services/keep-web/s3.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package main
   6
   7 import (
   8         "encoding/xml"
   9         "errors"
  10         "fmt"
  11         "io"
  12         "net/http"
  13         "os"
  14         "path/filepath"
  15         "sort"
  16         "strconv"
  17         "strings"
  18
  19         "git.arvados.org/arvados.git/sdk/go/arvados"
  20         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  21         "github.com/AdRoll/goamz/s3"
  22 )
  23
  24 const s3MaxKeys = 1000
  25
  26 // serveS3 handles r and returns true if r is a request from an S3
  27 // client, otherwise it returns false.
  28 func (h *handler) serveS3(w http.ResponseWriter, r *http.Request) bool {
  29         var token string
  30         if auth := r.Header.Get("Authorization"); strings.HasPrefix(auth, "AWS ") {
  31                 split := strings.SplitN(auth[4:], ":", 2)
  32                 if len(split) < 2 {
  33                         w.WriteHeader(http.StatusUnauthorized)
  34                         return true
  35                 }
  36                 token = split[0]
  37         } else if strings.HasPrefix(auth, "AWS4-HMAC-SHA256 ") {
  38                 for _, cmpt := range strings.Split(auth[17:], ",") {
  39                         cmpt = strings.TrimSpace(cmpt)
  40                         split := strings.SplitN(cmpt, "=", 2)
  41                         if len(split) == 2 && split[0] == "Credential" {
  42                                 keyandscope := strings.Split(split[1], "/")
  43                                 if len(keyandscope[0]) > 0 {
  44                                         token = keyandscope[0]
  45                                         break
  46                                 }
  47                         }
  48                 }
  49                 if token == "" {
  50                         w.WriteHeader(http.StatusBadRequest)
  51                         fmt.Println(w, "invalid V4 signature")
  52                         return true
  53                 }
  54         } else {
  55                 return false
  56         }
  57
  58         _, kc, client, release, err := h.getClients(r.Header.Get("X-Request-Id"), token)
  59         if err != nil {
  60                 http.Error(w, "Pool failed: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
  61                 return true
  62         }
  63         defer release()
  64
  65         fs := client.SiteFileSystem(kc)
  66         fs.ForwardSlashNameSubstitution(h.Config.cluster.Collections.ForwardSlashNameSubstitution)
  67
  68         objectNameGiven := strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 1
  69
  70         switch {
  71         case r.Method == http.MethodGet && !objectNameGiven:
  72                 // Path is "/{uuid}" or "/{uuid}/", has no object name
  73                 if _, ok := r.URL.Query()["versioning"]; ok {
  74                         // GetBucketVersioning
  75                         w.Header().Set("Content-Type", "application/xml")
  76                         io.WriteString(w, xml.Header)
  77                         fmt.Fprintln(w, `<VersioningConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/"/>`)
  78                 } else {
  79                         // ListObjects
  80                         h.s3list(w, r, fs)
  81                 }
  82                 return true
  83         case r.Method == http.MethodGet || r.Method == http.MethodHead:
  84                 fspath := "/by_id" + r.URL.Path
  85                 fi, err := fs.Stat(fspath)
  86                 if r.Method == "HEAD" && !objectNameGiven {
  87                         // HeadBucket
  88                         if err == nil && fi.IsDir() {
  89                                 w.WriteHeader(http.StatusOK)
  90                         } else if os.IsNotExist(err) {
  91                                 w.WriteHeader(http.StatusNotFound)
  92                         } else {
  93                                 http.Error(w, err.Error(), http.StatusBadGateway)
  94                         }
  95                         return true
  96                 }
  97                 if err == nil && fi.IsDir() && objectNameGiven && strings.HasSuffix(fspath, "/") && h.Config.cluster.Collections.S3FolderObjects {
  98                         w.Header().Set("Content-Type", "application/x-directory")
  99                         w.WriteHeader(http.StatusOK)
 100                         return true
 101                 }
 102                 if os.IsNotExist(err) ||
 103                         (err != nil && err.Error() == "not a directory") ||
 104                         (fi != nil && fi.IsDir()) {
 105                         http.Error(w, "not found", http.StatusNotFound)
 106                         return true
 107                 }
 108                 // shallow copy r, and change URL path
 109                 r := *r
 110                 r.URL.Path = fspath
 111                 http.FileServer(fs).ServeHTTP(w, &r)
 112                 return true
 113         case r.Method == http.MethodPut:
 114                 if !objectNameGiven {
 115                         http.Error(w, "missing object name in PUT request", http.StatusBadRequest)
 116                         return true
 117                 }
 118                 fspath := "by_id" + r.URL.Path
 119                 var objectIsDir bool
 120                 if strings.HasSuffix(fspath, "/") {
 121                         if !h.Config.cluster.Collections.S3FolderObjects {
 122                                 http.Error(w, "invalid object name: trailing slash", http.StatusBadRequest)
