This is a brief demonstration of the workflow required to mark string literals in your code as translatable, how to generate a template file that can then be used as the basis for all of your translations, and then how to automatically generate files that will be used for translations for the locales that you specify. I'll be using Python and the Babel tool for helping to make this process a lot easier.

Python has a builtin way of dealing with internationalized strings through the GNU gettext library and its own hooks for it through the gettext module. The gettext module provides the "lookup" service for finding the localized version of a string literal through something akin to a standard dictionary lookup. There is a minimal amount of setup needed to tell gettext where to look for translations and what to do if there is not a translation to get. I've put these steps in main.py but I will reproduce them below and walk through them.

import gettext
import os
import sys

locale_dir = os.path.join(os.path.abspath(os.path.dirname(__file__)), "locale")
translate = gettext.translation("messages", locale_dir, fallback=True)
_ = translate.gettext

The first things we have to do are import the gettext and os modules. These allow us to determine where the locale directory is as well as having the means to look up translations in that directory. We then set the path to the locale_dir by getting the path to that file and then appending locale to the end of it. We then specify in which file to look for the translations. In this case, we're looking for files named messages as this is the standard for GNU gettext. The final line is the "standard" shorthand for tranlating strings which is to "alias" the gettext method to _ as it is used frequently (and most other Python developers will know what you're doing when they see that.)

Once we've done that we can now do some simple Python to mark strings as translatable. Look at the following code snippet:

if __name__ == "__main__":
    args = sys.argv
    print(_("This is a translatable string."))
    print(_("Hello world."))
    print(_("Sphinx of black quartz, judge my vow!"))
    print(_("The args are {}.").format(args))

Note that we preface our string literals with _ to indicate that they should be translated if possible.

The file base.pot is the basis for all translations that will be done. We will use pybabel from Babel to generate this file by running the following commands.

# First make the directory where our translations will live
mkdir locale
pybabel extract <files_or_directories> -o locale/base.pot

This will extract the strings that we've marked with _ and create a file called base.pot. The .pot file extension stands for "Portable Object Template" as it will form the basis for any future translations that we will make.

Now that we have our template file, we can now use it to create a new message catalog to store our actual translations for different locales. We do this by using pybabel to create a new message catalog from our template. We specify the locale in the general style of language.country_code For example, if we wanted to specify American English, we would specify our locale as en_US or if we wanted Brazilian Portuguese, we would use pt_BR. You can look up the locale identifiers using this tool.

pybabel init -l <name_of_locale> -i locale/base.pot -d locale

This will create a new series of directories and a file as follows:

locale/<name_of_locale>/LC_MESSAGES/messages.po

The messages.po file is where our translations will go for the locale specified.

Before we're able to use our new translations, we have to compile our .po files into the binary format that gettext uses. The following command will compile our messages.po files in each locale directory under locale and generate the corresponding messages.mo files. The .mo extension stands for "Machine Object".

pybabel compile -d locale

To test our changes, we can set our locale on the command line to another one temporarily by setting the environment variable LANGUAGE before specifying the command we want to run.

LANGUAGE=de_DE python main.py

Simply run the pybabel extract command from above again!

If we've changed the strings in our code, we will need to update base.pot and then use that to update the individual message.po files in each of our supported locales.

pybabel extract <files_or_directories> -o locale/base.py
pybabel update -i locale/base.pot -d locale