abse4411 / projz_renpy_translation

RenPy翻译 rpy翻译 RenPy机翻工具 - A translator for RenPy, renpy translation, rpy translation, renpy translator

Projz - RenyPy Translation Toolkit

📘Document | 🛠Install | 💡New insights | 💖Sponsor

简体中文（中文版点这） | English

🚨Warning🚨

• It is currently under development, and the stored data of this version is not compatible with that before V0.4.0. To use the old version, please go here.
• We DO NOT provide any RenPy game files, and the program is designed only for the convenience of developers to manage translations. The user shall be held responsible for all the consequences arising from using this program.

Index

There are two tools to help you translate a RenPy Game:

Name	Features	Supported Translation Engines
RealTime Translator	Translate what you see. Very easy to use. See this.	☑️OpenAI Endpoint, ☑️UlionTse/translators
Commandline Translation Toolkit	Translate all texts in the game. It take a few steps to translate. It is mainly used to manage translations among RenPy games. There are many use command to manage these translations. See this.	☑️OpenAI Endpoint, ☑️UlionTse/translators, ☑️Google Translation, ☑️HTML Translation, ☑️Excel Translation, ☑️AI Translation Models

✨What's new

Click to show

1. Web Translation, only supports google translation: translate {index_or_name} -t web -n google -l {lang}

2. AI Translation: translate {index_or_name} -t ai -n mbart50 -l {lang}

3. Inspect translated texts: Use inspect command to find lost variables, style tags, and escape characters in translated texts (e.g., [var], {font}): inspect {index_or_name} -l {lang}. After you fix them in the generated Excel file, use updateexcel command to update these texts: updateexcel {index_or_name} -l {lang}

4. [0.4.1] Reuse pre-translated string texts: You can place some rpy files containing pre-translated string texts of language {lang} under resources/tl/{lang}. String texts are like:

   translate schinese strings:
   
       # renpy/common/00accessibility.rpy:28
       old "Self-voicing disabled."
       new "机器朗读已禁用。"
   
       # renpy/common/00accessibility.rpy:29
       old "Clipboard voicing enabled. "
       new "剪贴板朗读已启用。"
   

   In addition, we also have placed many rpy files with pre-translated string texts (These files are copied from RenPy). After executing import command (e.g., i {projrct} -l {lang}), the pre-translated string texts will be used to update imported string texts if the specified {lang} parameter matches the name of the subdirectory in resources/tl. If you don't want to reuse these translations, please append the -nr option to import command: i {projrct} -l {lang} -nr

   Note that, the source langauge of these provided rpy files is English. It means that there rpy files provide translations from English to other languages. The root dir of these files can be configed by index.recycle_dir in config.yaml.

5. [0.4.1] Open the location of RenPy game or save file (Windows OS Only): Use the new open command to pen the location of a RenPy game associated the TranslationIndex. Some commands for saving file (e.g., savehtml, saveexcel, dumpexcel), will automatically open the saved file's location after the file is saved. To prevent the behavior, please append the -nw option to these command.

6. [0.4.1] UlionTse/translators: translate {index_or_name} -t ts -n bing -l {lang}

7. [0.4.2] Use new_file command to create a FileTranslationIndex from a single file: nf {file_path} -s {type}, where {type} is the translation tool which generated the file, and you can pick from: mt(ManualTransFile.json generated by MTool.), xu(_AutoGeneratedTranslations.txt generated by XUnity Auto Translator), tp(xlsx/xls file generated by Translator++). Then, use other command to manage FileTranslationIndexes as usual. The following commands are not available for FileTranslationIndexes: inject,count. You learn more about it by nf -l and nf -h command.

8. [0.4.2] You can use -a option to enable AI translation and UlionTse/translators translation to load translation settings from the config file automatically. It reduces some args to enter for these commands.

   Auto UlionTse/translators translation: t {index_or_name} -t ts -l {lang} -a. Default args in config.yaml:

   translators:
      api_name: 'bing'
      # Language code can be found at: resources/translation/translators_langcode.txt
      from_language: 'auto'
      to_language: 'zh-Hans'

   Auto AI translation: t {index_or_name} -t ai -l {lang} -a. Default args in config.yaml:

    ai:
      # available models: 'm2m100', 'mbart50', 'nllb200'
      model_name: 'mbart50'
      batch_size: 2
      # Language code can be found at: resources/translation/dl-translate_langcode.txt
      from_language: 'English'
      to_language: 'Chinese'

9. [0.4.3] Now you can save translations as a TranslationIndex in the RealTime Translator, which means that you can process these translations toolkit to process these translations.

10. [0.4.4] Now you can use the OpenAI Endpoint to translate. See Use OpenAI Endpoint.

11. [0.4.5] New command 'llm translate' is available, which uses the OpenAI Endpoint to improve translation quality and reduce translation errors. Usage: llm_translate {index or name} -l {lang} -m {model_name} -t {target_lang}, or using default config: llm_translate {index or name} -l {lang} -a.

✨RealTime Translator (Free and Open Source)

Download exe.7z and libs.7z from Release, then unzip them. Open the server_ui.exe. You can see the following UI:

Then perform the following steps to translate a RenPy game:

1. Click the "Select" button to select your RenPy game path:

   

2. Click the "Injection" button for game detection and code injection:

   

3. Configure the translator and font in the following configuration pages (Note that, Provider=Foo is used for testing and does not support translation):

   

   

4. Click the "Start" button to start the translation server:

   

5. Click the "Start" button under GAME page to start the game, or manually start the game:

   

6. During running the game, you may feel a lag, which is caused by requesting translation.

   

You will also find that the translation result is not displayed when playing game for the first time. This is because a large number of translated texts are currently in the queue, so you need to wait for the value of the queue reduced to 0 and re-enter the game scene.

Finally, you can save the current translation results to the game by clicking the "Save translations" button. By doing this, even if you close the translator, you can still load translations from the translation cache.

This translation cache file projz_translations.json will be saved to the game root. Please note that, do not click the "Undo injection" button as this will disable our translation code.

