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 keepweb
   6
   7 import (
   8         "crypto/hmac"
   9         "crypto/sha256"
  10         "encoding/base64"
  11         "encoding/json"
  12         "encoding/xml"
  13         "errors"
  14         "fmt"
  15         "hash"
  16         "io"
  17         "mime"
  18         "net/http"
  19         "net/textproto"
  20         "net/url"
  21         "os"
  22         "path/filepath"
  23         "regexp"
  24         "sort"
  25         "strconv"
  26         "strings"
  27         "time"
  28
  29         "git.arvados.org/arvados.git/sdk/go/arvados"
  30         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  31 )
  32
  33 const (
  34         s3MaxKeys       = 1000
  35         s3SignAlgorithm = "AWS4-HMAC-SHA256"
  36         s3MaxClockSkew  = 5 * time.Minute
  37 )
  38
  39 type commonPrefix struct {
  40         Prefix string
  41 }
  42
  43 type listV1Resp struct {
  44         XMLName     string `xml:"http://s3.amazonaws.com/doc/2006-03-01/ ListBucketResult"`
  45         Name        string
  46         Prefix      string
  47         Delimiter   string
  48         Marker      string
  49         MaxKeys     int
  50         IsTruncated bool
  51         Contents    []s3Key
  52         // If we use a []string here, xml marshals an empty tag when
  53         // CommonPrefixes is nil, which confuses some clients.  Fix by
  54         // using this nested struct instead.
  55         CommonPrefixes []commonPrefix
  56         // Similarly, we need omitempty here, because an empty
  57         // tag confuses some clients (e.g.,
  58         // github.com/aws/aws-sdk-net never terminates its
  59         // paging loop).
  60         NextMarker string `xml:"NextMarker,omitempty"`
  61         // ListObjectsV2 has a KeyCount response field.
  62         KeyCount int
  63 }
  64
  65 type listV2Resp struct {
  66         XMLName               string `xml:"http://s3.amazonaws.com/doc/2006-03-01/ ListBucketResult"`
  67         IsTruncated           bool
  68         Contents              []s3Key
  69         Name                  string
  70         Prefix                string
  71         Delimiter             string
  72         MaxKeys               int
  73         CommonPrefixes        []commonPrefix
  74         EncodingType          string `xml:",omitempty"`
  75         KeyCount              int
  76         ContinuationToken     string `xml:",omitempty"`
  77         NextContinuationToken string `xml:",omitempty"`
  78         StartAfter            string `xml:",omitempty"`
  79 }
  80
  81 type s3Key struct {
  82         Key          string
  83         LastModified string
  84         Size         int64
  85         // The following fields are not populated, but are here in
  86         // case clients rely on the keys being present in xml
  87         // responses.
  88         ETag         string
  89         StorageClass string
  90         Owner        struct {
  91                 ID          string
  92                 DisplayName string
  93         }
  94 }
  95
  96 func hmacstring(msg string, key []byte) []byte {
  97         h := hmac.New(sha256.New, key)
  98         io.WriteString(h, msg)
  99         return h.Sum(nil)
 100 }
 101
 102 func hashdigest(h hash.Hash, payload string) string {
 103         io.WriteString(h, payload)
 104         return fmt.Sprintf("%x", h.Sum(nil))
 105 }
 106
 107 // Signing key for given secret key and request attrs.
 108 func s3signatureKey(key, datestamp, regionName, serviceName string) []byte {
 109         return hmacstring("aws4_request",
 110                 hmacstring(serviceName,
 111                         hmacstring(regionName,
 112                                 hmacstring(datestamp, []byte("AWS4"+key)))))
 113 }
 114
 115 // Canonical query string for S3 V4 signature: sorted keys, spaces
 116 // escaped as %20 instead of +, keyvalues joined with &.
