frostfs-http-gw/internal/handler/download.go
Denis Kirillov 9750521598 [#233] Fix browsing
Simplify tree listing (we need only nodes in exactly the same parent level)

Signed-off-by: Denis Kirillov <d.kirillov@yadro.com>
2025-04-23 13:03:58 +03:00

455 lines
14 KiB
Go

package handler
import (
"archive/tar"
"archive/zip"
"bufio"
"compress/gzip"
"context"
"errors"
"fmt"
"io"
"net/url"
"strings"
"time"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/data"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/logs"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/tree"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/utils"
"git.frostfs.info/TrueCloudLab/frostfs-observability/tracing"
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/valyala/fasthttp"
"go.uber.org/zap"
)
// DownloadByAddressOrBucketName handles download requests using simple cid/oid or bucketname/key format.
func (h *Handler) DownloadByAddressOrBucketName(req *fasthttp.RequestCtx) {
ctx, span := tracing.StartSpanFromContext(utils.GetContextFromRequest(req), "handler.DownloadByAddressOrBucketName")
defer span.End()
cidParam := req.UserValue("cid").(string)
oidParam := req.UserValue("oid").(string)
ctx = utils.SetReqLog(ctx, h.reqLogger(ctx).With(
zap.String("cid", cidParam),
zap.String("oid", oidParam),
))
path, err := url.QueryUnescape(oidParam)
if err != nil {
h.logAndSendError(ctx, req, logs.FailedToUnescapePath, err)
return
}
bktInfo, err := h.getBucketInfo(ctx, cidParam)
if err != nil {
h.logAndSendError(ctx, req, logs.FailedToGetBucketInfo, err)
return
}
checkS3Err := h.tree.CheckSettingsNodeExists(ctx, bktInfo)
if checkS3Err != nil && !errors.Is(checkS3Err, tree.ErrNodeNotFound) {
h.logAndSendError(ctx, req, logs.FailedToCheckIfSettingsNodeExist, checkS3Err)
return
}
prm := MiddlewareParam{
Context: ctx,
Request: req,
BktInfo: bktInfo,
Path: path,
}
indexPageEnabled := h.config.IndexPageEnabled()
if checkS3Err == nil {
run(prm, h.errorMiddleware(logs.ObjectNotFound, ErrObjectNotFound),
Middleware{Func: h.byS3PathMiddleware(h.receiveFile, noopFormer), Enabled: true},
Middleware{Func: h.byS3PathMiddleware(h.receiveFile, indexFormer), Enabled: indexPageEnabled},
Middleware{Func: h.browseIndexMiddleware(h.getDirObjectsS3), Enabled: indexPageEnabled},
)
} else {
slashFallbackEnabled := h.config.EnableFilepathSlashFallback()
fileNameFallbackEnabled := h.config.EnableFilepathFallback()
run(prm, h.errorMiddleware(logs.ObjectNotFound, ErrObjectNotFound),
Middleware{Func: h.byAddressMiddleware(h.receiveFile), Enabled: true},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFilePath, noopFormer), Enabled: true},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFilePath, reverseLeadingSlash), Enabled: slashFallbackEnabled},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFileName, noopFormer), Enabled: fileNameFallbackEnabled},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFileName, reverseLeadingSlash), Enabled: fileNameFallbackEnabled && slashFallbackEnabled},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFilePath, indexFormer), Enabled: indexPageEnabled},
Middleware{Func: h.byAttributeSearchMiddleware(h.receiveFile, object.AttributeFileName, indexFormer), Enabled: fileNameFallbackEnabled && indexPageEnabled},
Middleware{Func: h.browseIndexMiddleware(h.getDirObjectsNative), Enabled: indexPageEnabled},
)
}
}
type MiddlewareFunc func(param MiddlewareParam) bool
type MiddlewareParam struct {
Context context.Context
