diff --git a/modules/main.py b/modules/main.py index 9c15429..011b53d 100644 --- a/modules/main.py +++ b/modules/main.py @@ -9,7 +9,7 @@ from modules.rpgmakerace import handleACE from modules.csvtl import handleCSV from modules.textfile import handleTextfile -THREADS = 20 +THREADS = 20 # For GPT4 rate limit will be hit if you have more than 1 thread. # Info Message print(Fore.LIGHTYELLOW_EX + "WARNING: Once a translation starts do not close it unless you want to lose your\ diff --git a/modules/rpgmakermvmz.py b/modules/rpgmakermvmz.py index a761207..06e7cdd 100644 --- a/modules/rpgmakermvmz.py +++ b/modules/rpgmakermvmz.py @@ -23,10 +23,10 @@ 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 = 20 +THREADS = 20 # For GPT4 rate limit will be hit if you have more than 1 thread. LOCK = threading.Lock() -WIDTH = 70 -LISTWIDTH = 80 +WIDTH = 60 +LISTWIDTH = 32 MAXHISTORY = 10 ESTIMATE = '' TOTALCOST = 0 @@ -39,10 +39,10 @@ POSITION=0 LEAVE=False # Flags -CODE401 = True +CODE401 = False CODE405 = False -CODE102 = True -CODE122 = False +CODE102 = False +CODE122 = True CODE101 = False CODE355655 = False CODE357 = False @@ -52,6 +52,7 @@ CODE320 = False CODE324 = False CODE111 = False CODE408 = False +CODE108 = False def handleMVMZ(filename, estimate): global ESTIMATE, TOKENS, TOTALTOKENS, TOTALCOST @@ -424,6 +425,8 @@ def searchNames(name, pbar, context): name['profile'] = translatedText.strip('\"') translatedText = textwrap.fill(responseList[2], LISTWIDTH) name['nickname'] = translatedText.strip('\"') + if '<特徴1:' in name['note']: + tokens += translateNote(name, r'<特徴1:([^>]*)>') if 'Armors' in context or 'Weapons' in context: translatedText = textwrap.fill(responseList[1], LISTWIDTH) @@ -453,17 +456,20 @@ def searchCodes(page, pbar): ### IF these crash or fail your game will do the same. Use the flags to skip codes. ## Event Code: 401 Show Text - if page['list'][i]['code'] == 401 and CODE401 == True or page['list'][i]['code'] == 405 and CODE405: + if page['list'][i]['code'] == 401 and CODE401 == True or page['list'][i]['code'] == 405 and CODE405: + # Grab String jaString = page['list'][i]['parameters'][0] - if "peek inside" in jaString: - print('hi') + + # Catch Speaker + if jaString.startswith('　'): + jaString = jaString + ': ' + oldjaString = jaString jaString = jaString.replace('ﾞ', '') jaString = jaString.replace('。', '.') jaString = jaString.replace('・', '.') jaString = jaString.replace('‶', '') jaString = jaString.replace('”', '') - jaString = jaString.replace('ー', '-') jaString = jaString.replace('―', '-') jaString = jaString.replace('…', '...') jaString = re.sub(r'([\u3000-\uffef])\1{3,}', r'\1\1\1', jaString) @@ -479,7 +485,6 @@ def searchCodes(page, pbar): jaString = jaString.replace('・', '.') jaString = jaString.replace('‶', '') jaString = jaString.replace('”', '') - jaString = jaString.replace('ー', '-') jaString = jaString.replace('―', '-') jaString = jaString.replace('…', '...') jaString = re.sub(r'([\u3000-\uffef])\1{3,}', r'\1\1\1', jaString) @@ -500,8 +505,8 @@ def searchCodes(page, pbar): finalJAString = finalJAString.replace(match[0], speaker) # Need to remove outside code and put it back later - startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』a-zA-Z0-9Ａ-Ｚ０-９\\]+', finalJAString) - finalJAString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』a-zA-Z0-9Ａ-Ｚ０-９\\]+', '', finalJAString) + startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】（）[]<>「」『』a-zA-Z0-9Ａ-Ｚ０-９\\]+', finalJAString) + finalJAString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】（）[]<>「」『』a-zA-Z0-9Ａ-Ｚ０-９\\]+', '', finalJAString) if startString is None: startString = '' else: startString = startString.group() @@ -509,17 +514,20 @@ def searchCodes(page, pbar): finalJAString = re.sub(r'\n', ' ', finalJAString) # Translate - response = translateGPT(finalJAString, 'Previous Text for Context: ' + '\n\n'.join(textHistory), True) - tokens += response[1] - translatedText = response[0] + 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 + '\"') - # TextHistory is what we use to give GPT Context, so thats appended here. - # rawTranslatedText = re.sub(r'[\\<>]+[a-zA-Z]+\[[a-zA-Z0-9]+\]', '', translatedText) - textHistory.append('\"' + translatedText + '\"') - - # if speakerCaught == True: - # translatedText = speakerRaw + ':\n' + translatedText - # speakerCaught = False + # Remove added