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         "crypto/hmac"
   9         "crypto/sha256"
  10         "encoding/xml"
  11         "errors"
  12         "fmt"
  13         "hash"
  14         "io"
  15         "net/http"
  16         "net/url"
  17         "os"
  18         "path/filepath"
  19         "regexp"
  20         "sort"
  21         "strconv"
  22         "strings"
  23         "time"
  24
  25         "git.arvados.org/arvados.git/sdk/go/arvados"
  26         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  27         "github.com/AdRoll/goamz/s3"
  28 )
  29
  30 const (
  31         s3MaxKeys       = 1000
  32         s3SignAlgorithm = "AWS4-HMAC-SHA256"
  33         s3MaxClockSkew  = 5 * time.Minute
  34 )
  35
  36 func hmacstring(msg string, key []byte) []byte {
  37         h := hmac.New(sha256.New, key)
  38         io.WriteString(h, msg)
  39         return h.Sum(nil)
  40 }
  41
  42 func hashdigest(h hash.Hash, payload string) string {
  43         io.WriteString(h, payload)
  44         return fmt.Sprintf("%x", h.Sum(nil))
  45 }
  46
  47 // Signing key for given secret key and request attrs.
  48 func s3signatureKey(key, datestamp, regionName, serviceName string) []byte {
  49         return hmacstring("aws4_request",
  50                 hmacstring(serviceName,
  51                         hmacstring(regionName,
  52                                 hmacstring(datestamp, []byte("AWS4"+key)))))
  53 }
  54
  55 // Canonical query string for S3 V4 signature: sorted keys, spaces
  56 // escaped as %20 instead of +, keyvalues joined with &.
  57 func s3querystring(u *url.URL) string {
  58         keys := make([]string, 0, len(u.Query()))
  59         values := make(map[string]string, len(u.Query()))
  60         for k, vs := range u.Query() {
  61                 k = strings.Replace(url.QueryEscape(k), "+", "%20", -1)
  62                 keys = append(keys, k)
  63                 for _, v := range vs {
  64                         v = strings.Replace(url.QueryEscape(v), "+", "%20", -1)
  65                         if values[k] != "" {
  66                                 values[k] += "&"
  67                         }
  68                         values[k] += k + "=" + v
  69                 }
  70         }
  71         sort.Strings(keys)
  72         for i, k := range keys {
  73                 keys[i] = values[k]
  74         }
  75         return strings.Join(keys, "&")
  76 }
  77
  78 var reMultipleSlashChars = regexp.MustCompile(`//+`)
  79
  80 func s3stringToSign(alg, scope, signedHeaders string, r *http.Request) (string, error) {
  81         timefmt, timestr := "20060102T150405Z", r.Header.Get("X-Amz-Date")
  82         if timestr == "" {
  83                 timefmt, timestr = time.RFC1123, r.Header.Get("Date")
  84         }
  85         t, err := time.Parse(timefmt, timestr)
  86         if err != nil {
  87                 return "", fmt.Errorf("invalid timestamp %q: %s", timestr, err)
  88         }
  89         if skew := time.Now().Sub(t); skew < -s3MaxClockSkew || skew > s3MaxClockSkew {
  90                 return "", errors.New("exceeded max clock skew")
  91         }
  92
  93         var canonicalHeaders string
  94         for _, h := range strings.Split(signedHeaders, ";") {
  95                 if h == "host" {
  96                         canonicalHeaders += h + ":" + r.Host + "\n"
  97                 } else {
  98                         canonicalHeaders += h + ":" + r.Header.Get(h) + "\n"
  99                 }
 100         }
 101
 102         normalizedURL := *r.URL
 103         normalizedURL.RawPath = ""
 104         normalizedURL.Path = reMultipleSlashChars.ReplaceAllString(normalizedURL.Path, "/")
 105         canonicalRequest := fmt.Sprintf("%s\n%s\n%s\n%s\n%s\n%s", r.Method, normalizedURL.EscapedPath(), s3querystring(r.URL), canonicalHeaders, signedHeaders, r.Header.Get("X-Amz-Content-Sha256"))
 106         ctxlog.FromContext(r.Context()).Debugf("s3stringToSign: canonicalRequest %s", canonicalRequest)
 107         return fmt.Sprintf("%s\n%s\n%s\n%s", alg, r.Header.Get("X-Amz-Date"), scope, hashdigest(sha256.New(), canonicalRequest)), nil
 108 }
 109
 110 func s3signature(secretKey, scope, signedHeaders, stringToSign string) (string, error) {
 111         // scope is {datestamp}/{region}/{service}/aws4_request
 112         drs := strings.Split(scope, "/")
 113         if len(drs) != 4 {
 114                 return "", fmt.Errorf("invalid scope %q", scope)
 115         }
 116         key := s3signatureKey(secretKey, drs[0], drs[1], drs[2])
 117         return hashdigest(hmac.New(sha256.New, key), stringToSign), nil
 118 }
 119
 120 var v2tokenUnderscore = regexp.MustCompile(`^v2_[a-z0-9]{5}-gj3su-[a-z0-9]{15}_`)
 121
 122 func unescapeKey(key string) string {
 123         if v2tokenUnderscore.MatchString(key) {
 124                 // Entire Arvados token, with "/" replaced by "_" to
 125                 // avoid colliding with the Authorization header
 126                 // format.