 117 func s3querystring(u *url.URL) string {
 118         keys := make([]string, 0, len(u.Query()))
 119         values := make(map[string]string, len(u.Query()))
 120         for k, vs := range u.Query() {
 121                 k = strings.Replace(url.QueryEscape(k), "+", "%20", -1)
 122                 keys = append(keys, k)
 123                 for _, v := range vs {
 124                         v = strings.Replace(url.QueryEscape(v), "+", "%20", -1)
 125                         if values[k] != "" {
 126                                 values[k] += "&"
 127                         }
 128                         values[k] += k + "=" + v
 129                 }
 130         }
 131         sort.Strings(keys)
 132         for i, k := range keys {
 133                 keys[i] = values[k]
 134         }
 135         return strings.Join(keys, "&")
 136 }
 137
 138 var reMultipleSlashChars = regexp.MustCompile(`//+`)
 139
 140 func s3stringToSign(alg, scope, signedHeaders string, r *http.Request) (string, error) {
 141         timefmt, timestr := "20060102T150405Z", r.Header.Get("X-Amz-Date")
 142         if timestr == "" {
 143                 timefmt, timestr = time.RFC1123, r.Header.Get("Date")
 144         }
 145         t, err := time.Parse(timefmt, timestr)
 146         if err != nil {
 147                 return "", fmt.Errorf("invalid timestamp %q: %s", timestr, err)
 148         }
 149         if skew := time.Now().Sub(t); skew < -s3MaxClockSkew || skew > s3MaxClockSkew {
 150                 return "", errors.New("exceeded max clock skew")
 151         }
 152
 153         var canonicalHeaders string
 154         for _, h := range strings.Split(signedHeaders, ";") {
 155                 if h == "host" {
 156                         canonicalHeaders += h + ":" + r.Host + "\n"
 157                 } else {
 158                         canonicalHeaders += h + ":" + r.Header.Get(h) + "\n"
 159                 }
 160         }
 161
 162         normalizedPath := normalizePath(r.URL.Path)
 163         ctxlog.FromContext(r.Context()).Debugf("normalizedPath %q", normalizedPath)
 164         canonicalRequest := fmt.Sprintf("%s\n%s\n%s\n%s\n%s\n%s", r.Method, normalizedPath, s3querystring(r.URL), canonicalHeaders, signedHeaders, r.Header.Get("X-Amz-Content-Sha256"))
 165         ctxlog.FromContext(r.Context()).Debugf("s3stringToSign: canonicalRequest %s", canonicalRequest)
 166         return fmt.Sprintf("%s\n%s\n%s\n%s", alg, r.Header.Get("X-Amz-Date"), scope, hashdigest(sha256.New(), canonicalRequest)), nil
 167 }
 168
 169 func normalizePath(s string) string {
 170         // (url.URL).EscapedPath() would be incorrect here. AWS
 171         // documentation specifies the URL path should be normalized
 172         // according to RFC 3986, i.e., unescaping ALPHA / DIGIT / "-"
 173         // / "." / "_" / "~". The implication is that everything other
 174         // than those chars (and "/") _must_ be percent-encoded --
 175         // even chars like ";" and "," that are not normally
 176         // percent-encoded in paths.
 177         out := ""
 178         for _, c := range []byte(reMultipleSlashChars.ReplaceAllString(s, "/")) {
 179                 if (c >= 'a' && c <= 'z') ||
 180                         (c >= 'A' && c <= 'Z') ||
 181                         (c >= '0' && c <= '9') ||
 182                         c == '-' ||
 183                         c == '.' ||
 184                         c == '_' ||
 185                         c == '~' ||
 186                         c == '/' {
 187                         out += string(c)
 188                 } else {
 189                         out += fmt.Sprintf("%%%02X", c)
 190                 }
 191         }
 192         return out
 193 }
 194
 195 func s3signature(secretKey, scope, signedHeaders, stringToSign string) (string, error) {
 196         // scope is {datestamp}/{region}/{service}/aws4_request
 197         drs := strings.Split(scope, "/")
 198         if len(drs) != 4 {
 199                 return "", fmt.Errorf("invalid scope %q", scope)
 200         }
 201         key := s3signatureKey(secretKey, drs[0], drs[1], drs[2])
 202         return hashdigest(hmac.New(sha256.New, key), stringToSign), nil
 203 }
 204
 205 var v2tokenUnderscore = regexp.MustCompile(`^v2_[a-z0-9]{5}-gj3su-[a-z0-9]{15}_`)
 206
 207 func unescapeKey(key string) string {
 208         if v2tokenUnderscore.MatchString(key) {
 209                 // Entire Arvados token, with "/" replaced by "_" to
 210                 // avoid colliding with the Authorization header
 211                 // format.
 212                 return strings.Replace(key, "_", "/", -1)
 213         } else if s, err := url.PathUnescape(key); err == nil {
 214                 return s
 215         } else {
 216                 return key
 217         }
 218 }
 219
 220 // checks3signature verifies the given S3 V4 signature and returns the
 221 // Arvados token that corresponds to the given accessKey. An error is
 222 // returned if accessKey is not a valid token UUID or the signature
 223 // does not match.
 224 func (h *handler) checks3signature(r *http.Request) (string, error) {
 225         var key, scope, signedHeaders, signature string
 226         authstring := strings.TrimPrefix(r.Header.Get("Authorization"), s3SignAlgorithm+" ")
 227         for _, cmpt := range strings.Split(authstring, ",") {
 228                 cmpt = strings.TrimSpace(cmpt)
 229                 split := strings.SplitN(cmpt, "=", 2)
 230                 switch {
 231                 case len(split) != 2:
 232                         // (?) ignore
 233                 case split[0] == "Credential":
 234                         keyandscope := strings.SplitN(split[1], "/", 2)
 235                         if len(keyandscope) == 2 {
 236                                 key, scope = keyandscope[0], keyandscope[1]
 237                         }
 238                 case split[0] == "SignedHeaders":
 239                         signedHeaders = split[1]
 240                 case split[0] == "Signature":
 241                         signature = split[1]
 242                 }
 243         }
 244
 245         client := (&arvados.Client{
 246                 APIHost:  h.Cluster.Services.Controller.ExternalURL.Host,
 247                 Insecure: h.Cluster.TLS.Insecure,
 248         }).WithRequestID(r.Header.Get("X-Request-Id"))
 249         var aca arvados.APIClientAuthorization
 250         var secret string
 251         var err error
 252         if len(key) == 27 && key[5:12] == "-gj3su-" {
 253                 // Access key is the UUID of an Arvados token, secret
 254                 // key is the secret part.
