mirror of
https://gitee.com/milvus-io/milvus.git
synced 2025-12-28 14:35:27 +08:00
related: #45993 This commit extends nullable vector support to the proxy layer, querynode, and adds comprehensive validation, search reduce, and field data handling for nullable vectors with sparse storage. Proxy layer changes: - Update validate_util.go checkAligned() with getExpectedVectorRows() helper to validate nullable vector field alignment using valid data count - Update checkFloatVectorFieldData/checkSparseFloatVectorFieldData for nullable vector validation with proper row count expectations - Add FieldDataIdxComputer in typeutil/schema.go for logical-to-physical index translation during search reduce operations - Update search_reduce_util.go reduceSearchResultData to use idxComputers for correct field data indexing with nullable vectors - Update task.go, task_query.go, task_upsert.go for nullable vector handling - Update msg_pack.go with nullable vector field data processing QueryNode layer changes: - Update segments/result.go for nullable vector result handling - Update segments/search_reduce.go with nullable vector offset translation Storage and index changes: - Update data_codec.go and utils.go for nullable vector serialization - Update indexcgowrapper/dataset.go and index.go for nullable vector indexing Utility changes: - Add FieldDataIdxComputer struct with Compute() method for efficient logical-to-physical index mapping across multiple field data - Update EstimateEntitySize() and AppendFieldData() with fieldIdxs parameter - Update funcutil.go with nullable vector support functions <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Full support for nullable vector fields (float, binary, float16, bfloat16, int8, sparse) across ingest, storage, indexing, search and retrieval; logical↔physical offset mapping preserves row semantics. * Client: compaction control and compaction-state APIs. * **Bug Fixes** * Improved validation for adding vector fields (nullable + dimension checks) and corrected search/query behavior for nullable vectors. * **Chores** * Persisted validity maps with indexes and on-disk formats. * **Tests** * Extensive new and updated end-to-end nullable-vector tests. <sub>✏️ Tip: You can customize this high-level summary in your review settings.</sub> <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: marcelo-cjl <marcelo.chen@zilliz.com>
1536 lines
40 KiB
Go
1536 lines
40 KiB
Go
// Licensed to the LF AI & Data foundation under one
|
|
// or more contributor license agreements. See the NOTICE file
|
|
// distributed with this work for additional information
|
|
// regarding copyright ownership. The ASF licenses this file
|
|
// to you under the Apache License, Version 2.0 (the
|
|
// "License"); you may not use this file except in compliance
|
|
// with the License. You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
package storage
|
|
|
|
import (
|
|
"encoding/binary"
|
|
"fmt"
|
|
"io"
|
|
"math"
|
|
"strconv"
|
|
"strings"
|
|
|
|
"github.com/apache/arrow/go/v17/arrow"
|
|
"github.com/apache/arrow/go/v17/arrow/array"
|
|
"github.com/apache/arrow/go/v17/arrow/bitutil"
|
|
"github.com/apache/arrow/go/v17/parquet"
|
|
"github.com/apache/arrow/go/v17/parquet/compress"
|
|
"github.com/apache/arrow/go/v17/parquet/pqarrow"
|
|
"google.golang.org/protobuf/proto"
|
|
|
|
"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
|
|
"github.com/milvus-io/milvus/pkg/v2/common"
|
|
"github.com/milvus-io/milvus/pkg/v2/util/merr"
|
|
"github.com/milvus-io/milvus/pkg/v2/util/typeutil"
|
|
)
|
|
|
|
type Record interface {
|
|
Column(i FieldID) arrow.Array
|
|
Len() int
|
|
Release()
|
|
Retain()
|
|
}
|
|
|
|
type RecordWriter interface {
|
|
Write(r Record) error
|
|
GetWrittenUncompressed() uint64
|
|
Close() error
|
|
}
|
|
|
|
type (
|
|
Serializer[T any] func([]T) (Record, error)
|
|
Deserializer[T any] func(Record, []T) error
|
|
)
|
|
|
|
// compositeRecord is a record being composed of multiple records, in which each only have 1 column
|
|
type compositeRecord struct {
|
|
index map[FieldID]int16
|
|
recs []arrow.Array
|
|
}
|
|
|
|
var _ Record = (*compositeRecord)(nil)
|
|
|
|
func (r *compositeRecord) Column(i FieldID) arrow.Array {
|
|
if _, ok := r.index[i]; !ok {
|
|
return nil
|
|
}
|
|
return r.recs[r.index[i]]
|
|
}
|
|
|
|
func (r *compositeRecord) Len() int {
|
|
return r.recs[0].Len()
|
|
}
|
|
|
|
func (r *compositeRecord) Release() {
|
|
for _, rec := range r.recs {
|
|
rec.Release()
|
|
}
|
|
}
|
|
|
|
func (r *compositeRecord) Retain() {
|
|
for _, rec := range r.recs {
|
|
rec.Retain()
|
|
}
|
|
}
|
|
|
|
type serdeEntry struct {
|
|
// arrowType returns the arrow type for the given dimension and element type
|
|
// elementType is only used for ArrayOfVector
|
|
arrowType func(dim int, elementType schemapb.DataType) arrow.DataType
|
|
// deserialize deserializes the i-th element in the array, returns the value and ok.
|
|
// null is deserialized to nil without checking the type nullability.
|
|
// if shouldCopy is true, the returned value is copied rather than referenced from arrow array.
|
|
// elementType is only used for ArrayOfVector
|
|
deserialize func(a arrow.Array, i int, elementType schemapb.DataType, dim int, shouldCopy bool) (any, bool)
|
|
// serialize serializes the value to the builder, returns ok.
|
|
// nil is serialized to null without checking the type nullability.
