query-orchestration/api/docTextRunner/runner_test.go

package doctextrunner_test

import (
	"context"
	"testing"

	doctextrunner "queryorchestration/api/docTextRunner"
	"queryorchestration/internal/database/repository"
	"queryorchestration/internal/document"
	documenttext "queryorchestration/internal/document/text"
	"queryorchestration/internal/server/runner"
	"queryorchestration/internal/serviceconfig/aws"
	"queryorchestration/internal/serviceconfig/objectstore"
	"queryorchestration/internal/serviceconfig/queue/querysync"
	"queryorchestration/internal/serviceconfig/textract"
	objectstoremock "queryorchestration/mocks/objectstore"
	queuemock "queryorchestration/mocks/queue"
	textractmock "queryorchestration/mocks/textract"

	awstextract "github.com/aws/aws-sdk-go-v2/service/textract"
	"github.com/go-playground/validator/v10"
	"github.com/google/uuid"
	"github.com/pashagolub/pgxmock/v3"
	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/mock"
	"github.com/stretchr/testify/require"
)

type DocTextConfig struct {
	runner.BaseConfig[doctextrunner.Body]
	aws.AWSConfig
	querysync.QuerySyncConfig
	textract.TextractConfig
	objectstore.ObjectStoreConfig
}

func TestDocCleanRunner(t *testing.T) {
	ctx := context.Background()

	pool, err := pgxmock.NewPool()
	require.NoError(t, err)

	cfg := &DocTextConfig{}
	cfg.DBPool = pool
	cfg.DBQueries = repository.New(pool)
	mockSQS := queuemock.NewMockSQSClient(t)
	cfg.QueueClient = mockSQS
	cfg.QuerySyncURL = "/i/am/here"
	mockS3 := objectstoremock.NewMockS3Client(t)
	cfg.StoreClient = mockS3
	mockTextract := textractmock.NewMockTextractClient(t)
	cfg.TextractClient = mockTextract

	runner := doctextrunner.New(validator.New(), &doctextrunner.Services{
		Text: documenttext.New(cfg),
	})
	assert.NotNil(t, runner)

	t.Run("valid", func(t *testing.T) {
		doc := doctextrunner.Body{
			DocumentID: uuid.New(),
		}

		cleanId := uuid.New()
		inloc := document.Location{
			Bucket: "bucket_name",
			Key:    "/i/am/here",
		}
		hash := "example"

		pool.ExpectQuery("name: IsDocumentTextExtracted :one").WithArgs(doc.DocumentID).WillReturnRows(
			pgxmock.NewRows([]string{"isextracted"}).
				AddRow(false),
		)
		mimeType := "application/pdf"
		dbmimetype := repository.NullCleanmimetype{
			Valid:         true,
			Cleanmimetype: repository.Cleanmimetype(mimeType),
		}
		pool.ExpectQuery("name: GetCleanEntryByDocId :one").WithArgs(doc.DocumentID).WillReturnRows(
			pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail", "externalClientId"}).
				AddRow(cleanId, doc.DocumentID, &inloc.Bucket, &inloc.Key, int64(1), &hash, dbmimetype, repository.NullCleanfailtype{}, "clientId"),
		)
		jobId := "hello"
		triggerId := uuid.New()
		pool.ExpectBegin()
		pool.ExpectQuery("name: AddDocumentTextTrigger :one").WithArgs(cleanId, pgxmock.AnyArg()).WillReturnRows(
			pgxmock.NewRows([]string{"id"}).AddRow(triggerId),
		)

		mockTextract.EXPECT().
			StartDocumentTextDetection(
				mock.Anything,
				mock.MatchedBy(func(in *awstextract.StartDocumentTextDetectionInput) bool {
					return true
				}),
				mock.Anything,
			).
			Return(&awstextract.StartDocumentTextDetectionOutput{
				JobId: &jobId,
			}, nil)

		pool.ExpectExec("name: AddDocumentTextTriggerJobId :exec").WithArgs(&jobId, triggerId).
			WillReturnResult(pgxmock.NewResult("", 1))
		pool.ExpectCommit()

		assert.True(t, runner.Process(ctx, doc))
	})
}