werd

CLI tool for static website generation with a focus on translation to any number of languages via AI

• Free software: MIT license
• Documentation: https://werd.readthedocs.io

Features

• Translate content, marked up in markdown, into as many languages as desired (or at least as many as OpenAI's ChatGPT can translate into)
• Render content as HTML using Jinja2 templates (partially finished)

Installation

Install this tool using pip:

pip install werd

Usage

For help, run:

werd --help

You can also use:

python -m werd --help

Development

To contribute to this tool, first checkout the code. Then create a new virtual environment:

cd werd
python -m venv venv
source venv/bin/activate

Now install the dependencies and test dependencies:

pip install -e '.[test]'

To run the tests:

pytest

TODOs

In order of most likely to-do to least

• Commands
  • translate
    • sub-directories
    • Blog filename parser
    • Error page content
    • Parallel inference requests to speed up translations
    • force generate 1 language --flag
    • force generate all languages --flag
    • translate file names to be used for navigation.
    • content using OpenAI
  • render HTML using theme
    • Secondary content i.e. layout elements (not rendered as pages)
    • navigation
      • Hierarchy, directories as sub-menus
    • Non-content driven pages i.e.
      • General index rendering
        • Root
        • directory roots
        • blog index
      • Site index language selection list
      • site root redirect to language page
      • home/landing page
    • Error page, list content in all languages
    • Switch language
    • 1-to-1 rendering of content to a page using Jinja2 theme templates
  • go - translate and render in one command
  • server - local web server of HTML output
  • deploy
    • to S3 bucket
    • ftp?
    • Github pages? just a commit and push
  • scaffold - setup content, templates directories and config file.
• Documentation site (using werd and translated)
• Switch to SQLite for all data tracking?
• Refactor to use click

Credits

This project was bootstrapped from the cookiecutter template https://github.com/simonw/click-app.

The name werd was the result of the letters being close together on the keyboard and it not being taken as a package on pipy