|
|
// elementType is only used for ArrayOfVector
|
|
serialize func(b array.Builder, v any, elementType schemapb.DataType) bool
|
|
}
|
|
|
|
var serdeMap = func() map[schemapb.DataType]serdeEntry {
|
|
m := make(map[schemapb.DataType]serdeEntry)
|
|
m[schemapb.DataType_Bool] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.FixedWidthTypes.Boolean
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Boolean); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BooleanBuilder); ok {
|
|
if v, ok := v.(bool); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Int8] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Int8
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Int8); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Int8Builder); ok {
|
|
if v, ok := v.(int8); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Int16] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Int16
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Int16); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Int16Builder); ok {
|
|
if v, ok := v.(int16); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Int32] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Int32
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Int32); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Int32Builder); ok {
|
|
if v, ok := v.(int32); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Int64] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Int64
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Int64); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Int64Builder); ok {
|
|
if v, ok := v.(int64); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Float] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Float32
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Float32); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Float32Builder); ok {
|
|
if v, ok := v.(float32); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Double] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Float64
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Float64); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Float64Builder); ok {
|
|
if v, ok := v.(float64); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_Timestamptz] = serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.PrimitiveTypes.Int64
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, _ int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Int64); ok && i < arr.Len() {
|
|
return arr.Value(i), true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.Int64Builder); ok {
|
|
if v, ok := v.(int64); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
stringEntry := serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.BinaryTypes.String
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.String); ok && i < arr.Len() {
|
|
value := arr.Value(i)
|
|
if shouldCopy {
|
|
return strings.Clone(value), true
|
|
}
|
|
return value, true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.StringBuilder); ok {
|
|
if v, ok := v.(string); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
|
|
m[schemapb.DataType_VarChar] = stringEntry
|
|
m[schemapb.DataType_String] = stringEntry
|
|
m[schemapb.DataType_Text] = stringEntry
|
|
|
|
// We're not using the deserialized data in go, so we can skip the heavy pb serde.
|
|
// If there is need in the future, just assign it to m[schemapb.DataType_Array]
|
|
eagerArrayEntry := serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.BinaryTypes.Binary
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Binary); ok && i < arr.Len() {
|
|
v := &schemapb.ScalarField{}
|
|
if err := proto.Unmarshal(arr.Value(i), v); err == nil {
|
|
return v, true
|
|
}
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BinaryBuilder); ok {
|
|
if vv, ok := v.(*schemapb.ScalarField); ok {
|
|
if bytes, err := proto.Marshal(vv); err == nil {
|
|
builder.Append(bytes)
|
|
return true
|
|
}
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
_ = eagerArrayEntry
|
|
|
|
byteEntry := serdeEntry{
|
|
arrowType: func(_ int, _ schemapb.DataType) arrow.DataType {
|
|
return arrow.BinaryTypes.Binary
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.Binary); ok && i < arr.Len() {
|
|
value := arr.Value(i)
|
|
if shouldCopy {
|
|
result := make([]byte, len(value))
|
|
copy(result, value)
|
|
return result, true
|
|
}
|
|
return value, true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BinaryBuilder); ok {
|
|
if vv, ok := v.([]byte); ok {
|
|
builder.Append(vv)
|
|
return true
|
|
}
|
|
if vv, ok := v.(*schemapb.ScalarField); ok {
|
|
if bytes, err := proto.Marshal(vv); err == nil {
|
|
builder.Append(bytes)
|
|
return true
|
|
}
|
|
}
|
|
if vv, ok := v.(*schemapb.VectorField); ok {
|
|
if bytes, err := proto.Marshal(vv); err == nil {
|
|
builder.Append(bytes)
|
|
return true
|
|
}
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
|
|
m[schemapb.DataType_Array] = eagerArrayEntry
|
|
m[schemapb.DataType_JSON] = byteEntry
|
|
m[schemapb.DataType_Geometry] = byteEntry
|
|
|
|
// ArrayOfVector now implements the standard interface with elementType parameter
|
|
m[schemapb.DataType_ArrayOfVector] = serdeEntry{
|
|
arrowType: func(dim int, elementType schemapb.DataType) arrow.DataType {
|
|
return getArrayOfVectorArrowType(elementType, dim)
|
|
},
|
|
deserialize: func(a arrow.Array, i int, elementType schemapb.DataType, dim int, shouldCopy bool) (any, bool) {
|
|
return deserializeArrayOfVector(a, i, elementType, int64(dim), shouldCopy)
|
|
},
|
|
serialize: func(b array.Builder, v any, elementType schemapb.DataType) bool {
|
|
vf, ok := v.(*schemapb.VectorField)
|
|
if !ok {
|
|
return false
|
|
}
|
|
|
|
if vf == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
|
|
builder, ok := b.(*array.ListBuilder)
|
|
if !ok {
|
|
