services/keep-web/handler.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         "encoding/json"
   9         "errors"
  10         "fmt"
  11         "html"
  12         "html/template"
  13         "io"
  14         "net/http"
  15         "net/url"
  16         "os"
  17         "sort"
  18         "strconv"
  19         "strings"
  20         "sync"
  21
  22         "git.arvados.org/arvados.git/lib/cmd"
  23         "git.arvados.org/arvados.git/lib/webdavfs"
  24         "git.arvados.org/arvados.git/sdk/go/arvados"
  25         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  26         "git.arvados.org/arvados.git/sdk/go/auth"
  27         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  28         "git.arvados.org/arvados.git/sdk/go/httpserver"
  29         "git.arvados.org/arvados.git/sdk/go/keepclient"
  30         "github.com/sirupsen/logrus"
  31         "golang.org/x/net/webdav"
  32 )
  33
  34 type handler struct {
  35         Cache     cache
  36         Cluster   *arvados.Cluster
  37         setupOnce sync.Once
  38 }
  39
  40 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  41
  42 var notFoundMessage = "Not Found"
  43 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  44
  45 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  46 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  47 // otherwise "".
  48 func parseCollectionIDFromURL(s string) string {
  49         if arvadosclient.UUIDMatch(s) {
  50                 return s
  51         }
  52         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  53                 return pdh
  54         }
  55         return ""
  56 }
  57
  58 func (h *handler) setup() {
  59         keepclient.DefaultBlockCache.MaxBlocks = h.Cluster.Collections.WebDAVCache.MaxBlockEntries
  60 }
  61
  62 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  63         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  64 }
  65
  66 type errorWithHTTPStatus interface {
  67         HTTPStatus() int
  68 }
  69
  70 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  71 // sends an HTTP header indicating success, updateOnSuccess first
  72 // calls the provided update func. If the update func fails, an error
  73 // response is sent (using the error's HTTP status or 500 if none),
  74 // and the status code and body sent by the handler are ignored (all
  75 // response writes return the update error).
  76 type updateOnSuccess struct {
  77         httpserver.ResponseWriter
  78         logger     logrus.FieldLogger
  79         update     func() error
  80         sentHeader bool
  81         err        error
  82 }
  83
  84 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  85         if !uos.sentHeader {
  86                 uos.WriteHeader(http.StatusOK)
  87         }
  88         if uos.err != nil {
  89                 return 0, uos.err
  90         }
  91         return uos.ResponseWriter.Write(p)
  92 }
  93
  94 func (uos *updateOnSuccess) WriteHeader(code int) {
  95         if !uos.sentHeader {
  96                 uos.sentHeader = true
  97                 if code >= 200 && code < 400 {
  98                         if uos.err = uos.update(); uos.err != nil {
  99                                 code := http.StatusInternalServerError
 100                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 101                                         code = he.HTTPStatus()
 102                                 }
 103                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 104                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 105                                 return
 106                         }
 107                 }
 108         }
 109         uos.ResponseWriter.WriteHeader(code)
 110 }
 111
 112 var (
 113         corsAllowHeadersHeader = strings.Join([]string{
 114                 "Authorization", "Content-Type", "Range",
 115                 // WebDAV request headers:
 116                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 117         }, ", ")
 118         writeMethod = map[string]bool{
 119                 "COPY":      true,
 120                 "DELETE":    true,
 121                 "LOCK":      true,
 122                 "MKCOL":     true,
 123                 "MOVE":      true,
 124                 "PROPPATCH": true,
 125                 "PUT":       true,
 126                 "RMCOL":     true,
 127                 "UNLOCK":    true,
 128         }
 129         webdavMethod = map[string]bool{
 130                 "COPY":      true,
 131                 "DELETE":    true,
 132                 "LOCK":      true,
 133                 "MKCOL":     true,
 134                 "MOVE":      true,
 135                 "OPTIONS":   true,
 136                 "PROPFIND":  true,
 137                 "PROPPATCH": true,
 138                 "PUT":       true,
 139                 "RMCOL":     true,
 140                 "UNLOCK":    true,
 141         }
 142         browserMethod = map[string]bool{
 143                 "GET":  true,
 144                 "HEAD": true,
 145                 "POST": true,
 146         }
 147         // top-level dirs to serve with siteFS
 148         siteFSDir = map[string]bool{
 149                 "":      true, // root directory
 150                 "by_id": true,
 151                 "users": true,
 152         }
 153 )
 154
 155 func stripDefaultPort(host string) string {
 156         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 157         u := &url.URL{Host: host}
 158         if p := u.Port(); p == "80" || p == "443" {
 159                 return strings.ToLower(u.Hostname())
 160         } else {
 161                 return strings.ToLower(host)
 162         }
 163 }
 164
 165 // CheckHealth implements service.Handler.