 127                 return strings.Replace(key, "_", "/", -1)
 128         } else if s, err := url.PathUnescape(key); err == nil {
 129                 return s
 130         } else {
 131                 return key
 132         }
 133 }
 134
 135 // checks3signature verifies the given S3 V4 signature and returns the
 136 // Arvados token that corresponds to the given accessKey. An error is
 137 // returned if accessKey is not a valid token UUID or the signature
 138 // does not match.
 139 func (h *handler) checks3signature(r *http.Request) (string, error) {
 140         var key, scope, signedHeaders, signature string
 141         authstring := strings.TrimPrefix(r.Header.Get("Authorization"), s3SignAlgorithm+" ")
 142         for _, cmpt := range strings.Split(authstring, ",") {
 143                 cmpt = strings.TrimSpace(cmpt)
 144                 split := strings.SplitN(cmpt, "=", 2)
 145                 switch {
 146                 case len(split) != 2:
 147                         // (?) ignore
 148                 case split[0] == "Credential":
 149                         keyandscope := strings.SplitN(split[1], "/", 2)
 150                         if len(keyandscope) == 2 {
 151                                 key, scope = keyandscope[0], keyandscope[1]
 152                         }
 153                 case split[0] == "SignedHeaders":
 154                         signedHeaders = split[1]
 155                 case split[0] == "Signature":
 156                         signature = split[1]
 157                 }
 158         }
 159
 160         client := (&arvados.Client{
 161                 APIHost:  h.Config.cluster.Services.Controller.ExternalURL.Host,
 162                 Insecure: h.Config.cluster.TLS.Insecure,
 163         }).WithRequestID(r.Header.Get("X-Request-Id"))
 164         var aca arvados.APIClientAuthorization
 165         var secret string
 166         var err error
 167         if len(key) == 27 && key[5:12] == "-gj3su-" {
 168                 // Access key is the UUID of an Arvados token, secret
 169                 // key is the secret part.
 170                 ctx := arvados.ContextWithAuthorization(r.Context(), "Bearer "+h.Config.cluster.SystemRootToken)
 171                 err = client.RequestAndDecodeContext(ctx, &aca, "GET", "arvados/v1/api_client_authorizations/"+key, nil, nil)
 172                 secret = aca.APIToken
 173         } else {
 174                 // Access key and secret key are both an entire
 175                 // Arvados token or OIDC access token.
