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