Integrating summary generation into publishing process

[JohannesSeikowsky]

Right now summaries are generated after a book goes online on gutenberg.org. Which means it's online for a few days or more often a few weeks without summary. That's bad. It would be preferable if the summary generation would happen as part of the process of putting the book online. This has been partially discussed on Slack. Practically relevant bits from that discussion:

@rtonsing: "It would be better to add it to the processing that is done after dopush.sh sends a trigger file to Ibiblio. No humans involved."

@eshellman: "it would most easily be implemented as a "Writer" that runs along with the FacebookWriter, MastodonWriter and BlueskyWriter. You can find these in ebookconverter/writers/"

Whoever wants to have a go at this can find the summary generation code in summaries.py in the https://github.com/JohannesSeikowsky/gutenberg_updates

The code is pretty straightforward but I'm available for question if there are any.

[Rowan-McKereghan]

Accepted and working on it!

[eshellman]

two things that might not be obvious.

1. the Writer to subclass would be ebookmaker.writers.TxtWriter -don't need to load text from web.
2. ebookconverter.EbookConverter.DEPENDENCIES is where you require a 'txt' to be built before you write a 'summary'