Image dataset (#61)

* feat(dataset): Improve CSV file management in dataset dialog

Implements drag-and-drop reordering and deletion for CSV files
in the dataset creation/editing dialog.

Key changes:
- Uses dnd-kit for drag-and-drop functionality.
- Stores the ordered list of selected file names in the dataset's `data` field.
- Continues to send actual `File` objects for new/updated files to the API via the `files` field.
- Updates UI to display file names, sizes for new files, and allows reordering and deletion.
- Adds unit tests for new interactions and logic using Vitest.
- Ensures all build, lint, and type checks pass.

This addresses the issue of maintaining file order and allowing
more flexible file management for CSV datasets.

* add image type to dataset

* add image type

* feat: Add image type to dataset dialog

This commit introduces a new 'image' dataset type to the dataset creation and update dialog.

Key changes:
- Added "image" as an option in the dataset type selection.
- Implemented image file selection, allowing you to choose common image formats (PNG, JPG, GIF).
- Added functionality to generate and display small image thumbnails in the file list for selected images.
- Updated data structures, submission logic, and validation to support the new image type.
- Included unit tests to cover the new functionality, including image selection, thumbnail display (mocked), and submission.

The file input for images now shows a preview thumbnail for each selected image file in the list, enhancing your experience when working with image datasets.

* fix tests

* fix vision cli integration test

* update

* [Frontend] Add image dataset preview functionality

This commit introduces the ability to preview image datasets.
When a dataset with type 'image' is previewed, the images are
displayed as cards, with each card showing the image and its
filename below it. Multiple images are displayed per row in a
responsive grid layout.

Key changes:
- Modified `DatasetPreviewDialog` to handle `type: "image"`.
- Used `Card` components to display each image and its filename.
- Added unit tests for the new image preview functionality,
  covering successful rendering, empty lists, loading, and error states.

Note: Full verification of tests, tsc, and pnpm build was hindered
by persistent environmental issues.
The changes are submitted based on the successful implementation
of the code and test logic.

* update

* fix update image dataset

* fix lint

* style: Remove unnecessary comments from dataset tests

I removed explanatory code comments from `ui/src/components/dialog/dataset/dataset.test.tsx` to improve readability and maintainability.

The tests themselves remain unchanged in functionality. All tests (excluding the intentionally skipped image tests) continue to pass, and both `tsc` and `pnpm build` complete successfully.

* fix lint

* fix test

---------

Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>

Author: Yiling-J

api/dataset.go

@@ -3,7 +3,6 @@ package api
 import (
 	"errors"
 	"fmt"
-	"io"
 	"net/http"
 
 	"github.com/Yiling-J/tablepilot/ent"
@@ -12,8 +11,6 @@ import (
 	"github.com/gin-gonic/gin"
 )
 
-// Dataset-related handlers
-
 func (hs *HTTPServer) CreateDataset(ctx *gin.Context) {
 	var apiReq services_dataset.DatasetAPIRequest
 	if err := ctx.ShouldBind(&apiReq); err != nil {
@@ -28,19 +25,22 @@ func (hs *HTTPServer) CreateDataset(ctx *gin.Context) {
 		Data:        apiReq.Data,
 	}
 
-	if apiReq.Type == "csv" {
+	if apiReq.Type == "csv" || apiReq.Type == "image" {
 		if len(apiReq.Files) == 0 {
 			errorResponse(ctx, http.StatusBadRequest, errors.New("at least one file is required for CSV dataset type"))
 			return
 		}
-		var readers []io.Reader
+		var readers []services_dataset.CreateDatasetFile
 		for _, fh := range apiReq.Files {
 			f, err := fh.Open()
 			if err != nil {
 				errorResponse(ctx, http.StatusBadRequest, err)
 				return
 			}
-			readers = append(readers, f)
+			readers = append(readers, services_dataset.CreateDatasetFile{
+				Name:   fh.Filename,
+				Reader: f,
+			})
 		}
 		serviceReq.Files = readers
 	}
@@ -107,19 +107,22 @@ func (hs *HTTPServer) UpdateDataset(ctx *gin.Context) {
 	}
 
