fix(admin): don't pass untranslated plugin labels to Lingui #2430

Workflow file for this run

	name: Visual Regression

	# Untrusted measure job. Runs the visual-regression Playwright suite against the
	# committed baselines and, when snapshots differ, produces an artifact with the
	# candidate (accepted-state) baselines plus diff images. This workflow runs with
	# a read-only token and never comments or pushes. The companion workflows handle
	# that with elevated permissions, isolated from PR-authored code:
	# - "Visual Regression — Report" (workflow_run) posts the sticky comment.
	# - "Visual Regression — Apply" commits accepted baselines when a maintainer
	# comments /accept-baselines on the PR.
	#
	# Baselines are Linux/chromium PNGs regenerated here on the CI runner, so they
	# always match the environment they're diffed in. Contributors do NOT commit
	# baselines by hand (macOS PNGs would never match); the /accept-baselines flow
	# commits the correct Linux baselines for them.

	on:
	pull_request:
	branches: [main]
	paths:
	- "packages/admin/**"
	- "packages/core/**"
	- "e2e/**"
	- "playwright.config.ts"
	- ".github/workflows/visual.yml"
	- "pnpm-lock.yaml"
	- "package.json"

	permissions:
	contents: read

	jobs:
	measure:
	name: Measure
	runs-on: ubuntu-latest
	timeout-minutes: 20
	steps:
	- uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
	with:
	persist-credentials: false
	- uses: pnpm/action-setup@0e279bb959325dab635dd2c09392533439d90093 # v6.0.8
	- uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
	with:
	node-version: 22
	cache: pnpm
	- run: pnpm install --frozen-lockfile
	- run: pnpm locale:extract
	env:
	EMDASH_PSEUDO_LOCALE: "1"
	- run: pnpm run --filter emdash... build
	env:
	EMDASH_PSEUDO_LOCALE: "1"
	- uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
	id: playwright-cache
	with:
	path: ~/.cache/ms-playwright
	key: playwright-${{ hashFiles('pnpm-lock.yaml') }}
	- run: pnpm exec playwright install --with-deps chromium
	if: steps.playwright-cache.outputs.cache-hit != 'true'

	- name: Run visual regression
	id: run
	env:
	EMDASH_VISUAL: "1"
	EMDASH_PSEUDO_LOCALE: "1"
	run: \|
	set +e
	pnpm exec playwright test visual-regression --reporter=line
	code=$?
	set -e
	echo "exit=$code" >> "$GITHUB_OUTPUT"
	echo "Playwright exited with $code"

	# Always record PR metadata so the Report workflow can upsert or clear
	# the sticky comment. Candidate baselines + diff images are added only
	# when the regenerated baselines actually differ from the committed ones.
	# A non-zero exit alone is not drift: a flaky render can fail the diff
	# check on the first pass yet regenerate byte-identical baselines under
	# --update-snapshots, and an infra failure exits non-zero without
	# regenerating anything. Both leave the snapshots dir git-clean, so we
	# decide drift from `git status`, not from the exit code.
	- name: Collect result
	id: collect
	env:
	PR_NUMBER: ${{ github.event.pull_request.number }}
	HEAD_SHA: ${{ github.event.pull_request.head.sha }}
	RUN_EXIT: ${{ steps.run.outputs.exit }}
	SNAP_DIR: e2e/tests/visual-regression.spec.ts-snapshots
	run: \|
	mkdir -p visual-out
	echo "$PR_NUMBER" > visual-out/pr-number
	echo "$HEAD_SHA" > visual-out/head-sha

	if [ "$RUN_EXIT" = "0" ]; then
	echo "false" > visual-out/has-drift
	echo "has_drift=false" >> "$GITHUB_OUTPUT"
	echo "No visual drift."
	exit 0
	fi

	# Copy the diff/actual/expected triptychs Playwright writes on failure.
	mkdir -p visual-out/diff
	if [ -d test-results ]; then
	find test-results -type f \
	$ -name '-diff.png' -o -name '-actual.png' -o -name '*-expected.png' $ \
	-exec cp {} visual-out/diff/ \;
	fi

	# Regenerate the accepted-state baselines (the candidates the Apply
	# workflow commits verbatim on accept). These are Linux/chromium PNGs.
	EMDASH_PSEUDO_LOCALE=1 EMDASH_VISUAL=1 pnpm exec playwright test visual-regression --update-snapshots --reporter=line \|\| true

	# Real drift == the regeneration changed tracked baselines or added
	# new ones. Untracked new files count (bootstrap: a PR that adds a
	# page has no committed baseline yet). A clean tree means the failure
	# did not correspond to a stable pixel difference -- treat it as flake
	# or infra and keep the check green with a warning.
	if [ -z "$(git status --porcelain -- "$SNAP_DIR")" ]; then
	echo "false" > visual-out/has-drift
	echo "has_drift=false" >> "$GITHUB_OUTPUT"
	echo "::warning::Visual suite exited non-zero but regenerated baselines are identical to the committed ones (flaky render or infra failure, not a UI diff)."
	exit 0
	fi

	# Stage the candidate baselines for the Report/Apply workflows.
	mkdir -p visual-out/candidates
	if [ -d "$SNAP_DIR" ]; then
	cp -r "$SNAP_DIR/." visual-out/candidates/
	fi
	echo "true" > visual-out/has-drift
	echo "has_drift=true" >> "$GITHUB_OUTPUT"
	echo "Visual drift detected — candidates and diff images staged for the Report workflow."

	- name: Upload artifact
	uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
	with:
	name: visual-regression
	path: visual-out/
	retention-days: 7
	if-no-files-found: error

	# Make the check red so a UI PR gets a visible signal. This does not, by
	# itself, block merge -- keep it out of required checks until the false-
	# diff rate is near zero, then promote it.
	- name: Fail on drift
	if: steps.collect.outputs.has_drift == 'true'
	run: \|
	echo "::error::Visual snapshots changed. Review the diff in the sticky comment; a maintainer comments /accept-baselines to accept the new baselines."
	exit 1

Navigation Menu

fix(admin): don't pass untranslated plugin labels to Lingui #2430

Workflow file

fix(admin): don't pass untranslated plugin labels to Lingui #2430

Uh oh!

Workflow file for this run