db-refresh-dev: own dev DB as personal_dev, no prod grants; PG_CONTAINER mode

scripts/db-refresh-dev.sh

@@ -1,37 +1,58 @@
 #!/usr/bin/env bash
-# Refresh the local development database from a fresh production dump.
+# Refresh the development database (personal_dev) from a fresh production dump.
 #
 # Prod is the source of truth for data; dev is a disposable copy of it.
 #
 # Required env (put them in .env.dev-refresh or export them; never commit real values):
 #   PROD_DB_URL    read access to prod, e.g. postgresql://personal:...@127.0.0.1:5440/personal
-#   DEV_ADMIN_URL  superuser on the dev cluster, maintenance DB, e.g. postgresql://postgres:...@127.0.0.1:5432/postgres
+#   DEV_ADMIN_URL  superuser on the dev cluster, maintenance DB, e.g. postgresql://postgres:...@127.0.0.1:5440/postgres
+#                  (not needed with PG_CONTAINER, which uses the container's local superuser)
 # Optional:
 #   DEV_DB_NAME    target database (default personal_dev)
+#   DEV_OWNER      role that owns every object in the dev DB (default personal_dev). It is the
+#                  role the dev site logs in as, so it can also run migrations there. It must not
+#                  be a member of any prod role; prod's CONNECT is not granted to it.
 #   PG_DOCKER=1    use the postgres:17-alpine image instead of local pg_dump/pg_restore
+#   PG_CONTAINER   name of the Postgres container (e.g. postgres-db-1): run every tool inside it
+#                  via `docker exec` as the local superuser. URLs are then as seen from inside
+#                  the container, e.g. PROD_DB_URL=postgresql://personal:...@127.0.0.1:5432/personal
+#   DOCKER         docker command (default "docker"; e.g. "sudo -n docker")
 #
-# What it does: dump prod (custom format) -> drop + recreate DEV_DB_NAME -> create
-# missing roles (NOLOGIN) -> restore with ownership/grants intact -> clear sessions
-# and verification tokens so prod logins can't be reused in dev. It never writes to prod.
+# What it does: dump prod (custom format) -> drop + recreate DEV_DB_NAME owned by DEV_OWNER ->
+# restore without prod ownership/grants, every object owned by DEV_OWNER -> clear sessions and
+# verification tokens so prod logins can't be reused in dev. It never writes to prod.
 set -euo pipefail
 
 cd "$(dirname "$0")/.."
 if [[ -f .env.dev-refresh ]]; then set -a; . ./.env.dev-refresh; set +a; fi
 
 : "${PROD_DB_URL:?set PROD_DB_URL (source, read-only use)}"
-: "${DEV_ADMIN_URL:?set DEV_ADMIN_URL (superuser URL to the dev cluster maintenance DB)}"
 DEV_DB_NAME="${DEV_DB_NAME:-personal_dev}"
-
-if [[ ! "$DEV_DB_NAME" =~ ^[a-z_][a-z0-9_]*$ ]]; then
-	echo "DEV_DB_NAME must be a plain lowercase identifier" >&2
-	exit 1
+DEV_OWNER="${DEV_OWNER:-personal_dev}"
+DOCKER="${DOCKER:-docker}"
+if [[ -n "${PG_CONTAINER:-}" ]]; then
+	DEV_ADMIN_URL="${DEV_ADMIN_URL:-postgresql://postgres@localhost/postgres}"
 fi
+: "${DEV_ADMIN_URL:?set DEV_ADMIN_URL (superuser URL to the dev cluster maintenance DB)}"
+
+for ident in "$DEV_DB_NAME" "$DEV_OWNER"; do
+	if [[ ! "$ident" =~ ^[a-z_][a-z0-9_]*$ ]]; then
+		echo "DEV_DB_NAME / DEV_OWNER must be plain lowercase identifiers" >&2
+		exit 1
+	fi
+done
 case "$DEV_DB_NAME" in
 personal | personal_prod | postgres | template0 | template1)
 	echo "Refusing to overwrite '$DEV_DB_NAME'" >&2
 	exit 1
 	;;
 esac
+case "$DEV_OWNER" in
+personal | personal_app | personal_ddl | personal_ro | postgres)
+	echo "Refusing: DEV_OWNER '$DEV_OWNER' is a prod/superuser role" >&2
+	exit 1
+	;;
+esac
 