 255                 ctx := arvados.ContextWithAuthorization(r.Context(), "Bearer "+h.Cluster.SystemRootToken)
 256                 err = client.RequestAndDecodeContext(ctx, &aca, "GET", "arvados/v1/api_client_authorizations/"+key, nil, nil)
 257                 secret = aca.APIToken
 258         } else {
 259                 // Access key and secret key are both an entire
 260                 // Arvados token or OIDC access token.
 261                 ctx := arvados.ContextWithAuthorization(r.Context(), "Bearer "+unescapeKey(key))
 262                 err = client.RequestAndDecodeContext(ctx, &aca, "GET", "arvados/v1/api_client_authorizations/current", nil, nil)
 263                 secret = key
 264         }
 265         if err != nil {
 266                 ctxlog.FromContext(r.Context()).WithError(err).WithField("UUID", key).Info("token lookup failed")
 267                 return "", errors.New("invalid access key")
 268         }
 269         stringToSign, err := s3stringToSign(s3SignAlgorithm, scope, signedHeaders, r)
 270         if err != nil {
 271                 return "", err
 272         }
 273         expect, err := s3signature(secret, scope, signedHeaders, stringToSign)
 274         if err != nil {
 275                 return "", err
 276         } else if expect != signature {
 277                 return "", fmt.Errorf("signature does not match (scope %q signedHeaders %q stringToSign %q)", scope, signedHeaders, stringToSign)
 278         }
 279         return aca.TokenV2(), nil
 280 }
 281
 282 func s3ErrorResponse(w http.ResponseWriter, s3code string, message string, resource string, code int) {
 283         w.Header().Set("Content-Type", "application/xml")
 284         w.Header().Set("X-Content-Type-Options", "nosniff")
 285         w.WriteHeader(code)
 286         var errstruct struct {
 287                 Code      string
 288                 Message   string
 289                 Resource  string
 290                 RequestId string
 291         }
 292         errstruct.Code = s3code
 293         errstruct.Message = message
 294         errstruct.Resource = resource
 295         errstruct.RequestId = ""
 296         enc := xml.NewEncoder(w)
 297         fmt.Fprint(w, xml.Header)
 298         enc.EncodeElement(errstruct, xml.StartElement{Name: xml.Name{Local: "Error"}})
 299 }
 300
 301 var NoSuchKey = "NoSuchKey"
 302 var NoSuchBucket = "NoSuchBucket"
 303 var InvalidArgument = "InvalidArgument"
 304 var InternalError = "InternalError"
 305 var UnauthorizedAccess = "UnauthorizedAccess"
 306 var InvalidRequest = "InvalidRequest"
 307 var SignatureDoesNotMatch = "SignatureDoesNotMatch"
 308
 309 var reRawQueryIndicatesAPI = regexp.MustCompile(`^[a-z]+(&|$)`)
 310
 311 // serveS3 handles r and returns true if r is a request from an S3
 312 // client, otherwise it returns false.
 313 func (h *handler) serveS3(w http.ResponseWriter, r *http.Request) bool {
 314         var token string
 315         if auth := r.Header.Get("Authorization"); strings.HasPrefix(auth, "AWS ") {
 316                 split := strings.SplitN(auth[4:], ":", 2)
 317                 if len(split) < 2 {
 318                         s3ErrorResponse(w, InvalidRequest, "malformed Authorization header", r.URL.Path, http.StatusUnauthorized)
 319                         return true
 320                 }
 321                 token = unescapeKey(split[0])
 322         } else if strings.HasPrefix(auth, s3SignAlgorithm+" ") {
 323                 t, err := h.checks3signature(r)
 324                 if err != nil {
 325                         s3ErrorResponse(w, SignatureDoesNotMatch, "signature verification failed: "+err.Error(), r.URL.Path, http.StatusForbidden)
 326                         return true
 327                 }
 328                 token = t
 329         } else {
 330                 return false
 331         }
 332
 333         fs, sess, tokenUser, err := h.Cache.GetSession(token)
 334         if err != nil {
 335                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 336                 return true
 337         }
 338         defer sess.Release()
 339         readfs := fs
 340         if writeMethod[r.Method] {
 341                 // Create a FileSystem for this request, to avoid
 342                 // exposing incomplete write operations to concurrent
 343                 // requests.
 344                 client := sess.client.WithRequestID(r.Header.Get("X-Request-Id"))
 345                 fs = client.SiteFileSystem(sess.keepclient)
 346                 fs.ForwardSlashNameSubstitution(h.Cluster.Collections.ForwardSlashNameSubstitution)
 347         }
 348
 349         var objectNameGiven bool
 350         var bucketName string
 351         fspath := "/by_id"
 352         if id := arvados.CollectionIDFromDNSName(r.Host); id != "" {
 353                 fspath += "/" + id
 354                 bucketName = id
 355                 objectNameGiven = strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 0
 356         } else {
 357                 bucketName = strings.SplitN(strings.TrimPrefix(r.URL.Path, "/"), "/", 2)[0]
 358                 objectNameGiven = strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 1
 359         }
 360         fspath += reMultipleSlashChars.ReplaceAllString(r.URL.Path, "/")
 361
 362         switch {
 363         case r.Method == http.MethodGet && !objectNameGiven:
 364                 // Path is "/{uuid}" or "/{uuid}/", has no object name
 365                 if _, ok := r.URL.Query()["versioning"]; ok {
 366                         // GetBucketVersioning
 367                         w.Header().Set("Content-Type", "application/xml")
 368                         io.WriteString(w, xml.Header)
 369                         fmt.Fprintln(w, `<VersioningConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/"/>`)
 370                 } else if _, ok = r.URL.Query()["location"]; ok {
 371                         // GetBucketLocation
 372                         w.Header().Set("Content-Type", "application/xml")
 373                         io.WriteString(w, xml.Header)
 374                         fmt.Fprintln(w, `<LocationConstraint><LocationConstraint xmlns="http://s3.amazonaws.com/doc/2006-03-01/">`+
 375                                 h.Cluster.ClusterID+
 376                                 `</LocationConstraint></LocationConstraint>`)
 377                 } else if reRawQueryIndicatesAPI.MatchString(r.URL.RawQuery) {
 378                         // GetBucketWebsite ("GET /bucketid/?website"), GetBucketTagging, etc.