To use the translation cache file, you need to reopen the game and close our translator. In addition, our translator will automatically load this translation cache file from the game directory (after clicking the "Injection" button) to avoid duplicating translations.

At present, this translator is only a demo and will be integrated with our translation tools in the future.

You can also save the current translation as a TranslationIndex by clicking the "Save TranslationIndex" button, so that you can use various commands for FileTranslationIndex to quickly process these translations in our Commandline Translation Toolkit.

Customize your translation API in RealTime Translator

1. Create a py file in translation_provider. Then, create your class which inherits the Provider class in base.py, and implements these following methods (reload_config() is not a member method in Provider class, which is used to reload config to get the newest values if Reload config file button is clicked.):

from trans import Translator
from typing import List, Tuple
from trans.translators_api import TranslatorsTranslator
from translation_provider.base import Provider, register_provider
import translators as ts

class TranslatorsApi(Provider):

    def __init__(self):
        super().__init__()
        self.trans_kwargs = None
        self.tconfig = None
        self.reload_config()

    def reload_config(self):
        self.tconfig = self.config['translator']['translators']
        self.trans_kwargs = self.tconfig.get('translate_text', {})

    def api_names(self) -> List[str]:
        return list(ts.translators_pool)

    def default_api(self) -> str:
        self.reload_config()
        return self.tconfig.get('api_name', 'bing')

    def default_source_lang(self) -> str:
        self.reload_config()
        return self.tconfig.get('from_language', 'auto')

    def default_target_lang(self) -> str:
        self.reload_config()
        return self.tconfig['to_language']

    def languages_of(self, api: str) -> Tuple[List[str], List[str]]:
        langs = sorted(list(ts.get_languages(api).keys()))
        return ['auto'] + langs, langs

    def translator_of(self, api: str, source_lang: str, target_lang: str) -> Translator:
        if api in self.api_names():
            s, t = self.languages_of(api)
            if source_lang in s and target_lang in t:
                return TranslatorsTranslator(api, source_lang, target_lang, self.trans_kwargs)
        return None

# Register your translation API
register_provider('translators', TranslatorsApi())

2. Import your py file in __init__.py:

import logging
import translation_provider.base

# You should import it with a try-except block
try:
    import translation_provider.translators
except Exception as e:
    logging.exception(e)
try:
    import translation_provider.closeapi
except Exception as e:
    logging.exception(e)

3. Run: python3 server_ui.py

👀Before getting started

Note that, this Commandline Translation Toolkit is not a one-button translator for RenPy games, and it still requires a few steps to translate. It is mainly used to manage translations among RenPy games, and to translate texts. The main functions are as follows：

• Import and generate translations without the RenPy SDK.
• Manage translations of various languages among RenPy games.
• Export your translations to a Excel/Json/HTML file.
• Translate texts with free resources.
• Inspect translated texts for finding lost variables, style tags, and escape characters. Learn more: What's new 3.
• Provide the I18n plugin to change language or font in game.
• Customize your translation API. Learn more: Customize your translation API
• Reuse pre-translated string texts when importing translations. Learn more: What's new 4
• Import translations from a single file generated by other translation tools (MTool, Translator++, and XUnity Auto Translator). Learn more: What's new 7

All commands in this tool:

+--------------------+---------------------------------------------------------------------------+
|    Command name    |                                Description                                |
+--------------------+---------------------------------------------------------------------------+
|      new | n       |            Create a TranslationIndex from the given game path.            |
+--------------------+---------------------------------------------------------------------------+
|   new_file | nf    |          Create a FileTranslationIndex from the given file path.          |
+--------------------+---------------------------------------------------------------------------+
|     import | i     |   Import translations of the given language into this TranslationIndex.   |
|                    |                         (Base injection required)                         |
+--------------------+---------------------------------------------------------------------------+
|    generate | g    |  Generate translations of the given language from this TranslationIndex.  |
|                    |                         (Base injection required)                         |
+--------------------+---------------------------------------------------------------------------+
|     count | c      |        Print a count of missing translations of the given language.       |
|                    |                         (Base injection required)                         |
+--------------------+---------------------------------------------------------------------------+
|      open | o      | Open the location of the RenPy game associated with the TranslationIndex. |
|                    |                             (Windows OS Only)                             |
+--------------------+---------------------------------------------------------------------------+
|        lint        |                       Run lint for checking script.                       |
+--------------------+---------------------------------------------------------------------------+
|       launch       |        Launch the RenPy game associated with the TranslationIndex.        |
|                    |                             (Windows OS Only)                             |
+--------------------+---------------------------------------------------------------------------+
|    inject | ij     |               Inject our code or i18n plugins into the game.              |
+--------------------+---------------------------------------------------------------------------+
|   translate | t    |             Translate untranslated lines of the given language            |
|                    |                       using the specified translator.                     |
+--------------------+---------------------------------------------------------------------------+
| llm_translate | lt |             Translate untranslated lines of the given language            |
|                    |                     using the LLM Augment Translating.                    |
+--------------------+---------------------------------------------------------------------------+
|       ls | l       |                     List existing TranslationIndexes.                     |
+--------------------+---------------------------------------------------------------------------+
|        del         |                        Delete the TranslationIndex.                       |
+--------------------+---------------------------------------------------------------------------+
|       clear        |                    Clear all existing TranslationIndex.                   |
+--------------------+---------------------------------------------------------------------------+
|      discard       |                Discard translations of the given language.                |
+--------------------+---------------------------------------------------------------------------+
|       rename       |                  Rename a name of language translations.                  |
+--------------------+---------------------------------------------------------------------------+
|        copy        |                  Copy translations of the given language.                 |
+--------------------+---------------------------------------------------------------------------+
|        mark        |              Mark all untranslated lines as translated ones.              |
+--------------------+---------------------------------------------------------------------------+
|       unmark       |              Mark all translated lines as untranslated ones.              |
+--------------------+---------------------------------------------------------------------------+
|      upstats       |        Update translation stats of the specified TranslationIndex.        |
+--------------------+---------------------------------------------------------------------------+
|     merge | m      |  Merge translations of the given language from another TranslationIndex.  |
+--------------------+---------------------------------------------------------------------------+
|   savehtml | sh    |       Save untranslated lines of the given language to a html file.       |
+--------------------+---------------------------------------------------------------------------+
|   loadhtml | lh    |       Load translated lines of the given language from a html file.       |
+--------------------+---------------------------------------------------------------------------+
|   saveexcel | se   |       Save untranslated lines of the given language to a excel file.      |
+--------------------+---------------------------------------------------------------------------+
|   loadexcel | le   |       Load translated lines of the given language from a excel file.      |
+--------------------+---------------------------------------------------------------------------+
|   dumpexcel | de   |          Dump translations of the given language to a excel file.         |
+--------------------+---------------------------------------------------------------------------+
|      inspect       |         Inspect each translated line to find missing vars or tags,        |
|                    |        then save these error lines to a excel file. You can use the       |
|                    |       updateexcel command to update translations after you fix them.      |
+--------------------+---------------------------------------------------------------------------+
|  updateexcel | ue  |        Update translations of the given language from a excel file.       |
+--------------------+---------------------------------------------------------------------------+
|   savejson | sj    |       Save untranslated lines of the given language to a json file.       |
+--------------------+---------------------------------------------------------------------------+
|   loadjson | lj    |       Load translated lines of the given language from a json file.       |
+--------------------+---------------------------------------------------------------------------+
|      help | h      |                Print name and description of each command.                |
+--------------------+---------------------------------------------------------------------------+
|      quit | q      |                             Quit the program.                             |
+--------------------+---------------------------------------------------------------------------+
|      reconfig      |      Reload config from disk. It takes effect for most config items.      |
+--------------------+---------------------------------------------------------------------------+
|       about        |                                 About me.                                 |
+--------------------+---------------------------------------------------------------------------+
|        cls         |                               Clean screen.                               |
+--------------------+---------------------------------------------------------------------------+