 123                                 return true
 124                         }
 125                         n, err := r.Body.Read(make([]byte, 1))
 126                         if err != nil && err != io.EOF {
 127                                 http.Error(w, fmt.Sprintf("error reading request body: %s", err), http.StatusInternalServerError)
 128                                 return true
 129                         } else if n > 0 {
 130                                 http.Error(w, "cannot create object with trailing '/' char unless content is empty", http.StatusBadRequest)
 131                                 return true
 132                         } else if strings.SplitN(r.Header.Get("Content-Type"), ";", 2)[0] != "application/x-directory" {
 133                                 http.Error(w, "cannot create object with trailing '/' char unless Content-Type is 'application/x-directory'", http.StatusBadRequest)
 134                                 return true
 135                         }
 136                         // Given PUT "foo/bar/", we'll use "foo/bar/."
 137                         // in the "ensure parents exist" block below,
 138                         // and then we'll be done.
 139                         fspath += "."
 140                         objectIsDir = true
 141                 }
 142                 fi, err := fs.Stat(fspath)
 143                 if err != nil && err.Error() == "not a directory" {
 144                         // requested foo/bar, but foo is a file
 145                         http.Error(w, "object name conflicts with existing object", http.StatusBadRequest)
 146                         return true
 147                 }
 148                 if strings.HasSuffix(r.URL.Path, "/") && err == nil && !fi.IsDir() {
 149                         // requested foo/bar/, but foo/bar is a file
 150                         http.Error(w, "object name conflicts with existing object", http.StatusBadRequest)
 151                         return true
 152                 }
 153                 // create missing parent/intermediate directories, if any
 154                 for i, c := range fspath {
 155                         if i > 0 && c == '/' {
 156                                 dir := fspath[:i]
 157                                 if strings.HasSuffix(dir, "/") {
 158                                         err = errors.New("invalid object name (consecutive '/' chars)")
 159                                         http.Error(w, err.Error(), http.StatusBadRequest)
 160                                         return true
 161                                 }
 162                                 err = fs.Mkdir(dir, 0755)
 163                                 if err == arvados.ErrInvalidArgument {
 164                                         // Cannot create a directory
 165                                         // here.
 166                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 167                                         http.Error(w, err.Error(), http.StatusBadRequest)
 168                                         return true
 169                                 } else if err != nil && !os.IsExist(err) {
 170                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 171                                         http.Error(w, err.Error(), http.StatusInternalServerError)
 172                                         return true
 173                                 }
 174                         }
 175                 }
 176                 if !objectIsDir {
 177                         f, err := fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 178                         if os.IsNotExist(err) {
 179                                 f, err = fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 180                         }
 181                         if err != nil {
 182                                 err = fmt.Errorf("open %q failed: %w", r.URL.Path, err)
 183                                 http.Error(w, err.Error(), http.StatusBadRequest)
 184                                 return true
 185                         }
 186                         defer f.Close()
 187                         _, err = io.Copy(f, r.Body)
 188                         if err != nil {
 189                                 err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 190                                 http.Error(w, err.Error(), http.StatusBadGateway)
 191                                 return true
 192                         }
 193                         err = f.Close()
 194                         if err != nil {
 195                                 err = fmt.Errorf("write to %q failed: close: %w", r.URL.Path, err)
 196                                 http.Error(w, err.Error(), http.StatusBadGateway)
 197                                 return true
 198                         }
 199                 }
 200                 err = fs.Sync()
 201                 if err != nil {
 202                         err = fmt.Errorf("sync failed: %w", err)
 203                         http.Error(w, err.Error(), http.StatusInternalServerError)
 204                         return true
 205                 }
 206                 w.WriteHeader(http.StatusOK)
 207                 return true
 208         default:
 209                 http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
 210                 return true
 211         }
 212 }
 213
 214 // Call fn on the given path (directory) and its contents, in
 215 // lexicographic order.