 379                         s3ErrorResponse(w, InvalidRequest, "API not supported", r.URL.Path+"?"+r.URL.RawQuery, http.StatusBadRequest)
 380                 } else {
 381                         // ListObjects
 382                         h.s3list(bucketName, w, r, fs)
 383                 }
 384                 return true
 385         case r.Method == http.MethodGet || r.Method == http.MethodHead:
 386                 if reRawQueryIndicatesAPI.MatchString(r.URL.RawQuery) {
 387                         // GetObjectRetention ("GET /bucketid/objectid?retention&versionID=..."), etc.
 388                         s3ErrorResponse(w, InvalidRequest, "API not supported", r.URL.Path+"?"+r.URL.RawQuery, http.StatusBadRequest)
 389                         return true
 390                 }
 391                 fi, err := fs.Stat(fspath)
 392                 if r.Method == "HEAD" && !objectNameGiven {
 393                         // HeadBucket
 394                         if err == nil && fi.IsDir() {
 395                                 err = setFileInfoHeaders(w.Header(), fs, fspath)
 396                                 if err != nil {
 397                                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 398                                         return true
 399                                 }
 400                                 w.WriteHeader(http.StatusOK)
 401                         } else if os.IsNotExist(err) {
 402                                 s3ErrorResponse(w, NoSuchBucket, "The specified bucket does not exist.", r.URL.Path, http.StatusNotFound)
 403                         } else {
 404                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 405                         }
 406                         return true
 407                 }
 408                 if err == nil && fi.IsDir() && objectNameGiven && strings.HasSuffix(fspath, "/") && h.Cluster.Collections.S3FolderObjects {
 409                         err = setFileInfoHeaders(w.Header(), fs, fspath)
 410                         if err != nil {
 411                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 412                                 return true
 413                         }
 414                         w.Header().Set("Content-Type", "application/x-directory")
 415                         w.WriteHeader(http.StatusOK)
 416                         return true
 417                 }
 418                 if os.IsNotExist(err) ||
 419                         (err != nil && err.Error() == "not a directory") ||
 420                         (fi != nil && fi.IsDir()) {
 421                         s3ErrorResponse(w, NoSuchKey, "The specified key does not exist.", r.URL.Path, http.StatusNotFound)
 422                         return true
 423                 }
 424
 425                 if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 426                         http.Error(w, "Not permitted", http.StatusForbidden)
 427                         return true
 428                 }
 429                 h.logUploadOrDownload(r, sess.arvadosclient, fs, fspath, nil, tokenUser)
 430
 431                 // shallow copy r, and change URL path
 432                 r := *r
 433                 r.URL.Path = fspath
 434                 err = setFileInfoHeaders(w.Header(), fs, fspath)
 435                 if err != nil {
 436                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 437                         return true
 438                 }
 439                 http.FileServer(fs).ServeHTTP(w, &r)
 440                 return true
 441         case r.Method == http.MethodPut:
 442                 if reRawQueryIndicatesAPI.MatchString(r.URL.RawQuery) {
 443                         // PutObjectAcl ("PUT /bucketid/objectid?acl&versionID=..."), etc.
 444                         s3ErrorResponse(w, InvalidRequest, "API not supported", r.URL.Path+"?"+r.URL.RawQuery, http.StatusBadRequest)
 445                         return true
 446                 }
 447                 if !objectNameGiven {
 448                         s3ErrorResponse(w, InvalidArgument, "Missing object name in PUT request.", r.URL.Path, http.StatusBadRequest)
 449                         return true
 450                 }
 451                 var objectIsDir bool
 452                 if strings.HasSuffix(fspath, "/") {
 453                         if !h.Cluster.Collections.S3FolderObjects {
 454                                 s3ErrorResponse(w, InvalidArgument, "invalid object name: trailing slash", r.URL.Path, http.StatusBadRequest)
 455                                 return true
 456                         }
 457                         n, err := r.Body.Read(make([]byte, 1))
 458                         if err != nil && err != io.EOF {
 459                                 s3ErrorResponse(w, InternalError, fmt.Sprintf("error reading request body: %s", err), r.URL.Path, http.StatusInternalServerError)
 460                                 return true
 461                         } else if n > 0 {
 462                                 s3ErrorResponse(w, InvalidArgument, "cannot create object with trailing '/' char unless content is empty", r.URL.Path, http.StatusBadRequest)
 463                                 return true
 464                         } else if strings.SplitN(r.Header.Get("Content-Type"), ";", 2)[0] != "application/x-directory" {
 465                                 s3ErrorResponse(w, InvalidArgument, "cannot create object with trailing '/' char unless Content-Type is 'application/x-directory'", r.URL.Path, http.StatusBadRequest)
 466                                 return true
 467                         }
 468                         // Given PUT "foo/bar/", we'll use "foo/bar/."