You should know how to translate RenPy games (How to translate?). And, as you are familiar with using this tool, you can translate RenPy games quickly.

This tool supports to extract Voice statements or others, but the feature is turned off. It only extracts Say statements by default.

# game/script_21_1320.rpy:8
translate chinese scene_01_5f0ee2360:

    # voice "path/to/file"
    # a "text"
    voice "path/to/file"
    a "translated text"

For the above translation rpy, only a "translated text" will be extracted. To enable extracting voice "path/to/file", please set index.say_only as False in config.yaml.

✨Standalone EXE for Windows

You can download a packaged exe file for Windows to run this tool without installation of Python, see Release. To use AI translation, the installation for Python is still needed.

🛫Get started

1.Startup (Optional)

🚨Note🚨
Yon can get an main.exe in Release to run this tool. It only works on Windows OS.

After installing python 3.8, use pip to install the dependencies:

pip install -r requirements_full.txt

Once installed, launch the main program:

python main.py

2.Create a TranslationIndex

After starting the main program, run the following command in the opened console:

n D:\games\renpy_game_demo -n my_game

• D:\games\renpy_game_demo is the root dir of your RenPy game.

• -n my_game is optional，which enables you to take a nickname for the TranslationIndex. You can use this nickname to specify the TranslationIndex instead of an index when running a command. If no nickname is specified, it will be randomly generated.

  In addition, you can define an optional tag by providing the -t arg (e.g., -t 1.0). If no tag is provided, the default one is None.

• n is short name of new command. We have defined many short names for common commands. And you can run the command help -u to print the details of all commands (including their short names).

  🚨Note🚨
  If the path of RenPy game contains spaces, please use double quotes (or single quotes) to enclose the path. For example: new "D:\games\renpy game_demo" -n my_game

Then, enter ls command to see the TranslationIndex we created:

ls

It outputs like:

Note that: Translation Stats list translated/untranslated lines of dialogue and string for each language.
+-------+---------------+-------------------+------------------+-----------------------------------------------------+
| Index |  Nickname:tag | Translation Stats | Injection state  |                      Game info                      |
+-------+---------------+-------------------+------------------+-----------------------------------------------------+
|   1   |  my_game:None |                   |   Base   True    |       renpy_game_demo-V0.1, Ren'Py 7.4.11.2266      |
|       |               |                   |                  |                D:\games\renpy_game_demo             |
+-------+---------------+-------------------+------------------+-----------------------------------------------------+

Note that the Base True in Injection state means that we have successfully identified and injected the game.

3.Import translations of a language

🚨Note🚨
Before running this command, please make sure that all files are extracted from rpa files in the game dir (by useing rpatool or UnRPA). And all rpyc file have converted into rpy files (Otherwise, most of the translations cannot be scanned. You can use unrpyc). This tool integrates unrpa and unrypc: UnRen.

Unren your RenPy game (If needed)

If you find files whose name end with .rpa in your your_game_root/game dir (e.g, D:\games\renpy_game_demo\game), or there exists only rpyc files without corresponding rpy files which have same filenames, you should unren the game.

1. Download UnRen-forall.bat from UnRen Releases: Unzip it, you will get a UnRen-forall.bat, then put it into your game root dir:
2. Click it to run, and input: Enter key, 8, Enter key: Then, input: y: Just wait to be done, and you can close it:

Now, return to the import command

Next, run the command:

i my_game -l schinese

• my_game specifies the nickname of the TranslationIndex to import into, which can also use the Index: 1, or a combination of nickname and tag: my_game:None.
• -l schinese names the translations as schinese.
• This command will read both translated and untranslated texts in D:\games\renpy_game_demo\game\tl\schinese.

