[#151] index page: Add browse via native protocol
All checks were successful
/ DCO (pull_request) Successful in 1m14s
/ Vulncheck (pull_request) Successful in 1m32s
/ Builds (pull_request) Successful in 1m48s
/ Lint (pull_request) Successful in 2m8s
/ Tests (pull_request) Successful in 1m13s

Signed-off-by: Nikita Zinkevich <n.zinkevich@yadro.com>
This commit is contained in:
Nikita Zinkevich 2024-10-10 11:59:53 +03:00
parent 8fe8f2dcc2
commit 4fe34909aa
10 changed files with 359 additions and 111 deletions

View file

@ -219,20 +219,23 @@ func (a *app) loadIndexPageTemplate() {
if !a.settings.IndexPageEnabled() {
return
}
reader, err := os.Open(a.cfg.GetString(cfgIndexPageTemplatePath))
path := a.cfg.GetString(cfgIndexPageTemplatePath)
tmpl, err := a.readTemplate(path)
if err != nil {
a.settings.setIndexTemplate("")
a.log.Warn(logs.FailedToReadIndexPageTemplate, zap.Error(err))
return
} else {
a.settings.setIndexTemplate(string(tmpl))
a.log.Info(logs.SetCustomIndexPageTemplate, zap.String("path", path))
}
tmpl, err := io.ReadAll(reader)
}
func (a *app) readTemplate(path string) ([]byte, error) {
reader, err := os.Open(path)
if err != nil {
a.settings.setIndexTemplate("")
a.log.Warn(logs.FailedToReadIndexPageTemplate, zap.Error(err))
return
return nil, err
}
a.settings.setIndexTemplate(string(tmpl))
a.log.Info(logs.SetCustomIndexPageTemplate)
return io.ReadAll(reader)
}
func (s *appSettings) ClientCut() bool {

View file

@ -206,9 +206,6 @@ func settings() *viper.Viper {
// pool:
v.SetDefault(cfgPoolErrorThreshold, defaultPoolErrorThreshold)
v.SetDefault(cfgIndexPageEnabled, false)
v.SetDefault(cfgIndexPageTemplatePath, "")
// frostfs:
v.SetDefault(cfgBufferMaxSizeForPut, defaultBufferMaxSizeForPut)

View file

@ -351,7 +351,12 @@ resolve_bucket:
# `index_page` section
Parameters for index HTML-page output with S3-bucket or S3-subdir content for `Get object` request
Parameters for index HTML-page output. Activates if `GetObject` request returns `not found`. Two
index page modes available:
* `s3` mode uses tree service for listing objects,
* `native` sends requests to nodes via native protocol.
If request pass S3-bucket name instead of CID, `s3` mode will be used, otherwise `native`.
```yaml
index_page:

View file

@ -1,15 +1,21 @@
package handler
import (
"context"
"html/template"
"net/url"
"sort"
"strconv"
"strings"
"sync"
"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/utils"
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/docker/go-units"
"github.com/valyala/fasthttp"
"go.uber.org/zap"
@ -25,19 +31,66 @@ const (
type (
BrowsePageData struct {
BucketName,
Prefix string
Objects []ResponseObject
BucketInfo *data.BucketInfo
Prefix string
Objects []ResponseObject
IsNative bool
}
ResponseObject struct {
OID string
Created string
FileName string
FilePath string
Size string
IsDir bool
}
)
func newListObjectsResponseS3(attrs map[string]string) ResponseObject {
return ResponseObject{
OID: attrs[attrOID],
Created: attrs[attrCreated],
FileName: attrs[attrFileName],
Size: attrs[attrSize],
IsDir: attrs[attrOID] == "",
}
}
func newListObjectsResponseNative(attrs map[string]string) ResponseObject {
filename := lastPathElement(attrs[object.AttributeFilePath])
if filename == "" {
filename = attrs[attrFileName]
}
return ResponseObject{
OID: attrs[attrOID],
Created: attrs[object.AttributeTimestamp] + "000",
FileName: filename,
FilePath: attrs[object.AttributeFilePath],
Size: attrs[attrSize],
IsDir: false,
}
}
func getNextDir(filepath, prefix string) string {
restPath := strings.Replace(filepath, prefix, "", 1)
index := strings.Index(restPath, "/")
if index == -1 {
return ""
}