return false
|
|
}
|
|
|
|
builder.Append(true)
|
|
valueBuilder := builder.ValueBuilder().(*array.FixedSizeBinaryBuilder)
|
|
dim := vf.GetDim()
|
|
|
|
appendVectorChunks := func(data []byte, bytesPerVector int) bool {
|
|
numVectors := len(data) / bytesPerVector
|
|
for i := 0; i < numVectors; i++ {
|
|
start := i * bytesPerVector
|
|
end := start + bytesPerVector
|
|
valueBuilder.Append(data[start:end])
|
|
}
|
|
return true
|
|
}
|
|
|
|
switch elementType {
|
|
case schemapb.DataType_FloatVector:
|
|
if vf.GetFloatVector() == nil {
|
|
return false
|
|
}
|
|
floatData := vf.GetFloatVector().GetData()
|
|
numVectors := len(floatData) / int(dim)
|
|
// Convert float data to binary
|
|
for i := 0; i < numVectors; i++ {
|
|
start := i * int(dim)
|
|
end := start + int(dim)
|
|
vectorSlice := floatData[start:end]
|
|
|
|
bytes := make([]byte, dim*4)
|
|
for j, f := range vectorSlice {
|
|
binary.LittleEndian.PutUint32(bytes[j*4:], math.Float32bits(f))
|
|
}
|
|
valueBuilder.Append(bytes)
|
|
}
|
|
return true
|
|
|
|
case schemapb.DataType_BinaryVector:
|
|
if vf.GetBinaryVector() == nil {
|
|
return false
|
|
}
|
|
return appendVectorChunks(vf.GetBinaryVector(), int((dim+7)/8))
|
|
|
|
case schemapb.DataType_Float16Vector:
|
|
if vf.GetFloat16Vector() == nil {
|
|
return false
|
|
}
|
|
return appendVectorChunks(vf.GetFloat16Vector(), int(dim)*2)
|
|
|
|
case schemapb.DataType_BFloat16Vector:
|
|
if vf.GetBfloat16Vector() == nil {
|
|
return false
|
|
}
|
|
return appendVectorChunks(vf.GetBfloat16Vector(), int(dim)*2)
|
|
|
|
case schemapb.DataType_Int8Vector:
|
|
if vf.GetInt8Vector() == nil {
|
|
return false
|
|
}
|
|
return appendVectorChunks(vf.GetInt8Vector(), int(dim))
|
|
|
|
case schemapb.DataType_SparseFloatVector:
|
|
panic("SparseFloatVector in VectorArray not implemented yet")
|
|
default:
|
|
return false
|
|
}
|
|
},
|
|
}
|
|
|
|
fixedSizeDeserializer := func(a arrow.Array, i int, _ schemapb.DataType, _ int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.FixedSizeBinary); ok && i < arr.Len() {
|
|
value := arr.Value(i)
|
|
if shouldCopy {
|
|
result := make([]byte, len(value))
|
|
copy(result, value)
|
|
return result, true
|
|
}
|
|
return value, true
|
|
}
|
|
return nil, false
|
|
}
|
|
fixedSizeSerializer := func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if v, ok := v.([]byte); ok {
|
|
if builder, ok := b.(*array.FixedSizeBinaryBuilder); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BinaryBuilder); ok {
|
|
builder.Append(v)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
}
|
|
|
|
m[schemapb.DataType_BinaryVector] = serdeEntry{
|
|
arrowType: func(dim int, _ schemapb.DataType) arrow.DataType {
|
|
return &arrow.FixedSizeBinaryType{ByteWidth: (dim + 7) / 8}
|
|
},
|
|
deserialize: fixedSizeDeserializer,
|
|
serialize: fixedSizeSerializer,
|
|
}
|
|
m[schemapb.DataType_Float16Vector] = serdeEntry{
|
|
arrowType: func(dim int, _ schemapb.DataType) arrow.DataType {
|
|
return &arrow.FixedSizeBinaryType{ByteWidth: dim * 2}
|
|
},
|
|
deserialize: fixedSizeDeserializer,
|
|
serialize: fixedSizeSerializer,
|
|
}
|
|
m[schemapb.DataType_BFloat16Vector] = serdeEntry{
|
|
arrowType: func(dim int, _ schemapb.DataType) arrow.DataType {
|
|
return &arrow.FixedSizeBinaryType{ByteWidth: dim * 2}
|
|
},
|
|
deserialize: fixedSizeDeserializer,
|
|
serialize: fixedSizeSerializer,
|
|
}
|
|
m[schemapb.DataType_Int8Vector] = serdeEntry{
|
|
arrowType: func(dim int, _ schemapb.DataType) arrow.DataType {
|
|
return &arrow.FixedSizeBinaryType{ByteWidth: dim}
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, _ int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.FixedSizeBinary); ok && i < arr.Len() {
|
|
// convert to []int8
|
|
bytes := arr.Value(i)
|
|
int8s := make([]int8, len(bytes))
|
|
for i, b := range bytes {
|
|
int8s[i] = int8(b)
|
|
}
|
|
return int8s, true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
var bytesData []byte
|
|
if vv, ok := v.([]byte); ok {
|
|
bytesData = vv
|
|
} else if vv, ok := v.([]int8); ok {
|
|
bytesData = arrow.Int8Traits.CastToBytes(vv)
|
|
} else {
|
|
return false
|
|
}
|
|
if builder, ok := b.(*array.FixedSizeBinaryBuilder); ok {
|
|
builder.Append(bytesData)
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BinaryBuilder); ok {
|
|
builder.Append(bytesData)
|
|
return true
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_FloatVector] = serdeEntry{
|
|
arrowType: func(dim int, _ schemapb.DataType) arrow.DataType {
|
|
return &arrow.FixedSizeBinaryType{ByteWidth: dim * 4}
|
|
},
|
|
deserialize: func(a arrow.Array, i int, _ schemapb.DataType, _ int, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
if arr, ok := a.(*array.FixedSizeBinary); ok && i < arr.Len() {
|
|
vector := arrow.Float32Traits.CastFromBytes(arr.Value(i))
|
|
if shouldCopy {
|
|
vectorCopy := make([]float32, len(vector))
|
|
copy(vectorCopy, vector)
|
|
return vectorCopy, true
|
|
}
|
|
return vector, true
|
|
}
|
|
return nil, false
|
|
},
|
|
serialize: func(b array.Builder, v any, _ schemapb.DataType) bool {
|
|
if v == nil {
|
|
b.AppendNull()
|
|
return true
|
|
}
|
|
if vv, ok := v.([]float32); ok {
|
|
dim := len(vv)
|
|
byteLength := dim * 4
|
|
bytesData := make([]byte, byteLength)
|
|
for i, vec := range vv {
|
|
bytes := math.Float32bits(vec)
|
|
common.Endian.PutUint32(bytesData[i*4:], bytes)
|
|
}
|
|
if builder, ok := b.(*array.FixedSizeBinaryBuilder); ok {
|
|
builder.Append(bytesData)
|
|
return true
|
|
}
|
|
if builder, ok := b.(*array.BinaryBuilder); ok {
|
|
builder.Append(bytesData)
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
},
|
|
}
|
|
m[schemapb.DataType_SparseFloatVector] = byteEntry
|
|
return m
|
|
}()
|
|
|
|
// Since parquet does not support custom fallback encoding for now,
|
|
// we disable dict encoding for primary key.