 	if apiReq.Files != nil {
-		var readers []io.Reader
+		var readers []services_dataset.CreateDatasetFile
 		for _, fh := range apiReq.Files {
 			f, err := fh.Open()
 			if err != nil {
 				errorResponse(ctx, http.StatusBadRequest, err)
 				return
 			}
-			readers = append(readers, f)
+			readers = append(readers, services_dataset.CreateDatasetFile{
+				Name:   fh.Filename,
+				Reader: f,
+			})
 		}
 		serviceReq.Files = readers
 		serviceReq.Fields = append(serviceReq.Fields, "files")
 	} else {
-		serviceReq.Files = []io.Reader{}
+		serviceReq.Files = []services_dataset.CreateDatasetFile{}
 	}
 
 	err := hs.DatasetService.Update(ctx.Request.Context(), datasetID, serviceReq)

api/dataset_test.go

@@ -72,7 +72,7 @@ func TestAPI_CreateDatasetWithFiles(t *testing.T) {
 			require.Equal(t, expectedRequest.Description, req.Description)
 			require.Equal(t, expectedRequest.Type, req.Type)
 			require.Equal(t, 1, len(req.Files))
-			data, err := io.ReadAll(req.Files[0])
+			data, err := io.ReadAll(req.Files[0].Reader)
 			require.NoError(t, err)
 			require.Equal(t, "header1,header2,header3\nr1c1,r1c2,r1c3\n", string(data))
 			return "new_dataset_id", nil

cmd/cli/handler.go

@@ -181,16 +181,17 @@ func (h *Handler) CreateDataset(cmd *cobra.Command, args []string) error {
 			options = append(options, o...)
 		}
 		req.Data = options
-	case "csv":
+	case "csv", "image":
 		filePaths, err := cmd.Flags().GetStringArray("path")
 		if err != nil {
 			return fmt.Errorf("error getting file flag for csv type: %w", err)
 		}
 		if len(filePaths) == 0 {
-			return fmt.Errorf("at least one --path must be provided for type 'csv'")
+			return fmt.Errorf("at least one --path must be provided")
 		}
-		var readers []io.Reader
+		var readers []dataset.CreateDatasetFile
 		files, err := parsePaths(filePaths)
+		names := []string{}
 		if err != nil {
 			return err
 		}
@@ -199,12 +200,17 @@ func (h *Handler) CreateDataset(cmd *cobra.Command, args []string) error {
 			if err != nil {
 				return fmt.Errorf("failed to open file %s: %w", f, err)
 			}
-			readers = append(readers, file)
+			readers = append(readers, dataset.CreateDatasetFile{
+				Name:   filepath.Base(f),
+				Reader: file,
+			})
+			names = append(names, filepath.Base(f))
 		}
 		req.Files = readers
+		req.Data = names
 		defer func() {
 			for _, f := range readers {
-				if c, ok := f.(io.Closer); ok {
+				if c, ok := f.Reader.(io.Closer); ok {
 					c.Close()
 				}
 			}
@@ -307,32 +313,29 @@ func (h *Handler) UpdateDataset(cmd *cobra.Command, args []string) error {
 			options = append(options, o...)
 		}
 		req.Data = options
-	case "csv":
+	case "csv", "image":
 		filePaths, err := cmd.Flags().GetStringArray("file")
 		if err != nil {
 			return fmt.Errorf("error getting file flag for csv type: %w", err)
 		}
 		if len(filePaths) == 0 {
 			return fmt.Errorf("at least one --file path must be provided for type 'csv'")
 		}
-		var files []io.Reader
+		var files []dataset.CreateDatasetFile
 		for _, filePath := range filePaths {
 			file, err := os.Open(filePath)
 			if err != nil {
-				// Close already opened files if any
-				for _, f := range files {
-					if c, ok := f.(io.Closer); ok {
-						c.Close()
-					}
-				}
 				return fmt.Errorf("failed to open file %s: %w", filePath, err)
 			}
-			files = append(files, file)
+			files = append(files, dataset.CreateDatasetFile{
+				Name:   filepath.Base(filePath),
+				Reader: file,
+			})
 		}
 		req.Files = files
 		defer func() {
 			for _, f := range files {
-				if c, ok := f.(io.Closer); ok {
+				if c, ok := f.Reader.(io.Closer); ok {
 					c.Close()
 				}
 			}

ent/dataset/dataset.go

@@ -36,7 +36,7 @@ func (Dataset) Fields() []ent.Field {
 		field.String("name").Unique().NotEmpty(),
 		field.String("path").Optional(),
 		field.String("description").Default(""),
-		field.Enum("type").Values("list", "csv"),
+		field.Enum("type").Values("list", "csv", "image"),
 		field.JSON("indexer", CSVIndexer{}).Optional(),
 		field.Strings("values").Optional(),
 	}

ent/migrate/schema.go

@@ -7,13 +7,13 @@ import (
 	"io"
 	"os"
 	"path/filepath"
+	"slices"
 
