Work on Tyrano Script
This commit is contained in:
parent
47647abe40
commit
e2d47a9489
2 changed files with 310 additions and 1 deletions
|
|
@ -8,6 +8,7 @@ from modules.rpgmakermvmz import handleMVMZ
|
|||
from modules.rpgmakerace import handleACE
|
||||
from modules.csvtl import handleCSV
|
||||
from modules.textfile import handleTextfile
|
||||
from modules.tyrano import handleTyrano
|
||||
|
||||
THREADS = 10 # For GPT4 rate limit will be hit if you have more than 1 thread.
|
||||
|
||||
|
|
@ -29,7 +30,7 @@ def main():
|
|||
totalCost = 0
|
||||
version = ''
|
||||
while version == '':
|
||||
version = input('Select the RPGMaker Version:\n\n1. MV/MZ\n2. ACE\n3. CSV (From Translator++)\n4. Text (Custom)\n')
|
||||
version = input('Select the RPGMaker Version:\n\n1. MV/MZ\n2. ACE\n3. CSV (From Translator++)\n4. Text (Custom)\n5. Tyrano\n')
|
||||
match version:
|
||||
case '1':
|
||||
# Open File (Threads)
|
||||
|
|
@ -85,6 +86,20 @@ def main():
|
|||
tracebackLineNo = str(traceback.extract_tb(sys.exc_info()[2])[-1].lineno)
|
||||
print(Fore.RED + str(e) + '|' + tracebackLineNo + Fore.RESET)
|
||||
|
||||
case '5':
|
||||
# Open File (Threads)
|
||||
with ThreadPoolExecutor(max_workers=THREADS) as executor:
|
||||
futures = [executor.submit(handleTyrano, filename, estimate) \
|
||||
for filename in os.listdir("files") if filename.endswith('ks')]
|
||||
|
||||
for future in as_completed(futures):
|
||||
try:
|
||||
totalCost = future.result()
|
||||
|
||||
except Exception as e:
|
||||
tracebackLineNo = str(traceback.extract_tb(sys.exc_info()[2])[-1].lineno)
|
||||
print(Fore.RED + str(e) + '|' + tracebackLineNo + Fore.RESET)
|
||||
|
||||
case _:
|
||||
version = ''
|
||||
|
||||
|
|
|
|||
294
modules/tyrano.py
Normal file
294
modules/tyrano.py
Normal file
|
|
@ -0,0 +1,294 @@
|
|||
from concurrent.futures import ThreadPoolExecutor, as_completed
|
||||
import json
|
||||
import os
|
||||
from pathlib import Path
|
||||
import re
|
||||
import sys
|
||||
import textwrap
|
||||
import threading
|
||||
import time
|
||||
import traceback
|
||||
import tiktoken
|
||||
|
||||
from colorama import Fore
|
||||
from dotenv import load_dotenv
|
||||
import openai
|
||||
from retry import retry
|
||||
from tqdm import tqdm
|
||||
|
||||
#Globals
|
||||
load_dotenv()
|
||||
openai.organization = os.getenv('org')
|
||||
openai.api_key = os.getenv('key')
|
||||
|
||||
APICOST = .002 # Depends on the model https://openai.com/pricing
|
||||
PROMPT = Path('prompt.txt').read_text(encoding='utf-8')
|
||||
THREADS = 10 # For GPT4 rate limit will be hit if you have more than 1 thread.
