Files
query-orchestration/internal/document/text/extract_test.go
T
Michael McGuinness 81e7223560 Merged in feature/textextraction (pull request #110)
Text Extraction

* bases

* go

* splitting

* structure

* movetoasync

* movetoasync

* settinguptrigger

* reorder

* storevent

* standardisepollingvalidation

* unittests

* fixlint

* fixlint

* awscfg

* generatesample

* followthrough

* tests

* clena

* store

* externalidcleanup

* clientid

* local

* baseunittests

* putobjecttests

* tests
2025-04-02 18:50:03 +00:00

152 lines
4.3 KiB
Go

package documenttext
import (
"context"
"fmt"
"testing"
"queryorchestration/internal/database/repository"
"queryorchestration/internal/document"
"queryorchestration/internal/serviceconfig"
"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"
"github.com/stretchr/testify/require"
"github.com/aws/aws-sdk-go-v2/service/sqs"
awstextract "github.com/aws/aws-sdk-go-v2/service/textract"
"github.com/google/uuid"
"github.com/pashagolub/pgxmock/v3"
"github.com/stretchr/testify/mock"
)
type DocTextConfig struct {
aws.AWSConfig
serviceconfig.BaseConfig
querysync.QuerySyncConfig
textract.TextractConfig
objectstore.ObjectStoreConfig
}
func TestCreate(t *testing.T) {
ctx := context.Background()
pool, err := pgxmock.NewPool()
require.NoError(t, err)
mockSQS := queuemock.NewMockSQSClient(t)
cfg := &DocTextConfig{}
cfg.QueueClient = mockSQS
cfg.QuerySyncURL = "/i/am/here"
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
mockS3 := objectstoremock.NewMockS3Client(t)
cfg.StoreClient = mockS3
mockTextract := textractmock.NewMockTextractClient(t)
cfg.TextractClient = mockTextract
svc := Service{
cfg: cfg,
}
t.Run("valid", func(t *testing.T) {
docId := uuid.New()
cleanId := uuid.New()
inloc := document.Location{
Bucket: "bucket_name",
Key: "/i/am/here",
}
clientId := "AAA"
hash := "example"
pool.ExpectQuery("name: IsDocumentTextExtracted :one").WithArgs(docId).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(docId).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail", "externalClientId"}).
AddRow(cleanId, docId, &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()
err = svc.Extract(ctx, docId)
require.NoError(t, err)
})
t.Run("invalid clean", func(t *testing.T) {
id := uuid.New()
fail := repository.NullCleanfailtype{
Valid: true,
Cleanfailtype: repository.CleanfailtypeInvalidMimetype,
}
pool.ExpectQuery("name: IsDocumentTextExtracted :one").WithArgs(id).WillReturnRows(
pgxmock.NewRows([]string{"isextracted"}).AddRow(false),
)
pool.ExpectQuery("name: GetCleanEntryByDocId :one").WithArgs(id).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail", "externalClientId"}).
AddRow(id, uuid.New(), nil, nil, int64(1), nil, nil, fail, "AAA"),
)
err = svc.Extract(ctx, id)
require.NoError(t, err)
})
}
func TestInformClean(t *testing.T) {
ctx := context.Background()
mockSQS := queuemock.NewMockSQSClient(t)
cfg := &DocTextConfig{}
cfg.QueueClient = mockSQS
cfg.QuerySyncURL = "/i/am/here"
svc := Service{
cfg: cfg,
}
id := uuid.New()
mockSQS.EXPECT().
SendMessage(
mock.Anything,
mock.MatchedBy(func(in *sqs.SendMessageInput) bool {
return *in.QueueUrl == cfg.QuerySyncURL && *in.MessageBody == fmt.Sprintf("{\"id\":\"%s\"}", id)
}),
mock.Anything,
).
Return(&sqs.SendMessageOutput{}, nil)
err := svc.informExtraction(ctx, id)
require.NoError(t, err)
}