return restPath[:index]
}
func lastPathElement(path string) string {
if path == "" {
return path
}
index := strings.LastIndex(path, "/")
if index == len(path)-1 {
index = strings.LastIndex(path[:index], "/")
}
return path[index+1:]
}
func parseTimestamp(tstamp string) (time.Time, error) {
millis, err := strconv.ParseInt(tstamp, 10, 64)
if err != nil {
@ -47,16 +100,6 @@ func parseTimestamp(tstamp string) (time.Time, error) {
return time.UnixMilli(millis), nil
}
func NewResponseObject(nodes map[string]string) ResponseObject {
return ResponseObject{
OID: nodes[attrOID],
Created: nodes[attrCreated],
FileName: nodes[attrFileName],
Size: nodes[attrSize],
IsDir: nodes[attrOID] == "",
}
}
func formatTimestamp(strdate string) string {
date, err := parseTimestamp(strdate)
if err != nil || date.IsZero() {
@ -94,12 +137,9 @@ func trimPrefix(encPrefix string) string {
return prefix[:slashIndex]
}
func urlencode(prefix, filename string) string {
func urlencode(path string) string {
var res strings.Builder
path := filename
if prefix != "" {
path = strings.Join([]string{prefix, filename}, "/")
}
prefixParts := strings.Split(path, "/")
for _, prefixPart := range prefixParts {
prefixPart = "/" + url.PathEscape(prefixPart)
@ -112,28 +152,185 @@ func urlencode(prefix, filename string) string {
return res.String()
}
func (h *Handler) browseObjects(c *fasthttp.RequestCtx, bucketInfo *data.BucketInfo, prefix string) {
log := h.log.With(zap.String("bucket", bucketInfo.Name))
func (h *Handler) getDirObjectsS3(ctx context.Context, bucketInfo *data.BucketInfo, prefix string) ([]ResponseObject, error) {
nodes, _, err := h.tree.GetSubTreeByPrefix(ctx, bucketInfo, prefix, true)
if err != nil {
return nil, err
}
var objects = make([]ResponseObject, 0, len(nodes))
for _, node := range nodes {
meta := node.GetMeta()
if meta == nil {
continue
}
var attrs = make(map[string]string, len(meta))
for _, m := range meta {
attrs[m.GetKey()] = string(m.GetValue())
}
obj := newListObjectsResponseS3(attrs)
obj.FilePath = prefix + obj.FileName
objects = append(objects, obj)
}
return objects, nil
}
type headDirParams struct {
cnrID cid.ID
objectIDs ResObjectSearch
basePath string
objCh chan<- ResponseObject
errCh chan<- error
}
func (h *Handler) getDirObjectsNative(ctx context.Context, bucketInfo *data.BucketInfo, prefix string) ([]ResponseObject, error) {
const initialSliceCapacity = 100
log := h.log.With(
zap.String("cid", bucketInfo.CID.EncodeToString()),
zap.String("prefix", prefix),
)
basePath := strings.TrimRightFunc(prefix, func(r rune) bool {
return r != '/'
})
filters := []object.SearchMatchType{object.MatchCommonPrefix}
if basePath == "" {
filters = append(filters, object.MatchNotPresent)
}
objCh := make(chan ResponseObject)
errCh := make(chan error)
done := make(chan struct{})
objects := make([]ResponseObject, 0, initialSliceCapacity)
go func() {
for err := range errCh {
if err != nil {
log.Error(logs.FailedToHeadObject, zap.Error(err))
}
}
done <- struct{}{}
}()
go func() {
for obj := range objCh {
objects = append(objects, obj)
}
done <- struct{}{}
}()
wg := sync.WaitGroup{}
for _, filter := range filters {
wg.Add(1)
go func(filter object.SearchMatchType) {
defer wg.Done()
objectIDs, err := h.search(ctx, bucketInfo.CID, object.AttributeFilePath, prefix, filter)
if err != nil {
errCh <- err
return
}
defer objectIDs.Close()
h.headDirObjects(ctx, headDirParams{
cnrID: bucketInfo.CID,
objectIDs: objectIDs,
basePath: basePath,
objCh: objCh,
errCh: errCh,
})
}(filter)
}
wg.Wait()
close(errCh)
close(objCh)
<-done
<-done
return objects, nil
}
func (h *Handler) headDirObjects(ctx context.Context, p headDirParams) {