|
|
// It can be scale to all fields once parquet fallback encoding is available.
|
|
func getFieldWriterProps(field *schemapb.FieldSchema) *parquet.WriterProperties {
|
|
if field.GetIsPrimaryKey() {
|
|
return parquet.NewWriterProperties(
|
|
parquet.WithCompression(compress.Codecs.Zstd),
|
|
parquet.WithCompressionLevel(3),
|
|
parquet.WithDictionaryDefault(false),
|
|
)
|
|
}
|
|
return parquet.NewWriterProperties(
|
|
parquet.WithCompression(compress.Codecs.Zstd),
|
|
parquet.WithCompressionLevel(3),
|
|
)
|
|
}
|
|
|
|
type DeserializeReader[T any] interface {
|
|
NextValue() (*T, error)
|
|
Close() error
|
|
}
|
|
|
|
type DeserializeReaderImpl[T any] struct {
|
|
rr RecordReader
|
|
deserializer Deserializer[T]
|
|
rec Record
|
|
values []T
|
|
pos int
|
|
}
|
|
|
|
// Iterate to next value, return error or EOF if no more value.
|
|
func (deser *DeserializeReaderImpl[T]) NextValue() (*T, error) {
|
|
if deser.pos == 0 || deser.pos >= len(deser.values) {
|
|
r, err := deser.rr.Next()
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
deser.pos = 0
|
|
deser.rec = r
|
|
|
|
deser.values = make([]T, deser.rec.Len())
|
|
|
|
if err := deser.deserializer(deser.rec, deser.values); err != nil {
|
|
return nil, err
|
|
}
|
|
}
|
|
ret := &deser.values[deser.pos]
|
|
deser.pos++
|
|
return ret, nil
|
|
}
|
|
|
|
func (deser *DeserializeReaderImpl[T]) Close() error {
|
|
return deser.rr.Close()
|
|
}
|
|
|
|
func NewDeserializeReader[T any](rr RecordReader, deserializer Deserializer[T]) *DeserializeReaderImpl[T] {
|
|
return &DeserializeReaderImpl[T]{
|
|
rr: rr,
|
|
deserializer: deserializer,
|
|
}
|
|
}
|
|
|
|
var _ Record = (*selectiveRecord)(nil)
|
|
|
|
// selectiveRecord is a Record that only contains a single field, reusing existing Record.
|
|
type selectiveRecord struct {
|
|
r Record
|
|
fieldId FieldID
|
|
}
|
|
|
|
func (r *selectiveRecord) Column(i FieldID) arrow.Array {
|
|
if i == r.fieldId {
|
|
return r.r.Column(i)
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func (r *selectiveRecord) Len() int {
|
|
return r.r.Len()
|
|
}
|
|
|
|
func (r *selectiveRecord) Release() {
|
|
// do nothing.
|
|
}
|
|
|
|
func (r *selectiveRecord) Retain() {
|
|
// do nothing
|
|
}
|
|
|
|
func newSelectiveRecord(r Record, selectedFieldId FieldID) Record {
|
|
return &selectiveRecord{
|
|
r: r,
|
|
fieldId: selectedFieldId,
|
|
}
|
|
}
|
|
|
|
var _ RecordWriter = (*CompositeRecordWriter)(nil)
|
|
|
|
type CompositeRecordWriter struct {
|
|
writers map[FieldID]RecordWriter
|
|
}
|
|
|
|
func (crw *CompositeRecordWriter) GetWrittenUncompressed() uint64 {
|
|
s := uint64(0)
|
|
for _, w := range crw.writers {
|
|
s += w.GetWrittenUncompressed()
|
|
}
|
|
return s
|
|
}
|
|
|
|
func (crw *CompositeRecordWriter) Write(r Record) error {
|
|
for fieldId, w := range crw.writers {
|
|
sr := newSelectiveRecord(r, fieldId)
|
|
if err := w.Write(sr); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func (crw *CompositeRecordWriter) Close() error {
|
|
if crw != nil {
|
|
for _, w := range crw.writers {
|
|
if w != nil {
|
|
if err := w.Close(); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func NewCompositeRecordWriter(writers map[FieldID]RecordWriter) *CompositeRecordWriter {
|
|
return &CompositeRecordWriter{
|
|
writers: writers,
|
|
}
|
|
}
|
|
|
|
var _ RecordWriter = (*singleFieldRecordWriter)(nil)
|
|
|
|
type RecordWriterOptions func(*singleFieldRecordWriter)
|
|
|
|
func WithRecordWriterProps(writerProps *parquet.WriterProperties) RecordWriterOptions {
|
|
return func(w *singleFieldRecordWriter) {
|
|
w.writerProps = writerProps
|
|
}
|
|
}
|
|
|
|
type singleFieldRecordWriter struct {
|
|
fw *pqarrow.FileWriter
|
|
fieldId FieldID
|
|
schema *arrow.Schema
|
|
writerProps *parquet.WriterProperties
|
|
|
|
numRows int
|
|
writtenUncompressed uint64
|
|
memoryExpansionRatio int
|
|
}
|
|
|
|
func (sfw *singleFieldRecordWriter) Write(r Record) error {
|
|
sfw.numRows += r.Len()
|
|
a := r.Column(sfw.fieldId)
|
|
|
|
sfw.writtenUncompressed += calculateActualDataSize(a)
|
|
rec := array.NewRecord(sfw.schema, []arrow.Array{a}, int64(r.Len()))
|
|
defer rec.Release()
|
|
return sfw.fw.WriteBuffered(rec)
|
|
}
|
|
|
|
func (sfw *singleFieldRecordWriter) GetWrittenUncompressed() uint64 {
|
|
return sfw.writtenUncompressed * uint64(sfw.memoryExpansionRatio)
