from PIL import Image
import pytesseract
from pybraille import convertText
import matplotlib.pyplot as plt
import re

text_kwargs = dict(fontsize=8)

def imgToText(path):
    raw = pytesseract.image_to_string(path)
    # remove special characters
    raw = raw.replace('€', 'Euro')
    # remove special characters that do not fit into braille notation
    raw = re.sub(r"[^,.\na-zA-Z0-9]", ' ', raw)
    rows = raw.split('\n')
    # remove rows with less than 3 characters
    rows = [row for row in rows if len(row) > 3]
    # remove empty rows
    rows = [row for row in rows if row]
    # remove prefix and suffix spaces and double spaces
    rows = [' '.join(row.split()) for row in rows]
    # concat row
    return '\n'.join(rows)

def textToBraille(text):
    return convertText(text)

def brailleToImg(text, braille):
    f, ax = plt.subplots()
    plt.axis('off')
    plt.text(0, 0.5, text,
             horizontalalignment='left',
             verticalalignment='center',
             c="dodgerblue",
             transform=ax.transAxes)
    plt.text(0.5, 0.5, braille,
             horizontalalignment='left',
             verticalalignment='center',
             transform=ax.transAxes)
    plt.show()

if __name__ == '__main__':
    text = imgToText('sample.jpg')
    braille = textToBraille(text)
    print(braille)
    brailleToImg(text, braille)