PDF¶

With pdf: true, the action also builds the Sphinx project with sphinx-build -M latexpdf (the image has TeX Live), and copies the PDFs into the root of the site, where the pages can link to them. Without sphinx-source there is no PDF: opp_docgen writes none.

A PDF of a part of the site¶

What goes into a PDF is up to latex_documents in conf.py: each entry is a PDF, whose root document is the start of its toctree. The NED reference and the C++ API are pages for the web, so let the root be a document whose toctree does not include them, e.g. that of a guide:

latex_documents = [('guide/index', 'my-model.tex', 'My Model Guide', author, 'manual')]

The PDF of this site is built this way from its Getting Started pages, the toctree of guide/index:

latex_documents = [('guide/index', 'doc-builder.tex', 'OMNeT++ doc-builder', author, 'manual')]

Links in the PDF¶

Links to pages that are not in the PDF (e.g. from the guide to the NED reference) become their text. A :cppref: link to the C++ API on the site by a relative path (doxylink = {'cppref': ('api/tagfile.xml', 'api/')}) cannot work in a PDF either: it also becomes its text, with a warning, which fails the build with fail-on-warnings. Keep :cppref: out of the PDF’s documents, as this site does, or give doxylink the API’s URL, e.g. the published site’s, and the PDF links there:

doxylink = {'cppref': ('api/tagfile.xml', 'https://example.org/my-model/api/')}