[#170] Support tar.gz downloading

Split DownloadZip handler on methods. Add handler DownloadTar for downloading tar.gz archives. Make methods more universal for using in both implementations

Signed-off-by: Nikita Zinkevich <n.zinkevich@yadro.com>
This commit is contained in:
Nikita Zinkevich 2024-12-13 16:00:31 +03:00
parent a7617514d3
commit 7901d00924
Signed by: nzinkevich
GPG key ID: 748EA1D0B2E6420A
12 changed files with 214 additions and 94 deletions

View file

@ -1,20 +1,21 @@
package handler
import (
"archive/tar"
"archive/zip"
"bufio"
"compress/gzip"
"context"
"errors"
"fmt"
"io"
"net/http"
"net/url"
"time"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/data"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/layer"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/logs"
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/utils"
"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/bearer"
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"
@ -46,7 +47,7 @@ func (h *Handler) DownloadByAddressOrBucketName(c *fasthttp.RequestCtx) {
return
}
req := h.newRequest(c, log)
req := newRequest(c, log)
var objID oid.ID
if checkS3Err == nil && shouldDownload(oidParam, downloadParam) {
@ -62,13 +63,6 @@ func shouldDownload(oidParam string, downloadParam bool) bool {
return !isDir(oidParam) || downloadParam
}
func (h *Handler) newRequest(ctx *fasthttp.RequestCtx, log *zap.Logger) request {
return request{
RequestCtx: ctx,
log: log,
}
}
// DownloadByAttribute handles attribute-based download requests.
func (h *Handler) DownloadByAttribute(c *fasthttp.RequestCtx) {
h.byAttribute(c, h.receiveFile)
@ -90,13 +84,61 @@ func (h *Handler) search(ctx context.Context, cnrID cid.ID, key, val string, op
return h.frostfs.SearchObjects(ctx, prm)
}
func (h *Handler) addObjectToZip(zw *zip.Writer, obj *object.Object) (io.Writer, error) {
// DownloadZip handles zip by prefix requests.
func (h *Handler) DownloadZip(c *fasthttp.RequestCtx) {
scid, _ := c.UserValue("cid").(string)
ctx := utils.GetContextFromRequest(c)
log := utils.GetReqLogOrDefault(ctx, h.log)
bktInfo, err := h.getBucketInfo(ctx, scid, log)
if err != nil {
logAndSendBucketError(c, log, err)
return
}
resSearch, err := h.searchObjectsByPrefix(c, log, bktInfo.CID)
if err != nil {
return
}
c.Response.Header.Set(fasthttp.HeaderContentType, "application/zip")
c.Response.Header.Set(fasthttp.HeaderContentDisposition, "attachment; filename=\"archive.zip\"")
c.SetBodyStreamWriter(h.getZipResponseWriter(ctx, log, resSearch, bktInfo))
}
func (h *Handler) getZipResponseWriter(ctx context.Context, log *zap.Logger, 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, log, bktInfo.CID, buf,
func(obj *object.Object) (io.Writer, error) {
objectsWritten++
return h.createZipFile(zipWriter, obj)
}),
)
if errIter != nil {
log.Error(logs.IteratingOverSelectedObjectsFailed, zap.Error(errIter))
return
} else if objectsWritten == 0 {
log.Warn(logs.ObjectsNotFound)
}
if err := zipWriter.Close(); err != nil {
log.Error(logs.CloseZipWriter, zap.Error(err))
}
}
}
func (h *Handler) createZipFile(zw *zip.Writer, obj *object.Object) (io.Writer, error) {
method := zip.Store
if h.config.ZipCompression() {
method = zip.Deflate
}
filePath := getZipFilePath(obj)
filePath := getFilePath(obj)
if len(filePath) == 0 || filePath[len(filePath)-1] == '/' {
return nil, fmt.Errorf("invalid filepath '%s'", filePath)
}
@ -108,99 +150,139 @@ func (h *Handler) addObjectToZip(zw *zip.Writer, obj *object.Object) (io.Writer,
})
}
// DownloadZipped handles zip by prefix requests.