speaker + translatedText = translatedText.replace(speaker + ': ', '') + speaker = '' # Textwrap translatedText = textwrap.fill(translatedText, width=WIDTH) @@ -530,6 +538,7 @@ def searchCodes(page, pbar): # Set Data translatedText = translatedText.replace('ッ', '') translatedText = translatedText.replace('っ', '') + translatedText = translatedText.replace('ー', '') translatedText = translatedText.replace('\"', '') page['list'][i]['parameters'][0] = translatedText speaker = '' @@ -540,8 +549,13 @@ def searchCodes(page, pbar): textHistory.pop(0) currentGroup = [] - ## Event Code: 122 [Control Variables] [Optional] - if page['list'][i]['code'] == 122 and CODE122 == True: + ## Event Code: 122 [Set Variables] + if page['list'][i]['code'] == 122 and CODE122 == True: + # This is going to be the var being set. + varNum = page['list'][i]['parameters'][0] + if varNum != 1178: + continue + jaString = page['list'][i]['parameters'][4] if type(jaString) != str: continue @@ -553,37 +567,24 @@ def searchCodes(page, pbar): # If there isn't any Japanese in the text just skip if not re.search(r'[一-龠]+|[ぁ-ゔ]+|[ァ-ヴー]+', jaString): continue + + # Need to remove outside code and put it back later + matchList = re.findall(r"\'(.*?)\'", jaString) - # Definitely don't want to mess with files - if '\"' not in jaString: - continue + for match in matchList: + response = translateGPT(match, '', True) + translatedText = response[0] + tokens += response[1] - # Remove outside text - oldjaString = jaString - startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー\<\>【】\\]+', jaString) - jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー\<\>【】\\]+', '', jaString) - endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー\<\>【】。！？\\]+$', jaString) - jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー\<\>【】。！？\\]+$', '', jaString) - if startString is None: startString = '' - else: startString = startString.group() - if endString is None: endString = '' - else: endString = endString.group() + # Remove characters that may break scripts + charList = ['.', '\"', '\'', '\\n'] + for char in charList: + translatedText = translatedText.replace(char, '') - # Translate - response = translateGPT(jaString, '', True) - tokens += response[1] - translatedText = response[0] - - # Remove characters that may break scripts - charList = ['.', '\"', "\'"] - for char in charList: - translatedText = translatedText.replace(char, '') - - # Proper Formatting - translatedText = translatedText.replace('"', '\"') + jaString = jaString.replace(match, translatedText) # Set Data - translatedText = startString + translatedText + endString + translatedText = jaString page['list'][i]['parameters'][4] = translatedText ## Event Code: 357 [Picture Text] [Optional] @@ -759,14 +760,53 @@ def searchCodes(page, pbar): continue # Want to translate this script - if page['list'][i]['code'] == 408 and '\\>' not in jaString: + if 'ans:' not in jaString: continue # Need to remove outside code and put it back later startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】]+', jaString) jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】]+', '', jaString) - endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】。！？]+$', jaString) - jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】。！？]+$', '', jaString) + endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', jaString) + jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', '', jaString) + if startString is None: startString = '' + else: startString = startString.group() + if endString is None: endString = '' + else: endString = endString.group() + + # Translate + response = translateGPT(jaString, '', True) + tokens += response[1] + translatedText = response[0] + + # Remove characters that may break scripts + charList = ['.', '\"'] + for char in charList: + translatedText = translatedText.replace(char, '') + + translatedText = startString + translatedText + endString + + translatedText = translatedText.replace('"', '\"') + + # Set Data + page['list'][i]['parameters'][0] = translatedText + + ## Event Code: 108 (Script) + if (page['list'][i]['code'] == 108) and CODE108 == True: + jaString = page['list'][i]['parameters'][0] + + # If there isn't any Japanese in the text just skip + if not re.search(r'[一-龠]+|[ぁ-ゔ]+|[ァ-ヴー]+', jaString): + continue + + # Want to translate this script + if 'ask:' not