 166 func (h *handler) CheckHealth() error {
 167         return nil
 168 }
 169
 170 // Done implements service.Handler.
 171 func (h *handler) Done() <-chan struct{} {
 172         return nil
 173 }
 174
 175 // ServeHTTP implements http.Handler.
 176 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 177         h.setupOnce.Do(h.setup)
 178
 179         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 180                 r.URL.Scheme = xfp
 181         }
 182
 183         w := httpserver.WrapResponseWriter(wOrig)
 184
 185         if method := r.Header.Get("Access-Control-Request-Method"); method != "" && r.Method == "OPTIONS" {
 186                 if !browserMethod[method] && !webdavMethod[method] {
 187                         w.WriteHeader(http.StatusMethodNotAllowed)
 188                         return
 189                 }
 190                 w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
 191                 w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
 192                 w.Header().Set("Access-Control-Allow-Origin", "*")
 193                 w.Header().Set("Access-Control-Max-Age", "86400")
 194                 return
 195         }
 196
 197         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 198                 w.WriteHeader(http.StatusMethodNotAllowed)
 199                 return
 200         }
 201
 202         if r.Header.Get("Origin") != "" {
 203                 // Allow simple cross-origin requests without user
 204                 // credentials ("user credentials" as defined by CORS,
 205                 // i.e., cookies, HTTP authentication, and client-side
 206                 // SSL certificates. See
 207                 // http://www.w3.org/TR/cors/#user-credentials).
 208                 w.Header().Set("Access-Control-Allow-Origin", "*")
 209                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 210         }
 211
 212         if h.serveS3(w, r) {
 213                 return
 214         }
 215
 216         webdavPrefix := ""
 217         arvPath := r.URL.Path
 218         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 219                 // Enable a proxy (e.g., container log handler in
 220                 // controller) to satisfy a request for path
 221                 // "/foo/bar/baz.txt" using content from
 222                 // "//abc123-4.internal/bar/baz.txt", by adding a
 223                 // request header "X-Webdav-Prefix: /foo"
 224                 if !strings.HasPrefix(arvPath, prefix) {
 225                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 226                         return
 227                 }
 228                 arvPath = r.URL.Path[len(prefix):]
 229                 if arvPath == "" {
 230                         arvPath = "/"
 231                 }
 232                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 233                 webdavPrefix = prefix
 234         }
 235         pathParts := strings.Split(arvPath[1:], "/")
 236
 237         var stripParts int
 238         var collectionID string
 239         var tokens []string
 240         var reqTokens []string
 241         var pathToken bool
 242         var attachment bool
 243         var useSiteFS bool
 244         credentialsOK := h.Cluster.Collections.TrustAllContent
 245         reasonNotAcceptingCredentials := ""
 246
 247         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 248                 credentialsOK = true
 249                 attachment = true
 250         } else if r.FormValue("disposition") == "attachment" {
 251                 attachment = true
 252         }
 253
 254         if !credentialsOK {
 255                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 256                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 257         }
 258
 259         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 260                 // http://ID.collections.example/PATH...
 261                 credentialsOK = true
 262         } else if r.URL.Path == "/status.json" {
 263                 h.serveStatus(w, r)
 264                 return
 265         } else if siteFSDir[pathParts[0]] {
 266                 useSiteFS = true
 267         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 268                 // /c=ID[/PATH...]
 269                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 270                 stripParts = 1
 271         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 272                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 273                         // /collections/download/ID/TOKEN/PATH...
 274                         collectionID = parseCollectionIDFromURL(pathParts[2])
 275                         tokens = []string{pathParts[3]}
 276                         stripParts = 4
 277                         pathToken = true
 278                 } else {
 279                         // /collections/ID/PATH...
 280                         collectionID = parseCollectionIDFromURL(pathParts[1])
 281                         stripParts = 2
 282                         // This path is only meant to work for public
 283                         // data. Tokens provided with the request are
 284                         // ignored.