 216 //
 217 // If isRoot==true and path is not a directory, return nil.
 218 //
 219 // If fn returns filepath.SkipDir when called on a directory, don't
 220 // descend into that directory.
 221 func walkFS(fs arvados.CustomFileSystem, path string, isRoot bool, fn func(path string, fi os.FileInfo) error) error {
 222         if isRoot {
 223                 fi, err := fs.Stat(path)
 224                 if os.IsNotExist(err) || (err == nil && !fi.IsDir()) {
 225                         return nil
 226                 } else if err != nil {
 227                         return err
 228                 }
 229                 err = fn(path, fi)
 230                 if err == filepath.SkipDir {
 231                         return nil
 232                 } else if err != nil {
 233                         return err
 234                 }
 235         }
 236         f, err := fs.Open(path)
 237         if os.IsNotExist(err) && isRoot {
 238                 return nil
 239         } else if err != nil {
 240                 return fmt.Errorf("open %q: %w", path, err)
 241         }
 242         defer f.Close()
 243         if path == "/" {
 244                 path = ""
 245         }
 246         fis, err := f.Readdir(-1)
 247         if err != nil {
 248                 return err
 249         }
 250         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 251         for _, fi := range fis {
 252                 err = fn(path+"/"+fi.Name(), fi)
 253                 if err == filepath.SkipDir {
 254                         continue
 255                 } else if err != nil {
 256                         return err
 257                 }
 258                 if fi.IsDir() {
 259                         err = walkFS(fs, path+"/"+fi.Name(), false, fn)
 260                         if err != nil {
 261                                 return err
 262                         }
 263                 }
 264         }
 265         return nil
 266 }
 267
 268 var errDone = errors.New("done")
 269
 270 func (h *handler) s3list(w http.ResponseWriter, r *http.Request, fs arvados.CustomFileSystem) {
 271         var params struct {
 272                 bucket    string
 273                 delimiter string
 274                 marker    string
 275                 maxKeys   int
 276                 prefix    string
 277         }
 278         params.bucket = strings.SplitN(r.URL.Path[1:], "/", 2)[0]
 279         params.delimiter = r.FormValue("delimiter")
 280         params.marker = r.FormValue("marker")
 281         if mk, _ := strconv.ParseInt(r.FormValue("max-keys"), 10, 64); mk > 0 && mk < s3MaxKeys {
 282                 params.maxKeys = int(mk)
 283         } else {
 284                 params.maxKeys = s3MaxKeys
 285         }
 286         params.prefix = r.FormValue("prefix")
 287
 288         bucketdir := "by_id/" + params.bucket
 289         // walkpath is the directory (relative to bucketdir) we need
 290         // to walk: the innermost directory that is guaranteed to
 291         // contain all paths that have the requested prefix. Examples:
 292         // prefix "foo/bar"  => walkpath "foo"
 293         // prefix "foo/bar/" => walkpath "foo/bar"
 294         // prefix "foo"      => walkpath ""
 295         // prefix ""         => walkpath ""
 296         walkpath := params.prefix
 297         if cut := strings.LastIndex(walkpath, "/"); cut >= 0 {
 298                 walkpath = walkpath[:cut]
 299         } else {
 300                 walkpath = ""
 301         }
 302
 303         resp := s3.ListResp{
 304                 Name:      strings.SplitN(r.URL.Path[1:], "/", 2)[0],
 305                 Prefix:    params.prefix,
 306                 Delimiter: params.delimiter,
 307                 Marker:    params.marker,
 308                 MaxKeys:   params.maxKeys,
 309         }
 310         commonPrefixes := map[string]bool{}
 311         err := walkFS(fs, strings.TrimSuffix(bucketdir+"/"+walkpath, "/"), true, func(path string, fi os.FileInfo) error {
 312                 if path == bucketdir {
 313                         return nil
 314                 }
 315                 path = path[len(bucketdir)+1:]
 316                 filesize := fi.Size()
 317                 if fi.IsDir() {
 318                         path += "/"
 319                         filesize = 0
 320                 }
 321                 if len(path) <= len(params.prefix) {
 322                         if path > params.prefix[:len(path)] {