 # host:port/db of a URL (no credentials), for the "same database" guard.
 endpoint() { bun -e 'const u=new URL(process.argv[1]); console.log(`${u.hostname}:${u.port||5432}${u.pathname}`)' "$1"; }
@@ -44,8 +65,12 @@ if [[ "$(endpoint "$dev_url")" == "$prod_endpoint" ]]; then
 	exit 1
 fi
 
-if [[ "${PG_DOCKER:-0}" == 1 ]]; then
-	pg() { docker run --rm -i --network host -v "$work:$work" postgres:17-alpine "$@"; }
+# Dump goes to stdout and restore reads stdin, so the same commands work locally,
+# in a throwaway postgres image, or inside the Postgres container.
+if [[ -n "${PG_CONTAINER:-}" ]]; then
+	pg() { $DOCKER exec -i -u postgres "$PG_CONTAINER" "$@"; }
+elif [[ "${PG_DOCKER:-0}" == 1 ]]; then
+	pg() { $DOCKER run --rm -i --network host postgres:17-alpine "$@"; }
 else
 	pg() { "$@"; }
 fi
@@ -55,27 +80,30 @@ trap 'rm -rf "$work"' EXIT
 dump="$work/prod.dump"
 
 echo "Dumping prod ($prod_endpoint)..."
-pg pg_dump --format=custom --no-password --file="$dump" "$PROD_DB_URL"
+pg pg_dump --format=custom --no-password "$PROD_DB_URL" >"$dump"
+[[ -s "$dump" ]] || { echo "Empty dump; aborting before touching $DEV_DB_NAME" >&2; exit 1; }
+pg pg_restore --list <"$dump" >/dev/null
 
-echo "Recreating $DEV_DB_NAME..."
+echo "Recreating $DEV_DB_NAME (owner $DEV_OWNER)..."
 pg psql -v ON_ERROR_STOP=1 -q "$DEV_ADMIN_URL" <<SQL
-SELECT pg_terminate_backend(pid) FROM pg_stat_activity WHERE datname = '$DEV_DB_NAME' AND pid <> pg_backend_pid();
+SELECT count(pg_terminate_backend(pid)) AS terminated_sessions FROM pg_stat_activity WHERE datname = '$DEV_DB_NAME' AND pid <> pg_backend_pid();
 DROP DATABASE IF EXISTS "$DEV_DB_NAME";
 DO \$\$
-DECLARE r text;
 BEGIN
-  FOREACH r IN ARRAY ARRAY['personal','personal_app','personal_ddl','personal_ro'] LOOP
-    IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = r) THEN
-      EXECUTE format('CREATE ROLE %I NOLOGIN', r);
-      RAISE NOTICE 'created role % (NOLOGIN; give it a password + LOGIN for dev use)', r;
-    END IF;
-  END LOOP;
+  IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = '$DEV_OWNER') THEN
+    EXECUTE format('CREATE ROLE %I NOLOGIN', '$DEV_OWNER');
+    RAISE NOTICE 'created role $DEV_OWNER (NOLOGIN; give it a password + LOGIN for dev use)';
+  END IF;
 END \$\$;
-CREATE DATABASE "$DEV_DB_NAME" OWNER personal;
+CREATE DATABASE "$DEV_DB_NAME" OWNER "$DEV_OWNER";
+REVOKE CONNECT, TEMPORARY ON DATABASE "$DEV_DB_NAME" FROM PUBLIC;
 SQL
 
 echo "Restoring into $DEV_DB_NAME..."
-pg pg_restore --no-password --exit-on-error --dbname="$dev_url" "$dump"
+# --no-owner/--no-privileges + --role: everything ends up owned by DEV_OWNER and none of
+# prod's grants (personal_app, personal_ro, ...) are copied into dev.
+pg pg_restore --no-password --exit-on-error --no-owner --no-privileges --role="$DEV_OWNER" \
+	--dbname="$dev_url" <"$dump"
 
 echo "Clearing sessions and verification tokens..."
 pg psql -v ON_ERROR_STOP=1 -q "$dev_url" -c 'TRUNCATE auth.session, auth.verification'