 176                 ctx := arvados.ContextWithAuthorization(r.Context(), "Bearer "+unescapeKey(key))
 177                 err = client.RequestAndDecodeContext(ctx, &aca, "GET", "arvados/v1/api_client_authorizations/current", nil, nil)
 178                 secret = key
 179         }
 180         if err != nil {
 181                 ctxlog.FromContext(r.Context()).WithError(err).WithField("UUID", key).Info("token lookup failed")
 182                 return "", errors.New("invalid access key")
 183         }
 184         stringToSign, err := s3stringToSign(s3SignAlgorithm, scope, signedHeaders, r)
 185         if err != nil {
 186                 return "", err
 187         }
 188         expect, err := s3signature(secret, scope, signedHeaders, stringToSign)
 189         if err != nil {
 190                 return "", err
 191         } else if expect != signature {
 192                 return "", fmt.Errorf("signature does not match (scope %q signedHeaders %q stringToSign %q)", scope, signedHeaders, stringToSign)
 193         }
 194         return aca.TokenV2(), nil
 195 }
 196
 197 func s3ErrorResponse(w http.ResponseWriter, s3code string, message string, resource string, code int) {
 198         w.Header().Set("Content-Type", "application/xml")
 199         w.Header().Set("X-Content-Type-Options", "nosniff")
 200         w.WriteHeader(code)
 201         var errstruct struct {
 202                 Code      string
 203                 Message   string
 204                 Resource  string
 205                 RequestId string
 206         }
 207         errstruct.Code = s3code
 208         errstruct.Message = message
 209         errstruct.Resource = resource
 210         errstruct.RequestId = ""
 211         enc := xml.NewEncoder(w)
 212         fmt.Fprint(w, xml.Header)
 213         enc.EncodeElement(errstruct, xml.StartElement{Name: xml.Name{Local: "Error"}})
 214 }
 215
 216 var NoSuchKey = "NoSuchKey"
 217 var NoSuchBucket = "NoSuchBucket"
 218 var InvalidArgument = "InvalidArgument"
 219 var InternalError = "InternalError"
 220 var UnauthorizedAccess = "UnauthorizedAccess"
 221 var InvalidRequest = "InvalidRequest"
 222 var SignatureDoesNotMatch = "SignatureDoesNotMatch"
 223
 224 // serveS3 handles r and returns true if r is a request from an S3
 225 // client, otherwise it returns false.
 226 func (h *handler) serveS3(w http.ResponseWriter, r *http.Request) bool {
 227         var token string
 228         if auth := r.Header.Get("Authorization"); strings.HasPrefix(auth, "AWS ") {
 229                 split := strings.SplitN(auth[4:], ":", 2)
 230                 if len(split) < 2 {
 231                         s3ErrorResponse(w, InvalidRequest, "malformed Authorization header", r.URL.Path, http.StatusUnauthorized)
 232                         return true
 233                 }
 234                 token = unescapeKey(split[0])
 235         } else if strings.HasPrefix(auth, s3SignAlgorithm+" ") {
 236                 t, err := h.checks3signature(r)
 237                 if err != nil {
 238                         s3ErrorResponse(w, SignatureDoesNotMatch, "signature verification failed: "+err.Error(), r.URL.Path, http.StatusForbidden)
 239                         return true
 240                 }
 241                 token = t
 242         } else {
 243                 return false
 244         }
 245
 246         var err error
 247         var fs arvados.CustomFileSystem
 248         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 249                 // Use a single session (cached FileSystem) across
 250                 // multiple read requests.
 251                 fs, err = h.Config.Cache.GetSession(token)
 252                 if err != nil {
 253                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 254                         return true
 255                 }
 256         } else {
 257                 // Create a FileSystem for this request, to avoid
 258                 // exposing incomplete write operations to concurrent
 259                 // requests.
 260                 _, kc, client, release, err := h.getClients(r.Header.Get("X-Request-Id"), token)
 261                 if err != nil {
 262                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 263                         return true
 264                 }
 265                 defer release()
 266                 fs = client.SiteFileSystem(kc)
 267                 fs.ForwardSlashNameSubstitution(h.Config.cluster.Collections.ForwardSlashNameSubstitution)
 268         }
 269
 270         var objectNameGiven bool
 271         var bucketName string
 272         fspath := "/by_id"
 273         if id := parseCollectionIDFromDNSName(r.Host); id != "" {
 274                 fspath += "/" + id
 275                 bucketName = id
 276                 objectNameGiven = strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 0
 277         } else {
 278                 bucketName = strings.SplitN(strings.TrimPrefix(r.URL.Path, "/"), "/", 2)[0]
