enhance: (knowledge) only preload files on datasets if asked for (#336)

obot-platform · Jan 13, 2025 · 62a74c4 · 62a74c4
1 parent fab4dda
commit 62a74c4
Show file tree

Hide file tree

Showing 20 changed files with 55 additions and 25 deletions.
diff --git a/knowledge/pkg/client/client.go b/knowledge/pkg/client/client.go
@@ -35,7 +35,7 @@ type IngestPathsOpts struct {
 type Client interface {
 	CreateDataset(ctx context.Context, datasetID string, opts *types2.DatasetCreateOpts) (*types2.Dataset, error)
 	DeleteDataset(ctx context.Context, datasetID string) error
-	GetDataset(ctx context.Context, datasetID string) (*types2.Dataset, error)
+	GetDataset(ctx context.Context, datasetID string, opts *types2.DatasetGetOpts) (*types2.Dataset, error)
 	FindFile(ctx context.Context, searchFile types2.File) (*types2.File, error)
 	DeleteFile(ctx context.Context, datasetID, fileID string) error
 	ListDatasets(ctx context.Context) ([]types2.Dataset, error)

diff --git a/knowledge/pkg/client/common.go b/knowledge/pkg/client/common.go
@@ -277,7 +277,7 @@ func AskDir(ctx context.Context, c Client, path string, query string, opts *Inge
 func getOrCreateDataset(ctx context.Context, c Client, datasetID string, create bool) (*types.Dataset, error) {
 	var ds *types.Dataset
 	var err error
-	ds, err = c.GetDataset(ctx, datasetID)
+	ds, err = c.GetDataset(ctx, datasetID, nil)
 	if err != nil {
 		return nil, err
 	}

diff --git a/knowledge/pkg/client/standalone.go b/knowledge/pkg/client/standalone.go
@@ -53,8 +53,8 @@ func (c *StandaloneClient) DeleteDataset(ctx context.Context, datasetID string)
 	return c.Datastore.DeleteDataset(ctx, datasetID)
 }
 
-func (c *StandaloneClient) GetDataset(ctx context.Context, datasetID string) (*types2.Dataset, error) {
-	return c.Datastore.GetDataset(ctx, datasetID)
+func (c *StandaloneClient) GetDataset(ctx context.Context, datasetID string, opts *types2.DatasetGetOpts) (*types2.Dataset, error) {
+	return c.Datastore.GetDataset(ctx, datasetID, opts)
 }
 
 func (c *StandaloneClient) ListDatasets(ctx context.Context) ([]types2.Dataset, error) {

diff --git a/knowledge/pkg/cmd/edit_dataset.go b/knowledge/pkg/cmd/edit_dataset.go
@@ -33,7 +33,7 @@ func (s *ClientEditDataset) Run(cmd *cobra.Command, args []string) error {
 	datasetID := args[0]
 
 	// Get current dataset
-	dataset, err := c.GetDataset(cmd.Context(), datasetID)
+	dataset, err := c.GetDataset(cmd.Context(), datasetID, nil)
 	if err != nil {
 		return fmt.Errorf("failed to get dataset: %w", err)
 	}

diff --git a/knowledge/pkg/cmd/export.go b/knowledge/pkg/cmd/export.go
@@ -42,7 +42,7 @@ func (s *ClientExportDatasets) Run(cmd *cobra.Command, args []string) error {
 		}
 	} else {
 		for _, datasetID := range dsnames {
-			ds, err := c.GetDataset(cmd.Context(), datasetID)
+			ds, err := c.GetDataset(cmd.Context(), datasetID, nil)
 			if err != nil {
 				return err
 			}

diff --git a/knowledge/pkg/cmd/get_dataset.go b/knowledge/pkg/cmd/get_dataset.go
@@ -4,6 +4,7 @@ import (
 	"encoding/json"
 	"fmt"
 
+	"github.com/gptscript-ai/knowledge/pkg/index/types"
 	"github.com/spf13/cobra"
 )
 
@@ -28,7 +29,7 @@ func (s *ClientGetDataset) Run(cmd *cobra.Command, args []string) error {
 
 	datasetID := args[0]
 
-	ds, err := c.GetDataset(cmd.Context(), datasetID)
+	ds, err := c.GetDataset(cmd.Context(), datasetID, &types.DatasetGetOpts{IncludeFiles: true})
 	if err != nil {
 		return fmt.Errorf("failed to get dataset: %w", err)
 	}

diff --git a/knowledge/pkg/datastore/dataset.go b/knowledge/pkg/datastore/dataset.go
@@ -41,8 +41,8 @@ func (s *Datastore) DeleteDataset(ctx context.Context, datasetID string) error {
 	return nil
 }
 
