descrybe/apps/api/internal/processing/retention_cleanup_integration_test.go

package processing

import (
	"context"
	"os"
	"testing"
	"time"

	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"
)

func TestCleanupExpiredDeletesTerminalJobsAndCascadesProducts(t *testing.T) {
	dsn := os.Getenv("DATABASE_URL")
	if dsn == "" {
		t.Skip("DATABASE_URL not set")
	}

	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
	defer cancel()

	pg, err := pgxpool.New(ctx, dsn)
	if err != nil {
		t.Fatal(err)
	}
	defer pg.Close()

	var companyID uuid.UUID
	err = pg.QueryRow(ctx, `
		SELECT company_id
		FROM raw_products
		WHERE company_id IS NOT NULL
		ORDER BY updated_at DESC
		LIMIT 1`).Scan(&companyID)
	if errorsIsNoRows(err) {
		t.Skip("no raw_products rows available")
	}
	if err != nil {
		t.Fatal(err)
	}

	var userID uuid.UUID
	err = pg.QueryRow(ctx, `SELECT id FROM users ORDER BY created_at DESC LIMIT 1`).Scan(&userID)
	if errorsIsNoRows(err) {
		t.Skip("no users rows available")
	}
	if err != nil {
		t.Fatal(err)
	}

	var rawID uuid.UUID
	err = pg.QueryRow(ctx, `
		SELECT id
		FROM raw_products
		WHERE company_id = $1
		ORDER BY updated_at DESC
		LIMIT 1`, companyID).Scan(&rawID)
	if errorsIsNoRows(err) {
		t.Skip("no raw_products for selected company")
	}
	if err != nil {
		t.Fatal(err)
	}

	insertJob := func(status string, age string) uuid.UUID {
		t.Helper()
		var id uuid.UUID
		err := pg.QueryRow(ctx, `
			INSERT INTO processing_jobs (
				company_id, user_id, status, total_products, processed_products,
				processing_type, started_at, completed_at, updated_at, created_at
			) VALUES (
				$1, $2, $3, 1, 1, 'full',
				now() - interval '`+age+`',
				now() - interval '`+age+`',
				now() - interval '`+age+`',
				now() - interval '`+age+`'
			)
			RETURNING id`, companyID, userID, status).Scan(&id)
		if err != nil {
			t.Fatal(err)
		}
		return id
	}

	oldCompleted := insertJob("completed", "45 days")
	oldFailed := insertJob("failed", "45 days")
	oldRunning := insertJob("running", "45 days")
	freshCompleted := insertJob("completed", "1 day")

	defer func() {
		for _, id := range []uuid.UUID{oldCompleted, oldFailed, oldRunning, freshCompleted} {
			_, _ = pg.Exec(context.Background(), `DELETE FROM processing_job_products WHERE job_id = $1`, id)
			_, _ = pg.Exec(context.Background(), `DELETE FROM processing_jobs WHERE id = $1`, id)
		}
	}()

	if _, err := pg.Exec(ctx, `
		INSERT INTO processing_job_products (job_id, raw_product_id, status, updated_at, created_at)
		VALUES ($1, $2, 'processed', now() - interval '45 days', now() - interval '45 days')`,
		oldCompleted, rawID); err != nil {
		t.Fatal(err)
	}

	res, err := CleanupExpired(ctx, pg)
	if err != nil {
		t.Fatal(err)
	}
	if res.JobsDeleted < 2 {
		t.Fatalf("jobs_deleted=%d want >= 2", res.JobsDeleted)
	}

	assertGone := func(id uuid.UUID, label string) {
		t.Helper()
		var n int
		if err := pg.QueryRow(ctx, `SELECT COUNT(*) FROM processing_jobs WHERE id = $1`, id).Scan(&n); err != nil {
			t.Fatal(err)
		}
		if n != 0 {
			t.Fatalf("%s job still present", label)
		}
	}
	assertPresent := func(id uuid.UUID, label string) {
		t.Helper()
		var n int
		if err := pg.QueryRow(ctx, `SELECT COUNT(*) FROM processing_jobs WHERE id = $1`, id).Scan(&n); err != nil {
			t.Fatal(err)
		}
		if n != 1 {
			t.Fatalf("%s job missing", label)
		}
	}

	assertGone(oldCompleted, "old completed")
	assertGone(oldFailed, "old failed")
	assertPresent(oldRunning, "old running")
	assertPresent(freshCompleted, "fresh completed")

	var productCount int
	if err := pg.QueryRow(ctx, `
		SELECT COUNT(*) FROM processing_job_products WHERE job_id = $1`, oldCompleted).Scan(&productCount); err != nil {
		t.Fatal(err)
	}
	if productCount != 0 {
		t.Fatalf("cascaded products remaining=%d want 0", productCount)
	}
}