 279                 objectNameGiven = strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 1
 280         }
 281         fspath += reMultipleSlashChars.ReplaceAllString(r.URL.Path, "/")
 282
 283         switch {
 284         case r.Method == http.MethodGet && !objectNameGiven:
 285                 // Path is "/{uuid}" or "/{uuid}/", has no object name
 286                 if _, ok := r.URL.Query()["versioning"]; ok {
 287                         // GetBucketVersioning
 288                         w.Header().Set("Content-Type", "application/xml")
 289                         io.WriteString(w, xml.Header)
 290                         fmt.Fprintln(w, `<VersioningConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/"/>`)
 291                 } else if _, ok = r.URL.Query()["location"]; ok {
 292                         // GetBucketLocation
 293                         w.Header().Set("Content-Type", "application/xml")
 294                         io.WriteString(w, xml.Header)
 295                         fmt.Fprintln(w, `<LocationConstraint xmlns="http://s3.amazonaws.com/doc/2006-03-01/">`+h.Config.cluster.ClusterID+`</LocationConstraint>`)
 296                 } else {
 297                         // ListObjects
 298                         h.s3list(bucketName, w, r, fs)
 299                 }
 300                 return true
 301         case r.Method == http.MethodGet || r.Method == http.MethodHead:
 302                 fi, err := fs.Stat(fspath)
 303                 if r.Method == "HEAD" && !objectNameGiven {
 304                         // HeadBucket
 305                         if err == nil && fi.IsDir() {
 306                                 w.WriteHeader(http.StatusOK)
 307                         } else if os.IsNotExist(err) {
 308                                 s3ErrorResponse(w, NoSuchBucket, "The specified bucket does not exist.", r.URL.Path, http.StatusNotFound)
 309                         } else {
 310                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 311                         }
 312                         return true
 313                 }
 314                 if err == nil && fi.IsDir() && objectNameGiven && strings.HasSuffix(fspath, "/") && h.Config.cluster.Collections.S3FolderObjects {
 315                         w.Header().Set("Content-Type", "application/x-directory")
 316                         w.WriteHeader(http.StatusOK)
 317                         return true
 318                 }
 319                 if os.IsNotExist(err) ||
 320                         (err != nil && err.Error() == "not a directory") ||
 321                         (fi != nil && fi.IsDir()) {
 322                         s3ErrorResponse(w, NoSuchKey, "The specified key does not exist.", r.URL.Path, http.StatusNotFound)
 323                         return true
 324                 }
 325                 // shallow copy r, and change URL path
 326                 r := *r
 327                 r.URL.Path = fspath
 328                 http.FileServer(fs).ServeHTTP(w, &r)
 329                 return true
 330         case r.Method == http.MethodPut:
 331                 if !objectNameGiven {
 332                         s3ErrorResponse(w, InvalidArgument, "Missing object name in PUT request.", r.URL.Path, http.StatusBadRequest)
 333                         return true
 334                 }
 335                 var objectIsDir bool
 336                 if strings.HasSuffix(fspath, "/") {
 337                         if !h.Config.cluster.Collections.S3FolderObjects {
 338                                 s3ErrorResponse(w, InvalidArgument, "invalid object name: trailing slash", r.URL.Path, http.StatusBadRequest)
 339                                 return true
 340                         }
 341                         n, err := r.Body.Read(make([]byte, 1))
 342                         if err != nil && err != io.EOF {
 343                                 s3ErrorResponse(w, InternalError, fmt.Sprintf("error reading request body: %s", err), r.URL.Path, http.StatusInternalServerError)
 344                                 return true
 345                         } else if n > 0 {
 346                                 s3ErrorResponse(w, InvalidArgument, "cannot create object with trailing '/' char unless content is empty", r.URL.Path, http.StatusBadRequest)
 347                                 return true
 348                         } else if strings.SplitN(r.Header.Get("Content-Type"), ";", 2)[0] != "application/x-directory" {
 349                                 s3ErrorResponse(w, InvalidArgument, "cannot create object with trailing '/' char unless Content-Type is 'application/x-directory'", r.URL.Path, http.StatusBadRequest)
 350                                 return true
 351                         }
 352                         // Given PUT "foo/bar/", we'll use "foo/bar/."
 353                         // in the "ensure parents exist" block below,
 354                         // and then we'll be done.
 355                         fspath += "."
