JCS-pub/common/pkgs/ioswitch2/ops2/ec.go

package ops2

import (
	"fmt"
	"io"

	"gitlink.org.cn/cloudream/common/pkgs/future"
	"gitlink.org.cn/cloudream/common/utils/io2"
	"gitlink.org.cn/cloudream/common/utils/sync2"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ec"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ioswitch/dag"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ioswitch/exec"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ioswitch/utils"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/storage/pool"
	"gitlink.org.cn/cloudream/jcs-pub/common/pkgs/storage/types"
	jcstypes "gitlink.org.cn/cloudream/jcs-pub/common/types"
)

func init() {
	exec.UseOp[*ECMultiply]()

	exec.UseOp[*CallECMultiplier]()
}

type ECMultiply struct {
	Coef      [][]byte     `json:"coef"`
	Inputs    []exec.VarID `json:"inputs"`
	Outputs   []exec.VarID `json:"outputs"`
	ChunkSize int          `json:"chunkSize"`
}

func (o *ECMultiply) Execute(ctx *exec.ExecContext, e *exec.Executor) error {
	inputs, err := exec.BindArray[*exec.StreamValue](e, ctx.Context, o.Inputs)
	if err != nil {
		return err
	}
	defer func() {
		for _, s := range inputs {
			s.Stream.Close()
		}
	}()

	outputWrs := make([]*io.PipeWriter, len(o.Outputs))
	outputVars := make([]*exec.StreamValue, len(o.Outputs))

	for i := range o.Outputs {
		rd, wr := io.Pipe()
		outputVars[i] = &exec.StreamValue{Stream: rd}
		outputWrs[i] = wr
	}

	/// !!! 缓冲区大小必须是ChunkSize大小，因为Chunk数据很有可能来自于一个被Split的完整文件，此时必须保证按顺序读取每一个Chunk的数据 !!!

	inputChunks := make([][]byte, len(o.Inputs))
	for i := range o.Inputs {
		inputChunks[i] = make([]byte, o.ChunkSize)
	}

	// 输出用两个缓冲轮换
	outputBufPool := sync2.NewBucketPool[[][]byte]()
	for i := 0; i < 2; i++ {
		outputChunks := make([][]byte, len(o.Outputs))
		for i := range o.Outputs {
			outputChunks[i] = make([]byte, o.ChunkSize)
		}
		outputBufPool.PutEmpty(outputChunks)
	}

	fut := future.NewSetVoid()
	go func() {
		mul := ec.GaloisMultiplier().BuildGalois()
		defer outputBufPool.Close()
		for {
			err := sync2.ParallelDo(inputs, func(s *exec.StreamValue, i int) error {
				_, err := io.ReadFull(s.Stream, inputChunks[i])
				return err
			})
			if err == io.EOF {
				return
			}
			if err != nil {
				fut.SetError(err)
				return
			}

			outputBuf, ok := outputBufPool.GetEmpty()
			if !ok {
				return
			}

			err = mul.Multiply(o.Coef, inputChunks, outputBuf)
			if err != nil {
				fut.SetError(err)
				return
			}

			outputBufPool.PutFilled(outputBuf)
		}
	}()

	go func() {
		defer outputBufPool.Close()

		for {
			outputChunks, ok := outputBufPool.GetFilled()
			if !ok {
				fut.SetVoid()
				return
			}

			for i := range o.Outputs {
				err := io2.WriteAll(outputWrs[i], outputChunks[i])
				if err != nil {
					fut.SetError(err)
					return
				}
			}

			outputBufPool.PutEmpty(outputChunks)
		}
	}()

	exec.PutArray(e, o.Outputs, outputVars)

	err = fut.Wait(ctx.Context)
	if err != nil {
		for _, wr := range outputWrs {
			wr.CloseWithError(err)
		}
		return err
	}

	for _, wr := range outputWrs {
		wr.Close()
	}
	return nil
}

func (o *ECMultiply) String() string {
	return fmt.Sprintf(
		"ECMultiply(coef=%v) (%v) -> (%v)",
		o.Coef,
		utils.FormatVarIDs(o.Inputs),
		utils.FormatVarIDs(o.Outputs),
	)
}

type CallECMultiplier struct {
	UserSpace jcstypes.UserSpaceDetail
	Coef      [][]byte
	Inputs    []exec.VarID
	Outputs   []exec.VarID
	ChunkSize int
}

func (o *CallECMultiplier) Execute(ctx *exec.ExecContext, e *exec.Executor) error {
	stgPool, err := exec.GetValueByType[*pool.Pool](ctx)
	if err != nil {
		return fmt.Errorf("getting storage pool: %w", err)
	}

	ecMul, err := stgPool.GetECMultiplier(&o.UserSpace)
	if err != nil {
		return err
	}
	defer ecMul.Close()

