test: add unit, API, component and end-to-end test suites

The repo had no automated tests. Every change was verified by hand.

Adds four layers, 2023 tests in total, runnable with one command:

- 1191 unit tests over the pure logic in lib/, including the full
  computeProjectAccess permission matrix and the billing gate
- 167 component and hook tests in jsdom, covering the hooks that hold
  real logic rather than presentational wrappers
- 647 API integration tests against a real Postgres, with only auth()
  mocked, including a data-driven sweep asserting that none of the 60
  route modules answers 2xx to an unauthenticated caller
- 18 Playwright specs driving a real browser against a real build

Infrastructure: vitest.config.ts with three projects, a disposable
Postgres and MinIO in docker-compose.test.yml, factories and helpers
under tests/, scripts/test.sh as the single entry point, a pre-push
hook running bun run verify, and CI split into check, test and e2e jobs.

The test database is built with prisma db push plus a replay of the
hand-written SQL, because prisma migrate deploy cannot build this schema
from empty: the migration history has no captured baseline. This mirrors
what scripts/docker-db-bootstrap.ts already does in production, and
tests/setup/db-global.ts carries a drift guard so a new migration fails
the run until someone reviews it.

Production code is unchanged apart from one pure-function extraction out
of use-video-player.ts, which was too large to test in jsdom.

