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         "time"
  22
  23         "git.arvados.org/arvados.git/lib/cmd"
  24         "git.arvados.org/arvados.git/lib/webdavfs"
  25         "git.arvados.org/arvados.git/sdk/go/arvados"
  26         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  27         "git.arvados.org/arvados.git/sdk/go/auth"
  28         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  29         "git.arvados.org/arvados.git/sdk/go/httpserver"
  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         metrics *metrics
  38
  39         lockMtx    sync.Mutex
  40         lock       map[string]*sync.RWMutex
  41         lockTidied time.Time
  42 }
  43
  44 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  45
  46 var notFoundMessage = "Not Found"
  47 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  48
  49 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  50 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  51 // otherwise "".
  52 func parseCollectionIDFromURL(s string) string {
  53         if arvadosclient.UUIDMatch(s) {
  54                 return s
  55         }
  56         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  57                 return pdh
  58         }
  59         return ""
  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         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 178                 r.URL.Scheme = xfp
 179         }
 180
 181         w := httpserver.WrapResponseWriter(wOrig)
 182
 183         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 184                 return
 185         }
 186
 187         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 188                 w.WriteHeader(http.StatusMethodNotAllowed)
 189                 return
 190         }
 191
 192         if r.Header.Get("Origin") != "" {
 193                 // Allow simple cross-origin requests without user
 194                 // credentials ("user credentials" as defined by CORS,
 195                 // i.e., cookies, HTTP authentication, and client-side
 196                 // SSL certificates. See
 197                 // http://www.w3.org/TR/cors/#user-credentials).
 198                 w.Header().Set("Access-Control-Allow-Origin", "*")
 199                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 200         }
 201
 202         if h.serveS3(w, r) {
 203                 return
 204         }
 205
 206         webdavPrefix := ""
 207         arvPath := r.URL.Path
 208         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 209                 // Enable a proxy (e.g., container log handler in
 210                 // controller) to satisfy a request for path
 211                 // "/foo/bar/baz.txt" using content from
 212                 // "//abc123-4.internal/bar/baz.txt", by adding a
 213                 // request header "X-Webdav-Prefix: /foo"
 214                 if !strings.HasPrefix(arvPath, prefix) {
 215                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 216                         return
 217                 }
 218                 arvPath = r.URL.Path[len(prefix):]
 219                 if arvPath == "" {
 220                         arvPath = "/"
 221                 }
 222                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 223                 webdavPrefix = prefix
 224         }
 225         pathParts := strings.Split(arvPath[1:], "/")
 226
 227         var stripParts int
 228         var collectionID string
 229         var tokens []string
 230         var reqTokens []string
 231         var pathToken bool
 232         var attachment bool
 233         var useSiteFS bool
 234         credentialsOK := h.Cluster.Collections.TrustAllContent
 235         reasonNotAcceptingCredentials := ""
 236
 237         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 238                 credentialsOK = true
 239                 attachment = true
 240         } else if r.FormValue("disposition") == "attachment" {
 241                 attachment = true
 242         }
 243
 244         if !credentialsOK {
 245                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 246                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 247         }
 248
 249         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 250                 // http://ID.collections.example/PATH...
 251                 credentialsOK = true
 252         } else if r.URL.Path == "/status.json" {
 253                 h.serveStatus(w, r)
 254                 return
 255         } else if siteFSDir[pathParts[0]] {
 256                 useSiteFS = true
 257         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 258                 // /c=ID[/PATH...]
 259                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 260                 stripParts = 1
 261         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 262                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 263                         // /collections/download/ID/TOKEN/PATH...
 264                         collectionID = parseCollectionIDFromURL(pathParts[2])
 265                         tokens = []string{pathParts[3]}
 266                         stripParts = 4
 267                         pathToken = true
 268                 } else {
 269                         // /collections/ID/PATH...
