Turn into a python package

[drkane]

Quite a large refactor of the code to turn into a python package.

Main changes are:

Refactoring

• code lives in the src/tweet2embed directory
• added a pyproject.toml file to hold dependencies - so can now be installed, eg in a virtual environment (I've been using this to test out using uv). I've added selenium as an optional dependency as it has a bit of extra overhead
• command line interface is managed through click

CLI

• I've changed to a unified interface which is run through python -m tweet2embed. You can then follow it with any number of tweet IDs, tweet URLs or mastodon URLs.
• You can specify the outputs using the -o or --output flag - the options are html, img or json - you can specify more than one
• I've added -s/--save option to specify the output directory for saved files
• The -b/--clipboard/--no-clipboard option can switch copy to clipboard on or off (this is obviously less useful for multiple tweets)
• the -a/--archive/--no-archive option can switch off submission to archive.org
• all the other flags are still present

Templating

• I've added the jinja2 templating library. The main advantage is you get to separate out the html from the code. The templates are kept in src/tweet2embed/templates/
• The other advantage is you get html escaping by default
• I've also separated out the CSS into a separate file in that directory.
• @todo at the moment only the tweet_to_html() function (and mastodon equivalent) uses the template. Ideally the get_card_html, get_media, etc would move to a template too

Others

• I've tried to reuse requests.Session objects - this should in theory make it quicker with lots of tweets, but does cause a lot of annoying boilerplate.
• some of the repeated functions are now in src/tweet2embed/utils

The code that does the actual work is still 99.9% original - it may have been moved around a bit. Let me know what you think - I may have moved it too far from the "do one thing well" scripts that you created...

[edent]

This all looks excellent. I hope to have some time to test this soon.

[edent]

Really good. A few issues.

Currently requires python = ">=3.12" - my system has 3.10, can this be lowered?

If I manual lower the version, it says

• gftools 0.9.68 requires brotli, which is not installed.
• gftools 0.9.68 requires tabulate, which is not installed.

Manually installing gftools works.

Can copying to the clipboard be default on? Even using -b doesn't seem to copy data to the clipboard.

The HTML output could do with being minified. Seems to be lots of whitespace in there. (Not sure if that's a regression or whether I didn't update the GitHub copy).

Converting Mastodon to JSON gives TypeError: a bytes-like object is required, not 'str'

If no output directory is specified, what happens to the downloaded data? It seems to disappear.