frostfs-s3-gw/api/layer/object.go

package layer

import (
	"bytes"
	"context"
	"crypto/md5"
	"crypto/sha256"
	"encoding/base64"
	"encoding/hex"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"mime"
	"path/filepath"
	"strconv"
	"strings"

	"git.frostfs.info/TrueCloudLab/frostfs-s3-gw/api"
	"git.frostfs.info/TrueCloudLab/frostfs-s3-gw/api/auth"
	"git.frostfs.info/TrueCloudLab/frostfs-s3-gw/api/data"
	apiErrors "git.frostfs.info/TrueCloudLab/frostfs-s3-gw/api/errors"
	"git.frostfs.info/TrueCloudLab/frostfs-s3-gw/internal/logs"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/client"
	cid "git.frostfs.info/TrueCloudLab/frostfs-sdk-go/container/id"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/object"
	oid "git.frostfs.info/TrueCloudLab/frostfs-sdk-go/object/id"
	"github.com/minio/sio"
	"go.uber.org/zap"
)

type (
	getParams struct {
		// payload range
		off, ln uint64

		objInfo *data.ObjectInfo
		bktInfo *data.BucketInfo
	}

	getFrostFSParams struct {
		// payload range
		off, ln uint64

		oid     oid.ID
		bktInfo *data.BucketInfo
	}

	DeleteMarkerError struct {
		ErrorCode apiErrors.ErrorCode
	}
)

func (e DeleteMarkerError) Error() string {
	return "object is delete marker"
}

const (
	continuationToken = "<continuation-token>"
)

func newAddress(cnr cid.ID, obj oid.ID) oid.Address {
	var addr oid.Address
	addr.SetContainer(cnr)
	addr.SetObject(obj)
	return addr
}

// objectHead returns all object's headers.
func (n *Layer) objectHead(ctx context.Context, bktInfo *data.BucketInfo, idObj oid.ID) (*object.Object, error) {
	prm := PrmObjectRead{
		Container:  bktInfo.CID,
		Object:     idObj,
		WithHeader: true,
	}

	n.prepareAuthParameters(ctx, &prm.PrmAuth, bktInfo.Owner)

	res, err := n.frostFS.ReadObject(ctx, prm)
	if err != nil {
		return nil, err
	}

	return res.Head, nil
}

func (n *Layer) initObjectPayloadReader(ctx context.Context, p getParams) (io.Reader, error) {
	if _, isCombined := p.objInfo.Headers[MultipartObjectSize]; !isCombined {
		return n.initFrostFSObjectPayloadReader(ctx, getFrostFSParams{
			off:     p.off,
			ln:      p.ln,
			oid:     p.objInfo.ID,
			bktInfo: p.bktInfo,
		})
	}

	combinedObj, err := n.objectGet(ctx, p.bktInfo, p.objInfo.ID)
	if err != nil {
		return nil, fmt.Errorf("get combined object '%s': %w", p.objInfo.ID.EncodeToString(), err)
	}

	var parts []*data.PartInfo
	if err = json.Unmarshal(combinedObj.Payload(), &parts); err != nil {
		return nil, fmt.Errorf("unmarshal combined object parts: %w", err)
	}

	isEncrypted := FormEncryptionInfo(p.objInfo.Headers).Enabled
	objParts := make([]partObj, len(parts))
	for i, part := range parts {
		size := part.Size
		if isEncrypted {
			if size, err = sio.EncryptedSize(part.Size); err != nil {
				return nil, fmt.Errorf("compute encrypted size: %w", err)
			}
		}

		objParts[i] = partObj{
			OID:  part.OID,
			Size: size,
		}
	}

	return newMultiObjectReader(ctx, multiObjectReaderConfig{
		layer:   n,
		off:     p.off,
		ln:      p.ln,
		parts:   objParts,
		bktInfo: p.bktInfo,
	})
}

