Feat Add index insert data callback (#144)

examples/embeddings/jsondb/main.go

@@ -19,9 +19,12 @@ import (
 func main() {
 
 	index := index.New(
-		jsondb.New("db.json"),
+		jsondb.New().WithPersist("db.json"),
 		openaiembedder.New(openaiembedder.AdaEmbeddingV2),
-	).WithIncludeContents(true)
+	).WithIncludeContents(true).WithAddDataCallback(func(data *index.Data) error {
+		data.Metadata["contentLen"] = len(data.Metadata["content"].(string))
+		return nil
+	})
 
 	indexIsEmpty, _ := index.IsEmpty(context.Background())
 

examples/embeddings/knowledge_base/main.go

@@ -26,7 +26,7 @@ const (
 func main() {
 
 	index := index.New(
-		jsondb.New("db.json"),
+		jsondb.New().WithPersist("db.json"),
 		openaiembedder.New(openaiembedder.AdaEmbeddingV2),
 	).WithIncludeContents(true)
 

examples/embeddings/simplekb/main.go

@@ -15,7 +15,7 @@ import (
 
 func main() {
 	docs, _ := loader.NewPDFToTextLoader("./kb").WithTextSplitter(textsplitter.NewRecursiveCharacterTextSplitter(2000, 200)).Load(context.Background())
-	index := index.New(jsondb.New("db.json"), openaiembedder.New(openaiembedder.AdaEmbeddingV2)).WithIncludeContents(true)
+	index := index.New(jsondb.New(), openaiembedder.New(openaiembedder.AdaEmbeddingV2)).WithIncludeContents(true)
 	index.LoadFromDocuments(context.Background(), docs)
 	qapipeline.New(openai.NewChat().WithVerbose(true)).WithIndex(index).Query(context.Background(), "What is the NATO purpose?", option.WithTopK(1))
 }

examples/llm/cache/main.go

@@ -18,7 +18,7 @@ func main() {
 
 	embedder := openaiembedder.New(openaiembedder.AdaEmbeddingV2)
 	index := index.New(
-		jsondb.New("db.json"),
+		jsondb.New().WithPersist("db.json"),
 		embedder,
 	)
 	llm := openai.NewCompletion().WithCompletionCache(cache.New(embedder, index).WithTopK(3))

index/index.go

@@ -24,6 +24,8 @@ const (
 	defaultIncludeContent  = true
 )
 
+type AddDataCallback func(data *Data) error
+
 type Data struct {
 	ID       string
 	Values   []float64
@@ -45,6 +47,7 @@ type Index struct {
 	embedder        Embedder
 	batchInsertSize int
 	includeContent  bool
+	addDataCallback AddDataCallback
 }
 
 func New(vectorDB VectorDB, embedder Embedder) *Index {
@@ -53,6 +56,7 @@ func New(vectorDB VectorDB, embedder Embedder) *Index {
 		embedder:        embedder,
 		batchInsertSize: defaultBatchInsertSize,
 		includeContent:  defaultIncludeContent,
+		addDataCallback: nil,
 	}
 }
 
@@ -66,6 +70,13 @@ func (i *Index) WithBatchInsertSize(batchInsertSize int) *Index {
 	return i
 }
 
+// WithAddDataCallback allows to modify the data before it is added to the index.
+// This can be useful to add additional metadata to the vector.
+func (i *Index) WithAddDataCallback(callback AddDataCallback) *Index {
+	i.addDataCallback = callback
+	return i
+}
+
 func (i *Index) LoadFromDocuments(ctx context.Context, documents []document.Document) error {
 	err := i.batchUpsert(ctx, documents)
 	if err != nil {
@@ -78,6 +89,14 @@ func (i *Index) Add(ctx context.Context, data *Data) error {
 	if data == nil {
 		return nil
 	}
+
+	if i.addDataCallback != nil {
+		err := i.addDataCallback(data)
+		if err != nil {
+			return fmt.Errorf("%w: %w", ErrInternal, err)
+		}
+	}
+
 	return i.vectorDB.Insert(ctx, []Data{*data})
 }
 
@@ -126,6 +145,15 @@ func (i *Index) batchUpsert(ctx context.Context, documents []document.Document)
 			return err
 		}
 
+		if i.addDataCallback != nil {
+			for j := range data {
+				callbackErr := i.addDataCallback(&data[j])
+				if callbackErr != nil {
+					return fmt.Errorf("%w: %w", ErrInternal, callbackErr)
+				}
+			}
+		}
+
 		err = i.vectorDB.Insert(ctx, data)
 		if err != nil {
 			return err

index/option/option.go

@@ -4,7 +4,7 @@ type Option func(*Options)
 
 type Options struct {
 	TopK   int
-	Filter interface{}
+	Filter any
 }
 
 func WithTopK(topK int) Option {
@@ -13,7 +13,7 @@ func WithTopK(topK int) Option {
 	}
 }
 
-func WithFilter(filter interface{}) Option {
+func WithFilter(filter any) Option {
 	return func(opts *Options) {
 		opts.Filter = filter
 	}

index/vectordb/jsondb/jsondb.go

@@ -22,23 +22,34 @@ type data struct {
 	Values   []float64  `json:"values"`
 }
 
+// DB is a simple in-memory vector database
+// that stores the data in a json file only
+// if the persist option is enabled.
 type DB struct {
 	data   []data
 	dbPath string
 }
 
 type FilterFn func([]index.SearchResult) []index.SearchResult
 
-func New(dbPath string) *DB {
+func New() *DB {
 	index := &DB{
-		data:   []data{},
-		dbPath: dbPath,
+		data: []data{},
 	}
 
 	return index
 }
 
-func (i DB) save() error {
+func (i *DB) WithPersist(dbPath string) *DB {
+	i.dbPath = dbPath
+	return i
+}
+
+func (i *DB) save() error {
+	if i.dbPath == "" {
+		return nil
+	}
+
 	jsonContent, err := json.Marshal(i.data)
 	if err != nil {
 		return err
@@ -48,6 +59,10 @@ func (i DB) save() error {
 }
 
 func (i *DB) load() error {
+	if i.dbPath == "" {
+		return nil
+	}
+
 	if len(i.data) > 0 {
 		return nil
 	}