|
|
}
|
|
|
|
func (sfw *singleFieldRecordWriter) Close() error {
|
|
return sfw.fw.Close()
|
|
}
|
|
|
|
// getArrayOfVectorArrowType returns the appropriate Arrow type for ArrayOfVector based on element type
|
|
func getArrayOfVectorArrowType(elementType schemapb.DataType, dim int) arrow.DataType {
|
|
switch elementType {
|
|
case schemapb.DataType_FloatVector:
|
|
return arrow.ListOf(&arrow.FixedSizeBinaryType{ByteWidth: dim * 4})
|
|
case schemapb.DataType_BinaryVector:
|
|
return arrow.ListOf(&arrow.FixedSizeBinaryType{ByteWidth: (dim + 7) / 8})
|
|
case schemapb.DataType_Float16Vector:
|
|
return arrow.ListOf(&arrow.FixedSizeBinaryType{ByteWidth: dim * 2})
|
|
case schemapb.DataType_BFloat16Vector:
|
|
return arrow.ListOf(&arrow.FixedSizeBinaryType{ByteWidth: dim * 2})
|
|
case schemapb.DataType_Int8Vector:
|
|
return arrow.ListOf(&arrow.FixedSizeBinaryType{ByteWidth: dim})
|
|
case schemapb.DataType_SparseFloatVector:
|
|
return arrow.ListOf(arrow.BinaryTypes.Binary)
|
|
default:
|
|
panic(fmt.Sprintf("unsupported element type for ArrayOfVector: %s", elementType.String()))
|
|
}
|
|
}
|
|
|
|
// deserializeArrayOfVector deserializes ArrayOfVector data with known element type
|
|
func deserializeArrayOfVector(a arrow.Array, i int, elementType schemapb.DataType, dim int64, shouldCopy bool) (any, bool) {
|
|
if a.IsNull(i) {
|
|
return nil, true
|
|
}
|
|
|
|
arr, ok := a.(*array.List)
|
|
if !ok || i >= arr.Len() {
|
|
return nil, false
|
|
}
|
|
|
|
start, end := arr.ValueOffsets(i)
|
|
totalElements := end - start
|
|
if totalElements == 0 {
|
|
return nil, false
|
|
}
|
|
|
|
valuesArray := arr.ListValues()
|
|
binaryArray, ok := valuesArray.(*array.FixedSizeBinary)
|
|
if !ok {
|
|
// empty array
|
|
return nil, true
|
|
}
|
|
|
|
numVectors := int(totalElements)
|
|
|
|
// Helper function to extract byte vectors from FixedSizeBinary array
|
|
extractByteVectors := func(bytesPerVector int64) []byte {
|
|
totalBytes := numVectors * int(bytesPerVector)
|
|
data := make([]byte, totalBytes)
|
|
for j := 0; j < numVectors; j++ {
|
|
vectorIndex := int(start) + j
|
|
vectorData := binaryArray.Value(vectorIndex)
|
|
copy(data[j*int(bytesPerVector):], vectorData)
|
|
}
|
|
return data
|
|
}
|
|
|
|
switch elementType {
|
|
case schemapb.DataType_FloatVector:
|
|
totalFloats := numVectors * int(dim)
|
|
floatData := make([]float32, totalFloats)
|
|
for j := 0; j < numVectors; j++ {
|
|
vectorIndex := int(start) + j
|
|
binaryData := binaryArray.Value(vectorIndex)
|
|
vectorFloats := arrow.Float32Traits.CastFromBytes(binaryData)
|
|
copy(floatData[j*int(dim):], vectorFloats)
|
|
}
|
|
|
|
return &schemapb.VectorField{
|
|
Dim: dim,
|
|
Data: &schemapb.VectorField_FloatVector{
|
|
FloatVector: &schemapb.FloatArray{
|
|
Data: floatData,
|
|
},
|
|
},
|
|
}, true
|
|
|
|
case schemapb.DataType_BinaryVector:
|
|
return &schemapb.VectorField{
|
|
Dim: dim,
|
|
Data: &schemapb.VectorField_BinaryVector{
|
|
BinaryVector: extractByteVectors((dim + 7) / 8),
|
|
},
|
|
}, true
|
|
|
|
case schemapb.DataType_Float16Vector:
|
|
return &schemapb.VectorField{
|
|
Dim: dim,
|
|
Data: &schemapb.VectorField_Float16Vector{
|
|
Float16Vector: extractByteVectors(dim * 2),
|
|
},
|
|
}, true
|
|
|
|
case schemapb.DataType_BFloat16Vector:
|
|
return &schemapb.VectorField{
|
|
Dim: dim,
|
|
Data: &schemapb.VectorField_Bfloat16Vector{
|
|
Bfloat16Vector: extractByteVectors(dim * 2),
|
|
},
|
|
}, true
|
|
|
|
case schemapb.DataType_Int8Vector:
|
|
return &schemapb.VectorField{
|
|
Dim: dim,
|
|
Data: &schemapb.VectorField_Int8Vector{
|
|
Int8Vector: extractByteVectors(dim),
|
|
},
|
|
}, true
|
|
case schemapb.DataType_SparseFloatVector:
|
|
panic("SparseFloatVector in VectorArray deserialization not implemented yet")
|
|
default:
|
|
panic(fmt.Sprintf("unsupported element type for ArrayOfVector deserialization: %s", elementType.String()))
|
|
}
|
|
}
|
|
|
|
func newSingleFieldRecordWriter(field *schemapb.FieldSchema, writer io.Writer, opts ...RecordWriterOptions) (*singleFieldRecordWriter, error) {
|
|
// calculate memory expansion ratio
|
|
// arrays are serialized by protobuf, where int values may be compacted, see https://protobuf.dev/reference/go/size
|
|
// to correct the actual size, we need to multiply the memory expansion ratio accordingly.
