Remove const to enable move semantics and improve schema util functions (#25193)

Signed-off-by: Nbigsheeper <yihao.dai@zilliz.com>

internal/core/src/index/VectorDiskIndex.cpp

@@ -291,7 +291,7 @@ VectorDiskAnnIndex<T>::HasRawData() const {
 }
 
 template <typename T>
-const std::vector<uint8_t>
+std::vector<uint8_t>
 VectorDiskAnnIndex<T>::GetVector(const DatasetPtr dataset) const {
     auto res = index_.GetVectorByIds(*dataset);
     if (!res.has_value()) {

internal/core/src/index/VectorDiskIndex.h

@@ -73,7 +73,7 @@ class VectorDiskAnnIndex : public VectorIndex {
     const bool
     HasRawData() const override;
 
-    const std::vector<uint8_t>
+    std::vector<uint8_t>
     GetVector(const DatasetPtr dataset) const override;
 
     void

internal/core/src/index/VectorIndex.h

@@ -59,7 +59,7 @@ class VectorIndex : public IndexBase {
     virtual const bool
     HasRawData() const = 0;
 
-    virtual const std::vector<uint8_t>
+    virtual std::vector<uint8_t>
     GetVector(const DatasetPtr dataset) const = 0;
 
     IndexType

internal/core/src/index/VectorMemIndex.cpp

@@ -244,7 +244,7 @@ VectorMemIndex::HasRawData() const {
     return index_.HasRawData(GetMetricType());
 }
 
-const std::vector<uint8_t>
+std::vector<uint8_t>
 VectorMemIndex::GetVector(const DatasetPtr dataset) const {
     auto res = index_.GetVectorByIds(*dataset);
     if (!res.has_value()) {

internal/core/src/index/VectorMemIndex.h

@@ -65,7 +65,7 @@ class VectorMemIndex : public VectorIndex {
     const bool
     HasRawData() const override;
 
-    const std::vector<uint8_t>
+    std::vector<uint8_t>
     GetVector(const DatasetPtr dataset) const override;
 
     BinarySet

internal/core/src/segcore/FieldIndexing.cpp

@@ -67,7 +67,7 @@ VectorFieldIndexing::GetDataFromIndex(const int64_t* seg_offsets,
     ids_ds->SetIds(seg_offsets);
     ids_ds->SetIsOwner(false);
 
-    auto& vector = index_->GetVector(ids_ds);
+    auto vector = index_->GetVector(ids_ds);
 
     std::memcpy(output, vector.data(), count * element_size);
 }

internal/core/src/segcore/SegmentSealedImpl.cpp

@@ -605,7 +605,7 @@ SegmentSealedImpl::get_vector(FieldId field_id,
 
         if (has_raw_data) {
             auto ids_ds = GenIdsDataset(count, ids);
-            auto& vector = vec_index->GetVector(ids_ds);
+            auto vector = vec_index->GetVector(ids_ds);
             return segcore::CreateVectorDataArrayFrom(
                 vector.data(), count, filed_meta);
         }

internal/proxy/task_search.go

@@ -596,7 +596,7 @@ func (t *searchTask) Requery() error {
 		return err
 	}
 	offsets := make(map[any]int)
-	for i := 0; i < typeutil.GetDataSize(pkFieldData); i++ {
+	for i := 0; i < typeutil.GetPKSize(pkFieldData); i++ {
 		pk := typeutil.GetData(pkFieldData, i)
 		offsets[pk] = i
 	}

pkg/util/typeutil/schema.go

@@ -817,26 +817,12 @@ func GetSizeOfIDs(data *schemapb.IDs) int {
 	return result
 }
 
-func GetDataSize(fieldData *schemapb.FieldData) int {
+func GetPKSize(fieldData *schemapb.FieldData) int {
 	switch fieldData.GetType() {
-	case schemapb.DataType_Bool:
-		return len(fieldData.GetScalars().GetBoolData().GetData())
-	case schemapb.DataType_Int8, schemapb.DataType_Int16, schemapb.DataType_Int32:
-		return len(fieldData.GetScalars().GetIntData().GetData())
 	case schemapb.DataType_Int64:
 		return len(fieldData.GetScalars().GetLongData().GetData())
-	case schemapb.DataType_Float:
-		return len(fieldData.GetScalars().GetFloatData().GetData())
-	case schemapb.DataType_Double:
-		return len(fieldData.GetScalars().GetDoubleData().GetData())
-	case schemapb.DataType_String:
-		return len(fieldData.GetScalars().GetStringData().GetData())
 	case schemapb.DataType_VarChar:
 		return len(fieldData.GetScalars().GetStringData().GetData())
-	case schemapb.DataType_FloatVector:
-		return len(fieldData.GetVectors().GetFloatVector().GetData())
-	case schemapb.DataType_BinaryVector:
-		return len(fieldData.GetVectors().GetBinaryVector())
 	}
 	return 0
 }
@@ -874,8 +860,6 @@ func GetData(field *schemapb.FieldData, idx int) interface{} {
 		return field.GetScalars().GetFloatData().GetData()[idx]
 	case schemapb.DataType_Double:
 		return field.GetScalars().GetDoubleData().GetData()[idx]
-	case schemapb.DataType_String:
-		return field.GetScalars().GetStringData().GetData()[idx]
 	case schemapb.DataType_VarChar:
 		return field.GetScalars().GetStringData().GetData()[idx]
 	case schemapb.DataType_FloatVector:

pkg/util/typeutil/schema_test.go

@@ -1089,7 +1089,6 @@ func TestGetDataAndGetDataSize(t *testing.T) {
 	FloatArray := []float32{1.0, 2.0}
 	DoubleArray := []float64{11.0, 22.0}
 	VarCharArray := []string{"a", "b"}
-	StringArray := []string{"c", "d"}
 	BinaryVector := []byte{0x12, 0x34}
 	FloatVector := []float32{1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 11.0, 22.0, 33.0, 44.0, 55.0, 66.0, 77.0, 88.0}
 
@@ -1101,39 +1100,18 @@ func TestGetDataAndGetDataSize(t *testing.T) {
 	floatData := genFieldData(fieldName, fieldID, schemapb.DataType_Float, FloatArray, 1)
 	doubleData := genFieldData(fieldName, fieldID, schemapb.DataType_Double, DoubleArray, 1)
 	varCharData := genFieldData(fieldName, fieldID, schemapb.DataType_VarChar, VarCharArray, 1)
-	stringData := genFieldData(fieldName, fieldID, schemapb.DataType_String, StringArray, 1)
 	binVecData := genFieldData(fieldName, fieldID, schemapb.DataType_BinaryVector, BinaryVector, Dim)
 	floatVecData := genFieldData(fieldName, fieldID, schemapb.DataType_FloatVector, FloatVector, Dim)
 	invalidData := &schemapb.FieldData{
 		Type: schemapb.DataType_None,
 	}
 
-	t.Run("test GetDataSize", func(t *testing.T) {
-		boolDataRes := GetDataSize(boolData)
-		int8DataRes := GetDataSize(int8Data)
-		int16DataRes := GetDataSize(int16Data)
-		int32DataRes := GetDataSize(int32Data)
-		int64DataRes := GetDataSize(int64Data)
-		floatDataRes := GetDataSize(floatData)
-		doubleDataRes := GetDataSize(doubleData)
-		varCharDataRes := GetDataSize(varCharData)
-		stringDataRes := GetDataSize(stringData)
-		binVecDataRes := GetDataSize(binVecData)
-		floatVecDataRes := GetDataSize(floatVecData)
-		invalidDataRes := GetDataSize(invalidData)
-
-		assert.Equal(t, 2, boolDataRes)
-		assert.Equal(t, 2, int8DataRes)
-		assert.Equal(t, 2, int16DataRes)
-		assert.Equal(t, 2, int32DataRes)
+	t.Run("test GetPKSize", func(t *testing.T) {
+		int64DataRes := GetPKSize(int64Data)
+		varCharDataRes := GetPKSize(varCharData)
+
 		assert.Equal(t, 2, int64DataRes)
-		assert.Equal(t, 2, floatDataRes)
-		assert.Equal(t, 2, doubleDataRes)
 		assert.Equal(t, 2, varCharDataRes)
-		assert.Equal(t, 2, stringDataRes)
-		assert.Equal(t, 2*Dim/8, binVecDataRes)
-		assert.Equal(t, 2*Dim, floatVecDataRes)
-		assert.Equal(t, 0, invalidDataRes)
 	})
 
 	t.Run("test GetData", func(t *testing.T) {
@@ -1145,7 +1123,6 @@ func TestGetDataAndGetDataSize(t *testing.T) {
 		floatDataRes := GetData(floatData, 0)
 		doubleDataRes := GetData(doubleData, 0)
 		varCharDataRes := GetData(varCharData, 0)
-		stringDataRes := GetData(stringData, 0)
 		binVecDataRes := GetData(binVecData, 0)
 		floatVecDataRes := GetData(floatVecData, 0)
 		invalidDataRes := GetData(invalidData, 0)
@@ -1158,7 +1135,6 @@ func TestGetDataAndGetDataSize(t *testing.T) {
 		assert.Equal(t, FloatArray[0], floatDataRes)
 		assert.Equal(t, DoubleArray[0], doubleDataRes)
 		assert.Equal(t, VarCharArray[0], varCharDataRes)
-		assert.Equal(t, StringArray[0], stringDataRes)
 		assert.ElementsMatch(t, BinaryVector[:Dim/8], binVecDataRes)
 		assert.ElementsMatch(t, FloatVector[:Dim], floatVecDataRes)
 		assert.Nil(t, invalidDataRes)