Generate an eBook given a book repo

[sethwoodworth]

Take the main book source text file from a given repo and generate an eBook in epub and PDF formats. The main source file should be derivable from metadata.json. The generated books will use the manage downloads script to upload the generated files as github downloads.

This script will be run post-merge when book pull requests are accepted.

This will require handling of .txt, .tei, .TeX, and .rst files at a minimum. A default HTML and CSS template would help while generating .epub.

This feature is the major point of the project for me. But it should wait until after all the books go up online.

[sethwoodworth]

the .mobi format is easily generated from .epub using Amazon's KindleGen tool. It's not free software, which I am not sure I require for a toolchain for generationg .mobi files. There are also a few github repos that generate .mobi files, and at least one reader

[mpevner]

So, we could readily use Calibre's (http://calibre-ebook.com/) conversion utilities I think. They are independant of the GUI stuff, and are written in python too.

[sethwoodworth]

Calibre doesn't (and possibly wont) support .rst as an input format according to the developer's comments here: https://bugs.launchpad.net/calibre/+bug/803813?comments=all

Calibre probably will be a big part of the html >> {epub|mobi|etc} conversion, but it won't be the only part of the toolchain.

[ankitrgadiya]

Can you Pandoc which fulfills all the requirements.

[eshellman]

To document what we did (before closing the issue)
We implemented
asciidoc -> epub with asciidoctor-based code
html -> epub with PG Epubmaker
txt -> html -> epub with PG Epubmaker
epub -> mobi with Calibre ebook-convert
epub -> pdf with Calibre ebook-convert

for rst source we used the previously converted html
for tex source we used the previously converted pdf
for tei source (i think) we used previously converted html

PG Epubmaker (I think) shares a lot of code with pandoc and handles a lot of PG-specific stuff. It also will handle rst, but our ebook building machine does not reconvert rst. This piece should be relatively easy to activate if there's demand or need. The tex conversion should also be easy to add, however the tex software would add significantly to the time to build a machine (we spin up as many as 10 machines during conversions) and maybe should be a dedicated machine. A markdown conversion should also be straightforward if we use the Gitbook converter.

We do not use metadata.yaml (we're not using metadata.json, though json and yaml are interchangeable) for specifying the source file. available covers are specified in metadata.yaml