|
|
determineMemoryExpansionRatio := func(field *schemapb.FieldSchema) int {
|
|
if field.DataType == schemapb.DataType_Array {
|
|
switch field.GetElementType() {
|
|
case schemapb.DataType_Int16:
|
|
return 2
|
|
case schemapb.DataType_Int32:
|
|
return 4
|
|
case schemapb.DataType_Int64, schemapb.DataType_Timestamptz:
|
|
return 8
|
|
}
|
|
}
|
|
return 1
|
|
}
|
|
dim, _ := typeutil.GetDim(field)
|
|
|
|
var fieldMetadata arrow.Metadata
|
|
var arrowType arrow.DataType
|
|
elementType := schemapb.DataType_None
|
|
|
|
if field.DataType == schemapb.DataType_ArrayOfVector {
|
|
elementType = field.GetElementType()
|
|
fieldMetadata = arrow.NewMetadata(
|
|
[]string{"elementType", "dim"},
|
|
[]string{fmt.Sprintf("%d", int32(elementType)), fmt.Sprintf("%d", dim)},
|
|
)
|
|
}
|
|
|
|
if field.GetNullable() && typeutil.IsVectorType(field.DataType) && !typeutil.IsSparseFloatVectorType(field.DataType) {
|
|
arrowType = arrow.BinaryTypes.Binary
|
|
fieldMetadata = arrow.NewMetadata(
|
|
[]string{"dim"},
|
|
[]string{fmt.Sprintf("%d", dim)},
|
|
)
|
|
} else {
|
|
arrowType = serdeMap[field.DataType].arrowType(int(dim), elementType)
|
|
}
|
|
|
|
w := &singleFieldRecordWriter{
|
|
fieldId: field.FieldID,
|
|
schema: arrow.NewSchema([]arrow.Field{
|
|
{
|
|
Name: strconv.Itoa(int(field.FieldID)),
|
|
Type: arrowType,
|
|
Nullable: true, // No nullable check here.
|
|
Metadata: fieldMetadata,
|
|
},
|
|
}, nil),
|
|
writerProps: parquet.NewWriterProperties(
|
|
parquet.WithMaxRowGroupLength(math.MaxInt64), // No additional grouping for now.
|
|
parquet.WithCompression(compress.Codecs.Zstd),
|
|
parquet.WithCompressionLevel(3)),
|
|
memoryExpansionRatio: determineMemoryExpansionRatio(field),
|
|
}
|
|
for _, o := range opts {
|
|
o(w)
|
|
}
|
|
|
|
// Use appropriate Arrow writer properties for ArrayOfVector
|
|
arrowWriterProps := pqarrow.DefaultWriterProps()
|
|
if field.DataType == schemapb.DataType_ArrayOfVector {
|
|
// Ensure schema metadata is preserved for ArrayOfVector
|
|
arrowWriterProps = pqarrow.NewArrowWriterProperties(
|
|
pqarrow.WithStoreSchema(),
|
|
)
|
|
}
|
|
|
|
fw, err := pqarrow.NewFileWriter(w.schema, writer, w.writerProps, arrowWriterProps)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
w.fw = fw
|
|
return w, nil
|
|
}
|
|
|
|
var _ RecordWriter = (*multiFieldRecordWriter)(nil)
|
|
|
|
type multiFieldRecordWriter struct {
|
|
fw *pqarrow.FileWriter
|
|
fieldIDs []FieldID
|
|
schema *arrow.Schema
|
|
|
|
numRows int
|
|
writtenUncompressed uint64
|
|
}
|
|
|
|
func (mfw *multiFieldRecordWriter) Write(r Record) error {
|
|
mfw.numRows += r.Len()
|
|
columns := make([]arrow.Array, len(mfw.fieldIDs))
|
|
for i, fieldId := range mfw.fieldIDs {
|
|
columns[i] = r.Column(fieldId)
|
|
mfw.writtenUncompressed += calculateActualDataSize(columns[i])
|
|
}
|
|
rec := array.NewRecord(mfw.schema, columns, int64(r.Len()))
|
|
defer rec.Release()
|
|
return mfw.fw.WriteBuffered(rec)
|
|
}
|
|
|
|
func (mfw *multiFieldRecordWriter) GetWrittenUncompressed() uint64 {
|
|
return mfw.writtenUncompressed
|
|
}
|
|
|
|
func (mfw *multiFieldRecordWriter) Close() error {
|
|
return mfw.fw.Close()
|
|
}
|
|
|
|
func newMultiFieldRecordWriter(fieldIDs []FieldID, fields []arrow.Field, writer io.Writer) (*multiFieldRecordWriter, error) {
|
|
schema := arrow.NewSchema(fields, nil)
|
|
fw, err := pqarrow.NewFileWriter(schema, writer,
|
|
parquet.NewWriterProperties(parquet.WithMaxRowGroupLength(math.MaxInt64)), // No additional grouping for now.
