Files
query-orchestration/internal/document/clean/clean_test.go
T

447 lines
11 KiB
Go
Raw Normal View History

package documentclean
import (
"context"
2025-02-28 13:11:53 +00:00
"io"
2025-03-05 12:05:46 +00:00
"strings"
"testing"
"time"
2025-03-05 12:05:46 +00:00
"queryorchestration/internal/database/repository"
"queryorchestration/internal/document"
documenttypes "queryorchestration/internal/document/types"
"queryorchestration/internal/serviceconfig/objectstore"
2025-02-28 13:11:53 +00:00
objectstoremock "queryorchestration/mocks/objectstore"
"github.com/stretchr/testify/require"
2025-02-28 13:11:53 +00:00
"github.com/aws/aws-sdk-go-v2/service/s3"
"github.com/google/uuid"
"github.com/pashagolub/pgxmock/v3"
"github.com/stretchr/testify/assert"
2025-02-28 13:11:53 +00:00
"github.com/stretchr/testify/mock"
)
func TestClean(t *testing.T) {
ctx := context.Background()
pool, err := pgxmock.NewPool()
require.NoError(t, err)
cfg := &DocCleanConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
2025-02-28 13:11:53 +00:00
mockS3 := objectstoremock.NewMockS3Client(t)
cfg.StoreClient = mockS3
svc := Service{
cfg: cfg,
}
doc := document.DocumentSummary{
ID: uuid.New(),
ClientID: "yuyu",
Hash: "example_hash",
2025-02-28 13:11:53 +00:00
}
bucket := "example_bucket"
key := objectstore.BucketKey{
ClientID: doc.ClientID,
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
}
keyStr := key.String()
dbid := doc.ID
pool.ExpectQuery("name: GetDocumentSummary :one").WithArgs(dbid).
2025-02-28 13:11:53 +00:00
WillReturnRows(
pgxmock.NewRows([]string{"id", "clientId", "hash"}).
AddRow(dbid, doc.ClientID, doc.Hash),
2025-02-28 13:11:53 +00:00
)
pool.ExpectQuery("name: GetDocumentEntry :one").WithArgs(dbid).WillReturnRows(
pgxmock.NewRows([]string{"documentId", "bucket", "key"}).
AddRow(dbid, bucket, keyStr),
)
2025-02-28 13:11:53 +00:00
mimeType := "application/pdf"
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
2025-02-28 13:11:53 +00:00
}
pool.ExpectBegin()
pool.ExpectQuery("name: GetMostRecentDocumentCleanEntry :one").WithArgs(dbid).
WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail"}),
)
cleanid := uuid.New()
pool.ExpectQuery("name: AddDocumentClean :one").WithArgs(dbid, &bucket, &keyStr, &doc.Hash, dbmimetype, repository.NullCleanfailtype{}).
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(cleanid),
)
pool.ExpectExec("name: AddDocumentCleanEntry :exec").WithArgs(cleanid, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
2025-02-28 13:11:53 +00:00
var length int64 = 10
mockS3.EXPECT().
HeadObject(
mock.Anything,
mock.MatchedBy(func(in *s3.HeadObjectInput) bool {
return *in.Bucket == bucket && *in.Key == keyStr
2025-02-28 13:11:53 +00:00
}),
mock.Anything,
).
Return(&s3.HeadObjectOutput{
ContentType: &mimeType,
ContentLength: &length,
}, nil)
body := io.NopCloser(strings.NewReader(pdfHelloWorld))
mockS3.EXPECT().
GetObject(
mock.Anything,
mock.MatchedBy(func(in *s3.GetObjectInput) bool {
return *in.Bucket == bucket && *in.Key == keyStr
2025-02-28 13:11:53 +00:00
}),
mock.Anything,
).
Return(&s3.GetObjectOutput{
Body: body,
}, nil)
err = svc.clean(ctx, doc.ID)
2025-03-19 11:54:14 +00:00
require.NoError(t, err)
}
func TestExecuteCleanTasks(t *testing.T) {
2025-02-28 13:11:53 +00:00
ctx := context.Background()
cfg := &DocCleanConfig{}
mockS3 := objectstoremock.NewMockS3Client(t)
cfg.StoreClient = mockS3
svc := Service{
cfg: cfg,
}
id := uuid.New()
2025-02-28 13:11:53 +00:00
hash := "example_hash"
bucket := "example_bucket"
key := objectstore.BucketKey{
ClientID: "client",
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
2025-02-28 13:11:53 +00:00
}
keyStr := key.String()
2025-02-28 13:11:53 +00:00
mimeType := "application/pdf"
mockS3.EXPECT().