-func (s *Datastore) GetDataset(ctx context.Context, datasetID string) (*types.Dataset, error) {
-	return s.Index.GetDataset(ctx, datasetID)
+func (s *Datastore) GetDataset(ctx context.Context, datasetID string, opts *types.DatasetGetOpts) (*types.Dataset, error) {
+	return s.Index.GetDataset(ctx, datasetID, opts)
 }
 
 func (s *Datastore) ListDatasets(ctx context.Context) ([]types.Dataset, error) {
@@ -61,7 +61,7 @@ func (s *Datastore) UpdateDataset(ctx context.Context, updatedDataset types.Data
 		return origDS, fmt.Errorf("dataset ID is required")
 	}
 
-	origDS, err = s.GetDataset(ctx, updatedDataset.ID)
+	origDS, err = s.GetDataset(ctx, updatedDataset.ID, nil)
 	if err != nil {
 		return origDS, err
 	}

diff --git a/knowledge/pkg/datastore/ingest.go b/knowledge/pkg/datastore/ingest.go
@@ -39,7 +39,7 @@ func (s *Datastore) Ingest(ctx context.Context, datasetID string, filename strin
 	statusLog := log.FromCtx(ctx).With("phase", "store")
 
 	// Get dataset
-	ds, err := s.GetDataset(ctx, datasetID)
+	ds, err := s.GetDataset(ctx, datasetID, nil)
 	if err != nil {
 		return nil, err
 	}

diff --git a/knowledge/pkg/datastore/retrieve.go b/knowledge/pkg/datastore/retrieve.go
@@ -73,7 +73,7 @@ func (s *Datastore) Retrieve(ctx context.Context, datasetIDs []string, query str
 }
 
 func (s *Datastore) SimilaritySearch(ctx context.Context, query string, numDocuments int, datasetID string, where map[string]string, whereDocument []chromem.WhereDocument) ([]types2.Document, error) {
-	ds, err := s.GetDataset(ctx, datasetID)
+	ds, err := s.GetDataset(ctx, datasetID, nil)
 	if err != nil {
 		return nil, err
 	}

diff --git a/knowledge/pkg/datastore/retrievers/bm25.go b/knowledge/pkg/datastore/retrievers/bm25.go
@@ -44,7 +44,7 @@ func (r *BM25Retriever) Retrieve(ctx context.Context, store store.Store, query s
 	for _, datasetID := range datasetIDs {
 		// TODO: make configurable via RetrieveOpts
 		// silently ignore non-existent datasets
-		ds, err := store.GetDataset(ctx, datasetID)
+		ds, err := store.GetDataset(ctx, datasetID, nil)
 		if err != nil {
 			if strings.HasPrefix(err.Error(), "dataset not found") {
 				slog.Info("Dataset not found", "dataset", datasetID)

diff --git a/knowledge/pkg/datastore/retrievers/retrievers.go b/knowledge/pkg/datastore/retrievers/retrievers.go
@@ -86,7 +86,7 @@ func (r *BasicRetriever) Retrieve(ctx context.Context, store store.Store, query
 	for _, dataset := range datasetIDs {
 		// TODO: make configurable via RetrieveOpts
 		// silently ignore non-existent datasets
-		ds, err := store.GetDataset(ctx, dataset)
+		ds, err := store.GetDataset(ctx, dataset, nil)
 		if err != nil {
 			if strings.HasPrefix(err.Error(), "dataset not found") {
 				continue

diff --git a/knowledge/pkg/datastore/retrievers/routing.go b/knowledge/pkg/datastore/retrievers/routing.go
@@ -69,7 +69,7 @@ func (r *RoutingRetriever) Retrieve(ctx context.Context, store store.Store, quer
 
 	datasets := map[string]map[string]any{}
 	for _, dsID := range r.AvailableDatasets {
-		dataset, err := store.GetDataset(ctx, dsID)
+		dataset, err := store.GetDataset(ctx, dsID, nil)
 		if err != nil {
 			return nil, err
 		}

diff --git a/knowledge/pkg/datastore/retrievers/subquery.go b/knowledge/pkg/datastore/retrievers/subquery.go
@@ -91,7 +91,7 @@ func (s *SubqueryRetriever) Retrieve(ctx context.Context, store store.Store, que
 	for _, dataset := range datasetIDs {
 		// TODO: make configurable via RetrieveOpts
 		// silently ignore non-existent datasets
-		ds, err := store.GetDataset(ctx, dataset)
+		ds, err := store.GetDataset(ctx, dataset, nil)
 		if err != nil {
 			if strings.HasPrefix(err.Error(), "dataset not found") {
 				continue

diff --git a/knowledge/pkg/datastore/store/store.go b/knowledge/pkg/datastore/store/store.go
@@ -10,7 +10,7 @@ import (
 
 type Store interface {
 	ListDatasets(ctx context.Context) ([]types.Dataset, error)
-	GetDataset(ctx context.Context, datasetID string) (*types.Dataset, error)
+	GetDataset(ctx context.Context, datasetID string, opts *types.DatasetGetOpts) (*types.Dataset, error)
 	SimilaritySearch(ctx context.Context, query string, numDocuments int, collection string, where map[string]string, whereDocument []chromem.WhereDocument) ([]vs.Document, error)
 	GetDocuments(ctx context.Context, datasetID string, where map[string]string, whereDocument []chromem.WhereDocument) ([]vs.Document, error)
 }
diff --git a/knowledge/pkg/index/db.go b/knowledge/pkg/index/db.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log"
 	"log/slog"
+	"os"
 	"strings"
 	"time"
 