func (h *Handler) DownloadZipped(c *fasthttp.RequestCtx) {
// DownloadTar forms tar.gz from objects by prefix.
func (h *Handler) DownloadTar(c *fasthttp.RequestCtx) {
scid, _ := c.UserValue("cid").(string)
prefix, _ := c.UserValue("prefix").(string)
ctx := utils.GetContextFromRequest(c)
log := utils.GetReqLogOrDefault(ctx, h.log)
prefix, err := url.QueryUnescape(prefix)
if err != nil {
log.Error(logs.FailedToUnescapeQuery, zap.String("cid", scid), zap.String("prefix", prefix), zap.Error(err))
ResponseError(c, "could not unescape prefix: "+err.Error(), fasthttp.StatusBadRequest)
return
}
log = log.With(zap.String("cid", scid), zap.String("prefix", prefix))
bktInfo, err := h.getBucketInfo(ctx, scid, log)
if err != nil {
logAndSendBucketError(c, log, err)
return
}
resSearch, err := h.search(ctx, bktInfo.CID, object.AttributeFilePath, prefix, object.MatchCommonPrefix)
resSearch, err := h.searchObjectsByPrefix(c, log, bktInfo.CID)
if err != nil {
log.Error(logs.CouldNotSearchForObjects, zap.Error(err))
ResponseError(c, "could not search for objects: "+err.Error(), fasthttp.StatusBadRequest)
return
}
c.Response.Header.Set(fasthttp.HeaderContentType, "application/zip")
c.Response.Header.Set(fasthttp.HeaderContentDisposition, "attachment; filename=\"archive.zip\"")
c.Response.SetStatusCode(http.StatusOK)
c.Response.Header.Set(fasthttp.HeaderContentType, "application/gzip")
c.Response.Header.Set(fasthttp.HeaderContentDisposition, "attachment; filename=\"archive.tar.gz\"")
c.SetBodyStreamWriter(func(w *bufio.Writer) {
c.SetBodyStreamWriter(h.getTarResponseWriter(ctx, log, resSearch, bktInfo))
}
func (h *Handler) getTarResponseWriter(ctx context.Context, log *zap.Logger, resSearch ResObjectSearch, bktInfo *data.BucketInfo) func(w *bufio.Writer) {
return func(w *bufio.Writer) {
defer resSearch.Close()
zipWriter := zip.NewWriter(w)
compressionLevel := gzip.NoCompression
if h.config.ZipCompression() {
compressionLevel = gzip.DefaultCompression
}
var bufZip []byte
var addr oid.Address
// ignore error because it's not nil only if compressionLevel argument is invalid
gzipWriter, _ := gzip.NewWriterLevel(w, compressionLevel)
tarWriter := tar.NewWriter(gzipWriter)
empty := true
called := false
btoken := bearerToken(ctx)
addr.SetContainer(bktInfo.CID)
errIter := resSearch.Iterate(func(id oid.ID) bool {
called = true
if empty {
bufZip = make([]byte, 3<<20) // the same as for upload
defer func() {
if err := tarWriter.Close(); err != nil {
log.Error(logs.CloseTarWriter, zap.Error(err))
}
empty = false
addr.SetObject(id)
if err = h.zipObject(ctx, zipWriter, addr, btoken, bufZip); err != nil {
log.Error(logs.FailedToAddObjectToArchive, zap.String("oid", id.EncodeToString()), zap.Error(err))
if err := gzipWriter.Close(); err != nil {
log.Error(logs.CloseGzipWriter, zap.Error(err))
}
}()
return false
})
var objectsWritten int
buf := make([]byte, 3<<20) // the same as for upload
errIter := resSearch.Iterate(h.putObjectToArchive(ctx, log, bktInfo.CID, buf,
func(obj *object.Object) (io.Writer, error) {
objectsWritten++
