import numpy as np
import math
import matplotlib.patches as mpatches
from mpl_toolkits.mplot3d import Axes3D
from matplotlib.pyplot import figure, xticks, plot, title, xlabel, ylabel, show, xlim, ylim, legend, scatter, subplot, \
    tight_layout, grid, annotate


def readFile(fileString, numberOfAttributes, numberOfObservations):
    returnArray = np.full((numberOfObservations, numberOfAttributes), '', dtype='object')
    counter = 0  # Is a incrementing counter, keepin track of position in current row
    for i in range(numberOfObservations):  # Goes through all observations of the data set
        for j in range(numberOfAttributes):  # Goes through all attributes of the data set
            while (True):
                if counter == len(fileString):
                    return returnArray
                if fileString[counter] == ',':  # checks, if next attribute is present
                    counter += 1
                    break  # skips to next attribute
                else:
                    returnArray[i, j] = str(returnArray[i, j]) + str(fileString[counter])
                    counter += 1
    return returnArray


# Loading and reading file
observations = 194
attributes = 30
test = open('../Data/flag.txt')
text = test.read()
dataInArray = readFile(text, attributes, observations)


def outOfK(inputArray, numberOfAttributes, numberOfObservations):
    lengthOfOutOfK = np.ones(30)
    lengthOfOutOfK[1] = 6
    lengthOfOutOfK[2] = 4
    lengthOfOutOfK[5] = 10
    lengthOfOutOfK[6] = 7
    lengthOfOutOfK[17] = 7
    lengthOfOutOfK[28] = 7
    lengthOfOutOfK[29] = 7
    returnArray = np.zeros((194, 71))  # creates returnArray
    returnCount = 0
    for i in range(numberOfAttributes):
        if i == 0:
            returnCount = returnCount
        elif i in (1, 2, 5, 6, 17, 28, 29):  # defines the positions to do 1-out-of-K
            for j in range(int(lengthOfOutOfK[i])):
                if i in (1, 2, 5, 6):  # already defined
                    if i == 6:
                        for h in range(194):
                            returnArray[h, returnCount + int((inputArray[h, i]))] = 1
                    else:
                        for h in range(194):
                            returnArray[h, returnCount + int((inputArray[h, i])) - 1] = 1
                elif i in (17, 28, 29):  # colors
                    for h in range(194):  # goes through all observations
                        if inputArray[h, i] == 'red':
                            returnArray[h, returnCount] = 1
                        if inputArray[h, i] == 'green':
                            returnArray[h, returnCount + 1] = 1
                        if inputArray[h, i] == 'blue':
                            returnArray[h, returnCount + 2] = 1
                        if inputArray[h, i] == 'gold':
                            returnArray[h, returnCount + 3] = 1
                        if inputArray[h, i] == 'white':
                            returnArray[h, returnCount + 4] = 1
                        if inputArray[h, i] == 'black':
                            returnArray[h, returnCount + 5] = 1
                        if inputArray[h, i] in ('orange', 'brown'):
                            returnArray[h, returnCount + 6] = 1
        else:
            returnArray[:, returnCount] = inputArray[:, i]
        if i == 1:
            returnCount += 5
        elif i == 2:
            returnCount += 3
        elif i == 5:
            returnCount += 9
        elif i in (6, 17, 28, 29):
            returnCount += 6
        returnCount += 1
    returnArray = np.delete(returnArray, 0, 1)  # removes attribute of countries
    returnArray = np.delete(returnArray, np.array([5, 6, 7, 8, 9, 10, 11]), 1)  # removes attribute of religion
    returnArray = np.delete(returnArray, 3, 1)  # removes attribute of population

    return returnArray


dataOutOfK_ALL = outOfK(dataInArray, attributes, observations)

outOfKAttributes = np.array(
    ['Landmass', 'Geographic quadrant', 'Area[km2]', 'Population[millions]', 'Language', 'Religion', 'Bars'
        , 'Stripes', 'colours', 'red', 'green', 'blue', 'gold', 'white', 'black', 'orange/brown', 'predominant colour',
     'Circles', 'Crosses', 'Saltires', 'Quarters', 'Sunstars', 'Crecent moon', 'Triangle', 'Icon', 'Animate', 'text',
     'Topleft-colour', 'Bottomright-colour'])

definingLandmass = np.array(['North America', 'South America', 'Europe', 'Africa', 'Asia', 'Oceania'])
definingZone = np.array(['NE', 'SE', 'SW', 'NW'])
# definingArea
definingPopulation = np.array(['<1m', '1-5m', '6-10m', '11-20m', '21-50m', '>50m'])
definingLanguage = np.array(['English', 'Spanish', 'French', 'German', 'Slavic', 'Other European', 'Chineses', 'Arabic',
                             'Japenese, \nTurkish, \nMagyar', 'Others'])
definingReligion = np.array(['Catholic', 'Other Christian', 'Muslim', 'Buddhist', 'Hindu', 'Ethnic', 'Marxist'])

### SETUP FOR STATISTICS ###
# Set 'columnnumber' to the desired column to investigate
# Set 'tal' to:
#       1: if number
#       0: if text

columnnumber = 1
tal = 1

# Ejects the desired data
if (tal == 1):
    vektorTest = dataInArray[:, columnnumber].astype(np.int)
else:
    vektorTest = dataInArray[:, columnnumber]

# print means/variance etc.


### SETUP FOR PCA FOR ALL ###

# Doing PCA with normalising for 'Flag'
# 1) Subtracts mean from each and devides med standard deviation
def normalizeSet(data, M):
    temp = M -1
    for i in range(temp):
        if (i >= 0 and i <= 5):  # Landmass
            data[:, i] = (data[:, i]) / math.sqrt(6)
        elif (i >= 6 and i <= 9):  # zone
            data[:, i] = (data[:, i]) / math.sqrt(4)
        elif (i >= 11 and i <= 20):  # language
            data[:, i] = (data[:, i]) / math.sqrt(10)
        elif (i >= 30 and i <= 36):  # Mainhue
            data[:, i] = (data[:, i]) / math.sqrt(7)
        elif (i >= 59):
            data[:, i] = (data[:, i]) / math.sqrt(7)
        else:
            data[:, i] = (data[:, i]) / np.std(data[:, i])
    return data

#dataOutOfK_ALL[:,10] = dataOutOfK_ALL[:,10] / np.std(dataOutOfK_ALL[:, 10])