 285                         credentialsOK = false
 286                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 287                 }
 288         }
 289
 290         forceReload := false
 291         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 292                 forceReload = true
 293         }
 294
 295         if credentialsOK {
 296                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 297         }
 298
 299         formToken := r.FormValue("api_token")
 300         origin := r.Header.Get("Origin")
 301         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 302         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 303         safeAttachment := attachment && r.URL.Query().Get("api_token") == ""
 304         if formToken == "" {
 305                 // No token to use or redact.
 306         } else if safeAjax || safeAttachment {
 307                 // If this is a cross-origin request, the URL won't
 308                 // appear in the browser's address bar, so
 309                 // substituting a clipboard-safe URL is pointless.
 310                 // Redirect-with-cookie wouldn't work anyway, because
 311                 // it's not safe to allow third-party use of our
 312                 // cookie.
 313                 //
 314                 // If we're supplying an attachment, we don't need to
 315                 // convert POST to GET to avoid the "really resubmit
 316                 // form?" problem, so provided the token isn't
 317                 // embedded in the URL, there's no reason to do
 318                 // redirect-with-cookie in this case either.
 319                 reqTokens = append(reqTokens, formToken)
 320         } else if browserMethod[r.Method] {
 321                 // If this is a page view, and the client provided a
 322                 // token via query string or POST body, we must put
 323                 // the token in an HttpOnly cookie, and redirect to an
 324                 // equivalent URL with the query param redacted and
 325                 // method = GET.
 326                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 327                 return
 328         }
 329
 330         targetPath := pathParts[stripParts:]
 331         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 332                 // http://ID.example/t=TOKEN/PATH...
 333                 // /c=ID/t=TOKEN/PATH...
 334                 //
 335                 // This form must only be used to pass scoped tokens
 336                 // that give permission for a single collection. See
 337                 // FormValue case above.
 338                 tokens = []string{targetPath[0][2:]}
 339                 pathToken = true
 340                 targetPath = targetPath[1:]
 341                 stripParts++
 342         }
 343
 344         fsprefix := ""
 345         if useSiteFS {
 346                 if writeMethod[r.Method] {
 347                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 348                         return
 349                 }
 350                 if len(reqTokens) == 0 {
 351                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 352                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 353                         return
 354                 }
 355                 tokens = reqTokens
 356         } else if collectionID == "" {
 357                 http.Error(w, notFoundMessage, http.StatusNotFound)
 358                 return
 359         } else {
 360                 fsprefix = "by_id/" + collectionID + "/"
 361         }
 362
 363         if tokens == nil {
 364                 tokens = reqTokens
 365                 if h.Cluster.Users.AnonymousUserToken != "" {
 366                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 367                 }
 368         }
 369
 370         if len(targetPath) > 0 && targetPath[0] == "_" {
 371                 // If a collection has a directory called "t=foo" or
 372                 // "_", it can be served at
 373                 // //collections.example/_/t=foo/ or
 374                 // //collections.example/_/_/ respectively:
 375                 // //collections.example/t=foo/ won't work because
 376                 // t=foo will be interpreted as a token "foo".
 377                 targetPath = targetPath[1:]
 378                 stripParts++
 379         }
 380
 381         dirOpenMode := os.O_RDONLY
 382         if writeMethod[r.Method] {
 383                 dirOpenMode = os.O_RDWR
 384         }
 385
 386         var tokenValid bool
 387         var tokenScopeProblem bool
 388         var token string
 389         var tokenUser *arvados.User
 390         var sessionFS arvados.CustomFileSystem
 391         var session *cachedSession
 392         var collectionDir arvados.File
 393         for _, token = range tokens {
 394                 var statusErr errorWithHTTPStatus
 395                 fs, sess, user, err := h.Cache.GetSession(token)
 396                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 397                         // bad token
 398                         continue
 399                 } else if err != nil {
 400                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 401                         return
 402                 }
 403                 if token != h.Cluster.Users.AnonymousUserToken {
 404                         tokenValid = true
 405                 }
 406                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 407                 if errors.As(err, &statusErr) &&
 408                         statusErr.HTTPStatus() == http.StatusForbidden &&
 409                         token != h.Cluster.Users.AnonymousUserToken {
 410                         // collection id is outside scope of supplied
 411                         // token
 412                         tokenScopeProblem = true
 413                         continue
 414                 } else if os.IsNotExist(err) {
 415                         // collection does not exist or is not
 416                         // readable using this token
 417                         continue
 418                 } else if err != nil {
 419                         http.Error(w, err.Error(), http.StatusInternalServerError)
 420                         return
 421                 }
 422                 defer f.Close()
 423
 424                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 425                 break
 426         }
 427         if forceReload && collectionDir != nil {
 428                 err := collectionDir.Sync()
 429                 if err != nil {
 430                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 431                                 http.Error(w, err.Error(), he.HTTPStatus())
 432                         } else {
 433                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 434                         }
 435                         return
 436                 }
 437         }
 438         if session == nil {
 439                 if pathToken {
 440                         // The URL is a "secret sharing link" that
 441                         // didn't work out.  Asking the client for
 442                         // additional credentials would just be
 443                         // confusing.
