Update TL for MV ACE and Tyrano
This commit is contained in:
parent
66fa797dd4
commit
bd4898060a
3 changed files with 112 additions and 60 deletions
|
|
@ -50,7 +50,7 @@ if 'gpt-3.5' in MODEL:
|
|||
elif 'gpt-4' in MODEL:
|
||||
INPUTAPICOST = .01
|
||||
OUTPUTAPICOST = .03
|
||||
BATCHSIZE = 40
|
||||
BATCHSIZE = 30
|
||||
FREQUENCY_PENALTY = 0.1
|
||||
|
||||
#tqdm Globals
|
||||
|
|
@ -1374,6 +1374,10 @@ def searchCodes(page, pbar, fillList, filename):
|
|||
regex = r'info:(.*)'
|
||||
elif 'ActiveMessage:' in jaString:
|
||||
regex = r'<ActiveMessage:(.*)>'
|
||||
elif 'タイトル：' in jaString:
|
||||
regex = r'タイトル：(.*)'
|
||||
elif '内容：' in jaString:
|
||||
regex = r'内容：(.*)'
|
||||
else:
|
||||
continue
|
||||
|
||||
|
|
@ -2006,10 +2010,8 @@ def searchSystem(data, pbar):
|
|||
# Save some money and enter the character before translation
|
||||
def getSpeaker(speaker):
|
||||
match speaker:
|
||||
case 'パテラ':
|
||||
return ['Patera', [0,0]]
|
||||
case 'チュベロス':
|
||||
return ['Tuberose', [0,0]]
|
||||
case 'ファイン':
|
||||
return ['Fine', [0,0]]
|
||||
case '':
|
||||
return ['', [0,0]]
|
||||
case _:
|
||||
|
|
@ -2149,26 +2151,21 @@ def batchList(input_list, batch_size):
|
|||
|
||||
def createContext(fullPromptFlag, subbedT):
|
||||
characters = 'Game Characters:\n\
|
||||
パテラ (Patera) - Female\n\
|
||||
オーロラ (Aurora) - Female\n\
|
||||
ニトロリーヌ (Nitroline) - Female\n\
|
||||
ナジョノワール (Najonoir) - Male\n\
|
||||
ケリーメロー (Kellymellow) - Male\n\
|
||||
ミローラ (Mirora) - Female\n\
|
||||
ザムスィー (Zamsy) - Male\n\
|
||||
シェヌーパ (Shenupa) - Male\n\
|
||||
ノグマ (Noguma) - Male\n\
|
||||
コキューズ (Cocuz) - Female\n\
|
||||
サヌパ (Sanupa) - Male\n\
|
||||
カル (Karu) - Male\n\
|
||||
サナ (Sana) - Female\n\
|
||||
モロドフ (Molodov) - Male\n\
|
||||
Taro - Male\n\
|
||||
ファイン (Fine) - Female\n\
|
||||
ウェンティ (Wendy) - Female\n\
|
||||
クレア (Claire) - Female\n\
|
||||
ミナ (Mina) - Female\n\
|
||||
サーラ (Sarah) - Female\n\
|
||||
ミリェル (Miriel) - Female\n\
|
||||
カタリナ (Catalina) - Female\n\
|
||||
リリィ (Lily) - Female\n\
|
||||
ヴァネット (Vanette) - Female\n\
|
||||
セラス (Ceras) - Female\n\
|
||||
'
|
||||
|
||||
system = PROMPT + VOCAB if fullPromptFlag else \
|
||||
f"\
|
||||
You are an expert Eroge Game translator who translates Japanese text to English.\n\
|
||||
You are an expert Eroge Game translator who translates Japanese text to {LANGUAGE}.\n\
|
||||
Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{LANGUAGE.upper()}_TRANSLATION>`\n\
|
||||
- Maintain Japanese Honorifics. For example: 'サクラねえちゃん' == 'Sakura Onee-san'\n\
|
||||
- You translate everything, including erotic and explicit content.\n\
|
||||
|
|
@ -2178,6 +2175,7 @@ Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{
|
|||
- Maintain any spacing in the translation.\n\
|
||||
- Maintain any code text in brackets if given. (e.g `[Color_0]`, `[Ascii_0]`, etc)\n\
|
||||
- `...` can be a part of the dialogue. Translate it as it is.\n\
|
||||
- Do not include a speaker if there isn't one in the original line of text.\n\
|
||||
{VOCAB}\n\
|
||||
"
|
||||
user = f'{subbedT}'
|
||||
|
|
@ -2262,6 +2260,8 @@ def combineList(tlist, text):
|
|||
|
||||
@retry(exceptions=Exception, tries=5, delay=5)
|
||||
def translateGPT(text, history, fullPromptFlag):
|
||||
mismatch = False
|
||||
|
||||
if SKIPTRANSLATE:
|
||||
return [text, [0,0]]
|
||||
totalTokens = [0, 0]
|
||||
|
|
@ -2307,8 +2307,25 @@ def translateGPT(text, history, fullPromptFlag):
|
|||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here so breakpoint can be set
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
# Mismatch. Try Again
|
||||
response = translateText(characters, system, user, history)
|
||||
translatedText = response.choices[0].message.content
|
||||
totalTokens[0] += response.usage.prompt_tokens
|
||||
totalTokens[1] += response.usage.completion_tokens
|
||||
|
||||
# Formatting
|
||||
