Textbook-Arrange/main.py

import cv2
import pytesseract
import os
import numpy
directory = 'images'

# iterate over files in
# that directory
for filename in os.listdir(directory):
    f = os.path.join(directory, filename)
    # checking if it is a file
    if os.path.isfile(f):
        image = cv2.imread(f, 0)
        thresh = 255 - cv2.threshold(image, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
        #bottom 2 locations
        ROI1 = thresh[1383:1447,77:154] 
        ROI2 = thresh[1383:1447,950:1027]
        ROI = cv2.hconcat([ROI1, ROI2])
        data = pytesseract.image_to_string(ROI, lang='eng',config='--psm 7 -c tessedit_char_whitelist=0123456789i' )
        print(data)
        if os.path.isfile(f'images/{data}.jpg') == False:
            os.rename(f'{f}', f'images/{data}.jpg')
        elif os.path.isfile(f'images/{data}.jpg') == True:
            cv2.imshow(data, ROI)
            cv2.waitKey(1)
            manpage = input("please input the number on the page (if the title is correct, enter nothing)")
            #check if the new name already exists
            if manpage == "\n":
                rng = numpy.random.default_rng().random()
                os.rename (f'images/{data}.jpg', f'images/{data}.review-{rng}.jpg')
                os.rename(f'{f}', f'images/{data}.jpg')
            elif manpage != "\n":
                if os.path.isfile(f'images/{manpage}.jpg') == True:
                    rng = numpy.random.default_rng().random()
                    os.rename (f'images/{manpage}.jpg', f'images/{manpage}.review-{rng}.jpg')
                    os.rename(f'{f}', f'images/{manpage}.jpg')
                elif os.path.isfile(f'images/{manpage}.jpg') == False:
                    os.rename(f'{f}', f'images/{manpage}.jpg')
            cv2.destroyAllWindows()
        cv2.destroyAllWindows()