 444                         http.Error(w, notFoundMessage, http.StatusNotFound)
 445                         return
 446                 }
 447                 if tokenValid {
 448                         // The client provided valid token(s), but the
 449                         // collection was not found.
 450                         http.Error(w, notFoundMessage, http.StatusNotFound)
 451                         return
 452                 }
 453                 if tokenScopeProblem {
 454                         // The client provided a valid token but
 455                         // fetching a collection returned 401, which
 456                         // means the token scope doesn't permit
 457                         // fetching that collection.
 458                         http.Error(w, notFoundMessage, http.StatusForbidden)
 459                         return
 460                 }
 461                 // The client's token was invalid (e.g., expired), or
 462                 // the client didn't even provide one.  Redirect to
 463                 // workbench2's login-and-redirect-to-download url if
 464                 // this is a browser navigation request. (The redirect
 465                 // flow can't preserve the original method if it's not
 466                 // GET, and doesn't make sense if the UA is a
 467                 // command-line tool, is trying to load an inline
 468                 // image, etc.; in these cases, there's nothing we can
 469                 // do, so return 401 unauthorized.)
 470                 //
 471                 // Note Sec-Fetch-Mode is sent by all non-EOL
 472                 // browsers, except Safari.
 473                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 474                 //
 475                 // TODO(TC): This response would be confusing to
 476                 // someone trying (anonymously) to download public
 477                 // data that has been deleted.  Allow a referrer to
 478                 // provide this context somehow?
 479                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 480                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 481                         redirkey := "redirectToPreview"
 482                         if attachment {
 483                                 redirkey = "redirectToDownload"
 484                         }
 485                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 486                         // target.RawQuery = url.Values{redirkey:
 487                         // {target}}.Encode() would be the obvious
 488                         // thing to do here, but wb2 doesn't decode
 489                         // this as a query param -- it takes
 490                         // everything after "${redirkey}=" as the
 491                         // target URL. If we encode "/" as "%2F" etc.,
 492                         // the redirect won't work.
 493                         target.RawQuery = redirkey + "=" + callback
 494                         w.Header().Add("Location", target.String())
 495                         w.WriteHeader(http.StatusSeeOther)
 496                         return
 497                 }
 498                 if !credentialsOK {
 499                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 500                         return
 501                 }
 502                 // If none of the above cases apply, suggest the
 503                 // user-agent (which is either a non-browser agent
 504                 // like wget, or a browser that can't redirect through
 505                 // a login flow) prompt the user for credentials.
 506                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 507                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 508                 return
 509         }
 510
 511         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 512                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 513                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 514                         if !strings.HasSuffix(r.URL.Path, "/") {
 515                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 516                         } else {
 517                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 518                         }
 519                         return
 520                 }
 521         }
 522
 523         var basename string
 524         if len(targetPath) > 0 {
 525                 basename = targetPath[len(targetPath)-1]
 526         }
 527         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 528                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 529                 return
 530         }
 531         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 532                 http.Error(w, "Not permitted", http.StatusForbidden)
 533                 return
 534         }
 535         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 536
 537         if writeMethod[r.Method] {
 538                 // Save the collection only if/when all
 539                 // webdav->filesystem operations succeed --
 540                 // and send a 500 error if the modified
 541                 // collection can't be saved.