 469                         // in the "ensure parents exist" block below,
 470                         // and then we'll be done.
 471                         fspath += "."
 472                         objectIsDir = true
 473                 }
 474                 fi, err := fs.Stat(fspath)
 475                 if err != nil && err.Error() == "not a directory" {
 476                         // requested foo/bar, but foo is a file
 477                         s3ErrorResponse(w, InvalidArgument, "object name conflicts with existing object", r.URL.Path, http.StatusBadRequest)
 478                         return true
 479                 }
 480                 if strings.HasSuffix(r.URL.Path, "/") && err == nil && !fi.IsDir() {
 481                         // requested foo/bar/, but foo/bar is a file
 482                         s3ErrorResponse(w, InvalidArgument, "object name conflicts with existing object", r.URL.Path, http.StatusBadRequest)
 483                         return true
 484                 }
 485                 // create missing parent/intermediate directories, if any
 486                 for i, c := range fspath {
 487                         if i > 0 && c == '/' {
 488                                 dir := fspath[:i]
 489                                 if strings.HasSuffix(dir, "/") {
 490                                         err = errors.New("invalid object name (consecutive '/' chars)")
 491                                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 492                                         return true
 493                                 }
 494                                 err = fs.Mkdir(dir, 0755)
 495                                 if errors.Is(err, arvados.ErrInvalidArgument) || errors.Is(err, arvados.ErrInvalidOperation) {
 496                                         // Cannot create a directory
 497                                         // here.
 498                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 499                                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 500                                         return true
 501                                 } else if err != nil && !os.IsExist(err) {
 502                                         err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 503                                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 504                                         return true
 505                                 }
 506                         }
 507                 }
 508                 if !objectIsDir {
 509                         f, err := fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 510                         if os.IsNotExist(err) {
 511                                 f, err = fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 512                         }
 513                         if err != nil {
 514                                 err = fmt.Errorf("open %q failed: %w", r.URL.Path, err)
 515                                 s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 516                                 return true
 517                         }
 518                         defer f.Close()
 519
 520                         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 521                                 http.Error(w, "Not permitted", http.StatusForbidden)
 522                                 return true
 523                         }
 524                         h.logUploadOrDownload(r, sess.arvadosclient, fs, fspath, nil, tokenUser)
 525
 526                         _, err = io.Copy(f, r.Body)
 527                         if err != nil {
 528                                 err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 529                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 530                                 return true
 531                         }
 532                         err = f.Close()
 533                         if err != nil {
 534                                 err = fmt.Errorf("write to %q failed: close: %w", r.URL.Path, err)
 535                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusBadGateway)
 536                                 return true
 537                         }
 538                 }
 539                 err = h.syncCollection(fs, readfs, fspath)
 540                 if err != nil {
 541                         err = fmt.Errorf("sync failed: %w", err)
 542                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 543                         return true
 544                 }
 545                 w.WriteHeader(http.StatusOK)
 546                 return true
 547         case r.Method == http.MethodDelete:
 548                 if reRawQueryIndicatesAPI.MatchString(r.URL.RawQuery) {
 549                         // DeleteObjectTagging ("DELETE /bucketid/objectid?tagging&versionID=..."), etc.
 550                         s3ErrorResponse(w, InvalidRequest, "API not supported", r.URL.Path+"?"+r.URL.RawQuery, http.StatusBadRequest)
 551                         return true
 552                 }
 553                 if !objectNameGiven || r.URL.Path == "/" {
 554                         s3ErrorResponse(w, InvalidArgument, "missing object name in DELETE request", r.URL.Path, http.StatusBadRequest)
 555                         return true
 556                 }
 557                 if strings.HasSuffix(fspath, "/") {
 558                         fspath = strings.TrimSuffix(fspath, "/")
 559                         fi, err := fs.Stat(fspath)
 560                         if os.IsNotExist(err) {
 561                                 w.WriteHeader(http.StatusNoContent)
 562                                 return true
 563                         } else if err != nil {
 564                                 s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 565                                 return true
 566                         } else if !fi.IsDir() {
 567                                 // if "foo" exists and is a file, then
 568                                 // "foo/" doesn't exist, so we say
 569                                 // delete was successful.
 570                                 w.WriteHeader(http.StatusNoContent)
 571                                 return true
 572                         }
 573                 } else if fi, err := fs.Stat(fspath); err == nil && fi.IsDir() {
 574                         // if "foo" is a dir, it is visible via S3
 575                         // only as "foo/", not "foo" -- so we leave
 576                         // the dir alone and return 204 to indicate
 577                         // that "foo" does not exist.