// initializes payload reader of the FrostFS object.
// Zero range corresponds to full payload (panics if only offset is set).
func (n *Layer) initFrostFSObjectPayloadReader(ctx context.Context, p getFrostFSParams) (io.Reader, error) {
	prm := PrmObjectRead{
		Container:    p.bktInfo.CID,
		Object:       p.oid,
		WithPayload:  true,
		PayloadRange: [2]uint64{p.off, p.ln},
	}

	n.prepareAuthParameters(ctx, &prm.PrmAuth, p.bktInfo.Owner)

	res, err := n.frostFS.ReadObject(ctx, prm)
	if err != nil {
		return nil, err
	}

	return res.Payload, nil
}

// objectGet returns an object with payload in the object.
func (n *Layer) objectGet(ctx context.Context, bktInfo *data.BucketInfo, objID oid.ID) (*object.Object, error) {
	return n.objectGetBase(ctx, bktInfo, objID, PrmAuth{})
}

// objectGetWithAuth returns an object with payload in the object. Uses provided PrmAuth.
func (n *Layer) objectGetWithAuth(ctx context.Context, bktInfo *data.BucketInfo, objID oid.ID, auth PrmAuth) (*object.Object, error) {
	return n.objectGetBase(ctx, bktInfo, objID, auth)
}

func (n *Layer) objectGetBase(ctx context.Context, bktInfo *data.BucketInfo, objID oid.ID, auth PrmAuth) (*object.Object, error) {
	prm := PrmObjectRead{
		PrmAuth:     auth,
		Container:   bktInfo.CID,
		Object:      objID,
		WithHeader:  true,
		WithPayload: true,
	}

	n.prepareAuthParameters(ctx, &prm.PrmAuth, bktInfo.Owner)

	res, err := n.frostFS.ReadObject(ctx, prm)
	if err != nil {
		return nil, err
	}

	return res.Head, nil
}

// MimeByFilePath detect mime type by file path extension.
func MimeByFilePath(path string) string {
	ext := filepath.Ext(path)
	if len(ext) == 0 {
		return ""
	}
	return mime.TypeByExtension(ext)
}

func encryptionReader(r io.Reader, size uint64, key []byte) (io.Reader, uint64, error) {
	encSize, err := sio.EncryptedSize(size)
	if err != nil {
		return nil, 0, fmt.Errorf("failed to compute enc size: %w", err)
	}

	r, err = sio.EncryptReader(r, sio.Config{MinVersion: sio.Version20, MaxVersion: sio.Version20, Key: key, CipherSuites: []byte{sio.AES_256_GCM}})
	if err != nil {
		return nil, 0, fmt.Errorf("couldn't create encrypter: %w", err)
	}

	return r, encSize, nil
}

func ParseCompletedPartHeader(hdr string) (*Part, error) {
	// partInfo[0] -- part number, partInfo[1] -- part size, partInfo[2] -- checksum
	partInfo := strings.Split(hdr, "-")
	if len(partInfo) != 3 {
		return nil, fmt.Errorf("invalid completed part header")
	}
	num, err := strconv.Atoi(partInfo[0])
	if err != nil {
		return nil, fmt.Errorf("invalid completed part number '%s': %w", partInfo[0], err)
	}
	size, err := strconv.ParseUint(partInfo[1], 10, 64)
	if err != nil {
		return nil, fmt.Errorf("invalid completed part size '%s': %w", partInfo[1], err)
	}

	return &Part{
		ETag:       partInfo[2],
		PartNumber: num,
		Size:       size,
	}, nil
}

// PutObject stores object into FrostFS, took payload from io.Reader.
func (n *Layer) PutObject(ctx context.Context, p *PutObjectParams) (*data.ExtendedObjectInfo, error) {
	bktSettings, err := n.GetBucketSettings(ctx, p.BktInfo)
	if err != nil {
		return nil, fmt.Errorf("couldn't get versioning settings object: %w", err)
	}

	r := p.Reader
	if p.Encryption.Enabled() {
		p.Header[AttributeDecryptedSize] = strconv.FormatUint(p.Size, 10)
		if err = addEncryptionHeaders(p.Header, p.Encryption); err != nil {
			return nil, fmt.Errorf("add encryption header: %w", err)
		}

		var encSize uint64
		if r, encSize, err = encryptionReader(p.Reader, p.Size, p.Encryption.Key()); err != nil {
			return nil, fmt.Errorf("create encrypter: %w", err)
		}
		p.Size = encSize
	}

