User:Moussekateer/voicelines.py

# -*- coding: utf-8 -*-
import os, re

wavfilesLocation = """""" #Set location of wav files here.

game = 'Portal 2' #Set game here. Either 'Portal' or 'Portal 2'.
language = 'fi' #Set language code here. See scipts dictionary below for list of codes.
character = 'Wheatley' #Set Character here. See characters dictionary below for list of names. Not used for Portal.

portalscripts = {
		'da': 'closecaption_danish.txt',
		'nl': 'closecaption_dutch.txt',
		'en': 'closecaption_english.txt',
		'fi': 'closecaption_finnish.txt',
		'fr': 'closecaption_french.txt',
		'de': 'closecaption_german.txt',
		'it': 'closecaption_italian.txt',
		'ja': 'closecaption_japanese.txt',
		'ko': 'closecaption_korean.txt',
		'ka': 'closecaption_koreana.txt',
		'no': 'closecaption_norwegian.txt',
		'pl': 'closecaption_polish.txt',
		'po': 'closecaption_portuguese.txt',
		'ru': 'closecaption_russian.txt',
		'zh-hans': 'closecaption_schinese.txt',
		'es': 'closecaption_spanish.txt',
		'sw': 'closecaption_swedish.txt',
		'zh-hant': 'closecaption_tchinese.txt',
	}

portal2scripts = {
		'cz': 'subtitles_czech.txt',
		'da': 'subtitles_danish.txt',
		'nl': 'subtitles_dutch.txt',
		'en': 'subtitles_english.txt',
		'fi': 'subtitles_finnish.txt',
		'fr': 'subtitles_french.txt',
		'de': 'subtitles_german.txt',
		'hu': 'subtitles_hungarian.txt',
		'it': 'subtitles_italian.txt',
		'ja': 'subtitles_japanese.txt',
		'ko': 'subtitles_korean.txt',
		'ka': 'subtitles_koreana.txt',
		'no': 'subtitles_norwegian.txt',
		'pl': 'subtitles_polish.txt',
		'po': 'subtitles_portuguese.txt',
		'ro': 'subtitles_romanian.txt',
		'ru': 'subtitles_russian.txt',
		'zh-hans': 'subtitles_schinese.txt',
		'es': 'subtitles_spanish.txt',
		'sw': 'subtitles_swedish.txt',
		'zh-hant': 'subtitles_tchinese.txt',
		'th': 'subtitles_thai.txt',
		'tu': 'subtitles_turkish.txt'
	}

characters = {
		'Cave Johnson': 'cavejohnson',
		'Caroline': 'glados.caroline',
		'GLaDOS': 'glados',
		'Wheatley': 'sphere03',
		'Adventure core': 'core03',
		'Fact core': 'core02',
		'Space core': 'core01',
		'Turret': 'turret',
		'Floor Turret': 'npc_floorturret',
		'Defective Turret': 'turret',
		'Announcer': 'announcer'
	}

def utf8encode(file):
    s1 = open(file, "rb").read()
    s2 = unicode(s1, "utf-16")
    s3 = s2.encode("utf-8")
    open (file, "wb").write(s3)

if game == 'Portal':
	utf8encode(portalscripts[language])
	script = open(portalscripts[language], 'rb').read()
elif game == 'Portal 2':
	# utf8encode(portal2scripts[language])
	script = open(portal2scripts[language], 'rb').read()

files = os.listdir(wavfilesLocation)
files.sort()

failedmatches = []

