# Libraries import json import os import re import util.dazedwrap as dazedwrap import threading import time import traceback import tiktoken from pathlib import Path from colorama import Fore from dotenv import load_dotenv from retry import retry from tqdm import tqdm from util.translation import TranslationConfig, translateAI as sharedtranslateAI, getPricingConfig, calculateCost import tempfile # Globals MODEL = os.getenv("model") TIMEOUT = int(os.getenv("timeout")) LANGUAGE = os.getenv("language").capitalize() PROMPT = Path("prompt.txt").read_text(encoding="utf-8") VOCAB = Path("vocab.txt").read_text(encoding="utf-8") LOCK = threading.Lock() WIDTH = int(os.getenv("width")) LISTWIDTH = int(os.getenv("listWidth")) NOTEWIDTH = 70 MAXHISTORY = 10 ESTIMATE = "" TOKENS = [0, 0] NAMESLIST = [] NAMES = False # Output a list of all the character names found BRFLAG = False # If the game uses
instead FIXTEXTWRAP = True # Overwrites textwrap IGNORETLTEXT = False # Ignores all translated text. MISMATCH = [] # Lists files that throw a mismatch error (Length of GPT list response is wrong) FILENAME = None TIMETOTAL = 0 # Total Time Taken for all translations TIMETOTAL = 0 # Total Time Taken for all translations # tqdm Globals BAR_FORMAT = "{l_bar}{bar:10}{r_bar}{bar:-10b}" POSITION = 0 LEAVE = False PBAR = None # Regex - Need to change this if you want to translate from/to other languages. Default is Japanese Regex LANGREGEX = r"[一-龠ぁ-ゔァ-ヴーa-zA-Z0-9\uFF61-\uFF9F]+" # Get pricing configuration based on the model PRICING_CONFIG = getPricingConfig(MODEL) INPUTAPICOST = PRICING_CONFIG["inputAPICost"] OUTPUTAPICOST = PRICING_CONFIG["outputAPICost"] BATCHSIZE = PRICING_CONFIG["batchSize"] FREQUENCY_PENALTY = PRICING_CONFIG["frequencyPenalty"] # Initialize Translation Config TRANSLATION_CONFIG = TranslationConfig( model=MODEL, language=LANGUAGE, prompt=PROMPT, vocab=VOCAB, langRegex=LANGREGEX, batchSize=BATCHSIZE, maxHistory=MAXHISTORY, estimateMode=False # Will be set dynamically based on ESTIMATE ) LEAVE = False def handleRegex(filename, estimate): global ESTIMATE, TOKENS, FILENAME ESTIMATE = estimate FILENAME = filename # Translate start = time.time() translatedData = openFiles(filename) # Translate if not estimate: try: with open("translated/" + filename, "w", encoding="utf-8-sig") as outFile: outFile.writelines(translatedData[0]) except Exception: traceback.print_exc() return "Fail" # Print File end = time.time() tqdm.write(getResultString(translatedData, end - start, filename)) with LOCK: TOKENS[0] += translatedData[1][0] TOKENS[1] += translatedData[1][1] # Print Total totalString = getResultString(["", TOKENS, None], end - start, "TOTAL") # Print any errors on maps if len(MISMATCH) > 0: return totalString + Fore.RED + f"\nMismatch Errors: {MISMATCH}" + Fore.RESET else: return totalString def getResultString(translatedData, translationTime, filename): global TIMETOTAL # File Print String cost = calculateCost(translatedData[1][0], translatedData[1][1], MODEL) totalTokenstring = ( Fore.YELLOW + "[Input: " + str(translatedData[1][0]) + "]" "[Output: " + str(translatedData[1][1]) + "]" "[Cost: ${:,.4f}".format(cost) + "]" ) if filename != "TOTAL": timeString = Fore.BLUE + "[" + str(round(translationTime, 1)) + "s]" TIMETOTAL += round(translationTime, 1) else: timeString = Fore.BLUE + "[" + str(round(TIMETOTAL, 1)) + "s]" if translatedData[2] == None: # Success return filename + ": " + totalTokenstring + timeString + Fore.GREEN + " \u2713 " + Fore.RESET else: # Fail try: raise translatedData[2] except Exception