 356                         objectIsDir = true
 357                 }
 358                 fi, err := fs.Stat(fspath)
 359                 if err != nil && err.Error() == "not a directory" {
 360                         // requested foo/bar, but foo is a file
 361                         s3ErrorResponse(w, InvalidArgument, "object name conflicts with existing object", r.URL.Path, http.StatusBadRequest)
 362                         return true
 363                 }
 364                 if strings.HasSuffix(r.URL.Path, "/") && err == nil && !fi.IsDir() {
 365                         // requested foo/bar/, but foo/bar is a file
 366                         s3ErrorResponse(w, InvalidArgument, "object name conflicts with existing object", r.URL.Path, http.StatusBadRequest)
 367                         return true
 368                 }
 369                 // create missing parent/intermediate directories, if any
 370                 for i, c := range fspath {
 371                         if i > 0 && c == '/' {
 372                                 dir := fspath[:i]
 373                                 if strings.HasSuffix(dir, "/") {
 374                                         err = errors.New("invalid object name (consecutive '/' chars)")
 375                                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 376                                         return true
 377                                 }
 378                                 err = fs.Mkdir(dir, 0755)
 379                                 if err == arvados.ErrInvalidArgument {
 380                                         // Cannot create a directory
 381                                         // here.
 382                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 383                                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 384                                         return true
 385                                 } else if err != nil && !os.IsExist(err) {
 386                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 387                                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 388                                         return true
 389                                 }
 390                         }
 391                 }
 392                 if !objectIsDir {
 393                         f, err := fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 394                         if os.IsNotExist(err) {
 395                                 f, err = fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 396                         }
 397                         if err != nil {
 398                                 err = fmt.Errorf("open %q failed: %w", r.URL.Path, err)
 399                                 s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 400                                 return true
 401                         }
 402                         defer f.Close()
 403                         _, err = io.Copy(f, r.Body)
 404                         if err != nil {
 405                                 err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 406                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 407                                 return true
 408                         }
 409                         err = f.Close()
 410                         if err != nil {
 411                                 err = fmt.Errorf("write to %q failed: close: %w", r.URL.Path, err)
 412                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 413                                 return true
 414                         }
 415                 }
 416                 err = fs.Sync()
 417                 if err != nil {
 418                         err = fmt.Errorf("sync failed: %w", err)
 419                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 420                         return true
 421                 }
 422                 // Ensure a subsequent read operation will see the changes.
 423                 h.Config.Cache.ResetSession(token)
 424                 w.WriteHeader(http.StatusOK)
 425                 return true
 426         case r.Method == http.MethodDelete:
 427                 if !objectNameGiven || r.URL.Path == "/" {
 428                         s3ErrorResponse(w, InvalidArgument, "missing object name in DELETE request", r.URL.Path, http.StatusBadRequest)
 429                         return true
 430                 }
 431                 if strings.HasSuffix(fspath, "/") {
 432                         fspath = strings.TrimSuffix(fspath, "/")
 433                         fi, err := fs.Stat(fspath)
 434                         if os.IsNotExist(err) {
 435                                 w.WriteHeader(http.StatusNoContent)
 436                                 return true
 437                         } else if err != nil {
 438                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 439                                 return true
 440                         } else if !fi.IsDir() {
 441                                 // if "foo" exists and is a file, then
 442                                 // "foo/" doesn't exist, so we say
 443                                 // delete was successful.
 444                                 w.WriteHeader(http.StatusNoContent)
 445                                 return true
 446                         }
 447                 } else if fi, err := fs.Stat(fspath); err == nil && fi.IsDir() {
 448                         // if "foo" is a dir, it is visible via S3
 449                         // only as "foo/", not "foo" -- so we leave
 450                         // the dir alone and return 204 to indicate
 451                         // that "foo" does not exist.
 452                         w.WriteHeader(http.StatusNoContent)
 453                         return true
 454                 }
 455                 err = fs.Remove(fspath)
 456                 if os.IsNotExist(err) {
 457                         w.WriteHeader(http.StatusNoContent)
 458                         return true
 459                 }
 460                 if err != nil {
 461                         err = fmt.Errorf("rm failed: %w", err)
 462                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 463                         return true
 464                 }
 465                 err = fs.Sync()
 466                 if err != nil {
 467                         err = fmt.Errorf("sync failed: %w", err)
 468                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 469                         return true
 470                 }
 471                 // Ensure a subsequent read operation will see the changes.
 472                 h.Config.Cache.ResetSession(token)
 473                 w.WriteHeader(http.StatusNoContent)
 474                 return true
 475         default:
 476                 s3ErrorResponse(w, InvalidRequest, "method not allowed", r.URL.Path, http.StatusMethodNotAllowed)
 477                 return true
 478         }
 479 }
 480
 481 // Call fn on the given path (directory) and its contents, in
 482 // lexicographic order.