Request *fasthttp.RequestCtx
BktInfo *data.BucketInfo
Path string
}
type Middleware struct {
Func MiddlewareFunc
Enabled bool
}
func run(prm MiddlewareParam, defaultMiddleware MiddlewareFunc, middlewares ...Middleware) {
for _, m := range middlewares {
if m.Enabled && !m.Func(prm) {
return
}
}
defaultMiddleware(prm)
}
func indexFormer(path string) string {
indexPath := path
if indexPath != "" && !strings.HasSuffix(indexPath, "/") {
indexPath += "/"
}
return indexPath + "index.html"
}
func reverseLeadingSlash(path string) string {
if path == "" || path == "/" {
return path
}
if path[0] == '/' {
return path[1:]
}
return "/" + path
}
func noopFormer(path string) string {
return path
}
func (h *Handler) byS3PathMiddleware(handler func(context.Context, *fasthttp.RequestCtx, oid.Address), pathFormer func(string) string) MiddlewareFunc {
return func(prm MiddlewareParam) bool {
ctx, span := tracing.StartSpanFromContext(prm.Context, "handler.byS3Path")
defer span.End()
path := pathFormer(prm.Path)
foundOID, err := h.tree.GetLatestVersion(ctx, &prm.BktInfo.CID, path)
if err == nil {
if foundOID.IsDeleteMarker {
h.logAndSendError(ctx, prm.Request, logs.IndexWasDeleted, ErrObjectNotFound)
return false
}
addr := newAddress(prm.BktInfo.CID, foundOID.OID)
handler(ctx, prm.Request, addr)
return false
}
if !errors.Is(err, tree.ErrNodeNotFound) {
h.logAndSendError(ctx, prm.Request, logs.FailedToGetLatestVersionOfIndexObject, err, zap.String("path", path))
return false
}
return true
}
}
func (h *Handler) byAttributeSearchMiddleware(handler func(context.Context, *fasthttp.RequestCtx, oid.Address), attr string, pathFormer func(string) string) MiddlewareFunc {
return func(prm MiddlewareParam) bool {
ctx, span := tracing.StartSpanFromContext(prm.Context, "handler.byAttributeSearch")
defer span.End()
path := pathFormer(prm.Path)
res, err := h.search(ctx, prm.BktInfo.CID, attr, path, object.MatchStringEqual)
if err != nil {
h.logAndSendError(ctx, prm.Request, logs.FailedToFindObjectByAttribute, err)
return false
}
defer res.Close()
buf := make([]oid.ID, 1)
n, err := res.Read(buf)
if err == nil && n > 0 {
addr := newAddress(prm.BktInfo.CID, buf[0])
handler(ctx, prm.Request, addr)
return false
}
if !errors.Is(err, io.EOF) {
h.logAndSendError(ctx, prm.Request, logs.FailedToFindObjectByAttribute, err)
return false
}
return true
}
}
func (h *Handler) byAddressMiddleware(handler func(context.Context, *fasthttp.RequestCtx, oid.Address)) MiddlewareFunc {
return func(prm MiddlewareParam) bool {
ctx, span := tracing.StartSpanFromContext(prm.Context, "handler.byAddress")
defer span.End()
var objID oid.ID
if objID.DecodeString(prm.Path) == nil {
handler(ctx, prm.Request, newAddress(prm.BktInfo.CID, objID))
return false
}
return true
}
}
// DownloadByAttribute handles attribute-based download requests.
func (h *Handler) DownloadByAttribute(req *fasthttp.RequestCtx) {
ctx, span := tracing.StartSpanFromContext(utils.GetContextFromRequest(req), "handler.DownloadByAttribute")
defer span.End()
h.byAttribute(ctx, req, h.receiveFile)
}
func (h *Handler) search(ctx context.Context, cnrID cid.ID, key, val string, op object.SearchMatchType) (ResObjectSearch, error) {
filters := object.NewSearchFilters()
filters.AddRootFilter()
filters.AddFilter(key, val, op)
prm := PrmObjectSearch{
PrmAuth: PrmAuth{
BearerToken: bearerToken(ctx),
},
Container: cnrID,
Filters: filters,
}
return h.frostfs.SearchObjects(ctx, prm)
}
// DownloadZip handles zip by prefix requests.