	if r != nil {
		if len(p.Header[api.ContentType]) == 0 {
			if contentType := MimeByFilePath(p.Object); len(contentType) == 0 {
				d := newDetector(r)
				if contentType, err := d.Detect(); err == nil {
					p.Header[api.ContentType] = contentType
				}
				r = d.MultiReader()
			} else {
				p.Header[api.ContentType] = contentType
			}
		}
	}

	prm := PrmObjectCreate{
		Container:    p.BktInfo.CID,
		PayloadSize:  p.Size,
		Filepath:     p.Object,
		Payload:      r,
		CreationTime: TimeNow(ctx),
		CopiesNumber: p.CopiesNumbers,
	}

	prm.Attributes = make([][2]string, 0, len(p.Header))

	for k, v := range p.Header {
		prm.Attributes = append(prm.Attributes, [2]string{k, v})
	}

	size, id, hash, md5Hash, err := n.objectPutAndHash(ctx, prm, p.BktInfo)
	if err != nil {
		return nil, err
	}
	if len(p.ContentMD5) > 0 {
		headerMd5Hash, err := base64.StdEncoding.DecodeString(p.ContentMD5)
		if err != nil {
			return nil, apiErrors.GetAPIError(apiErrors.ErrInvalidDigest)
		}
		if !bytes.Equal(headerMd5Hash, md5Hash) {
			err = n.objectDelete(ctx, p.BktInfo, id)
			if err != nil {
				n.reqLogger(ctx).Debug(logs.FailedToDeleteObject, zap.Stringer("cid", p.BktInfo.CID), zap.Stringer("oid", id))
			}
			return nil, apiErrors.GetAPIError(apiErrors.ErrInvalidDigest)
		}
	}

	if !p.Encryption.Enabled() && len(p.ContentSHA256Hash) > 0 && !auth.IsStandardContentSHA256(p.ContentSHA256Hash) {
		contentHashBytes, err := hex.DecodeString(p.ContentSHA256Hash)
		if err != nil {
			return nil, apiErrors.GetAPIError(apiErrors.ErrContentSHA256Mismatch)
		}
		if !bytes.Equal(contentHashBytes, hash) {
			err = n.objectDelete(ctx, p.BktInfo, id)
			if err != nil {
				n.reqLogger(ctx).Debug(logs.FailedToDeleteObject, zap.Stringer("cid", p.BktInfo.CID), zap.Stringer("oid", id))
			}
			return nil, apiErrors.GetAPIError(apiErrors.ErrContentSHA256Mismatch)
		}
	}

	n.reqLogger(ctx).Debug(logs.PutObject, zap.Stringer("cid", p.BktInfo.CID), zap.Stringer("oid", id))
	now := TimeNow(ctx)
	newVersion := &data.NodeVersion{
		BaseNodeVersion: data.BaseNodeVersion{
			OID:      id,
			ETag:     hex.EncodeToString(hash),
			FilePath: p.Object,
			Size:     p.Size,
			Created:  &now,
			Owner:    &n.gateOwner,
		},
		IsUnversioned: !bktSettings.VersioningEnabled(),
		IsCombined:    p.Header[MultipartObjectSize] != "",
	}
	if len(p.CompleteMD5Hash) > 0 {
		newVersion.MD5 = p.CompleteMD5Hash
	} else {
		newVersion.MD5 = hex.EncodeToString(md5Hash)
	}

	if newVersion.ID, err = n.treeService.AddVersion(ctx, p.BktInfo, newVersion); err != nil {
		return nil, fmt.Errorf("couldn't add new verion to tree service: %w", err)
	}

	if p.Lock != nil && (p.Lock.Retention != nil || p.Lock.LegalHold != nil) {
		putLockInfoPrms := &PutLockInfoParams{
			ObjVersion: &data.ObjectVersion{
				BktInfo:    p.BktInfo,
				ObjectName: p.Object,
				VersionID:  id.EncodeToString(),
			},
			NewLock:       p.Lock,
			CopiesNumbers: p.CopiesNumbers,
			NodeVersion:   newVersion, // provide new version to make one less tree service call in PutLockInfo
		}

