#!/usr/bin/python2.7

import numpy as np
from matplotlib import pyplot as plt
from dbtools import raw_data
from utils import moving_sum

def moving_sum(array, window):
    if type(array) is not np.ndarray:
        raise TypeError('Expected one dimensional numpy array.')

    remainder = array.size % window
    if 0 != remainder:
        array = array[remainder:]
    array = array.reshape((array.size/window,window))
    sum_arr = np.sum(array,axis=1)

    return sum_arr

def run():
    window = 3
    y_lst = raw_data('2018-08-03 00:00:00', length=3600*24)
    raw_arr = np.array(y_lst)

    sum_arr = moving_sum(raw_arr,window)
    res = np.true_divide(sum_arr[1:],sum_arr[:-1])
    
    threshold = 0.5
    upper = np.array([1+threshold]*res.size)
    lower = np.array([1-threshold]*res.size)
    
    plt.plot(upper,lw=1,color='red',label='Upper')
    plt.plot(res,lw=1,color='blue',label='Trend')
    plt.plot(lower,lw=1,color='red',label='Lower')

    r_idx = np.argwhere(np.abs(res-1) > 0.5)
    plt.plot(r_idx, res[r_idx], 'ro')

    plt.legend()
    plt.show()

    return (r_idx,res[r_idx])