in jaString: + continue + + # Need to remove outside code and put it back later + startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】]+', jaString) + jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】]+', '', jaString) + endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', jaString) + jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', '', jaString) if startString is None: startString = '' else: startString = startString.group() if endString is None: endString = '' @@ -799,12 +839,12 @@ def searchCodes(page, pbar): continue # Want to translate this script - if 'addLog' not in jaString: + if 'ShowInfo' not in jaString: continue # Need to remove outside code and put it back later - startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』]+', jaString) - jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』]+', '', jaString) + startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』\\]+', jaString) + jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】（）「」『』\\]+', '', jaString) endString = re.search(r' [^一-龠ぁ-ゔァ-ヴー\<\>【】（）「」『』 。！？]+$', jaString) jaString = re.sub(r' [^一-龠ぁ-ゔァ-ヴー\<\>【】（）「」『』 。！？]+$', '', jaString) if startString is None: startString = '' @@ -813,7 +853,7 @@ def searchCodes(page, pbar): else: endString = endString.group() # Translate - response = translateGPT(jaString, 'Reply with only the English Translation of the text.', True) + response = translateGPT(jaString, '', True) tokens += response[1] translatedText = response[0] @@ -823,7 +863,7 @@ def searchCodes(page, pbar): translatedText = translatedText.replace(char, '') # Cant have spaces? - translatedText = translatedText.replace(' ', '　') + translatedText = translatedText.replace(' ', ' ') # Textwrap translatedText = textwrap.fill(translatedText, width=1000) @@ -872,27 +912,28 @@ def searchCodes(page, pbar): if type(jaString) != str: continue + # Only TL the Game Variable + if '$gameVariables' not in jaString: + continue + # Need to remove outside code and put it back later - startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー\<\>【】]+', jaString) - jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー\<\>【】]+', '', jaString) - endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】 。！？]+$', jaString) - jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】 。！？]+$', '', jaString) - if startString is None: startString = '' - else: startString = startString.group() - if endString is None: endString = '' - else: endString = endString.group() + matchList = re.findall(r"'(.*?)'", jaString) + + for match in matchList: + response = translateGPT(match, '', True) + translatedText = response[0] + tokens += response[1] - response = translateGPT(jaString, 'Reply with only the english translation.', True) - translatedText = response[0] + # Remove characters that may break scripts + charList = ['.', '\"', '\'', '\\n'] + for char in charList: + translatedText = translatedText.replace(char, '') - # Remove characters that may break scripts - charList = ['.', '\"', '\\n'] - for char in charList: - translatedText = translatedText.replace(char, '') + jaString = jaString.replace(match, translatedText) # Set Data - tokens += response[1] - page['list'][i]['parameters'][j] = startString + translatedText + endString + translatedText = jaString + page['list'][i]['parameters'][j] = translatedText ### Event Code: 320 Set Variable if page['list'][i]['code'] == 320 and CODE320 == True or page['list'][i]['code'] == 324 and CODE324 == True: @@ -901,8 +942,8 @@ def searchCodes(page, pbar): # Need to remove outside code and put it back later startString = re.search(r'^[^一-龠ぁ-ゔァ-ヴー【】a-zA-Z\\]+', jaString) jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー【】a-zA-Z\\]+', '', jaString) - endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】。！？]+$', jaString) - jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】。！？]+$', '', jaString) + endString = re.search(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', jaString) + jaString = re.sub(r'[^一-龠ぁ-ゔァ-ヴー【】。、…！？]