		if err = n.PutLockInfo(ctx, putLockInfoPrms); err != nil {
			return nil, err
		}
	}

	n.cache.CleanListCacheEntriesContainingObject(p.Object, p.BktInfo.CID)

	objInfo := &data.ObjectInfo{
		ID:  id,
		CID: p.BktInfo.CID,

		Owner:       n.gateOwner,
		Bucket:      p.BktInfo.Name,
		Name:        p.Object,
		Size:        size,
		Created:     prm.CreationTime,
		Headers:     p.Header,
		ContentType: p.Header[api.ContentType],
		HashSum:     newVersion.ETag,
		MD5Sum:      newVersion.MD5,
	}

	extendedObjInfo := &data.ExtendedObjectInfo{
		ObjectInfo:  objInfo,
		NodeVersion: newVersion,
	}

	n.cache.PutObjectWithName(n.BearerOwner(ctx), extendedObjInfo)

	return extendedObjInfo, nil
}

func (n *Layer) headLastVersionIfNotDeleted(ctx context.Context, bkt *data.BucketInfo, objectName string) (*data.ExtendedObjectInfo, error) {
	owner := n.BearerOwner(ctx)
	if extObjInfo := n.cache.GetLastObject(owner, bkt.Name, objectName); extObjInfo != nil {
		return extObjInfo, nil
	}

	node, err := n.treeService.GetLatestVersion(ctx, bkt, objectName)
	if err != nil {
		if errors.Is(err, ErrNodeNotFound) {
			return nil, fmt.Errorf("%w: %s", apiErrors.GetAPIError(apiErrors.ErrNoSuchKey), err.Error())
		}
		return nil, err
	}

	if node.IsDeleteMarker {
		return nil, DeleteMarkerError{ErrorCode: apiErrors.ErrNoSuchKey}
	}

	meta, err := n.objectHead(ctx, bkt, node.OID)
	if err != nil {
		if client.IsErrObjectNotFound(err) {
			return nil, fmt.Errorf("%w: %s; %s", apiErrors.GetAPIError(apiErrors.ErrNoSuchKey), err.Error(), node.OID.EncodeToString())
		}
		return nil, err
	}
	objInfo := objectInfoFromMeta(bkt, meta)
	objInfo.MD5Sum = node.MD5

	extObjInfo := &data.ExtendedObjectInfo{
		ObjectInfo:  objInfo,
		NodeVersion: node,
	}

	n.cache.PutObjectWithName(owner, extObjInfo)

	return extObjInfo, nil
}

func (n *Layer) headVersion(ctx context.Context, bkt *data.BucketInfo, p *HeadObjectParams) (*data.ExtendedObjectInfo, error) {
	var err error
	var foundVersion *data.NodeVersion
	if p.VersionID == data.UnversionedObjectVersionID {
		foundVersion, err = n.treeService.GetUnversioned(ctx, bkt, p.Object)
		if err != nil {
			if errors.Is(err, ErrNodeNotFound) {
				return nil, fmt.Errorf("%w: %s", apiErrors.GetAPIError(apiErrors.ErrNoSuchVersion), err.Error())
			}
			return nil, err
		}
	} else {
		versions, err := n.treeService.GetVersions(ctx, bkt, p.Object)
		if err != nil {
			return nil, fmt.Errorf("couldn't get versions: %w", err)
		}

		for _, version := range versions {
			if version.OID.EncodeToString() == p.VersionID {
				foundVersion = version
				break
			}
		}
		if foundVersion == nil {
			return nil, fmt.Errorf("%w: there isn't tree node with requested version id", apiErrors.GetAPIError(apiErrors.ErrNoSuchVersion))
		}
	}

	owner := n.BearerOwner(ctx)
	if extObjInfo := n.cache.GetObject(owner, newAddress(bkt.CID, foundVersion.OID)); extObjInfo != nil {
		return extObjInfo, nil
	}

	if foundVersion.IsDeleteMarker {
		return nil, DeleteMarkerError{ErrorCode: apiErrors.ErrMethodNotAllowed}
	}

