monument_string_android_ios_cleaning.py

# Python 3 !!
# folder structures:
'''
haominshi@MacBook-Pro-de-Haomin:~/Desktop$ tree diff
diff
├── Android_xml_content.txt
├── Android_xml_key.txt
├── Android_xml_key_stylized.txt
├── android_string_sorted.txt
├── android_string_sorted_final.txt
├── clean.py // this is this gist file
├── iOS_git.txt
├── iOS_git_cleaned.txt
├── iOS_git_en.txt
├── iOS_git_en_lower_case.txt
├── iOS_string_matched_android.txt
├── iOS_string_sorted.txt
├── langs
│   ├── english.txt
│   ├── english_sort.txt
│   ├── franch.txt
│   ├── franch_sort.txt
│   ├── german.txt
│   ├── german_sort.txt
│   └── strings_comparisons_with_title.txt
├── original.txt
├── text2.txt
└── text2_cleaned_all.txt

1 directory, 22 files
'''

import re

# remove all the blank lines in the stream


def clean_spaces_in_file(read_file_dir, write_file_dir):
	file_r = open(read_file_dir, "r")
	file_w = open(write_file_dir, "w")
	for line in file_r:
		if (line == "\n"):
			continue
		file_w.write(line)

# remove all the spaces and "" in the string e.g: "ddafasfdasf" = adsfsafasf


def clean_file_format_sorted(read_file_dir, write_file_dir):
	file_r = open(read_file_dir, "r")
	file_w = open(write_file_dir, "w")
	arrToWrite = []
	for line in file_r:
		cleaned_string = re.search('"(.*)" ', line)
		if (cleaned_string is None):
			file_w.write("||||||||||||||||||||||||||||||||||||||||||||||||||| \n")
		else:
			cleaned_string = cleaned_string.group()
			cleaned_string = cleaned_string.replace('"', '')
			cleaned_string = cleaned_string.replace(' ', '')
			arrToWrite.append(cleaned_string)

	arrToWrite.sort()
	for item in arrToWrite:
		if item is "\n":
			continue
		file_w.write(item + "\n")


def sort_strings(read_file_dir, write_file_dir):
	file_r = open(read_file_dir, "r")
	file_w = open(write_file_dir, "w")
	arrToWrite = []
	for line in file_r:
		arrToWrite.append(line)
	arrToWrite.sort()
	for item in arrToWrite:
		file_w.write(item + "\n")

# This method will take sorted ./lang/english_sort.txt ... etc
# and output them with each others translatrion to "./langs/strings_comparisons_with_title.txt"
# group by titles


def group_different_langs_via_title(de_dir, fr_dir, en_dir):
	de_r = open(de_dir, "r")
	fr_r = open(fr_dir, "r")
	en_r = open(en_dir, "r")

	file_w = open("./langs/strings_comparisons_with_title.txt", "w")

	en_arr = []
	fr_arr = []
	de_arr = []

	for line in de_r:
		if line is None:
			continue
		de_arr.append(line)
	for line in fr_r:
		if line is None:
			continue
		fr_arr.append(line)
	for line in en_r:
		if line is None:
		    continue
		en_arr.append(line)

	if ((en_arr.__len__() == de_arr.__len__()) and (en_arr.__len__() == fr_arr.__len__())):
		array_length = en_arr.__len__()

	output_arr = []
	for row in range(array_length):
		all_strings = en_arr[row] + fr_arr[row] + de_arr[row]
		output_arr.append(all_strings)
	for row in output_arr:
		file_w.write(row)

def concatanate():
	''''''
	basefile = open("./original.txt", "r")
	strings_comparison_w_titles = open("./iOS_git_en.txt", "r")
	# step 1: save all the english translation to an arr
	original_arr_title = []
	for line in basefile:
		if line is "\n": continue
		if line is None: continue
		original_arr_title.append(line.strip())
	# step 2: use the strings from the original_arr_title[] arr to get their titles
	title_arr = []
	for i in original_arr_title: print(i)
	for sentence in original_arr_title:
		for line in strings_comparison_w_titles:
			if sentence in line:
				cleaned_string = re.search('"(.*)" = ', line)
				if (cleaned_string is None):
					title_arr.append("||||||||||||||||||||||||||||||||||||||||||||||||||| \n")
				else:
					cleaned_string = cleaned_string.group()
					title_arr.append(cleaned_string)
			#else:
				#print(line + " xxxx " + sentence)
	for i in title_arr: print(i)

def everythingToLowerCase(intput_filename, output_filename):
	intput_file_r = open(intput_filename, 'r')
	output_file_w = open(output_filename, 'w')
	for line in intput_file_r:
		output_file_w.write(line.lower())

