vtt_formatter.py

#!/usr/bin/python
# -*- coding: utf-8 -*-
__author__ = 'AxelAli'
import os
import sys


#MADE BY AXEL ALI
#https://github.com/AxelAli



#USAGE:
# $ VTT Formatter.py [DIR]
# $ VTT Formatter.py ./subs

import re
print "Searching inside : "+sys.argv[1]

newdirectory = sys.argv[1]+"txt"
print "newdirectory", newdirectory
if not os.path.exists(newdirectory):
    os.makedirs(newdirectory) #Creates a newdirectory
combined = open(os.path.join(newdirectory,"allcombined.txt"),'w') #Creates a file of all the subs combined for dataset

for file in os.listdir(sys.argv[1]): #Gets All the files inside DIR (Argument1)
    if file.endswith(".vtt"):
        print "Formating : "+file #Shows current File
        newfile = open(os.path.join(newdirectory,file.replace(".vtt", ".txt")),'w') #Creates a newfile
        with open(os.path.join(sys.argv[1], file)) as f:
            contents_reached = False
            for line in f: #a line for each file
                # skip all lines meta deta till time line
                 if '-->' in line : #Couldnt Get the "or" working
                     contents_reached = True
                     continue
                 if contents_reached:
                     line = re.sub('<[^>]*>', '', line)
                     #print line, len(line)
                     if len(line) > 1:
                         newfile.write("%s" % line) #Add new line to new file
                         combined.write("%s" % line) #Add new line to combinedfile
            print "DONE!" #NEXT ONE!