func (h *Handler) DownloadZip(req *fasthttp.RequestCtx) {
ctx, span := tracing.StartSpanFromContext(utils.GetContextFromRequest(req), "handler.DownloadZip")
defer span.End()
scid, _ := req.UserValue("cid").(string)
prefix, _ := req.UserValue("prefix").(string)
ctx = utils.SetReqLog(ctx, h.reqLogger(ctx).With(zap.String("cid", scid), zap.String("prefix", prefix)))
bktInfo, err := h.getBucketInfo(ctx, scid)
if err != nil {
h.logAndSendError(ctx, req, logs.FailedToGetBucketInfo, err)
return
}
resSearch, err := h.searchObjectsByPrefix(ctx, bktInfo.CID, prefix)
if err != nil {
return
}
req.Response.Header.Set(fasthttp.HeaderContentType, "application/zip")
req.Response.Header.Set(fasthttp.HeaderContentDisposition, "attachment; filename=\"archive.zip\"")
req.SetBodyStreamWriter(h.getZipResponseWriter(ctx, resSearch, bktInfo))
}
func (h *Handler) getZipResponseWriter(ctx context.Context, resSearch ResObjectSearch, bktInfo *data.BucketInfo) func(w *bufio.Writer) {
return func(w *bufio.Writer) {
defer resSearch.Close()
buf := make([]byte, 3<<20)
zipWriter := zip.NewWriter(w)
var objectsWritten int
errIter := resSearch.Iterate(h.putObjectToArchive(ctx, bktInfo.CID, buf,
func(obj *object.Object) (io.Writer, error) {
objectsWritten++
return h.createZipFile(zipWriter, obj)
}),
)
if errIter != nil {
h.reqLogger(ctx).Error(logs.IteratingOverSelectedObjectsFailed, zap.Error(errIter), logs.TagField(logs.TagDatapath))
return
} else if objectsWritten == 0 {
h.reqLogger(ctx).Warn(logs.ObjectsNotFound, logs.TagField(logs.TagDatapath))
}
if err := zipWriter.Close(); err != nil {
h.reqLogger(ctx).Error(logs.CloseZipWriter, zap.Error(err), logs.TagField(logs.TagDatapath))
}
}
}
func (h *Handler) createZipFile(zw *zip.Writer, obj *object.Object) (io.Writer, error) {
method := zip.Store
if h.config.ArchiveCompression() {
method = zip.Deflate
}
filePath := getFilePath(obj)
if len(filePath) == 0 || filePath[len(filePath)-1] == '/' {
return nil, fmt.Errorf("invalid filepath '%s'", filePath)
}
return zw.CreateHeader(&zip.FileHeader{
Name: filePath,
Method: method,
Modified: time.Now(),
})
}
// DownloadTar forms tar.gz from objects by prefix.
func (h *Handler) DownloadTar(req *fasthttp.RequestCtx) {
ctx, span := tracing.StartSpanFromContext(utils.GetContextFromRequest(req), "handler.DownloadTar")
defer span.End()
scid, _ := req.UserValue("cid").(string)
prefix, _ := req.UserValue("prefix").(string)
ctx = utils.SetReqLog(ctx, h.reqLogger(ctx).With(zap.String("cid", scid), zap.String("prefix", prefix)))
bktInfo, err := h.getBucketInfo(ctx, scid)
if err != nil {
h.logAndSendError(ctx, req, logs.FailedToGetBucketInfo, err)
return
}
resSearch, err := h.searchObjectsByPrefix(ctx, bktInfo.CID, prefix)
if err != nil {
return
}
req.Response.Header.Set(fasthttp.HeaderContentType, "application/gzip")
req.Response.Header.Set(fasthttp.HeaderContentDisposition, "attachment; filename=\"archive.tar.gz\"")
req.SetBodyStreamWriter(h.getTarResponseWriter(ctx, resSearch, bktInfo))
}
func (h *Handler) getTarResponseWriter(ctx context.Context, resSearch ResObjectSearch, bktInfo *data.BucketInfo) func(w *bufio.Writer) {
return func(w *bufio.Writer) {
defer resSearch.Close()
compressionLevel := gzip.NoCompression
if h.config.ArchiveCompression() {
