mirror of
https://github.com/tiennm99/thptqg.git
synced 2026-10-11 03:13:48 +00:00
tiennm99
160a20125e
The group was the literal string "pages", so every run of this workflow shared one lane regardless of branch, and cancel-in-progress meant the newest arrival won. A pull-request run therefore cancels an in-flight deploy of main. That is not theoretical: the deploy of #10 was killed 3m22s in by a pull-request run that started after it, and the site quietly stayed on the previous build. Nothing reported a failure — the PR checks were green and the deploy showed "cancelled", which reads like something someone chose. Keyed by ref, a push still cancels its own superseded run, which is the case worth cancelling, and a branch can no longer interrupt a deploy. Also drops "compress it" from the pipeline comment: the databases have shipped uncompressed since they started being read by range request.
124 lines
4.6 KiB
YAML
124 lines
4.6 KiB
YAML
name: Deploy to GitHub Pages
|
|
|
|
on:
|
|
push:
|
|
branches: [main]
|
|
# Pull requests run the build job only: the deploy job is guarded to main, so
|
|
# a branch can be verified end to end without touching the live site.
|
|
pull_request:
|
|
workflow_dispatch:
|
|
|
|
permissions:
|
|
contents: read
|
|
pages: write
|
|
id-token: write
|
|
|
|
# Keyed by ref, not just "pages". With one shared group a pull-request run and
|
|
# a main deploy compete for the same lane, and cancel-in-progress means the
|
|
# newer one wins: the deploy of #10 was killed 3m22s in by a PR run that
|
|
# started after it, and the site silently stayed on the previous build while
|
|
# every check stayed green. Per ref, a push still cancels its own superseded
|
|
# run, which is the case where cancelling is worth having.
|
|
concurrency:
|
|
group: pages-${{ github.ref }}
|
|
cancel-in-progress: true
|
|
|
|
jobs:
|
|
build:
|
|
runs-on: ubuntu-latest
|
|
env:
|
|
# Every Go module here is cgo-free — grate, excelize, yaml.v3, x/net,
|
|
# x/text and modernc.org/sqlite — so no C toolchain is needed. Set
|
|
# explicitly rather than relying on the default.
|
|
CGO_ENABLED: '0'
|
|
steps:
|
|
- uses: actions/checkout@v7
|
|
|
|
# The version comes from go.mod rather than a literal here. setup-go sets
|
|
# GOTOOLCHAIN=local, so Go will not fetch the toolchain a module asks for
|
|
# — the one installed has to satisfy it. A literal '1.26' resolves to
|
|
# whatever patch the runner manifest has, which was 1.26.5 against a
|
|
# go.mod requiring 1.26.6, and the build stopped there. All three modules
|
|
# are bumped together, so parser/go.mod speaks for them.
|
|
- uses: actions/setup-go@v7
|
|
with:
|
|
go-version-file: parser/go.mod
|
|
cache-dependency-path: |
|
|
parser/go.sum
|
|
crawler/go.sum
|
|
assembler/go.sum
|
|
|
|
# web/ is the only npm project in the repository; the other stages are Go.
|
|
- uses: actions/setup-node@v7
|
|
with:
|
|
node-version: '24'
|
|
cache: 'npm'
|
|
cache-dependency-path: web/package-lock.json
|
|
|
|
- name: Install web dependencies
|
|
working-directory: web
|
|
run: npm ci
|
|
|
|
# The reader-fidelity suite compares every real input file against a
|
|
# committed hash oracle, so it is the regression guard for the whole
|
|
# reader. Runs before anything is built.
|
|
- name: Test parser
|
|
run: go -C parser test ./...
|
|
|
|
# The crawler is not part of the build — it only refreshes data/ by hand.
|
|
# It is still tested here so it cannot rot unnoticed, and because its
|
|
# fixture test guards the parser: input filenames decide which row
|
|
# survives a duplicate exam number.
|
|
- name: Test crawler
|
|
run: go -C crawler test ./...
|
|
|
|
# The assembler owns every guard between a built database and the
|
|
# published site, so its tests are the ones that prove a short or missing
|
|
# database cannot ship.
|
|
- name: Test assembler
|
|
run: go -C assembler test ./...
|
|
|
|
# The tests cover the framework-free modules, including the ASCII fold
|
|
# that has to match the Go parser.
|
|
- name: Test web
|
|
working-directory: web
|
|
run: npm test
|
|
|
|
- name: Lint web
|
|
working-directory: web
|
|
run: npm run lint
|
|
|
|
# excelize carries an open advisory, and the 2017 refresh runbook feeds
|
|
# network-downloaded spreadsheets straight into the parser.
|
|
- name: Vulnerability scan
|
|
run: |
|
|
go install golang.org/x/vuln/cmd/govulncheck@latest
|
|
GOVULNCHECK="$(go env GOPATH)/bin/govulncheck"
|
|
for m in parser crawler assembler; do (cd "$m" && "$GOVULNCHECK" ./...); done
|
|
|
|
# One command runs the whole pipeline: compile the parser, build and
|
|
# verify each database against its registry row count and size, build the
|
|
# web app, and assemble _site — refusing to continue if a database is
|
|
# short, an artifact looks truncated, or one is missing entirely.
|
|
- name: Build site
|
|
run: go -C assembler run ./cmd/assemble
|
|
|
|
- uses: actions/upload-pages-artifact@v5
|
|
with:
|
|
path: _site
|
|
|
|
deploy:
|
|
# Deploy only from main. pull_request and workflow_dispatch both run on
|
|
# other branches, and publishing one would put that branch's output on the
|
|
# live site while concurrency cancel-in-progress killed an in-flight good
|
|
# deploy on the way.
|
|
if: github.ref == 'refs/heads/main'
|
|
needs: build
|
|
runs-on: ubuntu-latest
|
|
environment:
|
|
name: github-pages
|
|
url: ${{ steps.deployment.outputs.page_url }}
|
|
steps:
|
|
- id: deployment
|
|
uses: actions/deploy-pages@v5
|