if game == 'Portal 2':
	for f in files:
		if character == 'Defective Turret' or character == 'Turret' or character == 'Floor Turret':
			pattern = "\"" + characters[character] + "." + f[:-4] + "\"\s\"<clr:(.[^>]+)><low>(.[^:]+): (.[^\"]+)\""
		else:
			pattern = "\"" + characters[character] + "." + f[:-4] + "\"\s\"<clr:(.[^>]+)>(.[^:]+): (.[^\"]+)\""
		regexp = re.compile(pattern, re.IGNORECASE)
		result = regexp.search(script)
		try:
			if language == 'en':
				print ('*[[Media:' + character + '_' + f + '|\"' + result.group(3) + '"]]')
			else:
				print ('*[[Media:' + character + '_' + f[:-4] + '_' + language + f[-4:] + '|\"' + result.group(3) + '"]]')
		except AttributeError:
			failedmatches.append(f)
			if language == 'en':
				print ('*[[Media:' + character + '_' + f + '|\"\"]]')
			else:
				print ('*[[Media:' + character + '_' + f[:-4] + '_' + language + f[-4:] + '|\"\"]]')
elif game == 'Portal':
	for f in files:
		pattern = "\"portal." + f[:-4] + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
		regexp = re.compile(pattern, re.MULTILINE)
		result = regexp.search(script)
		#BRACE YOURSELF
		if result is None:
			pattern = "\"portal." + f[:-4] + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			pattern = "\"" + f[:-4] + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			pattern = "\"" + f[:-4] + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			pattern = "\"portal.glados." + f[:-4] + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			pattern = "\"portal.glados." + f[:-4] + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[16:-4]
			pattern = "\"portal.glados.core_" + g.replace('-', '_') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[16:-4]
			pattern = "\"portal.glados.core_" + g.replace('-', '_') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[17:-4]
			pattern = "\"portal.glados_core." + g.replace('-', '_') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[17:-4]
			pattern = "\"portal.glados_core." + g.replace('-', '_') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[17:-4]
			pattern = "\"portal.glados_core." + g.replace('-', '_').replace('0', '') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None:
			g = f[17:-4]
			pattern = "\"portal.glados_core." + g.replace('-', '_').replace('0', '') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		#Some funny file -> caption mappings
		if result is None and f[:25] == 'escape_02_sphere_cakemix-':
			pattern = "\"portal.glados_core." + f[:-4].replace('escape_02_sphere_cakemix-', 'crazy_') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f[:25] == 'escape_02_sphere_cakemix-':
			pattern = "\"portal.glados_core." + f[:-4].replace('escape_02_sphere_cakemix-', 'crazy_') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f[:21] == 'escape_02_spherenag1-':
			stripzeropattern = "escape_02_spherenag1-0(\d).wav"
			regexpzero = re.compile(stripzeropattern, re.IGNORECASE)
			g = f
			if regexpzero.search(f):
				g = "escape_02_spherenag1-" + regexpzero.search(f).group(1) + ".wav"
			pattern = "\"portal.glados." + g[:-4].replace('escape_02_spherenag1-', 'core_nag_') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f[:21] == 'escape_02_spherenag1-':
			stripzeropattern = "escape_02_spherenag1-0(\d).wav"
			regexpzero = re.compile(stripzeropattern, re.IGNORECASE)
			g = f
			if regexpzero.search(f):
				g = "escape_02_spherenag1-" + regexpzero.search(f).group(1) + ".wav"
			pattern = "\"portal.glados." + g[:-4].replace('escape_02_spherenag1-', 'core_nag_') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f[:22] == 'escape_02_spheredrop1-':
			stripzeropattern = "escape_02_spheredrop1-0(\d).wav"
			regexpzero = re.compile(stripzeropattern, re.IGNORECASE)
			g = "escape_02_spheredrop1-" + regexpzero.search(f).group(1) + ".wav"
			pattern = "\"portal.glados." + g[:-4].replace('escape_02_spheredrop1-', 'core_drop_') + "\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f[:22] == 'escape_02_spheredrop1-':
			stripzeropattern = "escape_02_spheredrop1-0(\d).wav"
			regexpzero = re.compile(stripzeropattern, re.IGNORECASE)
			g = "escape_02_spheredrop1-" + regexpzero.search(f).group(1) + ".wav"
			pattern = "\"portal.glados." + g[:-4].replace('escape_02_spheredrop1-', 'core_drop_') + "\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-00.wav':
			pattern = "\"portal.glados.found_me\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-00.wav':
			pattern = "\"portal.glados.found_me\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-01.wav':
			pattern = "\"portal.glados.broke_my_heart\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-01.wav':
			pattern = "\"portal.glados.broke_my_heart\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-02.wav':
			pattern = "\"portal.glados.call_it_a_day\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-02.wav':
			pattern = "\"portal.glados.call_it_a_day\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-03.wav':
			pattern = "\"portal.glados.not_gonna_happen\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-03.wav':
			pattern = "\"portal.glados.not_gonna_happen\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-04.wav':
			pattern = "\"portal.glados.supprise_3_2_1\"[\t]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		if result is None and f == 'escape_02_entry-04.wav':
			pattern = "\"portal.glados.supprise_3_2_1\".[^\"]+\"<clr:(.[^>]+)><I>(.[^\"]+)\""
			regexp = re.compile(pattern, re.MULTILINE)
			result = regexp.search(script)
		try:
			if language == 'en':
				print ('*[[Media:GLaDOS ' + f + '|\"' + result.group(2) + '"]]')
			else:
				print ('*[[Media:GLaDOS ' + f[:-4] + '_' + language + f[-4:] + '|\"' + result.group(2) + '"]]')
		except AttributeError:
			failedmatches.append(f)
			if language == 'en':
				print ('*[[Media:GLaDOS ' + f + '|\"\"]]')
			else:
				print ('*[[Media:GLaDOS ' + f[:-4] + '_' + language + f[-4:] + '|\"\"]]')
print "Number of failed matches: " + str(len(failedmatches))