Delete duplicated nikola meta data in translated posts (must be in nikola's translated paths)

nikola_translated.py

# Delete duplicated nikola meta data in translated posts (must be in nikola's translated paths)

import os, sys, codecs
import shutil

# recursive, paths from working dir
FROM_PATH = ""
OUTPUT_PATH = None #if none, replace file

def file_open_save(path, mode, text=False):
	if sys.version_info[0] == 3:
		if not mode == 'w':
			return open(path, mode, encoding='utf-8')
		else:
			try:
				# make dirs in path tree from path + filename
				os.makedirs(os.path.dirname(path))
			except OSError:
				pass
			with open(path, mode, encoding='utf-8') as outputFile:
				outputFile.write(text)
	else:
		if not mode == 'w':
			return codecs.open(path, mode, encoding='utf-8')
		else:
			try:
				# make dirs in path tree from path + filename
				os.makedirs(os.path.dirname(path))
			except OSError:
				pass
			with codecs.open(path, mode, encoding='utf-8') as outputFile:
				outputFile.write(text)

def find_output_path(path):
	if OUTPUT_PATH:
		return os.path.join(OUTPUT_PATH, os.path.relpath(path, FROM_PATH))
	else:
		return path


if not os.path.exists(FROM_PATH) and not os.path.isdir(FROM_PATH):
	print (" Something wrong")
	sys.exit(0)

if OUTPUT_PATH and (not os.path.exists(OUTPUT_PATH) or not os.path.isdir(OUTPUT_PATH)):
	print (" Output path must be an existing folder")
	sys.exit(0)

print (" Starting..")


# full paths 
FROM_PATH = os.path.abspath(FROM_PATH)
if OUTPUT_PATH:
	OUTPUT_PATH = os.path.abspath(OUTPUT_PATH)

# get filepaths
files_list = list()
for root, subFolders, files in os.walk(FROM_PATH):
	for filename in files:
		filePath = os.path.join(root, filename)

		if os.path.exists(filePath):
			files_list.append(filePath)

if len(files_list) == 0:
	print (" No files found")
	sys.exit(0)	

# "root" filename
pieces = files_list[0].split(".")
prev_file = pieces[0]
prev_file_path = files_list[0]

new_path = find_output_path(prev_file_path)
if OUTPUT_PATH:
	shutil.copy2(prev_file_path, new_path)

for this_path in files_list[1:]:
	if this_path.split(".")[0] == prev_file:

		tmp_prev = os.path.relpath(prev_file_path, FROM_PATH)
		tmp_this = os.path.relpath(this_path, FROM_PATH)
		print ("  Translation found: " + tmp_prev + " -> " + tmp_this)
		
		# find default lang meta, that's only what we need
		meta_main = list()
		with file_open_save(prev_file_path, 'r') as the_text:
			for line in the_text:
				if line and not line.isspace():
					meta_main.append(line)
				else:
					break 

		# modify translated post
		with file_open_save(this_path, 'r') as the_text:
			file_translaton = the_text.readlines()

		meta_trans = list()
		for l_number, line in enumerate(file_translaton):
			if line and not line.isspace():

				if line not in meta_main:
					meta_trans.append(line)

			else:
				file_new = meta_trans + file_translaton[l_number:]
				file_new = "".join(file_new)
				break

		#save
		new_path = find_output_path(this_path)
		file_open_save(new_path, 'w', file_new)

	else:
		prev_file = this_path.split(".")[0]
		prev_file_path = this_path

		if OUTPUT_PATH:
			new_path = find_output_path(this_path)
			shutil.copy2(this_path, new_path)

print (" Finished.")