+$', '', jaString) if startString is None: startString = '' else: startString = startString.group() if endString is None: endString = '' @@ -922,8 +963,10 @@ def searchCodes(page, pbar): tokens += response[1] page['list'][i]['parameters'][1] = startString + translatedText + endString - except IndexError: - # This is part of the logic so we just pass it. + except IndexError as e: + # This is part of the logic so we just pass it + tracebackLineNo = str(traceback.extract_tb(sys.exc_info()[2])[-1].lineno) + # raise Exception(str(e) + '|Line:' + tracebackLineNo) pass except Exception as e: tracebackLineNo = str(traceback.extract_tb(sys.exc_info()[2])[-1].lineno) @@ -937,13 +980,8 @@ def searchCodes(page, pbar): translatedText = response[0] # TextHistory is what we use to give GPT Context, so thats appended here. - # rawTranslatedText = re.sub(r'[\\<>]+[a-zA-Z]+\[[a-zA-Z0-9]+\]', '', translatedText) textHistory.append('\"' + translatedText + '\"') - # if speakerCaught == True: - # translatedText = speakerRaw + ':\n' + translatedText - # speakerCaught = False - # Textwrap translatedText = textwrap.fill(translatedText, width=WIDTH) @@ -1051,7 +1089,7 @@ def searchSystem(data, pbar): # Variables for i in range(len(data['variables'])): - response = translateGPT(data['variables'][i], 'Reply with only the english translation of the variable name.', False) + response = translateGPT(data['variables'][i], 'Reply with only the english translation of the name', False) tokens += response[1] data['variables'][i] = response[0].strip('.\"') pbar.update(1) @@ -1059,7 +1097,7 @@ def searchSystem(data, pbar): # Messages messages = (data['terms']['messages']) for key, value in messages.items(): - response = translateGPT(value, 'Reply with only the english translation of the text.', False) + response = translateGPT(value, 'Reply with only the english translation of the battle text.', False) translatedText = response[0] # Remove characters that may break scripts @@ -1074,13 +1112,13 @@ def searchSystem(data, pbar): return tokens def subVars(jaString): - varRegex = r'\\+[a-zA-Z]+\[[0-9a-zA-Z\\\[\]]+\]|[\\]+[#a-zA-Z]' + 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, '[' + str(count) + ']') + jaString = jaString.replace(var, '[v' + str(count) + ']') count += 1 return [jaString, varList] @@ -1090,7 +1128,7 @@ def resubVars(translatedText, varList): if len(varList) != 0: for var in varList: - translatedText = translatedText.replace('[' + str(count) + ']', var) + translatedText = translatedText.replace('[v' + str(count) + ']', var) count += 1 return translatedText @@ -1114,17 +1152,19 @@ def translateGPT(t, history, fullPromptFlag): return(t, 0) """Translate text using GPT""" + context = 'Character Context: 莉音 == Rio | Female, 結衣 == Yui | Female, 美雪 == Miyuki | Female, あかり == Akari | Female, カガミ == Kagami | Female, ミズキ == Mizuki | Female, スズカ == Suzuka | Female, シズク == Shizuku | Female, 太郎 == Taro | Male' if fullPromptFlag: system = PROMPT user = 'Text to Translate: ' + subbedT else: - system = 'Reply with only the English Translation of the text.' + system = 'You are an expert translator who translates everything to English. Reply with only the English Translation of the text.' user = 'Text to Translate: ' + subbedT response = openai.ChatCompletion.create( temperature=0, - model="gpt-3.5-turbo-16k", + model="gpt-4", messages=[ {"role": "system", "content": system}, + {"role": "user", "content": context}, {"role": "user", "content": history}, {"role": "user", "content": user} ], @@ -1134,18 +1174,15 @@ def translateGPT(t, history, fullPromptFlag): translatedText = response.choices[0].message.content tokens = response.usage.total_tokens - # Make sure translation didn't wonk out - mlen=len(response.choices[0].message.content) - elnt=10*len(subbedT) - - #Resub Vars + # Resub Vars translatedText = resubVars(translatedText, varResponse[1]) - if len(translatedText) > 13 * len(t) or "I'm sorry, but I'm unable to assist with that translation" in translatedText: + translatedText = translatedText.replace('English Translation: ', '') + translatedText = translatedText.replace('Translation: ', '') + translatedText = translatedText.replace('Text to Translate: ', '') + + 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: - translatedText = translatedText.replace('English Translation: ', '') - translatedText = translatedText.replace('Translation: ', '') - translatedText = translatedText.replace('Text to Translate: ', '') return [translatedText, tokens] \ No newline at end of file diff --git a/start.py b/start.py index 3956884..1973eb2 100644 --- a/start.py +++ b/start.py @@ -1,3 +1,3 @@ from modules.main import main -main() +main() \ No newline at end of file