Several tests pin behaviour that looks wrong, each marked KNOWN BUG in
place. TESTING.md section 12 records where the plan turned out to be
wrong, and AGENTS.md now states which layer a change needs a test in.
This commit is contained in:
yusufipk
2026-07-26 11:17:26 +07:00
parent 52b2c8d2a9
commit 1d099c68f2
101 changed files with 27625 additions and 122 deletions
+22
View File
@@ -0,0 +1,22 @@
/**
* Builds the test database schema outside of Vitest.
*
* The `api` Vitest project gets this for free through its globalSetup, but the
* end-to-end suite runs the real app against the same database and needs the
* schema in place before the server starts. Both paths therefore call the same
* setup function, so there is exactly one description of how a test database is
* built (including why it uses `prisma db push` rather than `migrate deploy`,
* which is documented at the top of tests/setup/db-global.ts).
*
* Usage: bun run test:db:bootstrap
*/
import { setup } from '../tests/setup/db-global';
setup()
.then(() => {
console.log('Test database schema is ready');
})
.catch((error: unknown) => {
console.error(error instanceof Error ? error.message : error);
process.exit(1);
});
+302
View File
@@ -0,0 +1,302 @@
#!/usr/bin/env sh
# One entry point for the OpenFrame test suites.
#
# scripts/test.sh unit vitest unit + component projects
# scripts/test.sh api vitest api project (needs the test database)
# scripts/test.sh e2e playwright specs (needs the test database)
# scripts/test.sh all unit, then api, then e2e
#
# Run the api and e2e suites one at a time, never side by side. They share one
# database, and the api suite empties every table after each of its tests, so a
# concurrent e2e run loses the rows it seeded and fails for no real reason.
# `all` runs them in sequence for exactly this reason.
#
# Every suite runs inside a container, so no package manager runs on the host.
# TESTING.md section 8 documents the raw podman commands this wraps.
set -eu
bun_image='docker.io/oven/bun:alpine'
# Pinned to the installed @playwright/test version. The image carries the
# matching browser build, and Playwright refuses a mismatched pair. Microsoft
# publishes the image some time after the npm release, so check the tag exists
# before bumping either half:
# curl -sI https://mcr.microsoft.com/v2/playwright/manifests/v1.61.1-noble
playwright_image='mcr.microsoft.com/playwright:v1.61.1-noble'
# Shared podman network, so the runner container reaches Postgres by service
# name instead of a published port.
network='openframe-test'
# Resolve the repo root from this script's own location, so the script behaves
# the same from any working directory.
script_dir=$(CDPATH='' cd -- "$(dirname -- "$0")" && pwd)
repo_root=$(CDPATH='' cd -- "$script_dir/.." && pwd)
compose_file="$repo_root/docker-compose.test.yml"
playwright_config="$repo_root/playwright.config.ts"
env_test="$repo_root/.env.test"
env_test_example="$repo_root/.env.test.example"
# Attach a TTY only when there is one, so the script also works from a hook,
# a pipe, or a CI runner.
if [ -t 1 ]; then
tty_flag='-t'
else
tty_flag=''
fi
usage() {
cat <<'EOF'
Usage: scripts/test.sh <unit|api|e2e|all>
unit Unit and component suites. No database, no browser.
api API integration suites. Starts the disposable test Postgres first.
e2e Playwright end-to-end specs. Starts the test Postgres and MinIO first,
then builds and starts the app itself on port 3100.
all unit, then api, then e2e.
The containers keep running afterwards so the next run is fast. Stop them
with: podman compose -f docker-compose.test.yml --profile e2e down -v
EOF
}
say() {
printf '\n==> %s\n' "$1"
}
# Each argument is printed on its own line, so a diagnostic can carry the fix
# right under the problem.
die() {
printf 'scripts/test.sh: %s\n' "$1" >&2
shift
for line in "$@"; do
printf ' %s\n' "$line" >&2
done
exit 1
}
show() {
printf '+'
for word in "$@"; do
# Quote the arguments that contain spaces, so the printed line reads like
# something you could paste back into a shell.
case $word in
*' '*) printf " '%s'" "$word" ;;
*) printf ' %s' "$word" ;;
esac
done
printf '\n'
}
# Prints the command, runs it, and exits with the command's own status so the
# caller (a hook, CI, or a shell) sees the real result.
run_cmd() {
show "$@"
set +e
"$@"
status=$?
set -e
if [ "$status" -ne 0 ]; then
printf '\nscripts/test.sh: %s exited with %s\n' "$1" "$status" >&2
exit "$status"
fi
}
# The tty_flag expansion below stays quoted when set and disappears entirely
# when empty, which a plain "$tty_flag" cannot do (it would pass an empty
# argument to podman).
run_in_bun_image() {
bun_network=$1
bun_command=$2
if [ -n "$bun_network" ]; then
run_cmd podman run --rm ${tty_flag:+"$tty_flag"} --network "$bun_network" \
-v "$repo_root:/workspace:z" -w /workspace "$bun_image" sh -c "$bun_command"
else
run_cmd podman run --rm ${tty_flag:+"$tty_flag"} \
-v "$repo_root:/workspace:z" -w /workspace "$bun_image" sh -c "$bun_command"
fi
}
# --ipc=host is Playwright's documented requirement for Chromium in a
# container; without it Chromium runs out of shared memory on larger pages.
run_in_playwright_image() {
run_cmd podman run --rm ${tty_flag:+"$tty_flag"} --ipc=host --network "$network" \
-v "$repo_root:/workspace:z" -w /workspace "$playwright_image" sh -c "$1"