@@ -15,14 +16,26 @@ import (
 )
 
 func New(ctx context.Context, dsn string, autoMigrate bool) (Index, error) {
+	gormLogLevel := logger.Silent
+	switch os.Getenv("GORM_LOG_LEVEL") {
+	case "silent":
+		gormLogLevel = logger.Silent
+	case "error":
+		gormLogLevel = logger.Error
+	case "warn":
+		gormLogLevel = logger.Warn
+	case "info":
+		gormLogLevel = logger.Info
+	}
+
 	var (
 		indexDB Index
 		err     error
 		gormCfg = &gorm.Config{
 			Logger: logger.New(log.Default(), logger.Config{
 				SlowThreshold: 200 * time.Millisecond,
 				Colorful:      true,
-				LogLevel:      logger.Silent,
+				LogLevel:      gormLogLevel,
 			}),
 			TranslateError: true,
 		}

diff --git a/knowledge/pkg/index/index.go b/knowledge/pkg/index/index.go
@@ -12,7 +12,7 @@ type Index interface {
 
 	// Fundamental Dataset Operations
 	CreateDataset(ctx context.Context, dataset types.Dataset, opts *types.DatasetCreateOpts) error
-	GetDataset(ctx context.Context, datasetID string) (*types.Dataset, error)
+	GetDataset(ctx context.Context, datasetID string, opts *types.DatasetGetOpts) (*types.Dataset, error)
 	ListDatasets(ctx context.Context) ([]types.Dataset, error)
 	DeleteDataset(ctx context.Context, datasetID string) error
 

diff --git a/knowledge/pkg/index/postgres/postgres.go b/knowledge/pkg/index/postgres/postgres.go
@@ -59,8 +59,8 @@ func (i *Index) CreateDataset(ctx context.Context, dataset types.Dataset, opts *
 	return i.DB.CreateDataset(ctx, dataset, opts)
 }
 
-func (i *Index) GetDataset(ctx context.Context, datasetID string) (*types.Dataset, error) {
-	return i.DB.GetDataset(ctx, datasetID)
+func (i *Index) GetDataset(ctx context.Context, datasetID string, opts *types.DatasetGetOpts) (*types.Dataset, error) {
+	return i.DB.GetDataset(ctx, datasetID, opts)
 }
 
 func (i *Index) ListDatasets(ctx context.Context) ([]types.Dataset, error) {

diff --git a/knowledge/pkg/index/sqlite/sqlite.go b/knowledge/pkg/index/sqlite/sqlite.go
@@ -143,8 +143,8 @@ func (i *Index) CreateDataset(ctx context.Context, dataset types.Dataset, opts *
 	return i.DB.CreateDataset(ctx, dataset, opts)
 }
 
-func (i *Index) GetDataset(ctx context.Context, datasetID string) (*types.Dataset, error) {
-	return i.DB.GetDataset(ctx, datasetID)
+func (i *Index) GetDataset(ctx context.Context, datasetID string, opts *types.DatasetGetOpts) (*types.Dataset, error) {
+	return i.DB.GetDataset(ctx, datasetID, opts)
 }
 
 func (i *Index) ListDatasets(ctx context.Context) ([]types.Dataset, error) {

diff --git a/knowledge/pkg/index/types/models.go b/knowledge/pkg/index/types/models.go
@@ -10,6 +10,10 @@ type DatasetCreateOpts struct {
 	ErrOnExists bool
 }
 
+type DatasetGetOpts struct {
+	IncludeFiles bool
+}
+
 // Dataset refers to a VectorDB data space.
 // @Description Dataset refers to a VectorDB data space.
 type Dataset struct {

diff --git a/knowledge/pkg/index/types/types.go b/knowledge/pkg/index/types/types.go
@@ -81,9 +81,21 @@ func (db *DB) DeleteDataset(ctx context.Context, datasetID string) error {
 	return nil
 }
 
-func (db *DB) GetDataset(ctx context.Context, datasetID string) (*Dataset, error) {
+func (db *DB) GetDataset(ctx context.Context, datasetID string, opts *DatasetGetOpts) (*Dataset, error) {
 	dataset := &Dataset{}
-	tx := db.WithContext(ctx).Preload("Files.Documents").First(dataset, "id = ?", datasetID)
+	tx := db.WithContext(ctx)
+
+	if opts == nil {
+		opts = &DatasetGetOpts{
+			IncludeFiles: false,
+		}
+	}
+
+	if opts.IncludeFiles {
+		tx = tx.Preload("Files.Documents")
+	}
+
+	tx = tx.First(dataset, "id = ?", datasetID)
 	if tx.Error != nil {
 		if errors.Is(tx.Error, gorm.ErrRecordNotFound) {
 			return nil, nil