 	"github.com/Yiling-J/tablepilot/config"
 	"github.com/Yiling-J/tablepilot/ent"
 	db_dataset "github.com/Yiling-J/tablepilot/ent/dataset"
 	"github.com/Yiling-J/tablepilot/services/source"
 	"github.com/Yiling-J/tablepilot/services/source/csvindexer"
-	"github.com/Yiling-J/tablepilot/utils"
 )
 
 //go:generate moq -rm -out dataset_moq.go . DatasetService
@@ -39,56 +39,43 @@ func NewDatasetService(db *ent.Client, cfg *config.Config) *DatasetServiceImpl {
 }
 
 func (s DatasetServiceImpl) buildCreateDatasetReq(ctx context.Context, req *CreateDatasetRequest, sr *ent.Dataset) error {
-	switch req.Type {
-	case db_dataset.TypeCsv:
-		relativePath := filepath.Join("datasets/shared", sr.Nanoid)
-		dirPath := filepath.Join(s.cfg.Common.DataDir, relativePath)
-		err := os.MkdirAll(dirPath, os.ModePerm)
-		if err != nil {
-			return fmt.Errorf("failed to create directory: %w", err)
-		}
+	relativePath := filepath.Join("datasets/shared", sr.Nanoid)
+	dirPath := filepath.Join(s.cfg.Common.DataDir, relativePath)
+	err := os.MkdirAll(dirPath, os.ModePerm)
+	if err != nil {
+		return fmt.Errorf("failed to create directory: %w", err)
+	}
 
-		if len(req.Files) == 0 {
-			return errors.New("dataset.Create: files should not be empty")
-		}
-		filePath := filepath.Join(dirPath, "data.csv")
-		outFile, err := os.Create(filePath)
+	for _, file := range req.Files {
+		outFile, err := os.Create(filepath.Join(dirPath, file.Name))
 		if err != nil {
-			return fmt.Errorf("failed to create file %s: %w", filePath, err)
+			return fmt.Errorf("failed to create file %w", err)
 		}
-		for i, file := range req.Files {
-			// skip csv headers
-			if i > 0 {
-				reader := utils.NewCsvReader(file)
-				_, err = reader.Read()
-				if err != nil {
-					return fmt.Errorf("failed to read csv %w", err)
-				}
-				offset := reader.InputOffset()
-				_, err = file.(io.ReadSeeker).Seek(offset, io.SeekStart)
-				if err != nil {
-					return fmt.Errorf("failed to seek csv file %w", err)
-				}
-			}
-			_, err = io.Copy(outFile, file)
-			if err != nil {
-				return fmt.Errorf("failed to write to file %s: %w", filePath, err)
-			}
+		defer outFile.Close()
+		_, err = io.Copy(outFile, file.Reader)
+		if err != nil {
+			return fmt.Errorf("failed to write to file %w", err)
 		}
-		outFile.Close()
-		// build index
-		indexer, err := csvindexer.NewCSVIndexer(os.DirFS(dirPath), []string{"data.csv"})
+	}
+	switch req.Type {
+	case db_dataset.TypeCsv:
+		indexer, err := csvindexer.NewCSVIndexer(os.DirFS(dirPath), req.Data)
 		if err != nil {
 			return fmt.Errorf("table.Create: build csv index: %w", err)
 		}
-		err = sr.Update().SetPath(relativePath).SetIndexer(indexer.CSVIndexer).Exec(ctx)
+		err = sr.Update().SetPath(relativePath).SetIndexer(indexer.CSVIndexer).SetValues(req.Data).Exec(ctx)
 		if err != nil {
-			return fmt.Errorf("table.Create: update dataset metadata: %w", err) // Clarified error
+			return fmt.Errorf("table.Create: update dataset metadata: %w", err)
+		}
+	case db_dataset.TypeImage:
+		err = sr.Update().SetPath(relativePath).SetValues(req.Data).Exec(ctx)
+		if err != nil {
+			return fmt.Errorf("table.Create: update dataset metadata: %w", err)
 		}
 	case db_dataset.TypeList:
 		err := sr.Update().SetValues(req.Data).Exec(ctx)
 		if err != nil {
-			return fmt.Errorf("table.Create: update dataset values: %w", err) // Clarified error
+			return fmt.Errorf("table.Create: update dataset values: %w", err)
 		}
 	}
 	return nil
@@ -119,6 +106,10 @@ func (s *DatasetServiceImpl) List(ctx context.Context) ([]*DatasetInfo, error) {
 	}
 	datasetInfos := []*DatasetInfo{}
 	for _, ds := range datasets {
+		// backward compatible
+		if ds.Type == db_dataset.TypeCsv && len(ds.Values) == 0 {
+			ds.Values = []string{"data.csv"}
+		}
 		datasetInfos = append(datasetInfos, &DatasetInfo{
 			ID:          ds.Nanoid,
 			Name:        ds.Name,
@@ -161,22 +152,38 @@ func (s *DatasetServiceImpl) Update(ctx context.Context, dataset string, req *Up
 		case "description":
 			updater.SetDescription(req.Description)
 		case "data", "files":
-			processDataRebuild = true
-			updater.ClearIndexer().ClearPath().SetValues(nil)
+			// new files or data slice change
+			if len(req.Files) > 0 || !slices.Equal(req.Data, ds.Values) {
+				processDataRebuild = true
+				updater.ClearIndexer().ClearPath().SetValues(nil)
+			}
 		}
 	}
 