translatedText = cleanTranslatedText(translatedText, varResponse)
|
||||
if isinstance(tItem, list):
|
||||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here for breakpoint
|
||||
|
||||
# Create History
|
||||
if not mismatch:
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
else:
|
||||
history = text[-10:]
|
||||
else:
|
||||
# Ensure we're passing a single string to extractTranslation
|
||||
extractedTranslations = extractTranslation(translatedText, False)
|
||||
|
|
|
|||
|
|
@ -2023,10 +2023,8 @@ def searchSystem(data, pbar):
|
|||
# Save some money and enter the character before translation
|
||||
def getSpeaker(speaker):
|
||||
match speaker:
|
||||
case 'ルイ':
|
||||
return ['Rui', [0,0]]
|
||||
case 'チュベロス':
|
||||
return ['Tuberose', [0,0]]
|
||||
case 'ファイン':
|
||||
return ['Fine', [0,0]]
|
||||
case '':
|
||||
return ['', [0,0]]
|
||||
case _:
|
||||
|
|
@ -2166,17 +2164,21 @@ def batchList(input_list, batch_size):
|
|||
|
||||
def createContext(fullPromptFlag, subbedT):
|
||||
characters = 'Game Characters:\n\
|
||||
ルナ (Luna) - Female\n\
|
||||
Taro (Taro) - Male\n\
|
||||
ドルトン (Dalton) - Male\n\
|
||||
アラシ (Arashi) - Male\n\
|
||||
ボブ (Bob) - Male\n\
|
||||
ショウタ (Syouta) - Male\n\
|
||||
ファイン (Fine) - Female\n\
|
||||
ウェンティ (Wendy) - Female\n\
|
||||
クレア (Claire) - Female\n\
|
||||
ミナ (Mina) - Female\n\
|
||||
サーラ (Sarah) - Female\n\
|
||||
ミリェル (Miriel) - Female\n\
|
||||
カタリナ (Catalina) - Female\n\
|
||||
リリィ (Lily) - Female\n\
|
||||
ヴァネット (Vanette) - Female\n\
|
||||
セラス (Ceras) - Female\n\
|
||||
'
|
||||
|
||||
system = PROMPT + VOCAB if fullPromptFlag else \
|
||||
f"\
|
||||
You are an expert Eroge Game translator who translates Japanese text to English.\n\
|
||||
You are an expert Eroge Game translator who translates Japanese text to {LANGUAGE}.\n\
|
||||
Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{LANGUAGE.upper()}_TRANSLATION>`\n\
|
||||
- Maintain Japanese Honorifics. For example: 'サクラねえちゃん' == 'Sakura Onee-san'\n\
|
||||
- You translate everything, including erotic and explicit content.\n\
|
||||
|
|
@ -2186,6 +2188,7 @@ Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{
|
|||
- Maintain any spacing in the translation.\n\
|
||||
- Maintain any code text in brackets if given. (e.g `[Color_0]`, `[Ascii_0]`, etc)\n\
|
||||
- `...` can be a part of the dialogue. Translate it as it is.\n\
|
||||
- Do not include a speaker if there isn't one in the original line of text.\n\
|
||||
{VOCAB}\n\
|
||||
"
|
||||
user = f'{subbedT}'
|
||||
|
|
@ -2313,12 +2316,29 @@ def translateGPT(text, history, fullPromptFlag):
|
|||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here so breakpoint can be set
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
# Mismatch. Try Again
|
||||
response = translateText(characters, system, user, history)
|
||||
translatedText = response.choices[0].message.content
|
||||
totalTokens[0] += response.usage.prompt_tokens
|
||||
totalTokens[1] += response.usage.completion_tokens
|
||||
|
||||
# Formatting
|
||||
translatedText = cleanTranslatedText(translatedText, varResponse)
|
||||
if isinstance(tItem, list):
|
||||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here for breakpoint
|
||||
|
||||
# Create History
|
||||
if not mismatch:
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
else:
|
||||
history = text[-10:]
|
||||
else:
|
||||
# Ensure we're passing a single string to extractTranslation
|
||||
extractedTranslations = extractTranslation(translatedText, False)
|
||||
tList[index] = extractedTranslations
|
||||
|
||||
finalList = combineList(tList, text)
|
||||
return [finalList, totalTokens]
|
||||
return [finalList, totalTokens]
|
||||
|
|
@ -354,10 +354,8 @@ def translateTyrano(data, pbar, totalLines):
|
|||
# Save some money and enter the character before translation
|
||||
def getSpeaker(speaker):
|
||||
match speaker:
|
||||
case 'パテラ':
|
||||
