myOCR/tools.py

import base64
import os
import re
from crnn.crnn import crnnOcr
from ocrapi import ocr


def get_image():
    image_list = []
    for i in os.listdir('./result/image'):
        if re.match('(\d+)-(\d+)-(\d+)-(\d+)',i):
            with open(os.path.join('./result/image',i), 'rb') as f:
                base64_data = base64.b64encode(f.read())
                s = base64_data.decode()
                y_min, y_max, x_min, x_max = re.match('(\d+)-(\d+)-(\d+)-(\d+)', i).groups()
                image_list.append({'left':x_min,'top':y_min,'b64':s})

    return image_list

def get_text():
    image_list = []
    for i in os.listdir('./result/text_img'):
        if re.match('(\d+)-(\d+)-(\d+)-(\d+)',i):
            with open(os.path.join('./result/text_img',i), 'rb') as f:

                s = crnnOcr(os.path.join('./result/text_img',i))
                # print(s)
                # s = ocr(os.path.join('./result/text_img',i))
                y_min, y_max, x_min, x_max = re.match('(\d+)-(\d+)-(\d+)-(\d+)', i).groups()
                image_list.append({'left':x_min,'top':y_min,
                       'w':int(x_max)-int(x_min),
                       'h':int(y_max)-int(y_min),
                       'size':int((int(y_max)-int(y_min))*0.6),
                       's':s})

                # print(i)
                # print({'left':x_min,'top':y_min,
                #        'w':int(x_max)-int(x_min),
                #        'h':int(y_max)-int(y_min),
                #        'size':int((int(y_max)-int(y_min))*0.6),
                #        's':s})

    return image_list
if __name__ == '__main__':
    get_text()