Enter ls command to see translations we imported:

ls

It outputs like:

Note that: Translation Stats list translated/untranslated lines of dialogue and string for each language.
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
| Index |  Nickname:tag |            Translation Stats            | Injection state  |                      Game info                      |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
|   1   |  my_game:None |   Language   Dialogue   String   Sum    |   Base   True    |       renpy_game_demo-V0.1, Ren'Py 7.4.11.2266      |
|       |               |   schinese    0/940     0/384    1324   |                  |                D:\games\renpy_game_demo             |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+

3.Translate with the translation command

For convenience, we use savehtml and loadhtml to perform quick translation. Other Translate commands are available at: Web Translation, AI Translation, ⚡Fast translating⚡ with saveexcel and loadexcel, Use Uliontse-translators

Now let's translate with savehtml and loadhtml command:

sh 1 -l schinese

It outputs like:

1320 untranslated lines are saved to ./projz\html\my_game_None_schinese.html.

If you think that the number of lines is too large, you can specify --limit {max_num} to set the max number of lines to save.

Then, open the HTML file with Chrome or Microsoft Edge. Right click to open context menu, and select the translation option to translate this page. After translating, please press Ctrl+S to save and overwrite the original html file. For more instructions, please refer to ⚡Fast translating⚡ with savehtml and loadhtml.

Next, use loadhtml command to update translations:

lh 1 -l schinese

It outputs like:

...
...
Find 1229 translated lines, and discord 91 lines
schinese: 854 updated dialogue translations, 375 updated string translations. [use:1229, discord:0, total:1229]

Enter ls command to see translations we updated:

Note that: Translation Stats list translated/untranslated lines of dialogue and string for each language.
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
| Index |  Nickname:tag |            Translation Stats            | Injection state  |                      Game info                      |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
|   1   |  my_game:None |   Language   Dialogue   String   Sum    |   Base   True    |       renpy_game_demo-V0.1, Ren'Py 7.4.11.2266      |
|       |               |   schinese    856/84    377/7    1324   |                  |                D:\games\renpy_game_demo             |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+

⏱What's the fastest translation commands?
Generally speaking, the ranking of translation speed (from fastest to slowest) is:

1. saveexcel and loadexcel (semi-automatically): Manually upload the excel file to Google Translate, and save the translated file to overwrite the original one.

2. savehtml and loadexcel (semi-automatically): Use the translation function from Microsoft Edge or Chrome (You need to scroll the page from top to bottom to make all text get translated), and save the translated file to overwrite the original one.

3. Uliontse/translators (automatically): translate {index_or_name} -t ts -n bing -l {lang}

4. Web translation (automatically): translate 1 -t web -n google -l {lang} Use the automation tool to enter text into the input box in the translation website, and automatically extract the translation result.

5. AI Translation (automatically): translate 1 -t ai -n mbart50 -l {lang} Use deep network models to translate (with GPU resources).

It's still difficult to assess the translation quality of each translation command.

3.5 Inspect translation result

Since we just send the raw text to the translator, the translation result may contain some content that cause a runtime error when playing game. For example, the following raw text:

Today are [day].

may be translated (into Chinese) as:

今天是[天]。

There exists an obvious error that the variable [day] is translated incorrectly as [天]. As a result, you will receive a KeyError when playing.

There also some other kinds of error in translation result that raise a runtime error when playing:

1. You can set it to a {size=30}fixed size{/size}.
 ->你可以将其设置为{大小=30}固定大小{/大小}。
Error: `{大小=30}` and `{/大小}` should be `{size=30}` and `{/size}`, respectively.

2. I have 100%% confidence.
 ->我有100%的信心。
Error: `100%` should be `100%%`.

...

To find these potential error, we can use inspect command to export these errors to an Excel file by:

inspect 1 -l schinese

Then, open it to correct these errors manually in the new_text column: The column message show the missing tags\variables\escape chars in new_text compared to the raw_text. After fixing them, use the updateexcel command to update translations:

up 1 -l schinese

4.Generate translation rpys

Use generate command to generate translation rpy files to the game:

g 1 -l schinese

It outputs like:

...
...
schinese: dialogue translation: using 856 and missing 84, string translation: using 377 and missing 7

Note that, if there exists rpy files in game/tl/{lang}, existing translations in these files will not be overwritten. generate command only append translations that don't exist in the game by default. If you want to apply all TranslationIndex's translated texts that exist in these rpy files, please append the -f option to generate command. That will delete all rpy/rpyc files in game/tl/{lang}.

5.Inject our I18N plugin

In order to display translated texts correctly after changing the language, we have provided three fonts (download links can be found in resources/fonts/readme.txt):

projz_renpy-translator/
    |–– resources/
        |–– fonts
           –– DejaVuSans.ttf
           –– SourceHanSansLite.ttf
           –– Roboto-Light.ttf
           –– readme.txt

You can manually place custom fonts in resources/fonts dir (please pay attention to the copyright issue of fonts). And add font paths in config.yaml, so that the program will copy these fonts to the game dir. You can find these added font configs in I18N interface.

Run inject command to inject the I18N plugin (which enables you to chainge language or font):

ij 1 -t I18n

Enter ls command to see injection states:

Note that: Translation Stats list translated/untranslated lines of dialogue and string for each language.
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
| Index |  Nickname:tag |            Translation Stats            | Injection state  |                      Game info                      |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+
|   1   |  my_game:None |   Language   Dialogue   String   Sum    |   Base   True    |       renpy_game_demo-V0.1, Ren'Py 7.4.11.2266      |
|       |               |   schinese    856/84    377/7    1324   |   I18n   True    |                D:\games\renpy_game_demo             |
+-------+---------------+-----------------------------------------+------------------+-----------------------------------------------------+

Finally, open the RenPy game and enjoy it. You can open the location of game by launch command (o 1), or launch it by open command (launch 1).

You can show the I18N menu by using the shortcut key "Ctrl+i", or click the I18N button in game's preference menu. (To show the I18N button, the "screens.rpyc" should be converted into "screens.rpy" to inject its code.)：

💡Tips💡
You can config languages and fonts in config.yaml for I18N plugin. The final available languages generated by inject command is determined by languages shared between lang_map in config.yaml and dirs in game/tl (except None).

You can config fonts to fonts in config.yaml.