 542                 //
 543                 // Perform the write in a separate sitefs, so
 544                 // concurrent read operations on the same
 545                 // collection see the previous saved
 546                 // state. After the write succeeds and the
 547                 // collection record is updated, we reset the
 548                 // session so the updates are visible in
 549                 // subsequent read requests.
 550                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 551                 sessionFS = client.SiteFileSystem(session.keepclient)
 552                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 553                 if err != nil {
 554                         http.Error(w, err.Error(), http.StatusInternalServerError)
 555                         return
 556                 }
 557                 defer writingDir.Close()
 558                 w = &updateOnSuccess{
 559                         ResponseWriter: w,
 560                         logger:         ctxlog.FromContext(r.Context()),
 561                         update: func() error {
 562                                 err := writingDir.Sync()
 563                                 var te arvados.TransactionError
 564                                 if errors.As(err, &te) {
 565                                         err = te
 566                                 }
 567                                 if err != nil {
 568                                         return err
 569                                 }
 570                                 // Sync the changes to the persistent
 571                                 // sessionfs for this token.
 572                                 snap, err := writingDir.Snapshot()
 573                                 if err != nil {
 574                                         return err
 575                                 }
 576                                 collectionDir.Splice(snap)
 577                                 return nil
 578                         }}
 579         }
 580         if r.Method == http.MethodGet {
 581                 applyContentDispositionHdr(w, r, basename, attachment)
 582         }
 583         if webdavPrefix == "" {
 584                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 585         }
 586         wh := webdav.Handler{
 587                 Prefix: webdavPrefix,
 588                 FileSystem: &webdavfs.FS{
 589                         FileSystem:    sessionFS,
 590                         Prefix:        fsprefix,
 591                         Writing:       writeMethod[r.Method],
 592                         AlwaysReadEOF: r.Method == "PROPFIND",
 593                 },
 594                 LockSystem: webdavfs.NoLockSystem,
 595                 Logger: func(r *http.Request, err error) {
 596                         if err != nil {
 597                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 598                         }
 599                 },
 600         }
 601         wh.ServeHTTP(w, r)
 602         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 603                 wrote := int64(w.WroteBodyBytes())
 604                 fnm := strings.Join(pathParts[stripParts:], "/")
 605                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 606                 if err == nil && fi.Size() != wrote {
 607                         var n int
 608                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 609                         if err == nil {
 610                                 n, err = f.Read(make([]byte, 1024))
 611                                 f.Close()
 612                         }
 613                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 614                 }
 615         }
 616 }
 617
 618 var dirListingTemplate = `<!DOCTYPE HTML>
 619 <HTML><HEAD>
 620   <META name="robots" content="NOINDEX">
 621   <TITLE>{{ .CollectionName }}</TITLE>
 622   <STYLE type="text/css">
 623     body {
 624       margin: 1.5em;
 625     }
 626     pre {
 627       background-color: #D9EDF7;
 628       border-radius: .25em;
 629       padding: .75em;
 630       overflow: auto;
 631     }
 632     .footer p {
 633       font-size: 82%;
 634     }
 635     ul {
 636       padding: 0;
 637     }
 638     ul li {
 639       font-family: monospace;
 640       list-style: none;