|
||||
LOCK = threading.Lock()
|
||||
WIDTH = 60
|
||||
LISTWIDTH = 60
|
||||
MAXHISTORY = 10
|
||||
ESTIMATE = ''
|
||||
TOTALCOST = 0
|
||||
TOKENS = 0
|
||||
TOTALTOKENS = 0
|
||||
NAMESLIST = []
|
||||
|
||||
#tqdm Globals
|
||||
BAR_FORMAT='{l_bar}{bar:10}{r_bar}{bar:-10b}'
|
||||
POSITION=0
|
||||
LEAVE=False
|
||||
|
||||
# Flags
|
||||
NAMES = True # Output a list of all the character names found
|
||||
BRFLAG = False # If the game uses <br> instead
|
||||
FIXTEXTWRAP = False
|
||||
|
||||
def handleTyrano(filename, estimate):
|
||||
global ESTIMATE, TOKENS, TOTALTOKENS, TOTALCOST
|
||||
ESTIMATE = estimate
|
||||
|
||||
if estimate:
|
||||
start = time.time()
|
||||
translatedData = openFiles(filename)
|
||||
|
||||
# Print Result
|
||||
end = time.time()
|
||||
tqdm.write(getResultString(['', TOKENS, None], end - start, filename))
|
||||
if NAMES == True:
|
||||
tqdm.write(str(NAMESLIST))
|
||||
with LOCK:
|
||||
TOTALCOST += TOKENS * .001 * APICOST
|
||||
TOTALTOKENS += TOKENS
|
||||
TOKENS = 0
|
||||
|
||||
return getResultString(['', TOTALTOKENS, None], end - start, 'TOTAL')
|
||||
|
||||
else:
|
||||
try:
|
||||
with open('translated/' + filename, 'w', encoding='UTF-8') as outFile:
|
||||
start = time.time()
|
||||
translatedData = openFiles(filename, outFile)
|
||||
|
||||
# Print Result
|
||||
outFile.writeLines(translatedData[0])
|
||||
end = time.time()
|
||||
tqdm.write(getResultString(translatedData, end - start, filename))
|
||||
with LOCK:
|
||||
TOTALCOST += translatedData[1] * .001 * APICOST
|
||||
TOTALTOKENS += translatedData[1]
|
||||
except Exception as e:
|
||||
traceback.print_exc()
|
||||
return 'Fail'
|
||||
|
||||
return getResultString(['', TOTALTOKENS, None], end - start, 'TOTAL')
|
||||
|
||||
def openFiles(filename, outFile):
|
||||
with open('files/' + filename, 'r', encoding='utf-8') as readFile:
|
||||
translatedData = parseTyrano(readFile, outFile, filename)
|
||||
|
||||
return translatedData
|
||||
|
||||
def parseTyrano(readFile, outFile, filename):
|
||||
totalTokens = 0
|
||||
totalLines = 0
|
||||
global LOCK
|
||||
|
||||
# Get total for progress bar
|
||||
data = readFile.readlines()
|
||||
totalLines = len(data)
|
||||
|
||||
with tqdm(bar_format=BAR_FORMAT, position=POSITION, total=totalLines, leave=LEAVE) as pbar:
|
||||
pbar.desc=filename
|
||||
pbar.total=totalLines
|
||||
|
||||
try:
|
||||
final = translateTyrano(data, outFile, pbar)
|
||||
except Exception as e:
|
||||
traceback.print_exc()
|
||||
return [data, totalTokens, e]
|
||||
return [readFile, totalTokens, None]
|
||||
|
||||
def translateTyrano(data, outFile, pbar):
|
||||
textHistory = []
|
||||
maxHistory = MAXHISTORY
|
||||
tokens = 0
|
||||
currentGroup = []
|
||||
syncIndex = 0
|
||||
global LOCK, ESTIMATE
|
||||
|
||||
for i in range(len(data)):
|
||||
if syncIndex > i:
|
||||
i = syncIndex
|
||||
|
||||
# Speaker
|
||||
if '#' in data[i]:
|
||||
matchList = re.findall(r'#(.+)', data[i])
|
||||
if len(matchList) != 0:
|
||||
response = translateGPT(matchList[0], 'Reply with only the english translation of the NPC name', True)
|
||||
speaker = response[0]
|
||||
tokens += response[1]
|
||||
data[i] = '#' + speaker
|
||||
else:
|
||||
speaker = ''
|
||||
|
||||
# Lines
|
||||
elif '[p]' in data[i]:
|
||||
matchList = re.findall(r'(.+?)\[p\]', data[i])
|
||||
if len(matchList) > 0:
|
||||
matchList[0] = matchList[0].replace('「', '')
|
||||
matchList[0] = matchList[0].replace('」', '')
|
||||
currentGroup.append(matchList[0])
|
||||
if len(data) > i+1:
|
||||
while '[p]' in data[i+1]:
|
||||
i += 1
|
||||
matchList = re.findall(r'(.+?)\[p\]', data[i])
|
||||
if len(matchList) > 0:
|
||||
matchList[0] = matchList[0].replace('「', '')
|
||||
matchList[0] = matchList[0].replace('」', '')
|
||||
currentGroup.append(matchList[0])