 	updatedDsEntity, err := updater.Save(ctx)
 	if err != nil {
-		return ent.Rollback(tx, fmt.Errorf("dataset.Update: save changes: %w", err)) // Clarified error
+		return ent.Rollback(tx, fmt.Errorf("dataset.Update: save changes: %w", err))
 	}
 
 	if processDataRebuild {
 		if originalPath != "" {
 			oldDirPath := filepath.Join(s.cfg.Common.DataDir, originalPath)
 			if _, statErr := os.Stat(oldDirPath); !os.IsNotExist(statErr) {
-				if removeErr := os.RemoveAll(oldDirPath); removeErr != nil {
-					return ent.Rollback(tx, fmt.Errorf("dataset.Update: failed to remove old directory %s: %w", oldDirPath, removeErr))
+				keep := map[string]bool{}
+				for _, file := range req.Data {
+					keep[file] = true
+				}
+				entries, err := os.ReadDir(oldDirPath)
+				if err != nil {
+					return ent.Rollback(tx, fmt.Errorf("dataset.Update: read dir: %w", err))
+				}
+				for _, e := range entries {
+					if _, ok := keep[e.Name()]; !ok {
+						err = os.Remove(filepath.Join(oldDirPath, e.Name()))
+						if err != nil {
+							return ent.Rollback(tx, fmt.Errorf("dataset.Update: remove file: %w", err))
+						}
+					}
 				}
 			}
 		}
@@ -217,7 +224,7 @@ func (s *DatasetServiceImpl) Delete(ctx context.Context, dataset string) error {
 
 	err = tx.Dataset.DeleteOne(ds).Exec(ctx)
 	if err != nil {
-		return ent.Rollback(tx, fmt.Errorf("dataset.Delete: execute delete: %w", err)) // Clarified error
+		return ent.Rollback(tx, fmt.Errorf("dataset.Delete: execute delete: %w", err))
 	}
 
 	return tx.Commit()
@@ -229,7 +236,7 @@ func (s *DatasetServiceImpl) Get(ctx context.Context, source string) (*DatasetIn
 		db_dataset.Nanoid(source),
 	)).Only(ctx)
 	if err != nil {
-		return nil, fmt.Errorf("dataset.Get: query dataset: %w", err) // Clarified error
+		return nil, fmt.Errorf("dataset.Get: query dataset: %w", err)
 	}
 	return &DatasetInfo{
 		Name:        sr.Name,
@@ -276,6 +283,11 @@ func (s *DatasetServiceImpl) Preview(ctx context.Context, dataset string) (*Data
 			Type: sr.Type,
 			Rows: rows,
 		}, nil
+	case db_dataset.TypeImage:
+		return &DatasetRows{
+			Type: sr.Type,
+			Data: sr.Values,
+		}, nil
 	case db_dataset.TypeList:
 		return &DatasetRows{
 			Type: sr.Type,