 483 //
 484 // If isRoot==true and path is not a directory, return nil.
 485 //
 486 // If fn returns filepath.SkipDir when called on a directory, don't
 487 // descend into that directory.
 488 func walkFS(fs arvados.CustomFileSystem, path string, isRoot bool, fn func(path string, fi os.FileInfo) error) error {
 489         if isRoot {
 490                 fi, err := fs.Stat(path)
 491                 if os.IsNotExist(err) || (err == nil && !fi.IsDir()) {
 492                         return nil
 493                 } else if err != nil {
 494                         return err
 495                 }
 496                 err = fn(path, fi)
 497                 if err == filepath.SkipDir {
 498                         return nil
 499                 } else if err != nil {
 500                         return err
 501                 }
 502         }
 503         f, err := fs.Open(path)
 504         if os.IsNotExist(err) && isRoot {
 505                 return nil
 506         } else if err != nil {
 507                 return fmt.Errorf("open %q: %w", path, err)
 508         }
 509         defer f.Close()
 510         if path == "/" {
 511                 path = ""
 512         }
 513         fis, err := f.Readdir(-1)
 514         if err != nil {
 515                 return err
 516         }
 517         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 518         for _, fi := range fis {
 519                 err = fn(path+"/"+fi.Name(), fi)
 520                 if err == filepath.SkipDir {
 521                         continue
 522                 } else if err != nil {
 523                         return err
 524                 }
 525                 if fi.IsDir() {
 526                         err = walkFS(fs, path+"/"+fi.Name(), false, fn)
 527                         if err != nil {
 528                                 return err
 529                         }
 530                 }
 531         }
 532         return nil
 533 }
 534
 535 var errDone = errors.New("done")
 536
 537 func (h *handler) s3list(bucket string, w http.ResponseWriter, r *http.Request, fs arvados.CustomFileSystem) {
 538         var params struct {
 539                 delimiter string
 540                 marker    string
 541                 maxKeys   int
 542                 prefix    string
 543         }
 544         params.delimiter = r.FormValue("delimiter")
 545         params.marker = r.FormValue("marker")
 546         if mk, _ := strconv.ParseInt(r.FormValue("max-keys"), 10, 64); mk > 0 && mk < s3MaxKeys {
 547                 params.maxKeys = int(mk)
 548         } else {
 549                 params.maxKeys = s3MaxKeys
 550         }
 551         params.prefix = r.FormValue("prefix")
 552
 553         bucketdir := "by_id/" + bucket
 554         // walkpath is the directory (relative to bucketdir) we need
 555         // to walk: the innermost directory that is guaranteed to
 556         // contain all paths that have the requested prefix. Examples:
 557         // prefix "foo/bar"  => walkpath "foo"
 558         // prefix "foo/bar/" => walkpath "foo/bar"
 559         // prefix "foo"      => walkpath ""
 560         // prefix ""         => walkpath ""
 561         walkpath := params.prefix
 562         if cut := strings.LastIndex(walkpath, "/"); cut >= 0 {
 563                 walkpath = walkpath[:cut]
 564         } else {
 565                 walkpath = ""
 566         }
 567
 568         type commonPrefix struct {
 569                 Prefix string
 570         }
 571         type listResp struct {
 572                 XMLName string `xml:"http://s3.amazonaws.com/doc/2006-03-01/ ListBucketResult"`
 573                 s3.ListResp
 574                 // s3.ListResp marshals an empty tag when
 575                 // CommonPrefixes is nil, which confuses some clients.
 576                 // Fix by using this nested struct instead.
 577                 CommonPrefixes []commonPrefix
 578                 // Similarly, we need omitempty here, because an empty
 579                 // tag confuses some clients (e.g.,
 580                 // github.com/aws/aws-sdk-net never terminates its
 581                 // paging loop).
 582                 NextMarker string `xml:"NextMarker,omitempty"`
 583                 // ListObjectsV2 has a KeyCount response field.