If there exists the font config in game/tl/{lang}/style.rpy, it will disable our font setting because of its higher priority. That font config may like:

translate schinese python:
    gui.system_font = gui.main_font = gui.text_font = gui.name_text_font = gui.interface_text_font = gui.button_text_font = gui.choice_button_text_font = "SourceHanSansLite.ttf"

If existing above configs, you can't change fonts in I18N menu.

Changes of configs of Developer Mode and Debug Console will apply after reloading the game. Shortcut keys of Debug Console (Shift+O) and Reload Game (Shift+R) work only when Developer Mode is set to True.

Help for commands

Enter help command, and it will print the description and usage of all commands. To see all args of a command, please append a -h option to the command:

new -h

This will print the detailed usage of new command:

🍻To the end🍻
We're glad to you integrate your translation implementation into our tool, or help us translate the documentation.

Other helpful tips

1. To apply changes in config.yaml, use reconfig command. It takes effect for most config items.
2. If you want to ignore the translations of certain rpy files when importing and generating translations, please add these files to index.ignore in config.yaml. Note that, the path splitter in Windows OS is "\", that means that if you want to ignore translations of script/demo.rpy, you should rewrite its path to script\demo.rpy. If there exists space in the path, just keep it.
3. The tool will automatically download models in local dir if translator.ai.model_path is empty in config.yaml.
4. You can usede {index_or_name} -l {lang} to export translation data (including translated and untranslated texts) of a given TranslationIndex to an Excel file, then update translations (ue {index_or_name} -l {lang}) from it after you modify them. In this way, you can alter translations or translate them manually.
5. You can strip style tags before translating by setting index.strip_tag to True in config.yaml.
6. You can mark all untranslated texts as translated ones by mark command: mark {index_or_name} -l {lang}
7. You can rename a name of translations by rename command: rename {index_or_name} -l {lang} -t {new_lang}, where {new_lang} is new one.
8. To merge translations from another TranslationIndex into current TranslationIndex, use merge command: merge {index_or_name} -l {lang} -s {source_index}, where {source_index} is the source index to merge from. This enables you to use translations of old game to that of new game when new one is released.

---

⚡Fast translating⚡ with saveexcel and loadexcel

By saving translations to an Excel file, use translation website to translate. Then update translations from the translated file.

Instructions

1. Run the command se {index_or_name} -l {lang} to export untranslated texts to an Excel file. Upload the file to any translation website that supports document translation:

2. Wait until it is finished, then download it and overwrite the original file:

   

3. Run the command le {index_or_name} -l {lang} to update translations in TranslationIndex.

😕Translation website doesn't support uploading Excel files?
You can paste content of the generated Excel file into a doc file, then upload it. After completion, copy content from translated doc file to the Excel file.

---

⚡Fast translating⚡ with savehtml and loadhtml

By saving translations to an HTML file, use Microsoft Edge or Chrome to translate. Then update translations from the translated file.

Instructions

1. Run the command sh {index_or_name} -l {lang}to export untranslated texts to an HTML file. Open it with Microsoft Edge or Chrome.

2. Click translation option in context menu or at address bar:

   

3. Scroll the page from top to bottom to get all the text translated, then save the file and overwrite the original HTML file.

4. Run the command lh {index_or_name} -l {lang} to update translations in TranslationIndex.

---

Use web translation

Install Chrome driver

Download and install Chrome浏览器. Find your Chrome version in Settings->About Chrome, then download the corresponding Chrome driver from following links:

• Chrome Ver 116.x.xxxx.xxx below
• Chrome Ver 116.x.xxxx.xxx or higher🆕

Download the file whose name contains "win64"/"win32" (It depends on your Windows system processor architecture. "win64" is common for most modern PCs), and unzip it to a dir.

Then config the path translator.web.chrome_driver_path for the Chrome driver in config.yaml:

projz:
  translator:
    web:
      chrome_driver_path: 'D:\Users\Surface Book2\Downloads\chromedriver_win32\chromedriver.exe'

Get started

1. Run the command t {index_or_name} -t web -n google -l {lang}
2. Wait until the page displays, and set your translation target manually:
3. Wait for the prompt in console, and enter Y or y (the others to quit) to continue. Then, the tool starts automatically translating.

---

Use AI translation

Install suitable Pytorch (Optional)

If you want to use AI translation with CPU, please skip this step. These following instructions guide you to install suitable Pytorch that compatible with your NVIDIA GPU. It suggests that your GPU memory is not less than 4 GB.

1. Open a CMD, and run the following command to see your CUDA version:

   nvidia-smi

   It may output like:

   +-----------------------------------------------------------------------------+
   | NVIDIA-SMI 517.48       Driver Version: 517.48       CUDA Version: 11.7     |
   |-------------------------------+----------------------+----------------------+
   | GPU  Name            TCC/WDDM | Bus-Id        Disp.A | Volatile Uncorr. ECC |
   | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
   |                               |                      |               MIG M. |
   |===============================+======================+======================|
   |   0  NVIDIA GeForce ... WDDM  | 00000000:02:00.0 Off |                  N/A |
   | N/A   33C    P0    21W /  N/A |      0MiB /  6144MiB |      1%      Default |
   |                               |                      |                  N/A |
   +-------------------------------+----------------------+----------------------+
   
   +-----------------------------------------------------------------------------+
   | Processes:                                                                  |
   |  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
   |        ID   ID                                                   Usage      |
   |=============================================================================|
   |  No running processes found                                                 |
   +-----------------------------------------------------------------------------+
   

   It is shown that the CUDA version is 11.7. If you have run the command pip install -r requirements_full.txt before, you can skip the following step and goto Get started part. That version is what we use in requirements_full.txt.