 578                         w.WriteHeader(http.StatusNoContent)
 579                         return true
 580                 }
 581                 err = fs.Remove(fspath)
 582                 if os.IsNotExist(err) {
 583                         w.WriteHeader(http.StatusNoContent)
 584                         return true
 585                 }
 586                 if err != nil {
 587                         err = fmt.Errorf("rm failed: %w", err)
 588                         s3ErrorResponse(w, InvalidArgument, err.Error(), r.URL.Path, http.StatusBadRequest)
 589                         return true
 590                 }
 591                 err = h.syncCollection(fs, readfs, fspath)
 592                 if err != nil {
 593                         err = fmt.Errorf("sync failed: %w", err)
 594                         s3ErrorResponse(w, InternalError, err.Error(), r.URL.Path, http.StatusInternalServerError)
 595                         return true
 596                 }
 597                 w.WriteHeader(http.StatusNoContent)
 598                 return true
 599         default:
 600                 s3ErrorResponse(w, InvalidRequest, "method not allowed", r.URL.Path, http.StatusMethodNotAllowed)
 601                 return true
 602         }
 603 }
 604
 605 // Save modifications to the indicated collection in srcfs, then (if
 606 // successful) ensure they are also reflected in dstfs.
 607 func (h *handler) syncCollection(srcfs, dstfs arvados.CustomFileSystem, path string) error {
 608         coll, _ := h.determineCollection(srcfs, path)
 609         if coll == nil || coll.UUID == "" {
 610                 return errors.New("could not determine collection to sync")
 611         }
 612         d, err := srcfs.OpenFile("by_id/"+coll.UUID, os.O_RDWR, 0777)
 613         if err != nil {
 614                 return err
 615         }
 616         defer d.Close()
 617         err = d.Sync()
 618         if err != nil {
 619                 return err
 620         }
 621         snap, err := d.Snapshot()
 622         if err != nil {
 623                 return err
 624         }
 625         dstd, err := dstfs.OpenFile("by_id/"+coll.UUID, os.O_RDWR, 0777)
 626         if err != nil {
 627                 return err
 628         }
 629         defer dstd.Close()
 630         return dstd.Splice(snap)
 631 }
 632
 633 func setFileInfoHeaders(header http.Header, fs arvados.CustomFileSystem, path string) error {
 634         maybeEncode := func(s string) string {
 635                 for _, c := range s {
 636                         if c > '\u007f' || c < ' ' {
 637                                 return mime.BEncoding.Encode("UTF-8", s)
 638                         }
 639                 }
 640                 return s
 641         }
 642         path = strings.TrimSuffix(path, "/")
 643         var props map[string]interface{}
 644         for {
 645                 fi, err := fs.Stat(path)
 646                 if err != nil {
 647                         return err
 648                 }
 649                 switch src := fi.Sys().(type) {
 650                 case *arvados.Collection:
 651                         props = src.Properties
 652                 case *arvados.Group:
 653                         props = src.Properties
 654                 default:
 655                         if err, ok := src.(error); ok {
 656                                 return err
 657                         }
 658                         // Try parent
 659                         cut := strings.LastIndexByte(path, '/')
 660                         if cut < 0 {
 661                                 return nil
 662                         }
 663                         path = path[:cut]
 664                         continue
 665                 }
 666                 break
 667         }
 668         for k, v := range props {
 669                 if !validMIMEHeaderKey(k) {
 670                         continue
 671                 }
 672                 k = "x-amz-meta-" + k
 673                 if s, ok := v.(string); ok {
 674                         header.Set(k, maybeEncode(s))
 675                 } else if j, err := json.Marshal(v); err == nil {
 676                         header.Set(k, maybeEncode(string(j)))
 677                 }
 678         }
 679         return nil
 680 }
 681
 682 func validMIMEHeaderKey(k string) bool {
 683         check := "z-" + k
 684         return check != textproto.CanonicalMIMEHeaderKey(check)
 685 }
 686
 687 // Call fn on the given path (directory) and its contents, in
 688 // lexicographic order.
 689 //
 690 // If isRoot==true and path is not a directory, return nil.
 691 //
 692 // If fn returns filepath.SkipDir when called on a directory, don't
 693 // descend into that directory.
