Offer the manual as PDF and EPUB, laid out for print
Benchmarks / CPU and I/O (per commit) (push) Successful in 3m5s
Benchmarks / Frame budget (on demand) (push) Skipped
Build and test / Desktop (Linux) (push) Successful in 1h5m29s
Build and test / Layer separation (push) Successful in 29s
🐳 Android image / Build and push (push) Successful in 5s
Build and test / android-image (push) Successful in 8s
🐳 Windows image / Build and push (push) Successful in 5s
Build and test / windows-image (push) Successful in 7s
Manual pages / Publish the manual (push) Successful in 1m2s
Traceability / Requirement traces (push) Successful in 1m1s
Build and test / Android (aarch64) (push) Successful in 23m57s
Build and test / Windows (x86_64, cross) (push) Successful in 55m49s
Build and test / Publish the release (push) Skipped

The page gains a print stylesheet: A4 with a cover, a contents page whose
entries carry their page numbers, a chapter to a page with its name at the
head of each page and the page count at the foot, figures and table rows
never split, and the light palette. A line above the title links the PDF
and the EPUB on the published site, absolutely, since an installed copy
has none beside it.

The Manual pages workflow makes both: tools/manual/pdf.py renders the page
with WeasyPrint from JPEG copies of its pictures, which takes the PDF from
28 MB to 8, and pandoc makes the EPUB from the Markdown.
This commit is contained in:
2026-10-08 22:41:07 -04:00
parent 90d0985db8
commit b53d4dc391
4 changed files with 186 additions and 1 deletions
+25 -1
View File
@@ -7,7 +7,8 @@ name: Manual pages
# No site generator. docs/manual/index.html is already the rendered page —
# `cargo run -p traceability -- manual` writes it from README.md, and the
# Traceability workflow fails a push whose page disagrees with the README —
# so this job only copies the page and its pictures onto the branch.
# so this job copies the page and its pictures onto the branch, with a PDF
# and an EPUB made from them.
on:
push:
@@ -56,6 +57,28 @@ jobs:
echo "LFS pointers where pictures should be:"; echo "$bad"; exit 1
fi
# The downloads the page links to (traceability's PAGES_SITE, PDF_NAME,
# EPUB_NAME). The PDF is the page itself through its print stylesheet —
# A4, a cover and a contents page with page numbers, a chapter a page —
# so it says what the page says; tools/manual/pdf.py says why it goes
# through JPEG copies of the pictures first.
# The EPUB comes from the Markdown, which pandoc reads better than a page.
- name: Make the PDF and the EPUB
run: |
set -e
apt-get update -qq
DEBIAN_FRONTEND=noninteractive apt-get install -y -qq --no-install-recommends \
pandoc python3-pip python3-venv libpango-1.0-0 libpangoft2-1.0-0 \
fonts-noto-core fonts-dejavu-core > /dev/null
python3 -m venv /tmp/wp
/tmp/wp/bin/pip install -q weasyprint
mkdir -p site
/tmp/wp/bin/python tools/manual/pdf.py docs/manual site/darkroom-manual.pdf
title=$(sed -n 's/^# //p' docs/manual/README.md | head -1)
(cd docs/manual && pandoc README.md -o ../../site/darkroom-manual.epub \
--metadata title="$title" --toc --toc-depth=2 --epub-chapter-level=2)
ls -l site
# One commit, force-pushed: the branch is a build output, and its history
# is in master's.
- name: Push to gitea-pages
@@ -64,6 +87,7 @@ jobs:
run: |
set -e
site=$(mktemp -d)
cp site/darkroom-manual.pdf site/darkroom-manual.epub "$site/"
cp docs/manual/index.html "$site/"
cp -r docs/manual/media "$site/"
cd "$site"