 584                 KeyCount int
 585         }
 586         resp := listResp{
 587                 ListResp: s3.ListResp{
 588                         Name:      bucket,
 589                         Prefix:    params.prefix,
 590                         Delimiter: params.delimiter,
 591                         Marker:    params.marker,
 592                         MaxKeys:   params.maxKeys,
 593                 },
 594         }
 595         commonPrefixes := map[string]bool{}
 596         err := walkFS(fs, strings.TrimSuffix(bucketdir+"/"+walkpath, "/"), true, func(path string, fi os.FileInfo) error {
 597                 if path == bucketdir {
 598                         return nil
 599                 }
 600                 path = path[len(bucketdir)+1:]
 601                 filesize := fi.Size()
 602                 if fi.IsDir() {
 603                         path += "/"
 604                         filesize = 0
 605                 }
 606                 if len(path) <= len(params.prefix) {
 607                         if path > params.prefix[:len(path)] {
 608                                 // with prefix "foobar", walking "fooz" means we're done
 609                                 return errDone
 610                         }
 611                         if path < params.prefix[:len(path)] {
 612                                 // with prefix "foobar", walking "foobag" is pointless
 613                                 return filepath.SkipDir
 614                         }
 615                         if fi.IsDir() && !strings.HasPrefix(params.prefix+"/", path) {
 616                                 // with prefix "foo/bar", walking "fo"
 617                                 // is pointless (but walking "foo" or
 618                                 // "foo/bar" is necessary)
 619                                 return filepath.SkipDir
 620                         }
 621                         if len(path) < len(params.prefix) {
 622                                 // can't skip anything, and this entry
 623                                 // isn't in the results, so just
 624                                 // continue descent
 625                                 return nil
 626                         }
 627                 } else {
 628                         if path[:len(params.prefix)] > params.prefix {
 629                                 // with prefix "foobar", nothing we
 630                                 // see after "foozzz" is relevant
 631                                 return errDone
 632                         }
 633                 }
 634                 if path < params.marker || path < params.prefix {
 635                         return nil
 636                 }
 637                 if fi.IsDir() && !h.Config.cluster.Collections.S3FolderObjects {
 638                         // Note we don't add anything to
 639                         // commonPrefixes here even if delimiter is
 640                         // "/". We descend into the directory, and
 641                         // return a commonPrefix only if we end up
 642                         // finding a regular file inside it.
 643                         return nil
 644                 }
 645                 if params.delimiter != "" {
 646                         idx := strings.Index(path[len(params.prefix):], params.delimiter)
 647                         if idx >= 0 {
 648                                 // with prefix "foobar" and delimiter
 649                                 // "z", when we hit "foobar/baz", we
 650                                 // add "/baz" to commonPrefixes and
 651                                 // stop descending.
 652                                 commonPrefixes[path[:len(params.prefix)+idx+1]] = true
 653                                 return filepath.SkipDir
 654                         }
 655                 }
 656                 if len(resp.Contents)+len(commonPrefixes) >= params.maxKeys {
 657                         resp.IsTruncated = true
 658                         if params.delimiter != "" {
 659                                 resp.NextMarker = path
 660                         }
 661                         return errDone
 662                 }
 663                 resp.Contents = append(resp.Contents, s3.Key{
 664                         Key:          path,
 665                         LastModified: fi.ModTime().UTC().Format("2006-01-02T15:04:05.999") + "Z",
 666                         Size:         filesize,
 667                 })
 668                 return nil
 669         })
 670         if err != nil && err != errDone {
 671                 http.Error(w, err.Error(), http.StatusInternalServerError)
 672                 return
 673         }
 674         if params.delimiter != "" {
 675                 resp.CommonPrefixes = make([]commonPrefix, 0, len(commonPrefixes))
 676                 for prefix := range commonPrefixes {
 677                         resp.CommonPrefixes = append(resp.CommonPrefixes, commonPrefix{prefix})
 678                 }
 679                 sort.Slice(resp.CommonPrefixes, func(i, j int) bool { return resp.CommonPrefixes[i].Prefix < resp.CommonPrefixes[j].Prefix })
 680         }
 681         resp.KeyCount = len(resp.Contents)
 682         w.Header().Set("Content-Type", "application/xml")
 683         io.WriteString(w, xml.Header)
 684         if err := xml.NewEncoder(w).Encode(resp); err != nil {
 685                 ctxlog.FromContext(r.Context()).WithError(err).Error("error writing xml response")
 686         }
 687 }