def stylizeXlsForDiff_1(intput_filename_1, intput_filename_2, output_filename, trailing = ""):
	# this one is for adding "" to the string
	intput_file_r_1 = open(intput_filename_1, 'r')
	intput_file_r_2 = open(intput_filename_2, 'r')
	output_file_w = open(output_filename, 'w')
	for line in intput_file_r_1:
		output_file_w.write('"')
		output_file_w.write(line.strip().lower())
		output_file_w.write('"')
		output_file_w.write(trailing)
		output_file_w.write('"')
		output_file_w.write(intput_file_r_2.readline().lower().strip())
		output_file_w.write('"')
		output_file_w.write('\n')

def sortTxt(intput_filename, output_filename):
	input_file_r = open(intput_filename, "r")
	output_file_w = open(output_filename, "w")
	array = []
	for line in input_file_r:
		array.append(line)
	array.sort()
	for item in array:
		output_file_w.write(item)

def extractAndroidStringFromIOS(intput_filename, input_keys, output_filename):
	input_file_r = open(intput_filename, 'r')
	input_keys_r = open(input_keys, 'r')
	output_file_w = open(output_filename, 'w')
	for line in input_file_r:
		for keyItem in input_keys_r:
			key = keyItem.strip()
			if key in line:
				output_file_w.write(line)
		
def stripAndAddTrailing():
	input_file_r = open("./android_string_sorted.txt", 'r')
	output_file_w = open("./android_string_sorted_final.txt", 'w')
	for line in input_file_r:
		output_file_w.write(line.strip() + ";\n")

# Control
run_steps = [9]

for step in run_steps:
	if (step is 1):
		# step 1 -
		# cleaned up iOS git Raw and sheets export
		# sorted to compare titles for IDing unmathced strings
		# text2.txt // source = sheets-export-ENGLISH
		# iOS_git.txt // source = iOS git repo
		clean_file_format_sorted("./text2.txt", "./text2_cleaned_all.txt")
		clean_file_format_sorted("./iOS_git.txt", "./iOS_git_cleaned.txt")
		# Above CHECKED: all TITLE match
	if (step is 2):
		# step 2 -
		# compare the different in english strings
		# we still uses text2.txt and iOS_git.txt
		# we first sorted the file and output it
		# text2.txt // source = sheets-export-ENGLISH
		sort_strings("./text2.txt", "./text2_sorted.txt")
		sort_strings("./iOS_git.txt", "./iOS_git_sorted.txt")
		# Above SORTED: now can do diff
	if (step is 3):
		# step 3 -
		# compare the titles of DE, FR, EN to ensure there is no differences
		# or to identify differences
		clean_file_format_sorted("./langs/english.txt",
		                         "./langs/english_title_sort.txt")
		clean_file_format_sorted(
			"./langs/franch.txt", "./langs/franch_title_sort.txt")
		clean_file_format_sorted(
			"./langs/german.txt", "./langs/german_title_sort.txt")
		# Above is SORTED TITLE export: now can do diff - CHECKS OUT
	if (step is 4):
		sort_strings("./langs/english.txt", "./langs/english_sort.txt")
		sort_strings("./langs/franch.txt", "./langs/franch_sort.txt")
		sort_strings("./langs/german.txt", "./langs/german_sort.txt")
		group_different_langs_via_title(
			"./langs/english_sort.txt", "./langs/franch_sort.txt", "./langs/german_sort.txt")
	if (step is 5):
		concatanate()
	if (step is 6):
		# set ios localize string output to lower case
		everythingToLowerCase("./iOS_git_en.txt", "./iOS_git_en_lower_case.txt")
	if (step is 7):
		# create file that has android localize string to lower case for diff
		# the params for this method is a file that linked to:
		# first exprot android studio xml string res: to xlsx via http://convertcsv.com/xml-to-csv.htm
		# put the keys to param 1 // just column 1
		# put the values to param 2 // just column 2
		# the method will concate these first 2 param to one single file "./Android_xml_key_stylized.txt",  with " = " seperate key vals
		stylizeXlsForDiff_1("./Android_xml_key.txt", "./Android_xml_content.txt", "./Android_xml_key_stylized.txt", " = ")
	if (step is 8):
		# sort the output k v pair of xml files to do comparisons
		sortTxt("./Android_xml_key_stylized.txt", "./android_string_sorted.txt")
		sortTxt("./iOS_git_en_lower_case.txt", "./iOS_string_sorted.txt")
	# if (step is 9): # FAILED, the KEYS are different
	# 	extractAndroidStringFromIOS("./iOS_string_sorted.txt", "./Android_xml_key.txt", "./iOS_string_matched_android.txt")
	if (step is 9):
		stripAndAddTrailing()