delete_empty_boxes

delete_empty_boxes.py

import os
import icecream

path = '/home/astraadmin/Desktop/rep/drawing_text_recognition/new/tesseract-tutorial/tesstrain/data/OKBM_NX-ground-truth'

def delete_empty_boxes(path):
    for i, file_name in enumerate(os.listdir(path)):
        if (i+1)%3==0:
            full_path_box = path + '/' + file_name
            #print(file_name)

            if (os.stat(full_path_box).st_size == 0):
                os.remove(full_path_box)
                print(f'removed{full_path_box}')
                full_path_gttxt = path + '/' + file_name.split('.')[0] + '.gt.txt'
                full_path_tif = path + '/' + file_name.split('.')[0] + '.tif'
                if os.path.exists(full_path_gttxt):
                    os.remove(full_path_gttxt)
                    print(f'removed{full_path_gttxt}')
                if os.path.exists(full_path_tif):
                    os.remove(full_path_tif)
                    print(f'removed{full_path_tif}')
#delete_empty_boxes(path)


def delete_files(path):
    list_files = os.listdir(path)#[10000:]
    #list_files.sort()
    icecream.ic(list_files)
    for i, file_name in enumerate(list_files):
        full_path_box = path + '/' + file_name
        full_path_gttxt = path + '/' + file_name.split('.')[0] + '.gt.txt'
        full_path_tif = path + '/' + file_name.split('.')[0] + '.tif'
        full_path_lstm = path + '/' + file_name.split('.')[0] + '.lstmf'
        
        #if not os.path.exists(full_path_lstm):
        if os.path.exists(full_path_box):
            os.remove(full_path_box)
            print(f'removed: {full_path_box}')
        if os.path.exists(full_path_gttxt):
            os.remove(full_path_gttxt)
            print(f'removed: {full_path_gttxt}')
        if os.path.exists(full_path_tif):
            os.remove(full_path_tif)
            print(f'removed: {full_path_tif}')

#delete_files(path)

#list_files = os.listdir(path)
#print(list_files)