|
|
pqarrow.DefaultWriterProps())
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
return &multiFieldRecordWriter{
|
|
fw: fw,
|
|
fieldIDs: fieldIDs,
|
|
schema: schema,
|
|
}, nil
|
|
}
|
|
|
|
type SerializeWriter[T any] interface {
|
|
WriteValue(value T) error
|
|
Flush() error
|
|
Close() error
|
|
}
|
|
|
|
type SerializeWriterImpl[T any] struct {
|
|
rw RecordWriter
|
|
serializer Serializer[T]
|
|
batchSize int
|
|
|
|
buffer []T
|
|
pos int
|
|
}
|
|
|
|
func (sw *SerializeWriterImpl[T]) Flush() error {
|
|
if sw.pos == 0 {
|
|
return nil
|
|
}
|
|
buf := sw.buffer[:sw.pos]
|
|
r, err := sw.serializer(buf)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer r.Release()
|
|
if err := sw.rw.Write(r); err != nil {
|
|
return err
|
|
}
|
|
sw.pos = 0
|
|
return nil
|
|
}
|
|
|
|
func (sw *SerializeWriterImpl[T]) WriteValue(value T) error {
|
|
if sw.buffer == nil {
|
|
sw.buffer = make([]T, sw.batchSize)
|
|
}
|
|
sw.buffer[sw.pos] = value
|
|
sw.pos++
|
|
if sw.pos == sw.batchSize {
|
|
if err := sw.Flush(); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func (sw *SerializeWriterImpl[T]) Close() error {
|
|
if err := sw.Flush(); err != nil {
|
|
return err
|
|
}
|
|
return sw.rw.Close()
|
|
}
|
|
|
|
func NewSerializeRecordWriter[T any](rw RecordWriter, serializer Serializer[T], batchSize int) *SerializeWriterImpl[T] {
|
|
return &SerializeWriterImpl[T]{
|
|
rw: rw,
|
|
serializer: serializer,
|
|
batchSize: batchSize,
|
|
}
|
|
}
|
|
|
|
type simpleArrowRecord struct {
|
|
r arrow.Record
|
|
|
|
field2Col map[FieldID]int
|
|
}
|
|
|
|
var _ Record = (*simpleArrowRecord)(nil)
|
|
|
|
func (sr *simpleArrowRecord) Column(i FieldID) arrow.Array {
|
|
colIdx, ok := sr.field2Col[i]
|
|
if !ok {
|
|
panic(fmt.Sprintf("no such field: %d, having %v", i, sr.field2Col))
|
|
}
|
|
return sr.r.Column(colIdx)
|
|
}
|
|
|
|
func (sr *simpleArrowRecord) Len() int {
|
|
return int(sr.r.NumRows())
|
|
}
|
|
|
|
func (sr *simpleArrowRecord) Release() {
|
|
sr.r.Release()
|
|
}
|
|
|
|
func (sr *simpleArrowRecord) Retain() {
|
|
sr.r.Retain()
|
|
}
|
|
|
|
func (sr *simpleArrowRecord) ArrowSchema() *arrow.Schema {
|
|
return sr.r.Schema()
|
|
}
|
|
|
|
func NewSimpleArrowRecord(r arrow.Record, field2Col map[FieldID]int) *simpleArrowRecord {
|
|
return &simpleArrowRecord{
|
|
r: r,
|
|
field2Col: field2Col,
|
|
}
|
|
}
|
|
|
|
func BuildRecord(b *array.RecordBuilder, data *InsertData, schema *schemapb.CollectionSchema) error {
|
|
if data == nil {
|
|
return nil
|
|
}
|
|
idx := 0
|
|
serializeField := func(field *schemapb.FieldSchema) error {
|
|
fBuilder := b.Field(idx)
|
|
idx++
|
|
typeEntry, ok := serdeMap[field.DataType]
|
|
if !ok {
|
|
panic("unknown type")
|
|
}
|
|
fieldData, exists := data.Data[field.FieldID]
|
|
if !exists {
|
|
return merr.WrapErrFieldNotFound(field.FieldID, fmt.Sprintf("field %s not found", field.Name))
|
|
}
|
|
|
|
if fieldData.RowNum() == 0 {
|
|
return merr.WrapErrServiceInternal(fmt.Sprintf("row num is 0 for field %s", field.Name))
|
|
}
|
|
|
|
// Get element type for ArrayOfVector, otherwise use None
|
|
elementType := schemapb.DataType_None
|
|
if field.DataType == schemapb.DataType_ArrayOfVector {
|
|
elementType = field.GetElementType()
|
|
}
|
|
|
|
if field.GetNullable() && typeutil.IsVectorType(field.DataType) {
|
|
var validData []bool
|
|
switch fd := fieldData.(type) {
|
|
case *FloatVectorFieldData:
|
|
validData = fd.ValidData
|
|
case *BinaryVectorFieldData:
|
|
validData = fd.ValidData
|
|
case *Float16VectorFieldData:
|
|
validData = fd.ValidData
|
|
case *BFloat16VectorFieldData:
|
|
validData = fd.ValidData
|
|
case *SparseFloatVectorFieldData:
|
|
validData = fd.ValidData
|
|
case *Int8VectorFieldData:
|
|
validData = fd.ValidData
|
|
}
|
|
// Use len(validData) as logical row count, GetRow takes logical index
|
|
for j := 0; j < len(validData); j++ {
|
|
if !validData[j] {
|
|
fBuilder.(*array.BinaryBuilder).AppendNull()
|
|
} else {
|
|
rowData := fieldData.GetRow(j)
|
|
ok = typeEntry.serialize(fBuilder, rowData, elementType)
|
|
if !ok {
|
|
return merr.WrapErrServiceInternal(fmt.Sprintf("serialize error on type %s", field.DataType.String()))
|
|
}
|
|
}
|
|
}
|
|
} else {
|
|
for j := 0; j < fieldData.RowNum(); j++ {
|
|
ok = typeEntry.serialize(fBuilder, fieldData.GetRow(j), elementType)
|
|
if !ok {
|
|
return merr.WrapErrServiceInternal(fmt.Sprintf("serialize error on type %s", field.DataType.String()))
|
|
}
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
for _, field := range schema.GetFields() {
|
|
if err := serializeField(field); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
for _, structField := range schema.GetStructArrayFields() {
|
|
for _, field := range structField.GetFields() {