2. Next, uninstall some libraries:

   pip uninstall torch torchaudio torchvision transformers

   Goto the Pytorch, and follow its instructions to install the Pytorch that matches your CUDA. The following content is what I found:

    # For Linux and Windows
    # ROCM 5.4.2 (Linux only)
    pip install torch==2.0.1+rocm5.4.2 torchvision==0.15.2+rocm5.4.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/rocm5.4.2
    # CUDA 11.7
    pip install torch==2.0.1+cu117 torchvision==0.15.2+cu117 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu117
    # CUDA 11.8
    pip install torch==2.0.1+cu118 torchvision==0.15.2+cu118 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118
    # CPU only
    pip install torch==2.0.1+cpu torchvision==0.15.2+cpu torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cpu

3. Reinstall the appropriate "transformers" library:

   pip install transformers

Prepare models (Optional)

If you can access huggingface page, it means that this tool can download models without any problem. You can set the path translator.web.chrome_driver_path to empty in config.yaml, which lets this tool download model automatically:

projz:
  translator:
    ai:
      model_path: ''

And just goto Get started part.

If you want to specify the location of downloaded models, or have some problems during using AI translation like:

The following instructions guide you to download models manually:

1. Suppose your models is saved in: 'D:\Download\New36\save_models'. There are some links to download them:

   • m2m100：https://huggingface.co/facebook/m2m100_418M/tree/main
   • mbart50：https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt/tree/main
   • nllb200：https://huggingface.co/facebook/nllb-200-distilled-600M/tree/main

2. Choose a model you want, and create a dir that has same name with that of the model (m2m100, mbart50, nllb200) in D:\Download\New36\save_models. I choose the m2m100 model here, so I have the dir D:\Download\New36\save_models\m2m100.

   Download all files in this page except rust_model.ot:

   

3. Once finished downloading, config the root dir in config.yaml:

   projz:
     translator:
       ai:
         model_path: 'D:\Download\New36\save_models'

   Get started

4. Run the command:

   t {index_or_name} -t ai -n {model_name} -l {lang} -b 4

   • -n specifies the model to use. Available models are :m2m100,mbart50,nllb200. I use the m2m100 model here.
   • -b specifies the batch size during translating. It determines how many lines of untranslated text feed into the model. The bigger the value is, the more GPU memory it uses.

5. Set the translation target. If you want to translate text from English into Chinese, just enter their indexes:19 109

6. Then, the tool starts automatically translating.

---

Use UlionTse-translators

Get started

1. Run the command:

   t {index_or_name} -t ts -n {API_name} -l {lang}

   • -n specifies the API service to use. You can run the command t -t ts -h to show available API services. Default as bing.

2. Set the translation target. This is similar to step 2 in AI translation.

3. Then, the tool starts automatically translating.

Note, args of translate_text and preaccelerate method using in this command, can be configed at translator.translators in config.yaml. Learn more: UlionTse/translators

---

Use OpenAI Endpoint

Config your OpenAI Endpoint in config.yaml:

You can config your chat.completions.model, init.api_key, and init.base_url. The {target_lang} in the prompt is the language to translate into, which is determined at run time. The {text} in the prompt is the text to translate when translating.

open_ai:
  target_lang: 'Chinese'
  user_role: &user_role 'user'
  max_turns: 8 # Specify the maximum number of rounds for the conversation.
  langs: [...]
  models: [...]
  # Template Args
  # Args for initializing OpenAI() client
  init:
    base_url: http://localhost:11434/v1/
    api_key: 'ollama'
  chat:
    # Your client should have an endpoint whose url ends with "chat/completions". E.g., http://localhost:11434/v1/chat/completions
    # Args for client.chat.completions.create(), see https://platform.openai.com/docs/api-reference/chat/create
    completions:
      # temperature: 0.2
      # top_p: 0.1
      stream: false
      model: "qwen:0.5b"
      messages:
        -
          role: "system"
          content: "You are a professional translator. You are now required to translate the text given by the user into {target_lang} based on the context of the chat. Content enclosed in square brackets (For example, [lisa_alias], [day]) does not need to be translated. NO EXPLANATION is needed for the translation result."
          # content: "你是一个专业的翻译家。现在要求你根据聊天上下文信息把用户给定的文本翻译为{target_lang}。以方括号括起来的内容(例如: [lisa_alias], [day])不需要翻译。不需要对翻译结果做任何解释。"
        -
          role: *user_role
          content: "{text}"

Get started

For Translation Toolkit

t {index_or_name} -t openai -n {API_name} -l {lang}

⭐Suggestion⭐
We strongly recommend that using the 'llm_translate' command instead of the above translation command. To use llm_translate command: lt {index or name} -l {lang} -m qwen:0.5b -t Chinese or lt {index or name} -l {lang} -a

For RealTime Translator

Select the "CloseAI" in provider list in UI.

💪Customize your translation API

You can integrate your translation API easily to this tool. Create a py file in translator, and let your translation class inherits one of template classes. CachedTranslatorTemplate class has a translation buffer which allows to write translations into TranslationIndex when reaching a certain number. The cache size can be configed by translator.write_cache_size in config.yaml. TranslatorTemplate class provides basic implementation that writes all translations into TranslationIndex at once.

from argparse import ArgumentParser
from translator.base import CachedTranslatorTemplate
from command.translation.base import register_cmd_translator
from typing import List, Tuple
from config.base import ProjzConfig


# The pipeline of a translator. The following code provide a simple version of DlTranslator
# 1. The users enter: translate 1 -l chinese -t ai --name mbart50
# 2. Create a new instance of DlTranslator, then call register_args method.（Note that, your class should use a non-parameter constructor）
# 3. If there exists '-h' or '--help' in user input, print help and goto 7.
# 4. Call do_init method (Put your init code in the method. The passed args and config are ready at this moment).
# If do_init() returned False, goto 7.
# 5. Call invoke method (The Base class CachedTranslatorTemplate or TranslatorTemplate has implemented it.)
# 6. The default implementation of invoke calls translate_batch method. And the default implementation of translate_batch calls translate method within a for-loop.
# 7.Done

class DlTranslator(CachedTranslatorTemplate):
   def register_args(self, parser: ArgumentParser):
      super().register_args(parser)
      # Register your args
      # Note that, any init code shouldn't place here as user may want to print help instead of using it.
      # You should put your init code in do_init()
      parser.add_argument('-n', '--name', choices=['m2m100', 'mbart50', 'nllb200'], default='mbart50',
                          help='The name of deep learning translation  model.')