 641     }
 642   </STYLE>
 643 </HEAD>
 644 <BODY>
 645
 646 <H1>{{ .CollectionName }}</H1>
 647
 648 <P>This collection of data files is being shared with you through
 649 Arvados.  You can download individual files listed below.  To download
 650 the entire directory tree with wget, try:</P>
 651
 652 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 653
 654 <H2>File Listing</H2>
 655
 656 {{if .Files}}
 657 <UL>
 658 {{range .Files}}
 659 {{if .IsDir }}
 660   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 661 {{else}}
 662   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 663 {{end}}
 664 {{end}}
 665 </UL>
 666 {{else}}
 667 <P>(No files; this collection is empty.)</P>
 668 {{end}}
 669
 670 <HR noshade>
 671 <DIV class="footer">
 672   <P>
 673     About Arvados:
 674     Arvados is a free and open source software bioinformatics platform.
 675     To learn more, visit arvados.org.
 676     Arvados is not responsible for the files listed on this page.
 677   </P>
 678 </DIV>
 679
 680 </BODY>
 681 `
 682
 683 type fileListEnt struct {
 684         Name  string
 685         Size  int64
 686         IsDir bool
 687 }
 688
 689 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 690         var files []fileListEnt
 691         var walk func(string) error
 692         if !strings.HasSuffix(base, "/") {
 693                 base = base + "/"
 694         }
 695         walk = func(path string) error {
 696                 dirname := base + path
 697                 if dirname != "/" {
 698                         dirname = strings.TrimSuffix(dirname, "/")
 699                 }
 700                 d, err := fs.Open(dirname)
 701                 if err != nil {
 702                         return err
 703                 }
 704                 ents, err := d.Readdir(-1)
 705                 if err != nil {
 706                         return err
 707                 }
 708                 for _, ent := range ents {
 709                         if recurse && ent.IsDir() {
 710                                 err = walk(path + ent.Name() + "/")
 711                                 if err != nil {
 712                                         return err
 713                                 }
 714                         } else {
 715                                 files = append(files, fileListEnt{
 716                                         Name:  path + ent.Name(),
 717                                         Size:  ent.Size(),
 718                                         IsDir: ent.IsDir(),
 719                                 })
 720                         }
 721                 }
 722                 return nil
 723         }
 724         if err := walk(""); err != nil {
 725                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 726                 return
 727         }
 728
 729         funcs := template.FuncMap{
 730                 "nbsp": func(s string) template.HTML {
 731                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 732                 },
 733         }
 734         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 735         if err != nil {
 736                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 737                 return
 738         }
 739         sort.Slice(files, func(i, j int) bool {
 740                 return files[i].Name < files[j].Name
 741         })
 742         w.WriteHeader(http.StatusOK)
 743         tmpl.Execute(w, map[string]interface{}{
 744                 "CollectionName": collectionName,
 745                 "Files":          files,
 746                 "Request":        r,
 747                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 748         })
 749 }
 750
 751 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 752         disposition := "inline"
 753         if isAttachment {
 754                 disposition = "attachment"
 755         }
 756         if strings.ContainsRune(r.RequestURI, '?') {
 757                 // Help the UA realize that the filename is just
 758                 // "filename.txt", not
 759                 // "filename.txt?disposition=attachment".
 760                 //
 761                 // TODO(TC): Follow advice at RFC 6266 appendix D
 762                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 763         }
 764         if disposition != "inline" {
 765                 w.Header().Set("Content-Disposition", disposition)
 766         }
 767 }
 768
 769 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 770         if formToken := r.FormValue("api_token"); formToken != "" {
 771                 if !credentialsOK {
 772                         // It is not safe to copy the provided token
 773                         // into a cookie unless the current vhost
 774                         // (origin) serves only a single collection or
 775                         // we are in TrustAllContent mode.
 776                         http.Error(w, "cannot serve inline content at this URL (possible configuration error; see https://doc.arvados.org/install/install-keep-web.html#dns)", http.StatusBadRequest)
 777                         return
 778                 }
 779
 780                 // The HttpOnly flag is necessary to prevent
 781                 // JavaScript code (included in, or loaded by, a page
 782                 // in the collection being served) from employing the
 783                 // user's token beyond reading other files in the same
 784                 // domain, i.e., same collection.
 785                 //
 786                 // The 303 redirect is necessary in the case of a GET
 787                 // request to avoid exposing the token in the Location
 788                 // bar, and in the case of a POST request to avoid
 789                 // raising warnings when the user refreshes the
 790                 // resulting page.
 791                 http.SetCookie(w, &http.Cookie{
 792                         Name:     "arvados_api_token",
 793                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 794                         Path:     "/",
 795                         HttpOnly: true,
 796                         SameSite: http.SameSiteLaxMode,
 797                 })
 798         }
 799
 800         // Propagate query parameters (except api_token) from
 801         // the original request.
 802         redirQuery := r.URL.Query()
 803         redirQuery.Del("api_token")
 804
 805         u := r.URL
 806         if location != "" {
 807                 newu, err := u.Parse(location)
 808                 if err != nil {
 809                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 810                         return
 811                 }
 812                 u = newu
 813         }
 814         redir := (&url.URL{
 815                 Scheme:   r.URL.Scheme,
 816                 Host:     r.Host,
 817                 Path:     u.Path,
 818                 RawQuery: redirQuery.Encode(),
 819         }).String()
 820
 821         w.Header().Add("Location", redir)
 822         w.WriteHeader(http.StatusSeeOther)
 823         io.WriteString(w, `<A href="`)
 824         io.WriteString(w, html.EscapeString(redir))
 825         io.WriteString(w, `">Continue</A>`)
 826 }
 827
 828 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 829         var permitDownload bool
 830         var permitUpload bool
 831         if tokenUser != nil && tokenUser.IsAdmin {
 832                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 833                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 834         } else {
 835                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 836                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 837         }
 838         if (method == "PUT" || method == "POST") && !permitUpload {
 839                 // Disallow operations that upload new files.
