frostfs-http-gw/internal/handler/download.go

package handler

import (
	"archive/zip"
	"bufio"
	"context"
	"fmt"
	"io"
	"net/http"
	"net/url"
	"time"

	"git.frostfs.info/TrueCloudLab/frostfs-http-gw/internal/logs"
	"git.frostfs.info/TrueCloudLab/frostfs-http-gw/response"
	"git.frostfs.info/TrueCloudLab/frostfs-http-gw/utils"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/bearer"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/client"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/container"
	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"
	"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/pool"
	"github.com/valyala/fasthttp"
	"go.uber.org/zap"
)

// DownloadByAddressOrBucketName handles download requests using simple cid/oid or bucketname/key format.
func (h *Handler) DownloadByAddressOrBucketName(c *fasthttp.RequestCtx) {
	test, _ := c.UserValue("oid").(string)
	var id oid.ID
	err := id.DecodeString(test)
	if err != nil {
		h.byBucketname(c, h.receiveFile)
	} else {
		h.byAddress(c, h.receiveFile)
	}
}

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)
}

func (h *Handler) search(ctx context.Context, cid *cid.ID, key, val string, op object.SearchMatchType) (pool.ResObjectSearch, error) {
	filters := object.NewSearchFilters()
	filters.AddRootFilter()
	filters.AddFilter(key, val, op)

	var prm pool.PrmObjectSearch
	prm.SetContainerID(*cid)
	prm.SetFilters(filters)
	if btoken := bearerToken(ctx); btoken != nil {
		prm.UseBearer(*btoken)
	}

	return h.pool.SearchObjects(ctx, prm)
}

func (h *Handler) getContainer(ctx context.Context, cnrID cid.ID) (container.Container, error) {
	var prm pool.PrmContainerGet
	prm.SetContainerID(cnrID)

	return h.pool.GetContainer(ctx, prm)
}

func (h *Handler) addObjectToZip(zw *zip.Writer, obj *object.Object) (io.Writer, error) {
	method := zip.Store
	if h.config.ZipCompression() {
		method = zip.Deflate
	}

	filePath := getZipFilePath(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(),
	})
}

// DownloadZipped handles zip by prefix requests.
func (h *Handler) DownloadZipped(c *fasthttp.RequestCtx) {
	scid, _ := c.UserValue("cid").(string)
	prefix, _ := url.QueryUnescape(c.UserValue("prefix").(string))
	log := h.log.With(zap.String("cid", scid), zap.String("prefix", prefix))

	ctx := utils.GetContextFromRequest(c)

	containerID, err := h.getContainerID(ctx, scid)
	if err != nil {
		log.Error(logs.WrongContainerID, zap.Error(err))
		response.Error(c, "wrong container id", fasthttp.StatusBadRequest)
		return
	}

	// check if container exists here to be able to return 404 error,
	// otherwise we get this error only in object iteration step
	// and client get 200 OK.
	if _, err = h.getContainer(ctx, *containerID); err != nil {
		log.Error(logs.CouldNotCheckContainerExistence, zap.Error(err))
		if client.IsErrContainerNotFound(err) {
			response.Error(c, "Not Found", fasthttp.StatusNotFound)
			return
		}
		response.Error(c, "could not check container existence: "+err.Error(), fasthttp.StatusBadRequest)
		return
	}

	resSearch, err := h.search(ctx, containerID, object.AttributeFilePath, prefix, object.MatchCommonPrefix)
	if err != nil {
		log.Error(logs.CouldNotSearchForObjects, zap.Error(err))
		response.Error(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.SetBodyStreamWriter(func(w *bufio.Writer) {
		defer resSearch.Close()

		zipWriter := zip.NewWriter(w)

		var bufZip []byte
		var addr oid.Address

		empty := true
		called := false
		btoken := bearerToken(ctx)
		addr.SetContainer(*containerID)

		errIter := resSearch.Iterate(func(id oid.ID) bool {
			called = true

			if empty {
				bufZip = make([]byte, 3<<20) // the same as for upload
			}
			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))
			}

			return false
		})
		if errIter != nil {
			log.Error(logs.IteratingOverSelectedObjectsFailed, zap.Error(errIter))
		} else if !called {
			log.Error(logs.ObjectsNotFound)
		}

		if err = zipWriter.Close(); err != nil {
			log.Error(logs.CloseZipWriter, zap.Error(err))
		}
	})
}

func (h *Handler) zipObject(ctx context.Context, zipWriter *zip.Writer, addr oid.Address, btoken *bearer.Token, bufZip []byte) error {
	var prm pool.PrmObjectGet
	prm.SetAddress(addr)
	if btoken != nil {
		prm.UseBearer(*btoken)
	}

	resGet, err := h.pool.GetObject(ctx, prm)
	if err != nil {
		return fmt.Errorf("get FrostFS object: %v", err)
	}

	objWriter, err := h.addObjectToZip(zipWriter, &resGet.Header)
	if err != nil {
		return fmt.Errorf("zip create header: %v", err)
	}

	if _, err = io.CopyBuffer(objWriter, resGet.Payload, bufZip); 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)
	}

	if err = zipWriter.Flush(); err != nil {
		return fmt.Errorf("flush zip writer: %v", err)
	}

	return nil
}

func getZipFilePath(obj *object.Object) string {
	for _, attr := range obj.Attributes() {
		if attr.Key() == object.AttributeFilePath {
			return attr.Value()
		}
	}

	return ""
}