directory = r'C:\Users\mnix0\Desktop\Stock Project Data\Data\nyse stocks'


# loads a a file in the form
# 20050225,4.1553,4.2213,4.148,4.1774,344615,0\n
# Where
# [date, open, high, low, close, volume, 0\n]
# and converts line contents to floats for processing
def load_list_from_file():
    input_file = open(r'C:\Users\mnix0\Desktop\Stock Project Data\agn.us.txt', 'r')

    data_list = []
    for line in input_file:
        # creates temporary list used for appending index
        # of each line once converted from string
        new_list = []
        # for some reason "line" was at times being read as '\n'
        # this fixes it, don't know why it happened
        if line != '\n':
            # splits string into list, removing
            # last index containing newline character
            value = line.split(',')[:-1]
            for item in value:
                try:
                    # appends float value to temporary list
                    new_list.append(float(item))
                # use to handle first line which consists of
                #
                # Date,Open,High,Low,Close,Volume,OpenInt
                #
                # in files used for data
                except ValueError:
                    print('OK')
            # appends to data list containing
            # contents of entire file
            data_list.append(new_list)


        # write function for testing
        # write_to_file(data_list)
    input_file.close()
    data_list= data_list[1:]
    add_price_move(data_list)

    def write_data_list_to_file(title_list, write_list):
    output_file = open(r'C:\Users\mnix0\Desktop\Stock Project Data\data_output.txt', 'w')
    sep = ','
    temp_str = ''
    # writes title list to file
    for index in title_list:
        temp_str += str(index) + ','
    output_file.write(temp_str+'\n')

    # joins each list found in each index
    # of main data list into string
    # by mapping index of of main data list
    # and joining that with , separator
    for index in write_list:
        temp_str = ''
        for item in index:
            temp_str += str(item).format('.4f')+','
        output_file.write(temp_str[:-1] + '\n')
    output_file.close()


# ITS POSSIBLY BROKEN
# This should work, it might be worth spot checking it again somewhere in the middle
# of a data file it hasn't been tested periods other than 5 as of 12/6/2019

# it returns a list with an average of the price increase / decrease
# for the length of time defined by period
# totals (current days close - previous days close)
# then divides it by the period

# IT WOULD BE A GOOD IDEA TO MODIFY THIS FOR THE LIST TO
# BE CONTAINING THE AVERAGE CHANGE BETWEEN CLOSES AS A
# PERCENTAGE OF THE STOCKS TOTAL PRICE
def calc_rolling_avg_price_move(data_list, period):
    avg_move_list = []
    # starts looping processes at index after range size
    # avoiding indexing errors
    for i in range(period+1, len(data_list)):
        temp_total = 0
        for x in range(period,0, -1):
            index = i-x
            prev_close_index = index-1
            change_for_day = data_list[index][4] - data_list[prev_close_index][4]
            temp_total += change_for_day

        temp_avg = temp_total/period
        avg_move_list.append([data_list[i][0],temp_avg])
        title_list = ["date", 'average move trailing ' + str(period)+ ' days']
    write_data_list_to_file(avg_move_list, title_list)


# currently accepts list and period
# calculates simple moving average for
# period default = 50
def calc_moving_average(data_list, period = 50):
    list_of_moving_averages = []
    for curr_date in range(period+1, len(data_list)):
        sma = 0
        sum_previous_days = 0
        for days_ago in range(period, 0, -1):
            sum_previous_days += data_list[curr_date-days_ago][4]
        sma = sum_previous_days/period
        temp_list = [data_list[curr_date][0], sma]
        list_of_moving_averages.append(temp_list)
    title_list = ['Date', 'SMA '+ str(period)]
    write_data_list_to_file(list_of_moving_averages, title_list)


def add_price_move(data_list, period = 14, di = 0, oi =1, hi = 2,
                   li = 3, ci = 4, vi = 5):
    move_list = [] # up, down

    for curr_date in range(1, len(data_list)):
        prev_date = curr_date - 1
        curr_day_close = data_list[curr_date][ci]
        prev_day_close = data_list[prev_date][ci]
        price_move = curr_day_close - prev_day_close
        if price_move > 0:
            move_up = price_move
            move_down = 0
        elif price_move < 0:
            move_up = 0
            move_down = -(price_move)
        else:
            move_up = 0
            move_down = 0
            # appends to list [DATE, UP MOVEMENT, DOWN MOVEMENT]
        date = data_list[curr_date][di]
        open = data_list[curr_date][oi]
        high = data_list[curr_date][hi]
        low = data_list[curr_date][li]
        close = data_list[curr_date][ci]
        vol = data_list[curr_date][vi]

        new_list = [date, open, high, low, close, move_up, move_down, vol]
        data_list[curr_date] = new_list

    title_list = ['Date', 'open','High','Low','Close','Move up', 'Move down','vol']

    write_data_list_to_file(title_list,data_list[1:])



def test_main():
    load_list_from_file()
    #crawl_directory(directory)

test_main()