Build site #45

Workflow file for this run

.github/workflows/build-site.yml at c2b732b

	name: Build site

	# ---------------------------------------------------------------------------
	# Builds the serverless full-text search static site and deploys it to
	# GitHub Pages, then verifies the deploy is byte-range-servable (memex
	# pitfall #1: without Accept-Encoding: identity, GH Pages gzips the .db
	# and every ranged read returns the full gzipped blob, which SQLite reads
	# as page 0 → "database disk image is malformed").
	#
	# The deploy verification is MANDATORY — if the live index.db doesn't
	# answer 206 Partial Content with a valid Content-Range for a 0-4095
	# range request, the workflow FAILS. This catches every regression of
	# the pitfall before it hits the browser.
	# ---------------------------------------------------------------------------

	on:
	push:
	branches: [main]
	schedule:
	# Daily backstop at 04:30 UTC.
	- cron: "30 4 * * *"
	workflow_dispatch: {}

	permissions:
	contents: write # required by actions/checkout for cache paths
	pages: write # for actions/deploy-pages
	id-token: write # for actions/deploy-pages (OIDC)

	concurrency:
	group: build-site
	cancel-in-progress: false

	jobs:
	# -------------------------------------------------------------------------
	# Job 1: build + deploy
	# -------------------------------------------------------------------------
	build:
	runs-on: ubuntu-latest
	environment:
	name: github-pages
	url: ${{ steps.deploy_pages.outputs.page_url }}
	outputs:
	page_url: ${{ steps.deploy_pages.outputs.page_url }}
	steps:
	- name: Checkout (full history + LFS)
	uses: actions/checkout@v4
	with:
	fetch-depth: 0
	lfs: true

	- name: Set up uv + Python 3.13
	uses: astral-sh/setup-uv@v3
	with:
	python-version: "3.13"

	- name: Create venv + install pypdf (extractor runtime dep)
	# Ubuntu 24.04 marks the system Python as externally managed
	# (PEP 668). We create a dedicated venv and put its bin on
	# PATH so every downstream step's `python` picks it up.
	run: \|
	uv venv .venv --python 3.13
	uv pip install --python .venv/bin/python pypdf
	echo "$PWD/.venv/bin" >> $GITHUB_PATH

	- name: Fetch memex WASM bundle
	run: python tools/fetch_memex.py

	- name: Extract PDF text (parallel, 4 workers)
	run: python tools/extract_pdf_text_parallel.py

	- name: Build memex index.db (asserts <= 40 MB budget)
	run: python tools/build_index.py

	- name: Build vendor prefix cache
	run: python tools/build_vendor_cache.py

	- name: Assemble site/ directory
	run: python tools/build_site.py

	- name: Configure Pages
	uses: actions/configure-pages@v5

	- name: Upload Pages artifact
	uses: actions/upload-pages-artifact@v3
	with:
	path: site

	- name: Deploy to GitHub Pages
	id: deploy_pages
	uses: actions/deploy-pages@v4

	- name: Print deploy URL
	run: \|
	echo "Deployed to: ${{ steps.deploy_pages.outputs.page_url }}"

	# -------------------------------------------------------------------------
	# Job 2: post-deploy verification. Runs the same range-request check as
	# tests/test_deploy_verification.py, but at the CDN edge, right after
	# deploy. This is the gate that catches memex pitfall #1 (gzipped .db)
	# BEFORE anyone loads the browser page.
	# -------------------------------------------------------------------------
	verify:
	needs: build
	runs-on: ubuntu-latest
	steps:
	- name: Wait for Pages propagation (up to 90 s)
	run: \|
	set -eu
	URL="https://fastled.github.io/datasheets/_meta.json"
	echo "Polling ${URL} for propagation..."
	for i in $(seq 1 30); do
	code=$(curl -sS -o /dev/null -w '%{http_code}' -I "${URL}" \|\| echo 000)
	echo " attempt ${i}: HTTP ${code}"
	if [ "${code}" = "200" ]; then
	echo "Pages is live."
	exit 0
	fi
	sleep 3
	done
	echo "::error::Pages did not respond 200 on ${URL} within 90 s"
	exit 1

	- name: Byte-range probe on index.db
	run: \|
	set -eu
	URL="https://fastled.github.io/datasheets/index.db"
	echo "Requesting bytes 0-4095 of ${URL} with Accept-Encoding: identity..."
	curl -sS \
	-o /tmp/db_range.bin \
	-D /tmp/db_headers.txt \
	-r 0-4095 \
	-H "Accept-Encoding: identity" \
	"${URL}"
	echo "--- response headers ---"
	cat /tmp/db_headers.txt
	echo "------------------------"

	- name: Verify 206 Partial Content
	run: \|
	set -eu
	if ! grep -qi "^HTTP/[0-9.]* 206" /tmp/db_headers.txt \
	&& ! grep -q "206 Partial Content" /tmp/db_headers.txt; then
	echo "::error::index.db did not return 206 Partial Content"
	echo "This means either Accept-Encoding: identity was ignored,"
	echo "or the CDN is serving a gzipped copy. See memex IMPLEMENT.md"
	echo "pitfall #1."
	exit 1
	fi
	echo "OK: 206 Partial Content confirmed."

	- name: Verify Content-Range header
	run: \|
	set -eu
	if ! grep -qiE "^Content-Range: bytes 0-4095/[0-9]+" /tmp/db_headers.txt; then
	echo "::error::Missing or malformed Content-Range header"
	grep -i "^Content-Range" /tmp/db_headers.txt \|\| true
	exit 1
	fi
	echo "OK: Content-Range header valid."

	- name: Verify SQLite magic (bytes 0..3 == 'SQLi')
	run: \|
	set -eu
	# SQLite format 3 header starts with the ASCII "SQLite format 3\0".
	# First 4 bytes in hex: 5351 4c69 ("SQLi").
	head -c 4 /tmp/db_range.bin \| xxd \| tee /tmp/db_magic.txt
	if ! grep -q "5351 4c69" /tmp/db_magic.txt; then
	echo "::error::index.db first 4 bytes are NOT 'SQLi'"
	echo "The DB is corrupt or was served gzipped despite headers."
	exit 1
	fi
	echo "OK: SQLite magic confirmed."

	- name: Deploy verification passed
	run: \|
	echo "index.db is byte-range servable with correct SQLite header."
	echo "Deployed URL: ${{ needs.build.outputs.page_url }}"

Navigation Menu

Uh oh!

Build site #45

Workflow file

Build site #45

Uh oh!

Workflow file for this run