as e: traceback.print_exc() errorString = str(e) + Fore.RED return filename + ": " + totalTokenstring + timeString + Fore.RED + " \u2717 " + errorString + Fore.RESET def openFiles(filename): with open("files/" + filename, "r", encoding="utf-8-sig") as readFile: translatedData = parseRegex(readFile, filename) # Delete lines marked for deletion finalData = [] for line in translatedData[0]: if line != "\\d\n": finalData.append(line) translatedData[0] = finalData return translatedData def parseRegex(readFile, filename): global PBAR totalTokens = [0, 0] # Read File into data data = readFile.readlines() # Create Progress Bar with tqdm(bar_format=BAR_FORMAT, position=POSITION, leave=LEAVE) as pbar: pbar.desc = filename PBAR = pbar try: result = translateRegex(data, filename, []) totalTokens[0] += result[0] totalTokens[1] += result[1] except Exception as e: traceback.print_exc() return [data, totalTokens, e] return [data, totalTokens, None] def save_progress_lines(lines, filename, encoding="utf-8-sig"): """Atomically save current line-based translation progress.""" try: if ESTIMATE: return os.makedirs("translated", exist_ok=True) tmp_fd, tmp_path = tempfile.mkstemp(prefix=f"{filename}.", suffix=".tmp", dir="translated") try: with os.fdopen(tmp_fd, "w", encoding=encoding, newline="\n", errors="ignore") as tmp_file: tmp_file.writelines(lines) os.replace(tmp_path, os.path.join("translated", filename)) finally: if os.path.exists(tmp_path): try: os.remove(tmp_path) except OSError: pass except Exception: traceback.print_exc() def translateRegex(data, filename, translatedList): if translatedList: stringList = translatedList[0] choiceList = translatedList[1] else: stringList = [] choiceList = [] tokens = [0, 0] speaker = "" global LOCK, ESTIMATE, FILENAME, PBAR, MISMATCH i = 0 while i < len(data): voice = False lineRegexText = r'●(?:「| |()+(.+?)(?:[」)]|$)+' lineRegexSpeaker = r"\\text\(\"(.+?)\"\)" choiceRegex = r"\$menu_item.+?,(.*?)," titleRegex = r"title\s'(.*)'$" setgamedatatitleRegex = r'\\setgamedatatitle\("(.+?)"\)' dlgRegex = r'\\dlg\("(.+?)"\)' selRegex = r'\\sel\((.+)\)' speaker = "" # Setgamedatatitle match = re.search(setgamedatatitleRegex, data[i]) if match: # Pass 1 - Translate immediately (not batched) if not translatedList: response = translateAI( match.group(1).replace('\\n', '\n'), f"Reply with the {LANGUAGE} translation of the chapter title", True, ) tokens[0] += response[1][0] tokens[1] += response[1][1] title = response[0] # Convert newlines back to escape sequences and escape quotes title = title.replace('\n', '\\n').replace('"', '\\"') data[i] = data[i].replace(match.group(1), title) save_progress_lines(data, filename) # Title match = re.search(titleRegex, data[i]) if match: # Pass 1 - Translate immediately (not batched) if not translatedList: response = translateAI( match.group(1), f"Reply with the {LANGUAGE} translation of the chapter title", True, ) tokens[0] += response[1][0] tokens[1] += response[1][1] title = response[0] # Set title = re.sub(r"(?", ")") # Textwrap translatedText = dazedwrap.wrapText(translatedText, width=WIDTH) # Set Data while "\\endtext" not in data[i].strip(): del data[i] data.insert(i, f"{translatedText}\n") save_progress_lines(data, filename) # Choices match = re.search(choiceRegex, data[i]) if match: # Pass 1 if not translatedList: choiceList.append(match.group(1)) # Pass 2 else: # Grab and Pop translatedText = choiceList[0] choiceList.pop(0) # Replace Spaces translatedText = translatedText.replace("\u3000", " ") # Escape Quotes translatedText = re.sub(r'(?