|
|
if err := serializeField(field); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func calculateActualDataSize(a arrow.Array) uint64 {
|
|
data := a.Data()
|
|
if data == nil {
|
|
return 0
|
|
}
|
|
|
|
return ActualSizeInBytes(data)
|
|
}
|
|
|
|
// calculate preciese data size of sliced ArrayData
|
|
func ActualSizeInBytes(data arrow.ArrayData) uint64 {
|
|
var size uint64
|
|
dt := data.DataType()
|
|
length := data.Len()
|
|
offset := data.Offset()
|
|
buffers := data.Buffers()
|
|
|
|
switch dt.ID() {
|
|
case arrow.NULL:
|
|
return 0
|
|
|
|
case arrow.BOOL:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
|
|
case arrow.UINT8, arrow.INT8:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length)
|
|
}
|
|
|
|
case arrow.UINT16, arrow.INT16, arrow.FLOAT16:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 2)
|
|
}
|
|
|
|
case arrow.UINT32, arrow.INT32, arrow.FLOAT32, arrow.DATE32, arrow.TIME32,
|
|
arrow.INTERVAL_MONTHS:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 4)
|
|
}
|
|
|
|
case arrow.UINT64, arrow.INT64, arrow.FLOAT64, arrow.DATE64, arrow.TIME64,
|
|
arrow.TIMESTAMP, arrow.DURATION, arrow.INTERVAL_DAY_TIME:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 8)
|
|
}
|
|
|
|
case arrow.INTERVAL_MONTH_DAY_NANO:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 16)
|
|
}
|
|
|
|
case arrow.DECIMAL128:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 16)
|
|
}
|
|
|
|
case arrow.DECIMAL256:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 32)
|
|
}
|
|
|
|
case arrow.FIXED_SIZE_BINARY:
|
|
fsbType := dt.(*arrow.FixedSizeBinaryType)
|
|
byteWidth := fsbType.ByteWidth
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * byteWidth)
|
|
}
|
|
|
|
case arrow.STRING, arrow.BINARY:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil && buffers[2] != nil {
|
|
size += uint64((length + 1) * 4)
|
|
offsets := arrow.Int32Traits.CastFromBytes(buffers[1].Bytes())
|
|
if offset+length < len(offsets) {
|
|
dataStart := offsets[offset]
|
|
dataEnd := offsets[offset+length]
|
|
size += uint64(dataEnd - dataStart)
|
|
}
|
|
}
|
|
|
|
case arrow.LARGE_STRING, arrow.LARGE_BINARY:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil && buffers[2] != nil {
|
|
size += uint64((length + 1) * 8)
|
|
offsets := arrow.Int64Traits.CastFromBytes(buffers[1].Bytes())
|
|
if offset+length < len(offsets) {
|
|
dataStart := offsets[offset]
|
|
dataEnd := offsets[offset+length]
|
|
size += uint64(dataEnd - dataStart)
|
|
}
|
|
}
|
|
|
|
case arrow.STRING_VIEW, arrow.BINARY_VIEW:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * arrow.ViewHeaderSizeBytes)
|
|
}
|
|
for i := 2; i < len(buffers); i++ {
|
|
if buffers[i] != nil {
|
|
size += uint64(buffers[i].Len())
|
|
}
|
|
}
|
|
|
|
case arrow.LIST, arrow.MAP:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64((length + 1) * 4)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.LARGE_LIST:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64((length + 1) * 8)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.LIST_VIEW:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 4)
|
|
}
|
|
if buffers[2] != nil {
|
|
size += uint64(length * 4)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.LARGE_LIST_VIEW:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 8)
|
|
}
|
|
if buffers[2] != nil {
|
|
size += uint64(length * 8)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.FIXED_SIZE_LIST:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.STRUCT:
|
|
if buffers[0] != nil {
|
|
size += uint64(bitutil.BytesForBits(int64(length)))
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.SPARSE_UNION:
|
|
if buffers[0] != nil {
|
|
size += uint64(length)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.DENSE_UNION:
|
|
if buffers[0] != nil {
|
|
size += uint64(length)
|
|
}
|
|
if buffers[1] != nil {
|
|
size += uint64(length * 4)
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.DICTIONARY:
|
|
for _, buf := range buffers {
|
|
if buf != nil {
|
|
size += uint64(buf.Len())
|
|
}
|
|
}
|
|
if dict := data.Dictionary(); dict != nil {
|
|
size += ActualSizeInBytes(dict)
|
|
}
|
|
|
|
case arrow.RUN_END_ENCODED:
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
|
|
case arrow.EXTENSION:
|
|
extType := dt.(arrow.ExtensionType)
|
|
storageData := array.NewData(extType.StorageType(), length, buffers, data.Children(), data.NullN(), offset)
|
|
size = ActualSizeInBytes(storageData)
|
|
storageData.Release()
|
|
|
|
default:
|
|
for _, buf := range buffers {
|
|
if buf != nil {
|
|
size += uint64(buf.Len())
|
|
}
|
|
}
|
|
for _, child := range data.Children() {
|
|
size += ActualSizeInBytes(child)
|
|
}
|
|
}
|
|
|
|
return size
|
|
}
|