compressionLevel = gzip.DefaultCompression
}
// ignore error because it's not nil only if compressionLevel argument is invalid
gzipWriter, _ := gzip.NewWriterLevel(w, compressionLevel)
tarWriter := tar.NewWriter(gzipWriter)
defer func() {
if err := tarWriter.Close(); err != nil {
h.reqLogger(ctx).Error(logs.CloseTarWriter, zap.Error(err), logs.TagField(logs.TagDatapath))
}
if err := gzipWriter.Close(); err != nil {
h.reqLogger(ctx).Error(logs.CloseGzipWriter, zap.Error(err), logs.TagField(logs.TagDatapath))
}
}()
var objectsWritten int
buf := make([]byte, 3<<20) // the same as for upload
errIter := resSearch.Iterate(h.putObjectToArchive(ctx, bktInfo.CID, buf,
func(obj *object.Object) (io.Writer, error) {
objectsWritten++
return h.createTarFile(tarWriter, obj)
}),
)
if errIter != nil {
h.reqLogger(ctx).Error(logs.IteratingOverSelectedObjectsFailed, zap.Error(errIter), logs.TagField(logs.TagDatapath))
} else if objectsWritten == 0 {
h.reqLogger(ctx).Warn(logs.ObjectsNotFound, logs.TagField(logs.TagDatapath))
}
}
}
func (h *Handler) createTarFile(tw *tar.Writer, obj *object.Object) (io.Writer, error) {
filePath := getFilePath(obj)
if len(filePath) == 0 || filePath[len(filePath)-1] == '/' {
return nil, fmt.Errorf("invalid filepath '%s'", filePath)
}
return tw, tw.WriteHeader(&tar.Header{
Name: filePath,
Mode: 0655,
Size: int64(obj.PayloadSize()),
})
}
func (h *Handler) putObjectToArchive(ctx context.Context, cnrID cid.ID, buf []byte, createArchiveHeader func(obj *object.Object) (io.Writer, error)) func(id oid.ID) bool {
return func(id oid.ID) bool {
logger := h.reqLogger(ctx).With(zap.String("oid", id.EncodeToString()))
prm := PrmObjectGet{
PrmAuth: PrmAuth{
BearerToken: bearerToken(ctx),
},
Address: newAddress(cnrID, id),
}
resGet, err := h.frostfs.GetObject(ctx, prm)
if err != nil {
logger.Error(logs.FailedToGetObject, zap.Error(err), logs.TagField(logs.TagExternalStorage))
return false
}
fileWriter, err := createArchiveHeader(&resGet.Header)
if err != nil {
logger.Error(logs.FailedToAddObjectToArchive, zap.Error(err), logs.TagField(logs.TagDatapath))
return false
}
if err = writeToArchive(resGet, fileWriter, buf); err != nil {
logger.Error(logs.FailedToAddObjectToArchive, zap.Error(err), logs.TagField(logs.TagDatapath))
return false
}
return false
}
}
func (h *Handler) searchObjectsByPrefix(ctx context.Context, cnrID cid.ID, prefix string) (ResObjectSearch, error) {
prefix, err := url.QueryUnescape(prefix)
if err != nil {
return nil, fmt.Errorf("unescape prefix: %w", err)
}
resSearch, err := h.search(ctx, cnrID, object.AttributeFilePath, prefix, object.MatchCommonPrefix)
if err != nil {
return nil, fmt.Errorf("search objects by prefix: %w", err)
}
return resSearch, nil
}
func writeToArchive(resGet *Object, objWriter io.Writer, buf []byte) error {
var err error
if _, err = io.CopyBuffer(objWriter, resGet.Payload, buf); err != nil {
return fmt.Errorf("copy object payload to zip file: %v", err)
}
if err = resGet.Payload.Close(); err != nil {
return fmt.Errorf("object body close error: %w", err)
}
return nil
}
func getFilePath(obj *object.Object) string {
for _, attr := range obj.Attributes() {
if attr.Key() == object.AttributeFilePath {
return attr.Value()
}
}
return ""
}