SIMDJSON S3 select input (#8401)

This commit is contained in:
Klaus Post
2020-02-13 14:03:52 -08:00
committed by GitHub
parent d1144c2c7e
commit e4020fb41f
16 changed files with 1116 additions and 73 deletions

View File

@@ -17,9 +17,10 @@
package sql
import (
"fmt"
"io"
"github.com/bcicen/jstream"
"github.com/minio/simdjson-go"
)
// SelectObjectFormat specifies the format of the underlying data
@@ -32,6 +33,8 @@ const (
SelectFmtCSV
// SelectFmtJSON - JSON format
SelectFmtJSON
// SelectFmtSIMDJSON - SIMD JSON format
SelectFmtSIMDJSON
// SelectFmtParquet - Parquet format
SelectFmtParquet
)
@@ -39,7 +42,10 @@ const (
// Record - is a type containing columns and their values.
type Record interface {
Get(name string) (*Value, error)
Set(name string, value *Value) error
// Set a value.
// Can return a different record type.
Set(name string, value *Value) (Record, error)
WriteCSV(writer io.Writer, fieldDelimiter rune) error
WriteJSON(writer io.Writer) error
@@ -51,5 +57,77 @@ type Record interface {
Raw() (SelectObjectFormat, interface{})
// Replaces the underlying data
Replace(k jstream.KVS) error
Replace(k interface{}) error
}
// IterToValue converts a simdjson Iter to its underlying value.
// Objects are returned as simdjson.Object
// Arrays are returned as []interface{} with parsed values.
func IterToValue(iter simdjson.Iter) (interface{}, error) {
switch iter.Type() {
case simdjson.TypeString:
v, err := iter.String()
if err != nil {
return nil, err
}
return v, nil
case simdjson.TypeFloat:
v, err := iter.Float()
if err != nil {
return nil, err
}
return v, nil
case simdjson.TypeInt:
v, err := iter.Int()
if err != nil {
return nil, err
}
return v, nil
case simdjson.TypeUint:
v, err := iter.Int()
if err != nil {
// Can't fit into int, convert to float.
v, err := iter.Float()
return v, err
}
return v, nil
case simdjson.TypeBool:
v, err := iter.Bool()
if err != nil {
return nil, err
}
return v, nil
case simdjson.TypeObject:
obj, err := iter.Object(nil)
if err != nil {
return nil, err
}
return *obj, err
case simdjson.TypeArray:
arr, err := iter.Array(nil)
if err != nil {
return nil, err
}
iter := arr.Iter()
var dst []interface{}
var next simdjson.Iter
for {
typ, err := iter.AdvanceIter(&next)
if err != nil {
return nil, err
}
if typ == simdjson.TypeNone {
break
}
v, err := IterToValue(next)
if err != nil {
return nil, err
}
dst = append(dst, v)
}
return dst, err
case simdjson.TypeNull:
return nil, nil
}
return nil, fmt.Errorf("IterToValue: unknown JSON type: %s", iter.Type().String())
}