Files
query-orchestration/internal/document/text/extract_test.go
T
Michael McGuinness 53ea7d34e6 Merged in feature/textextract (pull request #108)
Start adding Textract + UUID changes

* base

* startclient

* ts

* short

* tests
2025-03-20 11:06:41 +00:00

163 lines
4.4 KiB
Go

package documenttext
import (
"context"
"testing"
"queryorchestration/internal/database/repository"
"queryorchestration/internal/document"
"github.com/stretchr/testify/require"
"github.com/google/uuid"
"github.com/pashagolub/pgxmock/v3"
"github.com/stretchr/testify/assert"
)
func TestExtract(t *testing.T) {
ctx := context.Background()
pool, err := pgxmock.NewPool()
require.NoError(t, err)
cfg := &DocTextConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
svc := Service{
cfg: cfg,
}
t.Run("successful clean", func(t *testing.T) {
id := uuid.New()
inloc := document.Location{
Bucket: "bucket_name",
Key: "/i/am/here",
}
cleanId := uuid.New()
dbCleanId := cleanId
mimeType := "application/pdf"
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
}
pool.ExpectQuery("name: GetCleanEntry :one").WithArgs(dbCleanId).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "mimetype", "fail"}).
AddRow(dbCleanId, id, &inloc.Bucket, &inloc.Key, int64(1), dbmimetype, repository.NullCleanfailtype{}),
)
textId := uuid.New()
pool.ExpectBegin()
pool.ExpectQuery("name: GetDocumentTextExtractionByHash :one").WithArgs(dbCleanId, "example").WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "cleanEntryId"}),
)
pool.ExpectQuery("name: AddDocumentText :one").WithArgs(inloc.Bucket, inloc.Key, dbCleanId, "example").
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(textId),
)
pool.ExpectExec("name: AddDocumentTextEntry :exec").WithArgs(textId, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
err = svc.extract(ctx, cleanId)
require.NoError(t, err)
})
t.Run("fail clean", func(t *testing.T) {
id := uuid.New()
cleanId := uuid.New()
fail := repository.NullCleanfailtype{
Valid: true,
Cleanfailtype: repository.CleanfailtypeInvalidMimetype,
}
pool.ExpectQuery("name: GetCleanEntry :one").WithArgs(id).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "mimetype", "fail"}).
AddRow(cleanId, id, nil, nil, int64(1), nil, fail),
)
err = svc.extract(ctx, id)
assert.Error(t, err)
})
}
func TestExecuteExtraction(t *testing.T) {
pool, err := pgxmock.NewPool()
require.NoError(t, err)
cfg := &DocTextConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
svc := Service{
cfg: cfg,
}
ctx := context.Background()
id := uuid.New()
location := document.Location{
Bucket: "buck",
Key: "key",
}
extract := extraction{
TextLocation: location,
Hash: "example",
}
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
}
pool.ExpectQuery("name: GetCleanEntry :one").WithArgs(id).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "mimetype", "fail"}).
AddRow(uuid.New(), id, &location.Bucket, &location.Key, int32(1), dbmimetype, repository.NullCleanfailtype{}),
)
outloc, err := svc.executeExtraction(ctx, id)
require.NoError(t, err)
assert.EqualExportedValues(t, extract, *outloc)
}
func TestStoreExtraction(t *testing.T) {
pool, err := pgxmock.NewPool()
require.NoError(t, err)
cfg := &DocTextConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
svc := Service{
cfg: cfg,
}
ctx := context.Background()
t.Run("not existing", func(t *testing.T) {
cleanId := uuid.New()
dbCleanId := cleanId
textId := uuid.New()
extract := extraction{
TextLocation: document.Location{
Bucket: "buck",
Key: "key",
},
Hash: "example",
}
pool.ExpectBegin()
pool.ExpectQuery("name: GetDocumentTextExtractionByHash :one").WithArgs(dbCleanId, extract.Hash).WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "cleanEntryId"}),
)
pool.ExpectQuery("name: AddDocumentText :one").WithArgs(extract.TextLocation.Bucket, extract.TextLocation.Key, dbCleanId, "example").
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(textId),
)
pool.ExpectExec("name: AddDocumentTextEntry :exec").WithArgs(textId, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
err = svc.storeExtraction(ctx, cleanId, &extract)
require.NoError(t, err)
})
}