wg := sync.WaitGroup{}
dirs := sync.Map{}
ctx, cancel := context.WithCancel(ctx)
defer cancel()
err := p.objectIDs.Iterate(func(id oid.ID) bool {
wg.Add(1)
go func(id oid.ID) {
defer wg.Done()
h.headDirObject(ctx, id, p, &dirs, cancel)
}(id)
select {
case <-ctx.Done():
return true
default:
return false
}
})
wg.Wait()
if err != nil {
p.errCh <- err
return
}
}
func (h *Handler) headDirObject(ctx context.Context, id oid.ID, p headDirParams, dirs *sync.Map, cancel context.CancelFunc) {
addr := newAddress(p.cnrID, id)
obj, err := h.frostfs.HeadObject(ctx, PrmObjectHead{
PrmAuth: PrmAuth{BearerToken: bearerToken(ctx)},
Address: addr,
})
if err != nil {
p.errCh <- err
cancel()
return
}
attrs := loadAttributes(obj.Attributes())
attrs[attrOID] = id.EncodeToString()
attrs[attrSize] = strconv.FormatUint(obj.PayloadSize(), 10)
dirname := getNextDir(attrs[object.AttributeFilePath], p.basePath)
if dirname == "" {
p.objCh <- newListObjectsResponseNative(attrs)
} else if _, ok := dirs.Load(dirname); !ok {
p.objCh <- ResponseObject{
FileName: dirname,
FilePath: p.basePath + dirname,
IsDir: true,
}
dirs.Store(dirname, true)
}
}
type browseParams struct {
bucketInfo *data.BucketInfo
prefix string
isNative bool
listObjects func(ctx context.Context, bucketName *data.BucketInfo, prefix string) ([]ResponseObject, error)
}
func (h *Handler) browseObjects(c *fasthttp.RequestCtx, p browseParams) {
log := h.log.With(
zap.String("bucket", p.bucketInfo.Name),
zap.String("container", p.bucketInfo.CID.EncodeToString()),
zap.String("prefix", p.prefix),
)
ctx := utils.GetContextFromRequest(c)
nodes, err := h.listObjects(ctx, bucketInfo, prefix)
objects, err := p.listObjects(ctx, p.bucketInfo, p.prefix)
if err != nil {
logAndSendBucketError(c, log, err)
return
}
respObjects := make([]ResponseObject, len(nodes))
for i, node := range nodes {
respObjects[i] = NewResponseObject(node)
}
sort.Slice(respObjects, func(i, j int) bool {
if respObjects[i].IsDir == respObjects[j].IsDir {
return respObjects[i].FileName < respObjects[j].FileName
sort.Slice(objects, func(i, j int) bool {
if objects[i].IsDir == objects[j].IsDir {
return objects[i].FileName < objects[j].FileName
}
return respObjects[i].IsDir
return objects[i].IsDir
})
indexTemplate := h.config.IndexPageTemplate()
tmpl, err := template.New("index").Funcs(template.FuncMap{
"formatTimestamp": formatTimestamp,
@ -141,15 +338,16 @@ func (h *Handler) browseObjects(c *fasthttp.RequestCtx, bucketInfo *data.BucketI
"trimPrefix": trimPrefix,
"urlencode": urlencode,
"parentDir": parentDir,
}).Parse(indexTemplate)
}).Parse(h.config.IndexPageTemplate())
if err != nil {
logAndSendBucketError(c, log, err)
return
}
if err = tmpl.Execute(c, &BrowsePageData{
BucketName: bucketInfo.Name,
Prefix: prefix,
Objects: respObjects,
BucketInfo: p.bucketInfo,
Prefix: p.prefix,
IsNative: p.isNative,
Objects: objects,
}); err != nil {
logAndSendBucketError(c, log, err)
return

View file

@ -23,10 +23,9 @@ import (
// 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 {
cnrIDStr, _ := c.UserValue("cid").(string)
var cnrID cid.ID
if err := cnrID.DecodeString(cnrIDStr); err != nil {
h.byObjectName(c, h.receiveFile)
} else {
h.byAddress(c, h.receiveFile)
@ -45,7 +44,7 @@ func (h *Handler) DownloadByAttribute(c *fasthttp.RequestCtx) {
h.byAttribute(c, h.receiveFile)
}
func (h *Handler) search(ctx context.Context, cnrID *cid.ID, key, val string, op object.SearchMatchType) (ResObjectSearch, error) {
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)
@ -54,7 +53,7 @@ func (h *Handler) search(ctx context.Context, cnrID *cid.ID, key, val string, op
PrmAuth: PrmAuth{
BearerToken: bearerToken(ctx),
},
