Process.py

import os
import shutil
import cv2
from PIL import Image
def rename_files():
	# renames all files in folders recursively
	try:
		folders = sorted(os.listdir(os.getcwd()))
		print(folders)
		folder_idx=1
		for folder in folders:
			print("in folders")
			file_idx=1
			if folder.endswith(".py"):
				continue
			for file in sorted(os.listdir(os.getcwd()+'/'+folder)):
				print("in files",file)
				ext = file[-4:]
				print(ext,end= ' ')		
				os.rename(os.getcwd()+'/'+folder+'/'+file,os.getcwd()+'/'+folder+'/'+folder+'_'+str(file_idx)+ext)
				file_idx+=1
			folder_idx+=1
			
	except Exception as e:
		raise e


def accumulate(lang=None):
	# collect all files to a single language folder
	if lang == None:
		raise ValueError
		return
	try:
		for folder in sorted(os.listdir(os.getcwd())):
			if folder.endswith(".py"):
				continue
			for file in sorted(os.listdir(os.getcwd()+'/'+folder)):
				new_file = file.replace(' ','_')
				shutil.move(os.getcwd()+'/'+folder+'/'+file,os.getcwd()+'/'+lang+'/'+new_file)
	except Exception as e:
		raise e

def preprocess(home=None):
	# remove garbage files and preprocess images
	if home == None:
		raise ValueError
		return
	try:
		for file in sorted(os.listdir(os.getcwd()+'/'+home)):
			print(file[-3:])
			if file[-3:]=='.db':
				continue
			img = cv2.imread(os.getcwd()+'/'+home+'/'+file,0)
			ret,img = cv2.threshold(img,127,255,cv2.THRESH_BINARY)
			cv2.imwrite(os.getcwd()+'/'+home+'/'+'Binarized_'+file,img)
	except Exception as e:
		raise e

def main():
	# driver module
	rename_files()
  inp = input("Enter path ").strip()
	accumulate(inp)
	preprocess(inp)

if __name__ == '__main__':
	main()

Translate.py

from googletrans import Translator
import time
import os
import shutil

def file_names(path=None):
	# return list of files in path
	if path==None:
		raise ValueError
	return sorted(os.listdir(path))

def translate_files(path=None):
	# translates and writes file
	parent = []
	# list of translated strings to be written
	translator = Translator() 
	# create translator object
	files = file_names(path)

	for file in files[len(os.listdir(path[:-7]+'/translated-results')):]:

		parent = ''
		flag_err = 0 # flag for files
		f_in = open(path+'/'+file,"r")
		f_out = open('Translated_'+file,"w+")
		write=''
		for line in f_in:
			parent+=line[:-1]+' '
			# needed so that Google API does not block the current IP address
		try:
			write = translator.translate(parent).text
		except :
			print("JSON Decode error occurred!")
			flag_err = 1
		f_out.writelines(write)
		time.sleep(1)

		f_in.close()
		f_out.close()
		print('Translating '+file+' ... done.')
		shutil.move(os.getcwd()+'/Translated_'+file, path[:-7]+'/translated-results/'+'Translated_'+file)
		if flag_err==1:
			os.rename(path[:-7]+'/translated-results/'+'Translated_'+file,path[:-7]+'/translated-results/'+'ERR-Translated_'+file)

def main():
	# driver module
	# checklist : path to dataset
	path = '/'+input("Enter path ")+'/'
	try:
		os.makedirs(os.getcwd()+path+'translated-results')
	except:
		pass # dir exists
	translate_files(os.getcwd()+path+'results')

if __name__ == '__main__':
	main()