|
||||
# Join up 401 groups for better translation.
|
||||
if len(currentGroup) > 0:
|
||||
finalJAString = ''.join(currentGroup)
|
||||
oldjaString = finalJAString
|
||||
|
||||
#Check Speaker
|
||||
if speaker == '':
|
||||
response = translateGPT(finalJAString, 'Previous Dialogue: ' + '\n\n'.join(textHistory), True)
|
||||
tokens += response[1]
|
||||
translatedText = response[0]
|
||||
textHistory.append('\"' + translatedText + '\"')
|
||||
else:
|
||||
response = translateGPT(speaker + ': ' + finalJAString, 'Previous Dialogue: ' + '\n\n'.join(textHistory), True)
|
||||
tokens += response[1]
|
||||
translatedText = response[0]
|
||||
textHistory.append('\"' + translatedText + '\"')
|
||||
|
||||
# Remove added speaker
|
||||
translatedText = translatedText.replace(speaker + ': ', '')
|
||||
|
||||
# Set Data
|
||||
translatedText = translatedText.replace('\"', '')
|
||||
data[i] = translatedText + '[p]'
|
||||
|
||||
currentGroup = []
|
||||
pbar.update(1)
|
||||
syncIndex = i + 1
|
||||
|
||||
return [tokens, data]
|
||||
|
||||
def getResultString(translatedData, translationTime, filename):
|
||||
# File Print String
|
||||
tokenString = Fore.YELLOW + '[' + str(translatedData[1]) + \
|
||||
' Tokens/${:,.4f}'.format(translatedData[1] * .001 * APICOST) + ']'
|
||||
timeString = Fore.BLUE + '[' + str(round(translationTime, 1)) + 's]'
|
||||
|
||||
if translatedData[2] == None:
|
||||
# Success
|
||||
return filename + ': ' + tokenString + timeString + Fore.GREEN + u' \u2713 ' + Fore.RESET
|
||||
|
||||
else:
|
||||
# Fail
|
||||
try:
|
||||
raise translatedData[2]
|
||||
except Exception as e:
|
||||
traceback.print_exc()
|
||||
errorString = str(e) + Fore.RED
|
||||
return filename + ': ' + tokenString + timeString + Fore.RED + u' \u2717 ' +\
|
||||
errorString + Fore.RESET
|
||||
|
||||
def subVars(jaString):
|
||||
varRegex = r'\\+[a-zA-Z]+\[[0-9a-zA-Z\\\[\]]+\]+|[\\]+[#a-zA-Z]+|[\\.<>]+'
|
||||
count = 0
|
||||
|
||||
varList = re.findall(varRegex, jaString)
|
||||
if len(varList) != 0:
|
||||
for var in varList:
|
||||
jaString = jaString.replace(var, '[v' + str(count) + ']')
|
||||
count += 1
|
||||
|
||||
return [jaString, varList]
|
||||
|
||||
def resubVars(translatedText, varList):
|
||||
count = 0
|
||||
|
||||
# Fix Spacing and ChatGPT Nonsense
|
||||
matchList = re.findall(r'@\s?[0-9]+?', translatedText)
|
||||
if len(matchList) > 0:
|
||||
for match in matchList:
|
||||
text = match.replace(' ', '')
|
||||
translatedText = translatedText.replace(match, text)
|
||||
|
||||
if len(varList) != 0:
|
||||
for var in varList:
|
||||
translatedText = translatedText.replace('@' + str(count) + '', var)
|
||||
count += 1
|
||||
|
||||
# Remove Color Variables Spaces
|
||||
# if '\\c' in translatedText:
|
||||
# translatedText = re.sub(r'\s*(\\+c\[[1-9]+\])\s*', r' \1', translatedText)
|
||||
# translatedText = re.sub(r'\s*(\\+c\[0+\])', r'\1', translatedText)
|
||||
return translatedText
|
||||
|
||||
@retry(exceptions=Exception, tries=5, delay=5)
|
||||
def translateGPT(t, history, fullPromptFlag):
|
||||
with LOCK:
|
||||
# If ESTIMATE is True just count this as an execution and return.
|
||||
if ESTIMATE:
|
||||
global TOKENS
|
||||
enc = tiktoken.encoding_for_model("gpt-3.5-turbo")
|
||||
TOKENS += len(enc.encode(t)) * 2 + len(enc.encode(history)) + len(enc.encode(PROMPT))
|
||||
return (t, 0)
|
||||
|
||||
# Sub Vars
|
||||
varResponse = subVars(t)
|
||||
subbedT = varResponse[0]
|
||||
|
||||
# If there isn't any Japanese in the text just skip
|
||||
if not re.search(r'[一-龠]+|[ぁ-ゔ]+|[ァ-ヴー]+', subbedT):
|
||||
return(t, 0)
|
||||
|
||||
"""Translate text using GPT"""
|
||||
context = 'Eroge Names Context: 桐嶋 香織 == Kaori Kirishima | Female, 肉山 猛 == Takeshi Nikuyama'
|
||||
if fullPromptFlag:
|
||||
system = PROMPT
|
||||
user = 'Line to Translate: ' + subbedT
|
||||
else:
|
||||
system = 'You are an expert translator who translates everything to English. Reply with only the English Translation of the text.'
|
||||
user = 'Line to Translate: ' + subbedT
|
||||
response = openai.ChatCompletion.create(
|
||||
temperature=0,
|
||||
frequency_penalty=0.2,
|
||||
presence_penalty=0.2,
|
||||
model="gpt-3.5-turbo",
|
||||
messages=[
|
||||
{"role": "system", "content": system},
|
||||
{"role": "user", "content": context},
|
||||
{"role": "user", "content": history},
|
||||
{"role": "user", "content": user}
|
||||
],
|
||||
request_timeout=30,
|
||||
)
|
||||
|
||||
# Save Translated Text
|
||||
translatedText = response.choices[0].message.content
|
||||
tokens = response.usage.total_tokens
|
||||
|
||||
# Resub Vars
|
||||
translatedText = resubVars(translatedText, varResponse[1])
|
||||
|
||||
# Remove Placeholder Text
|
||||
translatedText = translatedText.replace('English Translation: ', '')
|
||||
translatedText = translatedText.replace('Translation: ', '')
|
||||
translatedText = translatedText.replace('Line to Translate: ', '')
|
||||
translatedText = translatedText.replace('English Translation:', '')
|
||||
translatedText = translatedText.replace('Translation:', '')
|
||||
translatedText = translatedText.replace('Line to Translate:', '')
|
||||
translatedText = re.sub(r'\n\nPast Translated Text:.*', '', translatedText, 0, re.DOTALL)
|
||||
translatedText = re.sub(r'Note:.*', '', translatedText)
|
||||
|
||||
# Return Translation
|
||||
if len(translatedText) > 15 * len(t) or "I'm sorry, but I'm unable to assist with that translation" in translatedText:
|
||||
return [t, response.usage.total_tokens]
|
||||
else:
|
||||
return [translatedText, tokens]
|
||||
Loading…
Reference in a new issue