 840                 // Permit webdav operations that move existing files around.
 841                 return false
 842         } else if method == "GET" && !permitDownload {
 843                 // Disallow downloading file contents.
 844                 // Permit webdav operations like PROPFIND that retrieve metadata
 845                 // but not file contents.
 846                 return false
 847         }
 848         return true
 849 }
 850
 851 func (h *handler) logUploadOrDownload(
 852         r *http.Request,
 853         client *arvadosclient.ArvadosClient,
 854         fs arvados.CustomFileSystem,
 855         filepath string,
 856         collection *arvados.Collection,
 857         user *arvados.User) {
 858
 859         log := ctxlog.FromContext(r.Context())
 860         props := make(map[string]string)
 861         props["reqPath"] = r.URL.Path
 862         var useruuid string
 863         if user != nil {
 864                 log = log.WithField("user_uuid", user.UUID).
 865                         WithField("user_full_name", user.FullName)
 866                 useruuid = user.UUID
 867         } else {
 868                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 869         }
 870         if collection == nil && fs != nil {
 871                 collection, filepath = h.determineCollection(fs, filepath)
 872         }
 873         if collection != nil {
 874                 log = log.WithField("collection_file_path", filepath)
 875                 props["collection_file_path"] = filepath
 876                 // h.determineCollection populates the collection_uuid
 877                 // prop with the PDH, if this collection is being
 878                 // accessed via PDH. For logging, we use a different
 879                 // field depending on whether it's a UUID or PDH.
 880                 if len(collection.UUID) > 32 {
 881                         log = log.WithField("portable_data_hash", collection.UUID)
 882                         props["portable_data_hash"] = collection.UUID
 883                 } else {
 884                         log = log.WithField("collection_uuid", collection.UUID)
 885                         props["collection_uuid"] = collection.UUID
 886                 }
 887         }
 888         if r.Method == "PUT" || r.Method == "POST" {
 889                 log.Info("File upload")
 890                 if h.Cluster.Collections.WebDAVLogEvents {
 891                         go func() {
 892                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 893                                         "object_uuid": useruuid,
 894                                         "event_type":  "file_upload",
 895                                         "properties":  props}}
 896                                 err := client.Create("logs", lr, nil)
 897                                 if err != nil {
 898                                         log.WithError(err).Error("Failed to create upload log event on API server")
 899                                 }
 900                         }()
 901                 }
 902         } else if r.Method == "GET" {
 903                 if collection != nil && collection.PortableDataHash != "" {
 904                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 905                         props["portable_data_hash"] = collection.PortableDataHash
 906                 }
 907                 log.Info("File download")
 908                 if h.Cluster.Collections.WebDAVLogEvents {
 909                         go func() {
 910                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 911                                         "object_uuid": useruuid,
 912                                         "event_type":  "file_download",
 913                                         "properties":  props}}
 914                                 err := client.Create("logs", lr, nil)
 915                                 if err != nil {
 916                                         log.WithError(err).Error("Failed to create download log event on API server")
 917                                 }
 918                         }()
 919                 }
 920         }
 921 }
 922
 923 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 924         target := strings.TrimSuffix(path, "/")
 925         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 926                 target = target[:cut]
 927                 fi, err := fs.Stat(target)
 928                 if os.IsNotExist(err) {
 929                         // creating a new file/dir, or download
 930                         // destined to fail
 931                         continue
 932                 } else if err != nil {
 933                         return nil, ""
 934                 }
 935                 switch src := fi.Sys().(type) {
 936                 case *arvados.Collection:
 937                         return src, strings.TrimPrefix(path[len(target):], "/")
 938                 case *arvados.Group:
 939                         return nil, ""
 940                 default:
 941                         if _, ok := src.(error); ok {
 942                                 return nil, ""
 943                         }
 944                 }
 945         }
 946         return nil, ""
 947 }