	meta, err := n.objectHead(ctx, bkt, foundVersion.OID)
	if err != nil {
		if client.IsErrObjectNotFound(err) {
			return nil, fmt.Errorf("%w: %s", apiErrors.GetAPIError(apiErrors.ErrNoSuchVersion), err.Error())
		}
		return nil, err
	}
	objInfo := objectInfoFromMeta(bkt, meta)
	objInfo.MD5Sum = foundVersion.MD5

	extObjInfo := &data.ExtendedObjectInfo{
		ObjectInfo:  objInfo,
		NodeVersion: foundVersion,
	}

	n.cache.PutObject(owner, extObjInfo)

	return extObjInfo, nil
}

// objectDelete puts tombstone object into frostfs.
func (n *Layer) objectDelete(ctx context.Context, bktInfo *data.BucketInfo, idObj oid.ID) error {
	return n.objectDeleteBase(ctx, bktInfo, idObj, PrmAuth{})
}

// objectDeleteWithAuth puts tombstone object into frostfs. Uses provided PrmAuth.
func (n *Layer) objectDeleteWithAuth(ctx context.Context, bktInfo *data.BucketInfo, idObj oid.ID, auth PrmAuth) error {
	return n.objectDeleteBase(ctx, bktInfo, idObj, auth)
}

func (n *Layer) objectDeleteBase(ctx context.Context, bktInfo *data.BucketInfo, idObj oid.ID, auth PrmAuth) error {
	prm := PrmObjectDelete{
		PrmAuth:   auth,
		Container: bktInfo.CID,
		Object:    idObj,
	}

	n.prepareAuthParameters(ctx, &prm.PrmAuth, bktInfo.Owner)

	n.cache.DeleteObject(newAddress(bktInfo.CID, idObj))

	return n.frostFS.DeleteObject(ctx, prm)
}

// objectPutAndHash prepare auth parameters and invoke frostfs.CreateObject.
// Returns object ID and payload sha256 hash.
func (n *Layer) objectPutAndHash(ctx context.Context, prm PrmObjectCreate, bktInfo *data.BucketInfo) (uint64, oid.ID, []byte, []byte, error) {
	n.prepareAuthParameters(ctx, &prm.PrmAuth, bktInfo.Owner)
	prm.ClientCut = n.features.ClientCut()
	prm.BufferMaxSize = n.features.BufferMaxSizeForPut()
	prm.WithoutHomomorphicHash = bktInfo.HomomorphicHashDisabled
	var size uint64
	hash := sha256.New()
	md5Hash := md5.New()
	prm.Payload = wrapReader(prm.Payload, 64*1024, func(buf []byte) {
		size += uint64(len(buf))
		hash.Write(buf)
		md5Hash.Write(buf)
	})
	id, err := n.frostFS.CreateObject(ctx, prm)
	if err != nil {
		if _, errDiscard := io.Copy(io.Discard, prm.Payload); errDiscard != nil {
			n.reqLogger(ctx).Warn(logs.FailedToDiscardPutPayloadProbablyGoroutineLeaks, zap.Error(errDiscard))
		}

		return 0, oid.ID{}, nil, nil, err
	}
	return size, id, hash.Sum(nil), md5Hash.Sum(nil), nil
}

type logWrapper struct {
	log *zap.Logger
}

func (l *logWrapper) Printf(format string, args ...interface{}) {
	l.log.Info(fmt.Sprintf(format, args...))
}

func IsSystemHeader(key string) bool {
	_, ok := api.SystemMetadata[key]
	return ok || strings.HasPrefix(key, api.FrostFSSystemMetadataPrefix)
}

func wrapReader(input io.Reader, bufSize int, f func(buf []byte)) io.Reader {
	if input == nil {
		return nil
	}

	r, w := io.Pipe()
	go func() {
		var buf = make([]byte, bufSize)
		for {
			n, err := input.Read(buf)
			if n > 0 {
				f(buf[:n])
				_, _ = w.Write(buf[:n]) // ignore error, input is not ReadCloser
			}
			if err != nil {
				_ = w.CloseWithError(err)
				break
			}
		}
	}()
	return r
}