query-orchestration/api/storeEventRunner/runner_test.go

package storeeventrunner

import (
	"regexp"
	"testing"
	"time"

	"queryorchestration/internal/database/repository"
	documentstore "queryorchestration/internal/document/store"
	"queryorchestration/internal/server/runner"
	"queryorchestration/internal/serviceconfig/objectstore"
	"queryorchestration/internal/serviceconfig/queue"
	"queryorchestration/internal/serviceconfig/queue/documentinit"
	"queryorchestration/internal/test"

	"github.com/google/uuid"
	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"
)

type DocInitConfig struct {
	runner.BaseConfig[S3EventNotification]
	documentinit.DocInitConfig
	queue.QueueConfig
	objectstore.ObjectStoreConfig
}

func TestStoreEventRunner(t *testing.T) {
	cfg := &DocInitConfig{}
	test.CreateDB(t, cfg)
	acfg := test.CreateAWSContainer(t, cfg)

	test.SetQueueClient(t, t.Context(), cfg, acfg.ExternalEndpoint)
	cfg.DocInitURL = test.CreateQueue(t, cfg, test.DocInitRunnerName)

	runner := New(&Services{
		documentstore.New(cfg),
	})

	clientId := "hi"
	bucketName := "bucketName"
	location := objectstore.BucketKey{
		Location:  objectstore.Import,
		ClientID:  clientId,
		CreatedAt: time.Now().UTC(),
		EntityID:  uuid.New(),
	}
	doc := S3EventNotification{
		Records: []S3EventRecord{
			{
				EventName: EventS3ObjectCreatedPut,
				S3: S3EventRecordDetails{
					Bucket: S3Bucket{
						Name: bucketName,
					},
					Object: S3Object{
						Key:  location.String(),
						ETag: "example_hash",
					},
				},
			},
		},
	}

	err := cfg.GetDBQueries().CreateClient(t.Context(), &repository.CreateClientParams{
		Clientid: clientId,
		Name:     "name",
	})
	require.NoError(t, err)

	assert.True(t, runner.Process(t.Context(), doc))

	test.AssertMessageBody(t, cfg, cfg.GetDocInitURL(), regexp.MustCompile(`^\{"bucket":"bucketName","key":"hi\/import\/.+\/0\/.+","hash":"example_hash"\}$`))
}

func TestStoreEventRunner_BatchFiltering(t *testing.T) {
	cfg := &DocInitConfig{}
	test.CreateDB(t, cfg)
	acfg := test.CreateAWSContainer(t, cfg)

	test.SetQueueClient(t, t.Context(), cfg, acfg.ExternalEndpoint)
	cfg.DocInitURL = test.CreateQueue(t, cfg, test.DocInitRunnerName)

	runner := New(&Services{
		documentstore.New(cfg),
	})

	bucketName := "bucketName"

	// Test that batch keys are filtered out
	batchDoc := S3EventNotification{
		Records: []S3EventRecord{
			{
				EventName: EventS3ObjectCreatedPut,
				S3: S3EventRecordDetails{
					Bucket: S3Bucket{
						Name: bucketName,
					},
					Object: S3Object{
						Key:  "batches/test_client/2025/10/06/20/batch-id/file.pdf",
						ETag: "example_hash",
					},
				},
			},
		},
	}

	// Should return true (success) but not process the batch message
	assert.True(t, runner.Process(t.Context(), batchDoc))

	// Verify no message was sent to the queue (batch was filtered)
	// Try to receive a message with a short wait time - should get nothing
	result, err := cfg.ReceiveFromQueue(t.Context(), &queue.ReceiveParams{
		QueueURL:          cfg.GetDocInitURL(),
		VisibilityTimeout: 1,
		WaitTimeSeconds:   1,
	})
	require.NoError(t, err)
	assert.Empty(t, result.Messages, "expected no messages in queue for batch key")
}

func TestStoreEventRunner_MixedMessages(t *testing.T) {
	cfg := &DocInitConfig{}
	test.CreateDB(t, cfg)
	acfg := test.CreateAWSContainer(t, cfg)

	test.SetQueueClient(t, t.Context(), cfg, acfg.ExternalEndpoint)
	cfg.DocInitURL = test.CreateQueue(t, cfg, test.DocInitRunnerName)

	runner := New(&Services{
		documentstore.New(cfg),
	})

	clientId := "hi"
	bucketName := "bucketName"
	location := objectstore.BucketKey{
		Location:  objectstore.Import,
		ClientID:  clientId,
		CreatedAt: time.Now().UTC(),
		EntityID:  uuid.New(),
	}

	err := cfg.GetDBQueries().CreateClient(t.Context(), &repository.CreateClientParams{
		Clientid: clientId,
		Name:     "name",
	})
	require.NoError(t, err)

	// Mix of batch and normal messages
	mixedDoc := S3EventNotification{
		Records: []S3EventRecord{
			{
				EventName: EventS3ObjectCreatedPut,
				S3: S3EventRecordDetails{
					Bucket: S3Bucket{
						Name: bucketName,
					},
					Object: S3Object{
						Key:  "batches/test_client/2025/10/06/20/batch-id/file.pdf",
						ETag: "batch_hash",
					},
				},
			},
			{
				EventName: EventS3ObjectCreatedPut,
				S3: S3EventRecordDetails{
					Bucket: S3Bucket{
						Name: bucketName,
					},
					Object: S3Object{
						Key:  location.String(),
						ETag: "example_hash",
					},
				},
			},
		},
	}

	// Should return true and only process the non-batch message
	assert.True(t, runner.Process(t.Context(), mixedDoc))

	// Verify only one message was sent (the non-batch one)
	test.AssertMessageBody(t, cfg, cfg.GetDocInitURL(), regexp.MustCompile(`^\{"bucket":"bucketName","key":"hi\/import\/.+\/0\/.+","hash":"example_hash"\}$`))
}