Files
query-orchestration/api/docTextRunner/runner_test.go
T

123 lines
3.7 KiB
Go
Raw Normal View History

package doctextrunner_test
import (
"context"
"encoding/json"
"fmt"
2025-03-05 12:05:46 +00:00
"testing"
doctextrunner "queryorchestration/api/docTextRunner"
"queryorchestration/internal/database/repository"
"queryorchestration/internal/document"
documenttext "queryorchestration/internal/document/text"
"queryorchestration/internal/serviceconfig"
"queryorchestration/internal/serviceconfig/queue/querysync"
"queryorchestration/internal/serviceconfig/textract"
queuemock "queryorchestration/mocks/queue"
"github.com/aws/aws-sdk-go-v2/service/sqs"
"github.com/aws/aws-sdk-go-v2/service/sqs/types"
"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 {
serviceconfig.BaseConfig
querysync.QuerySyncConfig
textract.TextractConfig
}
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"
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{
ID: uuid.New(),
}
bodyBytes, err := json.Marshal(doc)
2025-03-19 11:54:14 +00:00
require.NoError(t, err)
body := string(bodyBytes)
msg := &types.Message{
Body: &body,
}
inloc := document.Location{
Bucket: "bucket_name",
Key: "/i/am/here",
}
pool.ExpectQuery("name: IsDocumentTextExtracted :one").WithArgs(doc.ID).WillReturnRows(
pgxmock.NewRows([]string{"isextracted"}).
AddRow(false),
)
cleanId := uuid.New()
mimeType := "application/pdf"
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
}
pool.ExpectQuery("name: GetCleanEntryByDocId :one").WithArgs(doc.ID).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "mimetype", "fail"}).
AddRow(cleanId, doc.ID, &inloc.Bucket, &inloc.Key, int64(1), dbmimetype, repository.NullCleanfailtype{}),
)
pool.ExpectQuery("name: GetCleanEntry :one").WithArgs(cleanId).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "mimetype", "fail"}).
AddRow(cleanId, doc.ID, &inloc.Bucket, &inloc.Key, int64(1), dbmimetype, repository.NullCleanfailtype{}),
)
textId := uuid.New()
pool.ExpectBegin()
pool.ExpectQuery("name: GetDocumentTextExtractionByHash :one").WithArgs(cleanId, "example").WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "cleanEntryId"}),
)
pool.ExpectQuery("name: AddDocumentText :one").WithArgs(inloc.Bucket, inloc.Key, cleanId, "example").
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(textId),
)
pool.ExpectExec("name: AddDocumentTextEntry :exec").WithArgs(textId, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
mockSQS.EXPECT().
SendMessage(
mock.Anything,
mock.MatchedBy(func(in *sqs.SendMessageInput) bool {
return *in.QueueUrl == cfg.QuerySyncURL && *in.MessageBody == fmt.Sprintf("{\"id\":\"%s\"}", doc.ID.String())
}),
mock.Anything,
).
Return(&sqs.SendMessageOutput{}, nil)
assert.True(t, runner.Process(ctx, msg))
})
t.Run("invalid body", func(t *testing.T) {
body := "definitely invalid"
msgId := "id"
msg := &types.Message{
Body: &body,
MessageId: &msgId,
}
assert.True(t, runner.Process(ctx, msg))
})
}