Container: *cnrID,
Container: cnrID,
Filters: filters,
}
@ -101,7 +100,7 @@ func (h *Handler) DownloadZipped(c *fasthttp.RequestCtx) {
return
}
resSearch, err := h.search(ctx, &bktInfo.CID, object.AttributeFilePath, prefix, object.MatchCommonPrefix)
resSearch, err := h.search(ctx, bktInfo.CID, 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)

View file

@ -182,7 +182,7 @@ func New(params *AppParams, config Config, tree *tree.Tree) *Handler {
func (h *Handler) byAddress(c *fasthttp.RequestCtx, f func(context.Context, request, oid.Address)) {
var (
idCnr, _ = c.UserValue("cid").(string)
idObj, _ = c.UserValue("oid").(string)
idObj, _ = url.PathUnescape(c.UserValue("oid").(string))
log = h.log.With(zap.String("cid", idCnr), zap.String("oid", idObj))
)
@ -196,6 +196,18 @@ func (h *Handler) byAddress(c *fasthttp.RequestCtx, f func(context.Context, requ
objID := new(oid.ID)
if err = objID.DecodeString(idObj); err != nil {
if h.config.IndexPageEnabled() {
var addr oid.Address
addr.SetContainer(bktInfo.CID)
c.SetStatusCode(fasthttp.StatusNotFound)
h.browseObjects(c, browseParams{
bucketInfo: bktInfo,
prefix: idObj,
listObjects: h.getDirObjectsNative,
isNative: true,
})
return
}
log.Error(logs.WrongObjectID, zap.Error(err))
response.Error(c, "wrong object id", fasthttp.StatusBadRequest)
return
@ -237,7 +249,12 @@ func (h *Handler) byObjectName(c *fasthttp.RequestCtx, f func(context.Context, r
if isDir(unescapedKey) || isContainerRoot(unescapedKey) {
if code := checkErrorType(err); code == fasthttp.StatusNotFound || code == fasthttp.StatusOK {
c.SetStatusCode(code)
h.browseObjects(c, bktInfo, unescapedKey)
h.browseObjects(c, browseParams{
bucketInfo: bktInfo,
prefix: unescapedKey,
listObjects: h.getDirObjectsS3,
isNative: false,
})
return
}
}
@ -294,7 +311,7 @@ func (h *Handler) byAttribute(c *fasthttp.RequestCtx, f func(context.Context, re
return
}
res, err := h.search(ctx, &bktInfo.CID, key, val, object.MatchStringEqual)
res, err := h.search(ctx, bktInfo.CID, key, val, object.MatchStringEqual)
if err != nil {
log.Error(logs.CouldNotSearchForObjects, zap.Error(err))
response.Error(c, "could not search for objects: "+err.Error(), fasthttp.StatusBadRequest)
@ -390,25 +407,3 @@ func (h *Handler) readContainer(ctx context.Context, cnrID cid.ID) (*data.Bucket
return bktInfo, err
}
func (h *Handler) listObjects(ctx context.Context, bucketInfo *data.BucketInfo, prefix string) ([]map[string]string, error) {
nodes, _, err := h.tree.GetSubTreeByPrefix(ctx, bucketInfo, prefix, true)
if err != nil {
return nil, err
}
var objects = make([]map[string]string, 0, len(nodes))
for _, node := range nodes {
meta := node.GetMeta()
if meta == nil {
continue
}
var obj = make(map[string]string, len(meta))
for _, m := range meta {
obj[m.GetKey()] = string(m.GetValue())
}
objects = append(objects, obj)
}
return objects, nil
}

View file

@ -57,10 +57,10 @@ func (c *configMock) IndexPageEnabled() bool {
return false
}
func (c *configMock) IndexPageTemplatePath() string {
func (c *configMock) IndexPageTemplate() string {
return ""
}
func (c *configMock) IndexPageTemplate() string {
func (c *configMock) IndexPageNativeTemplate() string {
return ""
}

View file

@ -12,6 +12,7 @@ import (
"git.frostfs.info/TrueCloudLab/frostfs-http-gw/tree"
"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/bearer"
"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/client"
"git.frostfs.info/TrueCloudLab/frostfs-sdk-go/object"
"github.com/valyala/fasthttp"
"go.uber.org/zap"
)
@ -59,6 +60,14 @@ func checkErrorType(err error) int {
}
}
func loadAttributes(attrs []object.Attribute) map[string]string {
result := make(map[string]string)