HeadObject(
mock.Anything,
mock.MatchedBy(func(in *s3.HeadObjectInput) bool {
return *in.Bucket == bucket && *in.Key == keyStr
2025-02-28 13:11:53 +00:00
}),
mock.Anything,
).
Return(&s3.HeadObjectOutput{
ContentType: &mimeType,
}, nil)
2025-02-28 13:11:53 +00:00
body := io.NopCloser(strings.NewReader(pdfHelloWorld))
mockS3.EXPECT().
GetObject(
mock.Anything,
mock.MatchedBy(func(in *s3.GetObjectInput) bool {
return *in.Bucket == bucket && *in.Key == keyStr
2025-02-28 13:11:53 +00:00
}),
mock.Anything,
).
Return(&s3.GetObjectOutput{
Body: body,
}, nil)
outloc, err := svc.executeCleanTasks(ctx, &CleanParams{
ID: id,
Hash: hash,
Bucket: bucket,
Key: key,
})
2025-03-19 11:54:14 +00:00
require.NoError(t, err)
2025-02-28 13:11:53 +00:00
assert.EqualExportedValues(t, ExecuteCleanResponse{
hash: &hash,
2025-02-28 13:11:53 +00:00
}, *outloc)
}
2025-02-28 13:11:53 +00:00
func TestStoreClean(t *testing.T) {
ctx := context.Background()
pool, err := pgxmock.NewPool()
require.NoError(t, err)
cfg := &DocCleanConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
svc := Service{
cfg: cfg,
}
doc := document.DocumentSummary{
ID: uuid.New(),
}
bucket := "example_bucket"
key := objectstore.BucketKey{
ClientID: doc.ClientID,
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
}
keyStr := key.String()
t.Run("no prev entry", func(t *testing.T) {
mimeType := "application/pdf"
hash := "heyman"
params := &ExecuteCleanResponse{
hash: &hash,
mimetype: (*documenttypes.MimeType)(&mimeType),
key: &key,
bucket: &bucket,
}
dbid := doc.ID
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
}
pool.ExpectBegin()
pool.ExpectQuery("name: GetMostRecentDocumentCleanEntry :one").WithArgs(dbid).
WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail"}),
)
cleanid := uuid.New()
pool.ExpectQuery("name: AddDocumentClean :one").WithArgs(dbid, &bucket, &keyStr, params.hash, dbmimetype, repository.NullCleanfailtype{}).
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(cleanid),
)
pool.ExpectExec("name: AddDocumentCleanEntry :exec").WithArgs(cleanid, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
err = svc.storeClean(ctx, doc.ID, params)
2025-03-19 11:54:14 +00:00
require.NoError(t, err)
})
t.Run("diff prev entry", func(t *testing.T) {
mimeType := "application/pdf"
hash := "heyman"
params := &ExecuteCleanResponse{
hash: &hash,
bucket: &bucket,
key: &key,
mimetype: (*documenttypes.MimeType)(&mimeType),
}
dbid := doc.ID
dbmimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
}
pool.ExpectBegin()
pool.ExpectQuery("name: GetMostRecentDocumentCleanEntry :one").WithArgs(dbid).
WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail"}).
AddRow(uuid.New(), dbid, nil, nil, int64(1), nil, nil, repository.CleanfailtypeInvalidRead),
)
cleanid := uuid.New()
pool.ExpectQuery("name: AddDocumentClean :one").WithArgs(dbid, &bucket, &keyStr, params.hash, dbmimetype, repository.NullCleanfailtype{}).
WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(cleanid),
)
pool.ExpectExec("name: AddDocumentCleanEntry :exec").WithArgs(cleanid, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
err = svc.storeClean(ctx, doc.ID, params)
2025-03-19 11:54:14 +00:00
require.NoError(t, err)
})
t.Run("same fail", func(t *testing.T) {
reason := documenttypes.InvalidDocumentMimeType
params := &ExecuteCleanResponse{
failReason: &reason,
}
dbid := doc.ID
cleanid := uuid.New()
pool.ExpectBegin()
pool.ExpectQuery("name: GetMostRecentDocumentCleanEntry :one").WithArgs(dbid).
WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail"}).