}
require_compose_file() {
[ -f "$compose_file" ] && return 0
die "docker-compose.test.yml not found at $compose_file." \
'The test database ships with Phase 2 of TESTING.md (section 5), so the api' \
'and e2e suites cannot run until that lands.'
}
require_env_test() {
[ -f "$env_test" ] && return 0
if [ -f "$env_test_example" ]; then
die "$env_test not found." \
'Create it once with: cp .env.test.example .env.test'
fi
die "Neither $env_test nor $env_test_example exists." \
'Both ship with Phase 2 of TESTING.md (section 5).'
}
require_playwright_config() {
[ -f "$playwright_config" ] && return 0
die "playwright.config.ts not found at $playwright_config." \
'The end-to-end suite is Phase 3 of TESTING.md (section 6) and has not' \
'landed yet, so there is nothing for playwright to run.'
}
ensure_network() {
if podman network exists "$network"; then
return 0
fi
say "creating the $network podman network"
run_cmd podman network create "$network"
}
# `podman compose up -d --wait` is not usable here. With podman-compose 1.6.0 as
# the provider it does not block on the healthcheck at all when the service is
# starting, and worse, it never returns when the service is already up and
# healthy: an e2e run was observed wedged on it for 22 minutes with nothing to
# show for it. So every service is started without --wait and its readiness is
# polled here instead.
wait_for_test_db() {
say 'waiting for postgres-test to accept connections'
attempt=0
while [ "$attempt" -lt 60 ]; do
if podman compose -f "$compose_file" exec -T postgres-test \
pg_isready -U openframe -d openframe_test >/dev/null 2>&1; then
printf 'postgres-test is ready\n'
return 0
fi
attempt=$((attempt + 1))
sleep 1
done
die 'postgres-test did not become ready within 60 seconds.' \
"Inspect it with: podman compose -f $compose_file logs postgres-test"
}
wait_for_test_storage() {
say 'waiting for minio-test to report healthy'
attempt=0
while [ "$attempt" -lt 60 ]; do
if podman exec openframe-minio-test mc ready local >/dev/null 2>&1; then
printf 'minio-test is ready\n'
return 0
fi
attempt=$((attempt + 1))
sleep 1
done
die 'minio-test did not become ready within 60 seconds.' \
"Inspect it with: podman compose -f $compose_file logs minio-test"
}
start_test_db() {
ensure_network
say 'starting the test database'
run_cmd podman compose -f "$compose_file" up -d postgres-test
wait_for_test_db
}
# Object storage for the direct video upload flow. The browser PUTs the file
# straight at the presigned URL, so this has to be real; there is nothing to
# mock at that boundary from inside a browser.
#
# minio-test-init is a one-shot container that creates the bucket. Nothing at
# runtime does: ensureR2BucketExists() lives in scripts/self-host-bootstrap.ts,
# not on the request path.
start_test_storage() {
say 'starting object storage for the upload flow'
# Asking compose to start a container that is already up prints a red
# `cannot start an already running container` error and keeps going, which
# reads like a failure in the log of an otherwise clean run. Skip the call
# instead.
if podman exec openframe-minio-test mc ready local >/dev/null 2>&1; then
printf 'minio-test is already running\n'
else
run_cmd podman compose -f "$compose_file" --profile e2e up -d minio-test
wait_for_test_storage
fi
# --no-deps: the init container declares depends_on minio-test, and without
# this compose tries to start that dependency again and prints the same
# spurious `already running` error the guard above exists to avoid.
run_cmd podman compose -f "$compose_file" --profile e2e up --no-deps minio-test-init
}
# --frozen-lockfile keeps a test run from rewriting bun.lock as a side effect.
install_step='bun install --frozen-lockfile'
run_unit() {
say 'unit and component suites'
run_in_bun_image '' "$install_step && bun run test"
}
run_api() {
say 'api suites'
require_compose_file
require_env_test
start_test_db
run_in_bun_image "$network" "$install_step && bun run test:api"
}
run_e2e() {
say 'end-to-end specs'
require_compose_file
require_playwright_config
require_env_test
start_test_db
start_test_storage
# The official Playwright image carries node and the browsers but not bun.
# bun is needed for `bun run test:e2e`; the web server inside
# playwright.config.ts runs next through node_modules/.bin, so it works under
# either runtime. bun is installed into the throwaway container, never on the
# host, and `oven-sh/setup-bun` cannot be used because the image has no unzip.
#
# Playwright starts and stops the app itself (`webServer`), on port 3100 so it
# cannot attach to a dev server on 3000. The build output lands in the mounted
# .next, which is what keeps the second run fast.
run_in_playwright_image \
"npm install --global --silent bun && $install_step && bun run test:e2e"
}
case "${1-}" in
-h | --help | help)
usage
exit 0
;;
esac
if [ "$#" -ne 1 ]; then
printf 'scripts/test.sh: exactly one mode is required\n\n' >&2
usage >&2
exit 64
fi
if ! command -v podman >/dev/null 2>&1; then
die 'podman was not found on PATH.' \
'Every suite runs in a container, so podman is required.'
fi
case "$1" in
unit)
run_unit
;;
api)
run_api
;;
e2e)
run_e2e
;;
all)
run_unit
run_api
run_e2e
;;
*)
printf 'scripts/test.sh: unknown mode "%s"\n\n' "$1" >&2
usage >&2
exit 64
;;
esac