# -*- coding: utf-8 -*-
# https://github.com/bmoyles0117/Vehicle-Color-Detector/blob/master/image_processor.py

from PIL import Image

class Colors(object):
	class Color(object):
		def __init__(self, value):
			self.value = value

		def __str__(self):
			return "%s : %s" % (self.__class__.__name__, self.value)

	class Red(Color): pass
	class Blue(Color): pass
	class Green(Color): pass
	class Yellow(Color): pass
	class White(Color): pass
	class Gray(Color): pass
	class Black(Color): pass
	class Pink(Color): pass
	class Teal(Color): pass

class ColorWheel(object):
    def __init__(self, rgb):
        r, g, b = rgb

        self.rgb = (Colors.Red(r), Colors.Green(g), Colors.Blue(b), )
	
    def estimate_color(self):
        dominant_colors = self.get_dominant_colors()
        total_colors = len(dominant_colors)
        if total_colors == 1:
            return dominant_colors[0]
        elif total_colors == 2:
            color_classes = [x.__class__ for x in dominant_colors]
            if Colors.Red in color_classes and Colors.Green in color_classes:
                return Colors.Yellow(dominant_colors[0].value)
            elif Colors.Red in color_classes and Colors.Blue in color_classes:
                return Colors.Pink(dominant_colors[0].value)
            elif Colors.Blue in color_classes and Colors.Green in color_classes:
                return Colors.Teal(dominant_colors[0].value)
        elif total_colors == 3:
            if dominant_colors[0].value > 200:
                return Colors.White(dominant_colors[0].value)
            elif dominant_colors[0].value > 100:
                return Colors.Gray(dominant_colors[0].value)
            else:
                return Colors.Black(dominant_colors[0].value)
        else:
            print ("Dominant Colors : %s" % dominant_colors)
	
    def get_dominant_colors(self):
        max_color = max([x.value for x in self.rgb])
        return [x for x in self.rgb if x.value >= max_color * .85]

def process_image(image):
    image_color_quantities = {}
    width, height = image.size
    width_margin = int(width - (width * .65))
    height_margin = int(height - (height * .75))
    for x in range(width_margin, width - width_margin, 4):
        for y in range(height_margin, height - height_margin):
            r, g, b = image.getpixel((x, y))
            key = "%s:%s:%s" % (r, g, b, )
            key = (r, g, b, )
            image_color_quantities[key] = image_color_quantities.get(key, 0) + 1

    total_assessed_pixels = sum([v for k, v in image_color_quantities.items() if v > 10])
    strongest_color_wheels = [(ColorWheel(k), v / float(total_assessed_pixels) * 100, ) for k, v in image_color_quantities.items() if v > 10]

    final_colors = {}
    
    st_color = ''
    strong = 0

    for color_wheel, strength in strongest_color_wheels:
        color = color_wheel.estimate_color()
        final_colors[color.__class__] = final_colors.get(color.__class__, 0) + strength
            
    for color, strength in final_colors.items():
        #print ("%s - %s" % (color.__name__, strength, ))
        if strong<strength:
            strong=strength
            st_color = color.__name__

    #image.show()
    return st_color

if __name__ == '__main__':
    import pandas as pd
    from tqdm import tqdm
    import os
    
    meta = pd.read_csv('../input/metadata.csv')[:2000]
    uids = meta['id']
    
    test = '../input/test/'
    
    colors = []
    for car in tqdm(uids, miniters=10):
        if os.path.isfile(test+car+'_05.jpg') :
            path=test
        else:
            path='../input/train/'         
        image = Image.open(path+car+'_05.jpg')
        color = process_image(image)
        colors.append(color)
    
    meta['color'] = colors
    
    meta.to_csv('metadata2.csv', index=False)