   def do_init(self, args, config: ProjzConfig):
      super().do_init(args, config)
      # The method will be called when the user decides to use this translation API.
      # Put your init code here. The rgs and config are ready.
      self._model_name = args.name
      self._model_path = config['translator']['ai']['model_path']
      self._load_model()
      return True # Return True if everything is OK.

   def translate(self, text: str):
      # Your implementation for translating.
      # The method take a text and return a translated text.
      return self.mt.translate(text, self._source, self._source, batch_size=1, verbose=True)

   def translate_batch(self, texts: List[str]):
      # If your translation API supports to translate a batch of texts,
      # you can implement this method. The length of `texts` and that of returned texts should be the same.
      # The default implementation will call translate method within a for-loop if you don't override this method.
      # CachedTranslatorTemplate class writes translations into TranslationIndex after each call of translate_batch.
      # The max length of texts can be configed by translator.write_cache_size in config.yaml.
      return self.mt.translate(texts, self._source, self._source, batch_size=self._batch_size, verbose=True)


# Register your translation API to translation command
# User can run the command: translate 1 -l chinese -t ai --name mbart50
# where -t ai is the name of your translation API you register
# Note that, DlTranslator use a non-parameter constructor.
# Don't forget to add the call of  __init__ of super class to your non-parameter constructor if you have it.
register_cmd_translator('ai', DlTranslator)

Finally, import your translation API in translator/__init __.py:

import logging
import translator.base

try:
    import translator.web
except Exception as e:
    print(f'error: {e}')
    logging.exception(e)

try:
    # You should use the try-except block to wrap your import statement.
    # It allows users to launch the main program even if there exists missing libs in user's environment.
    import translator.ai
except Exception as e:
    print(f'error: {e}')
    logging.exception(e)

To see complete implementation of DlTranslator, please see translator/ai/impl.py.

💖Sponsor

This project is developed in my nonworking time. If you would like to support this project, just give me a star⭐, or sponsor me by , .

😘Thanks to my sponsors:

• 799190761
• ansan

If you find your name missed in this list, please accept my apologies and email me to add it.

🗒Todo list:

1. Add English document
2. GUI support
3. Text check when translating

🔗Acknowledgement

The codes or libs we use or refer to:

• Previous code of Web translation: Maooookai(Mirage), DrDRR
• AI translation: dl-translate, MIT License
• UlionTse/translators, GPL-3.0 License
• Pre-translated RPY file: RenPy, MIT License for these rpy files
• resources/codes/projz_injection.py: RenPy, MIT License for the code file
• UI: PyQt5, it contains dual licenses: the GNU GPL v3 and the Riverbank Commercial License. See here.
• UI theme: Qt-Material, BSD-2-Clause license
• Other python libs：requirements_full.txt

Star History

projz_renpy_translation's Issues

无法翻译问题，提示No translations of language {lang} No untranslated lines to update.

在使用AI翻译或UlionTse-translators翻译时，输入指令后立刻提示
No translations of language {lang}
No untranslated lines to update.
后没有翻译任何文本，sh和lh翻译使用正常

角色名内包含old导致文本不会被判断为新文本以进行翻译

角色名内包含old导致文本不会被判断为新文本以进行翻译

有一些游戏无法新建任务

n G:\AMothersLovePLUS-Part12PLUS-pc -n aml
Checking the skeleton for this RenPy game...
ERROR:root:Coundn't find a executable file file in G:\AMothersLovePLUS-Part12PLUS-pc\lib
Traceback (most recent call last):
File ".\main.py", line 62, in main
execute_cmd(cmd_name, cmd_line)
File "G:\0.Translate\projz_renpy_translation-master\command\manage.py", line 77, in execute_cmd
cmd.invoke()
File "G:\0.Translate\projz_renpy_translation-master\command\renpy\base.py", line 44, in invoke
index = TranslationIndex.from_dir(self.args.game_path, self.args.name, self.args.tag)
File "G:\0.Translate\projz_renpy_translation-master\store\index.py", line 834, in from_dir
project = Project.from_dir(project_path)
File "G:\0.Translate\projz_renpy_translation-master\injection\renpy.py", line 235, in from_dir
executable_path = check_python_exe(abs_path)
File "G:\0.Translate\projz_renpy_translation-master\injection\renpy.py", line 78, in check_python_exe
assert executable_path is not None, f'Coundn't find a executable file file in {lib_path}'
AssertionError: Coundn't find a executable file file in G:\AMothersLovePLUS-Part12PLUS-pc\lib

是因为这些游戏用的Renpy版本过老吗？

初步試用跟心得

我有两台电脑一台西文一台中文用新版的翻译程式，觉得旧版的准确率比较高，新版的速度比较快，可是有时候会翻译不到，或是没有加上@@ 或是 @$,還有一次出現三行出現相同翻譯（新版），不管原文是什麼，所以确定不是电脑的问题，因为一台刚好翻译西文，一台中文做有交叉比对，小弟我小小的心得目前还在努力研究中辛苦Abse4411大大了，

能否使用谷歌的文档翻译功能进行快速批量翻译

文本量较大的游戏使用API或者HTML的翻译速度都比较慢。
尤其是HTML的网页翻译会分段加载，可能需要使用鼠标滚轮翻动几千页来保证edge的页面翻译能够完全工作。
google提供了.docx文档翻译功能，是否可以添加功能，导出.docx文档，使用不会被翻译的纯数字串作为索引进行批量翻译？

apply命令提示error: module 'distutils' has no attribute 'util'

apply命令提示error: module 'distutils' has no attribute 'util'

执行loadexcel命令过程中报错AttributeError: 'float' object has no attribute 'strip'

