🌐 feat(i18n): overhaul translation system & add languages

[welpo]

TL;DR

Overhauled the translation system, loading strings from /i18n TOML files.

Overview

This PR aims to significantly improve the localisation system, streamline template structures, and add multiple new languages. It offers a more user-friendly approach to managing translations and addresses several issues related to the readability and maintainability of the codebase.

Advantages

1. Ease of Translation Management: Moved translations to an /i18n folder, allowing easier addition and modification.

2. User customisation: Theme users can easily override or add new translation strings without touching the core logic.

3. Flexible default language: Users can set a default language other than English.

4. Robustness: Missing translation strings will fall back to English, reducing the chance of errors.

5. Reduced configuration: Eliminated the need for config.extra.languages and copying/pasting strings into config files.

6. Modularity: Enhanced modularity by isolating translation and language-switching logic.

7. Simplicity: Simplified the translation macro, making it easier to read and maintain.

8. Performance: Potentially minor performance gains by reducing conditional checks.

Also fixes #134.

---

Changes

1. New translation system:

• Replaced the old translation macro with a more streamlined version that's easier to maintain and extend.
• Translation strings are now loaded from TOML files within the /i18n directory, allowing for easy management.

2. Updated macro calls:

• All existing macro calls have been updated to utilise the new language_strings variable, aligning them with the improved translation system. All translated strings are now optional, falling back to English.

3. Template refactoring:

• Removed content.html and integrated its logic into page.html. (Certain macros should be partials #141)
• Moved the language-switcher logic to its own partial template.

4. UI improvements:

• Fixed a minor spacing issue in the multilingual quotes shortcode.

5. Languages added:

• Chinese (Simplified) (machine translation)
• French (machine translation)
• Hindi (machine translation)
• Italian (machine translation)
• Japanese (machine translation)
• Korean (machine translation)
• Portuguese (European) (reviewed by native speaker)
• Russian (reviewed by native speaker)
• Ukranian (machine translation)

Credit

This overhaul was inspired by Thomas Weitzel's (@thomasweitzel) work on the i18n of his Zolarwind theme.

[Jieiku]

Oh my "All translated strings are now optional, falling back to English."

That alone is absolutely worth pursuing this! I will be following up with doing the same!

[welpo]

Finally found a way to do it!

[Jieiku]

Regarding the "machine translation" did you find a convenient way to automatically pull the translations into files? or was it copy paste into google translate. (which is how I did most of mine)

[welpo]

I asked GPT-4 for the translations, giving it the en.toml file, some context, and my requirements.

I validated the output (like the joined 404 keys, or the metadata in context) to make minor adjustments (with GPT-4), and finally I translated the keys on Google Translate and/or DeepL to verify it was reasonable.

[Jieiku]

I just tested using this: https://github.com/mgruner/deepl-api-rs
docs: https://mgruner.github.io/deepl-api-rs-docs/deepl/index.html

I signed up for the free account, and it seems to work:

echo "Recent Posts" | deepl translate --source-language EN --target-language FR | cat -

Messages récents

There is a 500,000 monthly character limit, so no idea how quickly I would hit that limit.

For a small site I would think your unlikely to hit the limit very fast. After I implement what you have done here, I would like to try automating some of the translations, possibly via a bash script.

I am going to see if there is a way to automate the use of GPT-4 as well, no idea if they have an API yet.