return h.createTarFile(tarWriter, obj)
}),
)
if errIter != nil {
log.Error(logs.IteratingOverSelectedObjectsFailed, zap.Error(errIter))
} else if !called {
log.Error(logs.ObjectsNotFound)
} else if objectsWritten == 0 {
log.Warn(logs.ObjectsNotFound)
}
}
}
if err = zipWriter.Close(); err != nil {
log.Error(logs.CloseZipWriter, zap.Error(err))
}
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) zipObject(ctx context.Context, zipWriter *zip.Writer, addr oid.Address, btoken *bearer.Token, bufZip []byte) error {
prm := PrmObjectGet{
PrmAuth: PrmAuth{
BearerToken: btoken,
},
Address: addr,
}
func (h *Handler) putObjectToArchive(ctx context.Context, log *zap.Logger, cnrID cid.ID, buf []byte, createArchiveHeader func(obj *object.Object) (io.Writer, error)) func(id oid.ID) bool {
return func(id oid.ID) bool {
log = log.With(zap.String("oid", id.EncodeToString()))
resGet, err := h.frostfs.GetObject(ctx, prm)
prm := PrmObjectGet{
PrmAuth: PrmAuth{
BearerToken: bearerToken(ctx),
},
Address: newAddress(cnrID, id),
}
resGet, err := h.frostfs.GetObject(ctx, prm)
if err != nil {
log.Error(logs.FailedToGetObject, zap.Error(err))
return false
}
fileWriter, err := createArchiveHeader(&resGet.Header)
if err != nil {
log.Error(logs.FailedToAddObjectToArchive, zap.Error(err))
return false
}
if err = writeToArchive(resGet, fileWriter, buf); err != nil {
log.Error(logs.FailedToAddObjectToArchive, zap.Error(err))
return false
}
return false
}
}
func (h *Handler) searchObjectsByPrefix(c *fasthttp.RequestCtx, log *zap.Logger, cnrID cid.ID) (ResObjectSearch, error) {
scid := cnrID.EncodeToString()
prefix, _ := c.UserValue("prefix").(string)
ctx := utils.GetContextFromRequest(c)
prefix, err := url.QueryUnescape(prefix)
if err != nil {
return fmt.Errorf("get FrostFS object: %v", err)
log.Error(logs.FailedToUnescapeQuery, zap.String("cid", scid), zap.String("prefix", prefix), zap.Error(err))
ResponseError(c, "could not unescape prefix: "+err.Error(), fasthttp.StatusBadRequest)
return nil, err
}
objWriter, err := h.addObjectToZip(zipWriter, &resGet.Header)
log = log.With(zap.String("cid", scid), zap.String("prefix", prefix))
resSearch, err := h.search(ctx, cnrID, object.AttributeFilePath, prefix, object.MatchCommonPrefix)
if err != nil {
return fmt.Errorf("zip create header: %v", err)
log.Error(logs.CouldNotSearchForObjects, zap.Error(err))
ResponseError(c, "could not search for objects: "+err.Error(), fasthttp.StatusBadRequest)
return nil, err
}
return resSearch, nil
}
if _, err = io.CopyBuffer(objWriter, resGet.Payload, bufZip); err != 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)
}
@ -208,14 +290,10 @@ func (h *Handler) zipObject(ctx context.Context, zipWriter *zip.Writer, addr oid
return fmt.Errorf("object body close error: %w", err)
}
if err = zipWriter.Flush(); err != nil {
return fmt.Errorf("flush zip writer: %v", err)
}
return nil
}
func getZipFilePath(obj *object.Object) string {
func getFilePath(obj *object.Object) string {
for _, attr := range obj.Attributes() {
if attr.Key() == object.AttributeFilePath {
return attr.Value()