详情：
大文本量的rpy合并后成功保存为excel文件，使用GPT云翻后用loadexcel导入时报错，之后我又尝试了导出excel后不做改动直接导入，也是一样的报错。

error: 'float' object has no attribute 'strip'
[2023-08-22 20:49:11][parse_console.py:333, tid17164][ERROR] - 'float' object has no attribute 'strip'
Traceback (most recent call last):
  File "C:\JulienSlink_Personal\Ind_Prog\github\projz_renpy-devp\parse_console.py", line 330, in main
    register_commands[cmd](*args[1:])
  File "C:\JulienSlink_Personal\Ind_Prog\github\projz_renpy-devp\parse_console.py", line 285, in loadexcel_cmd
    res = load_from_excel(excel_file, source_data)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\JulienSlink_Personal\Ind_Prog\github\projz_renpy-devp\store\file_store.py", line 124, in load_from_excel
    encrypted_tid, old_str, new_str = encrypted_tid.strip().upper(), old_str.strip(), new_str.strip()
                                                                     ^^^^^^^^^^^^^
AttributeError: 'float' object has no attribute 'strip'

附件：使用的rpy文件
scripts.zip

当角色名内有引号时翻译有误

# game/chapter1/ch1afterback.rpy:81
translate chinese ch1afterback_9b893d82:

    # "*Unknown Hooden Woman*" "Hello agent [pl]."
    "@@“你好 [pl]探员”。"

# game/chapter1/ch1afterback.rpy:83
translate chinese ch1afterback_97e0fbf8:

    # "*Unknown Hooded Woman*" "He's here."
    "@@“他来了。"

# game/chapter1/ch1afterback.rpy:84
translate chinese ch1afterback_b28faf94:

    # "*Unknown Hooded Woman*" "..."
    "@@* 未知兜帽女 * “ ...”。"

不好意思請問一下有關翻譯語言部分

Abse4411 大大 您好 不好意思，我剛學編程可是我想編輯英文到西文或是葡萄牙文不知道是要修改哪裡，我沒有找到所以才來詢問您 不好意思，希望您可以教教我，謝謝

功能建议

Note: [savehtml] 请用MS Edge进行网页翻译，不要使用Chrome。

针对保存为html文件进行翻译，请使用MS Edge进行网页翻译。

使用Chrome进行网页翻译会有问题，翻译完后会带有html标签，如果还想用Chrome翻译，请翻译完后手动去除html标签在使用loadhtml命令。

能否将自定义翻译做成类似于XUnity.AutoTranslator那样

仅需要在配置文件中填入API 和密钥

AI翻译功能反馈

大佬，我尝试使用了AI翻译功能，但我观察到我的显存占用并没有上来了反而是cpu占用达到了70%，内存占用到达了18G。这是GPU没有介入工作吗？我安装教程安装了全部的依赖，并且我是可以正常使用SD的，不知道是哪里出了问题。我应该怎么检查我是否正确配置了环境？感谢大佬！！！
配置：
GPU：3060 12G
CPU：AMD 5950X

蹲大佬

坐等作者大大更新导出excel功能，
由于我用的是AiNiee来用gpt翻译文本，但是它只支持excel或者json文件导入和导出，我试着把它导出的excel另存为html，但是这样的html没能成功识别，提示utf8编码不支持啥的
蹲个大佬教教我怎么excel文件转成支持的html，多谢多谢

REMOVE_MARKS功能無效

apply後還是帶有@@標示

打开提示问题

新版本0.4.5导入tl文件的几个问题

去年9月的时候用过大佬的项目，最近回来发现换了新版本，尝试的过程中发现几个问题，不知道是不是我的操作有错误
我的操作过程如图：

发现大部分文本都被标为已经翻译过的，使用unmark全部转为未翻译，用se导出为excel后，发现导出的只有直接放在GamesofDepravityTheMotel-1.0-pc\game\tl\schinese文件夹下的rpy文件内容，而在GamesofDepravityTheMotel-1.0-pc\game\tl\schinese\events内的rpy文件未被提取出来。文件结构如下图

不知道是我的哪一步操作有问题

翻译文本为空时不会自动跳过

生成的翻译文本中存在空文本时，脚本可能会多次使用上一句的翻译，并再放回时省略掉目标文本的引号

彩云小译似乎更新了页面前端导致脚本无法定位到文本框

如题

用generate命令生成翻译rpy文件内容不全

你好，我使用generate命令来生成翻译rpy文件，打开生成的翻译rpy文件全部只有old和new模块，而根本看不到那些对话块和加密码那些模块

是不是某些游戏不能导出对话

不知道是不是部分游戏的结构不够标准化还是什么，提取不出对话。

我对renpy游戏的结构不是很了解，提取对话失败的两个游戏似乎脚本都在game目录下的rpa文件里，能提取对话的那个游戏则全都是rpy文件。

尝试把提取失败的游戏rpa文件解包了，删除TranslationIndex后重新import好像还是没生效，不知道是哪里做得不对。之前试过github上另一个叫Renpy Translator的项目 ，也经常遇到这个问题，对话无法提取。那个工具我还没翻译成功过。不过他有个术语表的功能还挺眼馋的，不知道我们这个项目怎么实现，尝试过导出到excel后替换名词，导回去之后就变成已翻译内容了，能不能用ue更新回默认语言？请教大佬

当然大部分游戏使用即时翻译都没有问题，但是必须先过一遍游戏再退回来稍微有点剧透，才想着用之前的提取全部对话来翻译。
(遇到过一次游戏使用即时翻译注入后，启动游戏能进去，一点开始游戏就报错的，因为是测试随便下的游戏就没在意）

彩云小译似乎更新了页面前端导致脚本无法定位到文本框

Error in starting the web translator, 4212 untranslated line(s) will be not translated: Message: no such element: Unable to locate element: {"method":"xpath","selector":"//*[@id="app"]/div[2]/div[1]/div[2]/div/div[1]/div[3]"}

即时翻译server_ui打开之后提示错误，translator无法运行

计算机只懂一点点，不知道是不是系统变量配置问题，检查这个module是有的。还有我装了anaconda不知道有没有影响