argparser.add_argument("--output", help="number of clusters", required=False)

args = argparser.parse_args()
if(args.filename == None):
    filename = raw_input("Enter the name of your input data set (.csv) without file type: ")
    print("\n **Please make sure your data set include id feature** \n")
    measurements = raw_input("Are 10-measurements included in file as features T1 ... T10 (yes(y)/no(n))? ")
    print("\n **In case of \'No\', you should put *mean* and *std* (standard deviation) for each record. Header of features should be mean and std respectfully** \n")
    cluster_num= raw_input("Enter number of clusters to divide data set (default is 2): ")
    if(cluster_num == ""):
        cluster_numbers = 2
    else:
        try:
            cluster_numbers = int(cluster_num)
        except ValueError:
            print("That's not an int!")
    cluster_image = raw_input("Enter the name of plot file for clustering: ")
    cluster_output = raw_input("Enter the name of output data set (.csv) file without file type: ")

    df = pd.read_csv("Clustering_input/" + filename+".csv")
else:
    filename = args.filename
    df = pd.read_csv(filename)
    measurements = args.measurements
    cluster_num = args.clusters
    cluster_image = args.output
    cluster_output = args.output
    if(cluster_num == ""):
        cluster_numbers = 2
    else:
        try:
            cluster_numbers = int(cluster_num)
        except ValueError:
            print("Clusrer_number should be integer")


if(measurements == "yes" or measurements == "y" or measurements == ""):
    df_T = df[['T1','T2','T3','T4','T5','T6','T7','T8','T9','T10']]
    df['mean'] = (df['T1'] + df['T2'] + df['T3'] + df['T4'] + df['T5'] + df['T6'] + df['T7'] + df['T8'] + df['T9'] + df['T10'])/(10)
    std = np.array(df_T).std(1)
    df['std'] = pd.DataFrame(std)

Mean = np.array(df['mean'].reshape(-1,1))
np.round(Mean,2)
ST_DIV = np.array(df['std'].reshape(-1,1))
np.round(ST_DIV,2)

kmeans = KMeans(n_clusters=cluster_numbers)
kmeans.fit(Mean)

centroids = kmeans.cluster_centers_
labels = kmeans.labels_

if(cluster_numbers < 8):
    colors = ["g.","r.","c.","y.","b.","k.","m."]
else:
    colors = cluster_numbers*["g.","r.","c.","y.","b.","k.","m."]

for i in range(len(Mean)):
    plt.plot(i, Mean[i][0], colors[labels[i]], markersize = 10)

if(args.filename == None):
    plt.savefig("Clustering_results/" + cluster_image+".png")
    print("Cluster plot has generated. Please wait to generate final data set for Classification step. We are going to calculate weight and label for each record!")
else:
    plt.savefig(cluster_image+".png")

label_set = set(labels)

myMin = []
myMax = []
flag = True
for set in label_set:
    for i in range(len(Mean)):
        if labels[i]==set:
            if flag:
                min,max = Mean[i][0],Mean[i][0]
                flag = False
            else:
                temp = Mean[i][0]
                if(temp < min):
                    min = Mean[i][0]
                if(temp > max):
                    max = Mean[i][0]
    myMin.append(min)
    myMax.append(max)
    flag = True

minNP = np.array(myMin)
maxNP = np.array(myMax)

minNP = np.sort(minNP)
maxNP = np.sort(maxNP)
minNP = np.delete(minNP,0,0)
maxNP = np.delete(maxNP,-1,0)

intervals = [0]
for i in range(len(minNP)):
    temp = (minNP[i] + maxNP[i])/2
    intervals.append(temp)
intervals.append(intervals[-1]*1000)

rows, columns = len(Mean), cluster_numbers
Clusters = [[0 for x in range(columns)] for y in range(rows)]

if(measurements == "yes" or measurements == "y" or measurements == ""):
    df.drop(['T1','T2','T3','T4','T5','T6','T7','T8','T9','T10','std','mean'],1,inplace=True)
else:
    df.drop(['std','mean'],1,inplace=True)

header = list(df.columns.values)
X = np.array(df)
m, n = X.shape
X1 = np.zeros((m*cluster_numbers,n+2))
new_rows = 0
for i in range(len(Mean)):
    for j in range(len(intervals)-1):
        interval1 = intervals[j]
        interval2 = intervals[j+1]
        Clusters[i][j] = norm(Mean[i][0],ST_DIV[i][0]).cdf(interval2) - norm(Mean[i][0],ST_DIV[i][0]).cdf(interval1)
        Clusters[i][j] = round(Clusters[i][j],2)
        if(Clusters[i][j] > 0.99):
            Clusters[i][j] = 1
        elif(Clusters[i][j] < 0.01):
            Clusters[i][j] = 0
    for k in range(cluster_numbers):
        if(Clusters[i][k] > 0):
            X1[new_rows,:n] = X[i,:]
            X1[new_rows,n] = Clusters[i][k] * 100
            X1[new_rows,n+1] = k
            new_rows += 1

header.append('weight')
header.append('label')

i = 0
while i < len(X1):
    if(X1[i,n] == 0.00):
        X1 = X1[~np.all(X1 == 0, axis=1)]
    else:
        i += 1
df2 = pd.DataFrame(X1, columns=header)
df2.set_index('id', inplace=True)

if(args.filename == None):
    df2.to_csv("Clustering_results/" + cluster_output+".csv")
else:
    df2.to_csv(cluster_output+".csv")