 323                                 // with prefix "foobar", walking "fooz" means we're done
 324                                 return errDone
 325                         }
 326                         if path < params.prefix[:len(path)] {
 327                                 // with prefix "foobar", walking "foobag" is pointless
 328                                 return filepath.SkipDir
 329                         }
 330                         if fi.IsDir() && !strings.HasPrefix(params.prefix+"/", path) {
 331                                 // with prefix "foo/bar", walking "fo"
 332                                 // is pointless (but walking "foo" or
 333                                 // "foo/bar" is necessary)
 334                                 return filepath.SkipDir
 335                         }
 336                         if len(path) < len(params.prefix) {
 337                                 // can't skip anything, and this entry
 338                                 // isn't in the results, so just
 339                                 // continue descent
 340                                 return nil
 341                         }
 342                 } else {
 343                         if path[:len(params.prefix)] > params.prefix {
 344                                 // with prefix "foobar", nothing we
 345                                 // see after "foozzz" is relevant
 346                                 return errDone
 347                         }
 348                 }
 349                 if path < params.marker || path < params.prefix {
 350                         return nil
 351                 }
 352                 if fi.IsDir() && !h.Config.cluster.Collections.S3FolderObjects {
 353                         // Note we don't add anything to
 354                         // commonPrefixes here even if delimiter is
 355                         // "/". We descend into the directory, and
 356                         // return a commonPrefix only if we end up
 357                         // finding a regular file inside it.
 358                         return nil
 359                 }
 360                 if params.delimiter != "" {
 361                         idx := strings.Index(path[len(params.prefix):], params.delimiter)
 362                         if idx >= 0 {
 363                                 // with prefix "foobar" and delimiter
 364                                 // "z", when we hit "foobar/baz", we
 365                                 // add "/baz" to commonPrefixes and
 366                                 // stop descending.
 367                                 commonPrefixes[path[:len(params.prefix)+idx+1]] = true
 368                                 return filepath.SkipDir
 369                         }
 370                 }
 371                 if len(resp.Contents)+len(commonPrefixes) >= params.maxKeys {
 372                         resp.IsTruncated = true
 373                         if params.delimiter != "" {
 374                                 resp.NextMarker = path
 375                         }
 376                         return errDone
 377                 }
 378                 resp.Contents = append(resp.Contents, s3.Key{
 379                         Key:          path,
 380                         LastModified: fi.ModTime().UTC().Format("2006-01-02T15:04:05.999") + "Z",
 381                         Size:         filesize,
 382                 })
 383                 return nil
 384         })
 385         if err != nil && err != errDone {
 386                 http.Error(w, err.Error(), http.StatusInternalServerError)
 387                 return
 388         }
 389         if params.delimiter != "" {
 390                 for prefix := range commonPrefixes {
 391                         resp.CommonPrefixes = append(resp.CommonPrefixes, prefix)
 392                         sort.Strings(resp.CommonPrefixes)
 393                 }
 394         }
 395         wrappedResp := struct {
 396                 XMLName string `xml:"http://s3.amazonaws.com/doc/2006-03-01/ ListBucketResult"`
 397                 s3.ListResp
 398         }{"", resp}
 399         w.Header().Set("Content-Type", "application/xml")
 400         io.WriteString(w, xml.Header)
 401         if err := xml.NewEncoder(w).Encode(wrappedResp); err != nil {
 402                 ctxlog.FromContext(r.Context()).WithError(err).Error("error writing xml response")
 403         }
 404 }