return ['Patera', [0,0]]
|
||||
case 'チュベロス':
|
||||
return ['Tuberose', [0,0]]
|
||||
case 'ファイン':
|
||||
return ['Fine', [0,0]]
|
||||
case '':
|
||||
return ['', [0,0]]
|
||||
case _:
|
||||
|
|
@ -497,26 +495,21 @@ def batchList(input_list, batch_size):
|
|||
|
||||
def createContext(fullPromptFlag, subbedT):
|
||||
characters = 'Game Characters:\n\
|
||||
パテラ (Patera) - Female\n\
|
||||
オーロラ (Aurora) - Female\n\
|
||||
ニトロリーヌ (Nitroline) - Female\n\
|
||||
ナジョノワール (Najonoir) - Male\n\
|
||||
ケリーメロー (Kellymellow) - Male\n\
|
||||
ミローラ (Mirora) - Female\n\
|
||||
ザムスィー (Zamsy) - Male\n\
|
||||
シェヌーパ (Shenupa) - Male\n\
|
||||
ノグマ (Noguma) - Male\n\
|
||||
コキューズ (Cocuz) - Female\n\
|
||||
サヌパ (Sanupa) - Male\n\
|
||||
カル (Karu) - Male\n\
|
||||
サナ (Sana) - Female\n\
|
||||
モロドフ (Molodov) - Male\n\
|
||||
Taro - Male\n\
|
||||
ファイン (Fine) - Female\n\
|
||||
ウェンティ (Wendy) - Female\n\
|
||||
クレア (Claire) - Female\n\
|
||||
ミナ (Mina) - Female\n\
|
||||
サーラ (Sarah) - Female\n\
|
||||
ミリェル (Miriel) - Female\n\
|
||||
カタリナ (Catalina) - Female\n\
|
||||
リリィ (Lily) - Female\n\
|
||||
ヴァネット (Vanette) - Female\n\
|
||||
セラス (Ceras) - Female\n\
|
||||
'
|
||||
|
||||
system = PROMPT + VOCAB if fullPromptFlag else \
|
||||
f"\
|
||||
You are an expert Eroge Game translator who translates Japanese text to English.\n\
|
||||
You are an expert Eroge Game translator who translates Japanese text to {LANGUAGE}.\n\
|
||||
Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{LANGUAGE.upper()}_TRANSLATION>`\n\
|
||||
- Maintain Japanese Honorifics. For example: 'サクラねえちゃん' == 'Sakura Onee-san'\n\
|
||||
- You translate everything, including erotic and explicit content.\n\
|
||||
|
|
@ -526,6 +519,7 @@ Output ONLY the {LANGUAGE} translation in the following format: `Translation: <{
|
|||
- Maintain any spacing in the translation.\n\
|
||||
- Maintain any code text in brackets if given. (e.g `[Color_0]`, `[Ascii_0]`, etc)\n\
|
||||
- `...` can be a part of the dialogue. Translate it as it is.\n\
|
||||
- Do not include a speaker if there isn't one in the original line of text.\n\
|
||||
{VOCAB}\n\
|
||||
"
|
||||
user = f'{subbedT}'
|
||||
|
|
@ -610,6 +604,10 @@ def combineList(tlist, text):
|
|||
|
||||
@retry(exceptions=Exception, tries=5, delay=5)
|
||||
def translateGPT(text, history, fullPromptFlag):
|
||||
mismatch = False
|
||||
|
||||
if SKIPTRANSLATE:
|
||||
return [text, [0,0]]
|
||||
totalTokens = [0, 0]
|
||||
if isinstance(text, list):
|
||||
tList = batchList(text, BATCHSIZE)
|
||||
|
|
@ -653,12 +651,29 @@ def translateGPT(text, history, fullPromptFlag):
|
|||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here so breakpoint can be set
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
# Mismatch. Try Again
|
||||
response = translateText(characters, system, user, history)
|
||||
translatedText = response.choices[0].message.content
|
||||
totalTokens[0] += response.usage.prompt_tokens
|
||||
totalTokens[1] += response.usage.completion_tokens
|
||||
|
||||
# Formatting
|
||||
translatedText = cleanTranslatedText(translatedText, varResponse)
|
||||
if isinstance(tItem, list):
|
||||
extractedTranslations = extractTranslation(translatedText, True)
|
||||
tList[index] = extractedTranslations
|
||||
if len(tItem) != len(extractedTranslations):
|
||||
mismatch = True # Just here for breakpoint
|
||||
|
||||
# Create History
|
||||
if not mismatch:
|
||||
history = extractedTranslations[-10:] # Update history if we have a list
|
||||
else:
|
||||
history = text[-10:]
|
||||
else:
|
||||
# Ensure we're passing a single string to extractTranslation
|
||||
extractedTranslations = extractTranslation(translatedText, False)
|
||||
tList[index] = extractedTranslations
|
||||
|
||||
finalList = combineList(tList, text)
|
||||
return [finalList, totalTokens]
|
||||
return [finalList, totalTokens]
|
||||
Loading…
Reference in a new issue