 694 func walkFS(fs arvados.CustomFileSystem, path string, isRoot bool, fn func(path string, fi os.FileInfo) error) error {
 695         if isRoot {
 696                 fi, err := fs.Stat(path)
 697                 if os.IsNotExist(err) || (err == nil && !fi.IsDir()) {
 698                         return nil
 699                 } else if err != nil {
 700                         return err
 701                 }
 702                 err = fn(path, fi)
 703                 if err == filepath.SkipDir {
 704                         return nil
 705                 } else if err != nil {
 706                         return err
 707                 }
 708         }
 709         f, err := fs.Open(path)
 710         if os.IsNotExist(err) && isRoot {
 711                 return nil
 712         } else if err != nil {
 713                 return fmt.Errorf("open %q: %w", path, err)
 714         }
 715         defer f.Close()
 716         if path == "/" {
 717                 path = ""
 718         }
 719         fis, err := f.Readdir(-1)
 720         if err != nil {
 721                 return err
 722         }
 723         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 724         for _, fi := range fis {
 725                 err = fn(path+"/"+fi.Name(), fi)
 726                 if err == filepath.SkipDir {
 727                         continue
 728                 } else if err != nil {
 729                         return err
 730                 }
 731                 if fi.IsDir() {
 732                         err = walkFS(fs, path+"/"+fi.Name(), false, fn)
 733                         if err != nil {
 734                                 return err
 735                         }
 736                 }
 737         }
 738         return nil
 739 }
 740
 741 var errDone = errors.New("done")
 742
 743 func (h *handler) s3list(bucket string, w http.ResponseWriter, r *http.Request, fs arvados.CustomFileSystem) {
 744         var params struct {
 745                 v2                bool
 746                 delimiter         string
 747                 maxKeys           int
 748                 prefix            string
 749                 marker            string // decoded continuationToken (v2) or provided by client (v1)
 750                 startAfter        string // v2
 751                 continuationToken string // v2
 752                 encodingTypeURL   bool   // v2
 753         }
 754         params.delimiter = r.FormValue("delimiter")
 755         if mk, _ := strconv.ParseInt(r.FormValue("max-keys"), 10, 64); mk > 0 && mk < s3MaxKeys {
 756                 params.maxKeys = int(mk)
 757         } else {
 758                 params.maxKeys = s3MaxKeys
 759         }
 760         params.prefix = r.FormValue("prefix")
 761         switch r.FormValue("list-type") {
 762         case "":
 763         case "2":
 764                 params.v2 = true
 765         default:
 766                 http.Error(w, "invalid list-type parameter", http.StatusBadRequest)
 767                 return
 768         }
 769         if params.v2 {
 770                 params.continuationToken = r.FormValue("continuation-token")
 771                 marker, err := base64.StdEncoding.DecodeString(params.continuationToken)
 772                 if err != nil {
 773                         http.Error(w, "invalid continuation token", http.StatusBadRequest)
 774                         return
 775                 }
 776                 // marker and start-after perform the same function,
 777                 // but we keep them separate so we can repeat them
 778                 // back to the client in the response.
 779                 params.marker = string(marker)
 780                 params.startAfter = r.FormValue("start-after")
 781                 switch r.FormValue("encoding-type") {
 782                 case "":
 783                 case "url":
 784                         params.encodingTypeURL = true
 785                 default:
 786                         http.Error(w, "invalid encoding-type parameter", http.StatusBadRequest)
 787                         return
 788                 }
 789         } else {
 790                 // marker is functionally equivalent to start-after.
 791                 params.marker = r.FormValue("marker")
 792         }
 793
 794         // startAfter is params.marker or params.startAfter, whichever
 795         // comes last.
 796         startAfter := params.startAfter
 797         if startAfter < params.marker {
 798                 startAfter = params.marker
 799         }
 800
 801         bucketdir := "by_id/" + bucket
 802         // walkpath is the directory (relative to bucketdir) we need
 803         // to walk: the innermost directory that is guaranteed to
 804         // contain all paths that have the requested prefix. Examples:
 805         // prefix "foo/bar"  => walkpath "foo"
 806         // prefix "foo/bar/" => walkpath "foo/bar"
 807         // prefix "foo"      => walkpath ""
 808         // prefix ""         => walkpath ""
 809         walkpath := params.prefix
 810         if cut := strings.LastIndex(walkpath, "/"); cut >= 0 {
 811                 walkpath = walkpath[:cut]
 812         } else {
 813                 walkpath = ""
 814         }
 815
 816         resp := listV2Resp{
 817                 Name:              bucket,
 818                 Prefix:            params.prefix,
 819                 Delimiter:         params.delimiter,
 820                 MaxKeys:           params.maxKeys,
 821                 ContinuationToken: r.FormValue("continuation-token"),
 822                 StartAfter:        params.startAfter,
 823         }
 824
 825         // nextMarker will be the last path we add to either
 826         // resp.Contents or commonPrefixes.  It will be included in
 827         // the response as NextMarker or NextContinuationToken if
 828         // needed.
 829         nextMarker := ""
 830
 831         commonPrefixes := map[string]bool{}
 832         full := false
 833         err := walkFS(fs, strings.TrimSuffix(bucketdir+"/"+walkpath, "/"), true, func(path string, fi os.FileInfo) error {
 834                 if path == bucketdir {
 835                         return nil
 836                 }
 837                 path = path[len(bucketdir)+1:]
 838                 filesize := fi.Size()
 839                 if fi.IsDir() {
 840                         path += "/"
 841                         filesize = 0
 842                 }
 843                 if strings.HasPrefix(params.prefix, path) && params.prefix != path {
 844                         // Descend into subtree until we reach desired prefix
 845                         return nil
 846                 } else if path < params.prefix {
 847                         // Not an ancestor or descendant of desired
 848                         // prefix, therefore none of its descendants
 849                         // can be either -- skip
 850                         return filepath.SkipDir
 851                 } else if path > params.prefix && !strings.HasPrefix(path, params.prefix) {
 852                         // We must have traversed everything under
 853                         // desired prefix
 854                         return errDone
 855                 } else if path == startAfter {
 856                         // Skip startAfter itself, just descend into
 857                         // subtree
 858                         return nil
 859                 } else if strings.HasPrefix(startAfter, path) {
 860                         // Descend into subtree in case it contains
 861                         // something after startAfter
 862                         return nil
 863                 } else if path < startAfter {
 864                         // Skip ahead until we reach startAfter
 865                         return filepath.SkipDir
 866                 }
 867                 if fi.IsDir() && !h.Cluster.Collections.S3FolderObjects {
 868                         // Note we don't add anything to
 869                         // commonPrefixes here even if delimiter is
 870                         // "/". We descend into the directory, and
 871                         // return a commonPrefix only if we end up
 872                         // finding a regular file inside it.
