Merged in feature/preserve-folder-paths (pull request #179)

support folder paths

* support folder paths

* s3 storage docs

internal/document/init/create.go

@@ -15,9 +15,10 @@ import (
 )
 
 type Create struct {
-	Bucket string
-	Key    objectstore.BucketKey
-	Hash   string
+	Bucket   string
+	Key      objectstore.BucketKey
+	Hash     string
+	Filename *string // Optional filename for batch processing
 }
 
 func (s *Service) Create(ctx context.Context, doc *Create) (uuid.UUID, error) {
@@ -47,10 +48,11 @@ func (s *Service) Create(ctx context.Context, doc *Create) (uuid.UUID, error) {
 }
 
 type createDocumentParams struct {
-	ID     *uuid.UUID
-	Hash   string
-	Bucket string
-	Key    objectstore.BucketKey
+	ID       *uuid.UUID
+	Hash     string
+	Bucket   string
+	Key      objectstore.BucketKey
+	Filename *string
 }
 
 func (s *Service) getCreateParams(ctx context.Context, doc *Create) (*createDocumentParams, error) {
@@ -66,10 +68,11 @@ func (s *Service) getCreateParams(ctx context.Context, doc *Create) (*createDocu
 	}
 
 	return &createDocumentParams{
-		ID:     docID,
-		Hash:   doc.Hash,
-		Key:    doc.Key,
-		Bucket: doc.Bucket,
+		ID:       docID,
+		Hash:     doc.Hash,
+		Key:      doc.Key,
+		Bucket:   doc.Bucket,
+		Filename: doc.Filename,
 	}, nil
 }
 
@@ -83,6 +86,7 @@ func (s *Service) submitCreate(ctx context.Context, params *createDocumentParams
 				Clientid: params.Key.ClientID,
 				Hash:     params.Hash,
 				BatchID:  params.Key.BatchID,
+				Filename: params.Filename,
 			})
 			if err != nil {
 				return err

internal/document/init/create_test.go

@@ -52,7 +52,7 @@ func TestCreate(t *testing.T) {
 		pgxmock.NewRows([]string{"id"}),
 	)
 	pool.ExpectBegin()
-	pool.ExpectQuery("name: CreateDocument :one").WithArgs(doc.ClientID, doc.Hash, (*uuid.UUID)(nil)).
+	pool.ExpectQuery("name: CreateDocument :one").WithArgs(doc.ClientID, doc.Hash, (*uuid.UUID)(nil), (*string)(nil)).
 		WillReturnRows(
 			pgxmock.NewRows([]string{"id"}).
 				AddRow(doc.ID),
@@ -80,14 +80,14 @@ func TestCreate(t *testing.T) {
 			mock.MatchedBy(func(in *sqs.SendMessageInput) bool {
 				return *in.QueueUrl == cfg.DocumentSyncURL && *in.MessageBody == fmt.Sprintf("{\"id\":\"%s\"}", doc.ID.String())
 			}),
-			mock.Anything,
 		).
 		Return(&sqs.SendMessageOutput{}, nil)
 
 	id, err := svc.Create(ctx, &Create{
-		Hash:   doc.Hash,
-		Key:    key,
-		Bucket: bucket,
+		Hash:     doc.Hash,
+		Key:      key,
+		Bucket:   bucket,
+		Filename: nil, // No filename for this test
 	})
 	require.NoError(t, err)
 	assert.Equal(t, doc.ID, id)
@@ -235,7 +235,7 @@ func TestSubmitCreate(t *testing.T) {
 	}
 
 	pool.ExpectBegin()
-	pool.ExpectQuery("name: CreateDocument :one").WithArgs(doc.ClientID, doc.Hash, (*uuid.UUID)(nil)).
+	pool.ExpectQuery("name: CreateDocument :one").WithArgs(doc.ClientID, doc.Hash, (*uuid.UUID)(nil), (*string)(nil)).
 		WillReturnRows(
 			pgxmock.NewRows([]string{"id"}).
 				AddRow(doc.ID),
@@ -245,9 +245,10 @@ func TestSubmitCreate(t *testing.T) {
 	pool.ExpectCommit()
 
 	id, err := svc.submitCreate(ctx, &createDocumentParams{
-		Hash:   doc.Hash,
-		Bucket: bucket,
-		Key:    key,
+		Hash:     doc.Hash,
+		Bucket:   bucket,
+		Key:      key,
+		Filename: nil, // No filename for this test
 	})
 	require.NoError(t, err)
 	assert.Equal(t, doc.ID, id)

internal/document/upload/get.go

@@ -34,7 +34,14 @@ func detectContentType(filename string) string {
 }
 
 func (s *Service) Upload(ctx context.Context, file File) error {
-	return s.cfg.ExecuteDBTransaction(ctx, func(ctx context.Context, q *repository.Queries) error {
+	_, err := s.UploadAndReturnKey(ctx, file)
+	return err
+}
+
+// UploadAndReturnKey uploads a file and returns the BucketKey used
+func (s *Service) UploadAndReturnKey(ctx context.Context, file File) (objectstore.BucketKey, error) {
+	var key objectstore.BucketKey
+	err := s.cfg.ExecuteDBTransaction(ctx, func(ctx context.Context, q *repository.Queries) error {
 		now := time.Now().UTC()
 		part, err := s.cfg.GetDBQueries().GetDocumentUploadCurrentPart(ctx, &repository.GetDocumentUploadCurrentPartParams{
 			Clientid: &file.ClientID,
@@ -49,7 +56,7 @@ func (s *Service) Upload(ctx context.Context, file File) error {
 
 		uploadId := uuid.New()
 		newPart := s.cfg.GetDirectoryPart(part.Part, part.Count)
-		key := objectstore.BucketKey{
+		key = objectstore.BucketKey{
 			ClientID:  file.ClientID,
 			Location:  objectstore.Import,
 			CreatedAt: now,
@@ -70,6 +77,8 @@ func (s *Service) Upload(ctx context.Context, file File) error {
 				Valid: true,
 				Time:  now,
 			},
+			Filename: &file.Filename,
+			BatchID:  file.BatchID,
 		})
 		if err != nil {
 			return err
@@ -78,11 +87,18 @@ func (s *Service) Upload(ctx context.Context, file File) error {
 		// Detect content type from original filename
 		contentType := detectContentType(file.Filename)
 
+		// Prepare metadata for S3 object
+		metadata := make(map[string]string)
+		if file.Filename != "" {
+			metadata["original-filename"] = file.Filename
+		}
+
 		_, err = s.cfg.GetStoreClient().PutObject(ctx, &s3.PutObjectInput{
 			Bucket:      &bucket,
 			Key:         &keyStr,
 			Body:        file.Content,
 			ContentType: &contentType,
+			Metadata:    metadata,
 		})
 		if err != nil {
 			return err
@@ -90,4 +106,6 @@ func (s *Service) Upload(ctx context.Context, file File) error {
 
 		return nil
 	})
+
+	return key, err
 }