for _, attr := range attrs {
result[attr.Key()] = attr.Value()
}
return result
}
func isValidToken(s string) bool {
for _, c := range s {
if c <= ' ' || c > 127 {

View file

@ -71,6 +71,7 @@ const (
AddedStoragePeer = "added storage peer" // Info in ../../settings.go
CouldntGetBucket = "could not get bucket" // Error in ../handler/utils.go
CouldntPutBucketIntoCache = "couldn't put bucket info into cache" // Warn in ../handler/handler.go
FailedToHeadObject = "failed to HEAD object" // Error in ../handler/handler.go
InvalidCacheEntryType = "invalid cache entry type" // Warn in ../cache/buckets.go
InvalidLifetimeUsingDefaultValue = "invalid lifetime, using default value (in seconds)" // Error in ../../cmd/http-gw/settings.go
InvalidCacheSizeUsingDefaultValue = "invalid cache size, using default value" // Error in ../../cmd/http-gw/settings.go

View file

@ -1,10 +1,15 @@
{{$bucketName := .BucketName}}
{{$isNative := .IsNative}}
{{$container := .BucketInfo.Name}}
{{if $isNative}}
{{$container = .BucketInfo.CID}}
{{end}}
{{ $prefix := trimPrefix .Prefix }}
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8"/>
<title>Index of s3://{{$bucketName}}/{{if $prefix}}/{{$prefix}}/{{end}}</title>
<title>Index of {{if $isNative}}frostfs{{else}}s3{{end}}://{{$container}}
/{{if $prefix}}/{{$prefix}}/{{end}}</title>
<style>
table {
width: 80%;
@ -23,15 +28,21 @@
th {
background-color: #c3bcbc;
}
h1 {
font-size: 1.5em;
}
tr:nth-child(even) {background-color: #ebe7e7;}
</style>
</head>
<body>
<h1>Index of s3://{{$bucketName}}/{{if $prefix}}{{$prefix}}/{{end}}</h1>
<h1>Index of {{if $isNative}}frostfs{{else}}s3{{end}}://{{$container}}
/{{if $prefix}}{{$prefix}}/{{end}}</h1>
<table>
<thead>
<tr>
<th>Filename</th>
<th>OID</th>
<th>Size</th>
<th>Created</th>
<th>Download</th>
@ -42,47 +53,77 @@
{{if $trimmedPrefix }}
<tr>
<td>
⮐<a href="/get/{{$bucketName}}{{ urlencode $trimmedPrefix "" }}">..</a>
⮐<a href="/get/{{$container}}{{ urlencode $trimmedPrefix }}/">..</a>
</td>
<td></td>
<td></td>
<td></td>
<td></td>
</tr>
{{else}}
<tr>
<td>
⮐<a href="/get/{{ $bucketName }}/">..</a>
⮐<a href="/get/{{$container}}/">..</a>
</td>
<td></td>
<td></td>
<td></td>
<td></td>
</tr>
{{end}}
{{range .Objects}}
<tr>
<td>
{{if .IsDir}}
🗀
<a href="/get/{{ $bucketName }}{{ urlencode $prefix .FileName }}/">
{{.FileName}}/
</a>
{{else}}
🗎
<a href="/get/{{ $bucketName }}{{ urlencode $prefix .FileName }}">
{{.FileName}}
</a>
{{end}}
</td>
<td>{{if not .IsDir}}{{ formatSize .Size }}{{end}}</td>
<td>{{if not .IsDir}}{{ formatTimestamp .Created }}{{end}}</td>
<td>
{{ if not .IsDir }}
<a href="/get/{{ $bucketName}}{{ urlencode $prefix .FileName }}?download=true">
Link
</a>
{{ end }}
</td>
</tr>
{{if $isNative}}
<tr>
<td>
{{if .IsDir}}
🗀
<a href="/get/{{$container}}{{ urlencode .FilePath }}/">
{{.FileName}}/
</a>
{{else}}
🗎
<a href="/get/{{$container}}/{{ .OID }}">
{{.FileName}}
</a>
{{end}}
</td>
<td>{{.OID}}</td>
<td>{{if .Size}}{{ formatSize .Size }}{{end}}</td>
<td>{{if .Created}}{{ formatTimestamp .Created }}{{end}}</td>
<td>
{{ if .OID }}
<a href="/get/{{$container}}/{{ .OID }}?download=true">
Link
</a>
{{ end }}
</td>
</tr>
{{else}}
<tr>
<td>
{{if .IsDir}}
🗀
<a href="/get/{{$container}}{{ urlencode .FilePath }}/">
{{.FileName}}/
</a>
{{else}}
🗎
<a href="/get/{{$container}}{{ urlencode .FilePath }}">
{{.FileName}}
</a>
{{end}}
</td>
<td>{{if not .IsDir}}{{ formatSize .Size }}{{end}}</td>
<td>{{if not .IsDir}}{{ formatTimestamp .Created }}{{end}}</td>
<td>
{{ if not .IsDir }}
<a href="/get/{{$container}}{{ urlencode .FilePath }}?download=true">
Link
</a>
{{ end }}
</td>
</tr>
{{end}}
{{end}}
</tbody>
</table>