	inputs, err := exec.BindArray[*HTTPRequestValue](e, ctx.Context, o.Inputs)
	if err != nil {
		return err
	}

	reqs := make([]types.HTTPRequest, 0, len(inputs))
	for _, input := range inputs {
		reqs = append(reqs, input.HTTPRequest)
	}

	outputs, err := ecMul.Multiply(o.Coef, reqs, o.ChunkSize)
	if err != nil {
		return err
	}

	outputVals := make([]*FileInfoValue, 0, len(outputs))
	for _, output := range outputs {
		outputVals = append(outputVals, &FileInfoValue{
			FileInfo: output,
		})
	}
	exec.PutArray(e, o.Outputs, outputVals)

	return nil
}

func (o *CallECMultiplier) String() string {
	return fmt.Sprintf(
		"CallECMultiplier(userSpace=%v, coef=%v) (%v) -> (%v)",
		o.Coef,
		o.UserSpace,
		utils.FormatVarIDs(o.Inputs),
		utils.FormatVarIDs(o.Outputs),
	)
}

type ECMultiplyNode struct {
	dag.NodeBase
	EC            jcstypes.ECRedundancy
	InputIndexes  []int
	OutputIndexes []int
}

func (b *GraphNodeBuilder) NewECMultiply(ec jcstypes.ECRedundancy) *ECMultiplyNode {
	node := &ECMultiplyNode{
		EC: ec,
	}
	b.AddNode(node)
	return node
}

func (t *ECMultiplyNode) AddInput(str *dag.StreamVar, dataIndex int) {
	t.InputIndexes = append(t.InputIndexes, dataIndex)
	idx := t.InputStreams().EnlargeOne()
	str.To(t, idx)
}

func (t *ECMultiplyNode) RemoveAllInputs() {
	t.InputStreams().ClearAllInput(t)
	t.InputStreams().Slots.Resize(0)
	t.InputIndexes = nil
}

func (t *ECMultiplyNode) NewOutput(dataIndex int) *dag.StreamVar {
	t.OutputIndexes = append(t.OutputIndexes, dataIndex)
	return t.OutputStreams().AppendNew(t).Var()
}

func (t *ECMultiplyNode) GenerateOp() (exec.Op, error) {
	rs, err := ec.NewRs(t.EC.K, t.EC.N)
	if err != nil {
		return nil, err
	}
	coef, err := rs.GenerateMatrix(t.InputIndexes, t.OutputIndexes)
	if err != nil {
		return nil, err
	}

	return &ECMultiply{
		Coef:      coef,
		Inputs:    t.InputStreams().GetVarIDs(),
		Outputs:   t.OutputStreams().GetVarIDs(),
		ChunkSize: t.EC.ChunkSize,
	}, nil
}

// func (t *MultiplyType) String() string {
// 	return fmt.Sprintf("Multiply[]%v%v", formatStreamIO(node), formatValueIO(node))
// }

type CallECMultiplierNode struct {
	dag.NodeBase
	UserSpace     jcstypes.UserSpaceDetail
	EC            jcstypes.ECRedundancy
	InputIndexes  []int
	OutputIndexes []int
}

func (b *GraphNodeBuilder) NewCallECMultiplier(userSpace jcstypes.UserSpaceDetail) *CallECMultiplierNode {
	node := &CallECMultiplierNode{
		UserSpace: userSpace,
	}
	b.AddNode(node)
	return node
}

func (t *CallECMultiplierNode) InitFrom(node *ECMultiplyNode) {
	t.EC = node.EC
	t.InputIndexes = node.InputIndexes
	t.OutputIndexes = node.OutputIndexes

	t.InputValues().Init(len(t.InputIndexes))
	t.OutputValues().Init(t, len(t.OutputIndexes))
}

func (t *CallECMultiplierNode) HTTPRequestSlot(idx int) dag.ValueInputSlot {
	return dag.ValueInputSlot{
		Node:  t,
		Index: idx,
	}
}

func (t *CallECMultiplierNode) FileInfoVar(idx int) dag.ValueOutputSlot {
	return dag.ValueOutputSlot{
		Node:  t,
		Index: idx,
	}
}

func (t *CallECMultiplierNode) GenerateOp() (exec.Op, error) {
	rs, err := ec.NewRs(t.EC.K, t.EC.N)
	if err != nil {
		return nil, err
	}
	coef, err := rs.GenerateMatrix(t.InputIndexes, t.OutputIndexes)
	if err != nil {
		return nil, err
	}

	return &CallECMultiplier{
		UserSpace: t.UserSpace,
		Coef:      coef,
		Inputs:    t.InputValues().GetVarIDs(),
		Outputs:   t.OutputValues().GetVarIDs(),
		ChunkSize: t.EC.ChunkSize,
	}, nil
}