AddRow(cleanid, dbid, nil, nil, int64(1), nil, nil, repository.NullCleanfailtype{
Valid: true,
Cleanfailtype: repository.CleanfailtypeInvalidMimetype,
}),
)
pool.ExpectExec("name: AddDocumentCleanEntry :exec").WithArgs(cleanid, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
err = svc.storeClean(ctx, doc.ID, params)
assert.EqualError(t, err, "invalid_mimetype")
})
}
func TestIsNewClean(t *testing.T) {
svc := Service{}
t.Run("no last entry", func(t *testing.T) {
var lastEntry *repository.GetMostRecentDocumentCleanEntryRow = nil
params := &ExecuteCleanResponse{}
assert.True(t, svc.isNewClean(lastEntry, params))
})
t.Run("same fail", func(t *testing.T) {
var lastEntry *repository.GetMostRecentDocumentCleanEntryRow = &repository.GetMostRecentDocumentCleanEntryRow{
ID: uuid.New(),
Fail: repository.NullCleanfailtype{
Valid: true,
Cleanfailtype: repository.CleanfailtypeInvalidMimetype,
},
}
reason := documenttypes.InvalidDocumentMimeType
params := &ExecuteCleanResponse{
failReason: &reason,
}
assert.False(t, svc.isNewClean(lastEntry, params))
})
t.Run("different fail", func(t *testing.T) {
var lastEntry *repository.GetMostRecentDocumentCleanEntryRow = &repository.GetMostRecentDocumentCleanEntryRow{
ID: uuid.New(),
Fail: repository.NullCleanfailtype{
Valid: true,
Cleanfailtype: repository.CleanfailtypeInvalidMimetype,
},
}
reason := documenttypes.InvalidDocumentLargeDPI
params := &ExecuteCleanResponse{
failReason: &reason,
}
assert.True(t, svc.isNewClean(lastEntry, params))
})
t.Run("same location", func(t *testing.T) {
bucket := "example_bucket"
key := objectstore.BucketKey{
ClientID: "diw",
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
}
keyStr := key.String()
mimeType := documenttypes.MimeTypePDF
var lastEntry *repository.GetMostRecentDocumentCleanEntryRow = &repository.GetMostRecentDocumentCleanEntryRow{
ID: uuid.New(),
Bucket: &bucket,
Key: &keyStr,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
},
}
params := &ExecuteCleanResponse{
bucket: &bucket,
key: &key,
mimetype: &mimeType,
}
assert.False(t, svc.isNewClean(lastEntry, params))
})
t.Run("different location", func(t *testing.T) {
bucket := "example_bucket"
oldKey := objectstore.BucketKey{
ClientID: "hi",
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
}
oldKeyStr := oldKey.String()
newKey := objectstore.BucketKey{
ClientID: "hi",
CreatedAt: time.Now().UTC(),
EntityID: uuid.New(),
Location: objectstore.Import,
}
mimeType := documenttypes.MimeTypePDF
var lastEntry *repository.GetMostRecentDocumentCleanEntryRow = &repository.GetMostRecentDocumentCleanEntryRow{
ID: uuid.New(),
Bucket: &bucket,
Key: &oldKeyStr,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.Cleanmimetype(mimeType),
},
}
params := &ExecuteCleanResponse{
bucket: &bucket,
key: &newKey,
mimetype: &mimeType,
}
assert.True(t, svc.isNewClean(lastEntry, params))
})
}
2025-02-28 13:11:53 +00:00
const pdfHelloWorld = `%PDF-1.4
%
1 0 obj
<<
/Type /Catalog
/Pages 2 0 R
/Version /1.4
>>
endobj
2 0 obj
<<
/Type /Pages
/Kids [3 0 R]
/Count 1
>>
endobj
3 0 obj
<<
/Type /Page
/Parent 2 0 R
/MediaBox [0 0 612 792]
/Resources <<
/Font <<
/F1 <<
/Type /Font
/Subtype /Type1
/BaseFont /Helvetica
>>
>>
>>
/Contents 4 0 R
>>
endobj
4 0 obj
<<
/Length
44
>>
stream
BT
/F1 24 Tf
100 700 Td
(Hello World!) Tj
ET
endstream
endobj
xref
0 5
0000000000 65535 f
0000000015 00000 n
0000000086 00000 n
0000000151 00000 n
0000000376 00000 n
trailer
<<
/Size 5
/Root 1 0 R
>>
startxref
472
%%EOF`