 873                         return nil
 874                 }
 875                 if params.delimiter != "" {
 876                         idx := strings.Index(path[len(params.prefix):], params.delimiter)
 877                         if idx >= 0 {
 878                                 // with prefix "foobar" and delimiter
 879                                 // "z", when we hit "foobar/baz", we
 880                                 // add "/baz" to commonPrefixes and
 881                                 // stop descending.
 882                                 prefix := path[:len(params.prefix)+idx+1]
 883                                 if prefix == startAfter {
 884                                         return nil
 885                                 } else if prefix < startAfter && !strings.HasPrefix(startAfter, prefix) {
 886                                         return nil
 887                                 } else if full {
 888                                         resp.IsTruncated = true
 889                                         return errDone
 890                                 } else {
 891                                         commonPrefixes[prefix] = true
 892                                         nextMarker = prefix
 893                                         full = len(resp.Contents)+len(commonPrefixes) >= params.maxKeys
 894                                         return filepath.SkipDir
 895                                 }
 896                         }
 897                 }
 898                 if full {
 899                         resp.IsTruncated = true
 900                         return errDone
 901                 }
 902                 resp.Contents = append(resp.Contents, s3Key{
 903                         Key:          path,
 904                         LastModified: fi.ModTime().UTC().Format("2006-01-02T15:04:05.999") + "Z",
 905                         Size:         filesize,
 906                 })
 907                 nextMarker = path
 908                 full = len(resp.Contents)+len(commonPrefixes) >= params.maxKeys
 909                 return nil
 910         })
 911         if err != nil && err != errDone {
 912                 http.Error(w, err.Error(), http.StatusInternalServerError)
 913                 return
 914         }
 915         if params.delimiter == "" && !params.v2 || !resp.IsTruncated {
 916                 nextMarker = ""
 917         }
 918         if params.delimiter != "" {
 919                 resp.CommonPrefixes = make([]commonPrefix, 0, len(commonPrefixes))
 920                 for prefix := range commonPrefixes {
 921                         resp.CommonPrefixes = append(resp.CommonPrefixes, commonPrefix{prefix})
 922                 }
 923                 sort.Slice(resp.CommonPrefixes, func(i, j int) bool { return resp.CommonPrefixes[i].Prefix < resp.CommonPrefixes[j].Prefix })
 924         }
 925         resp.KeyCount = len(resp.Contents)
 926         var respV1orV2 interface{}
 927
 928         if params.encodingTypeURL {
 929                 // https://docs.aws.amazon.com/AmazonS3/latest/API/API_ListObjectsV2.html
 930                 // "If you specify the encoding-type request
 931                 // parameter, Amazon S3 includes this element in the
 932                 // response, and returns encoded key name values in
 933                 // the following response elements:
 934                 //
 935                 // Delimiter, Prefix, Key, and StartAfter.
 936                 //
 937                 //      Type: String
 938                 //
 939                 // Valid Values: url"
 940                 //
 941                 // This is somewhat vague but in practice it appears
 942                 // to mean x-www-form-urlencoded as in RFC1866 8.2.1
 943                 // para 1 (encode space as "+") rather than straight
 944                 // percent-encoding as in RFC1738 2.2.  Presumably,
 945                 // the intent is to allow the client to decode XML and
 946                 // then paste the strings directly into another URI
 947                 // query or POST form like "https://host/path?foo=" +
 948                 // foo + "&bar=" + bar.
 949                 resp.EncodingType = "url"
 950                 resp.Delimiter = url.QueryEscape(resp.Delimiter)
 951                 resp.Prefix = url.QueryEscape(resp.Prefix)
 952                 resp.StartAfter = url.QueryEscape(resp.StartAfter)
 953                 for i, ent := range resp.Contents {
 954                         ent.Key = url.QueryEscape(ent.Key)
 955                         resp.Contents[i] = ent
 956                 }
 957                 for i, ent := range resp.CommonPrefixes {
 958                         ent.Prefix = url.QueryEscape(ent.Prefix)
 959                         resp.CommonPrefixes[i] = ent
 960                 }
 961         }
 962
 963         if params.v2 {
 964                 resp.NextContinuationToken = base64.StdEncoding.EncodeToString([]byte(nextMarker))
 965                 respV1orV2 = resp
 966         } else {
 967                 respV1orV2 = listV1Resp{
 968                         CommonPrefixes: resp.CommonPrefixes,
 969                         NextMarker:     nextMarker,
 970                         KeyCount:       resp.KeyCount,
 971                         IsTruncated:    resp.IsTruncated,
 972                         Name:           bucket,
 973                         Prefix:         params.prefix,
 974                         Delimiter:      params.delimiter,
 975                         Marker:         params.marker,
 976                         MaxKeys:        params.maxKeys,
 977                         Contents:       resp.Contents,
 978                 }
 979         }
 980
 981         w.Header().Set("Content-Type", "application/xml")
 982         io.WriteString(w, xml.Header)
 983         if err := xml.NewEncoder(w).Encode(respV1orV2); err != nil {
 984                 ctxlog.FromContext(r.Context()).WithError(err).Error("error writing xml response")
 985         }
 986 }