 270                         collectionID = parseCollectionIDFromURL(pathParts[1])
 271                         stripParts = 2
 272                         // This path is only meant to work for public
 273                         // data. Tokens provided with the request are
 274                         // ignored.
 275                         credentialsOK = false
 276                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 277                 }
 278         }
 279
 280         forceReload := false
 281         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 282                 forceReload = true
 283         }
 284
 285         if credentialsOK {
 286                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 287         }
 288
 289         r.ParseForm()
 290         origin := r.Header.Get("Origin")
 291         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 292         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 293         // Important distinction: safeAttachment checks whether api_token exists
 294         // as a query parameter. haveFormTokens checks whether api_token exists
 295         // as request form data *or* a query parameter. Different checks are
 296         // necessary because both the request disposition and the location of
 297         // the API token affect whether or not the request needs to be
 298         // redirected. The different branch comments below explain further.
 299         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 300         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 301                 // No token to use or redact.
 302         } else if safeAjax || safeAttachment {
 303                 // If this is a cross-origin request, the URL won't
 304                 // appear in the browser's address bar, so
 305                 // substituting a clipboard-safe URL is pointless.
 306                 // Redirect-with-cookie wouldn't work anyway, because
 307                 // it's not safe to allow third-party use of our
 308                 // cookie.
 309                 //
 310                 // If we're supplying an attachment, we don't need to
 311                 // convert POST to GET to avoid the "really resubmit
 312                 // form?" problem, so provided the token isn't
 313                 // embedded in the URL, there's no reason to do
 314                 // redirect-with-cookie in this case either.
 315                 for _, tok := range formTokens {
 316                         reqTokens = append(reqTokens, tok)
 317                 }
 318         } else if browserMethod[r.Method] {
 319                 // If this is a page view, and the client provided a
 320                 // token via query string or POST body, we must put
 321                 // the token in an HttpOnly cookie, and redirect to an
 322                 // equivalent URL with the query param redacted and
 323                 // method = GET.
 324                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 325                 return
 326         }
 327
 328         targetPath := pathParts[stripParts:]
 329         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 330                 // http://ID.example/t=TOKEN/PATH...
 331                 // /c=ID/t=TOKEN/PATH...
 332                 //
 333                 // This form must only be used to pass scoped tokens
 334                 // that give permission for a single collection. See
 335                 // FormValue case above.
 336                 tokens = []string{targetPath[0][2:]}
 337                 pathToken = true
 338                 targetPath = targetPath[1:]
 339                 stripParts++
 340         }
 341
 342         fsprefix := ""
 343         if useSiteFS {
 344                 if writeMethod[r.Method] {
 345                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 346                         return
 347                 }
 348                 if len(reqTokens) == 0 {
 349                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 350                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 351                         return
 352                 }
 353                 tokens = reqTokens
 354         } else if collectionID == "" {
 355                 http.Error(w, notFoundMessage, http.StatusNotFound)
 356                 return
 357         } else {
 358                 fsprefix = "by_id/" + collectionID + "/"
 359         }
 360
 361         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 362                 fsprefix += src[1:]
 363         }
 364
 365         if tokens == nil {
 366                 tokens = reqTokens
 367                 if h.Cluster.Users.AnonymousUserToken != "" {
 368                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 369                 }
 370         }
 371
 372         if len(targetPath) > 0 && targetPath[0] == "_" {
 373                 // If a collection has a directory called "t=foo" or
 374                 // "_", it can be served at
 375                 // //collections.example/_/t=foo/ or
 376                 // //collections.example/_/_/ respectively:
 377                 // //collections.example/t=foo/ won't work because
 378                 // t=foo will be interpreted as a token "foo".
 379                 targetPath = targetPath[1:]
 380                 stripParts++
 381         }
 382
 383         dirOpenMode := os.O_RDONLY
 384         if writeMethod[r.Method] {
 385                 dirOpenMode = os.O_RDWR
 386         }
 387
 388         var tokenValid bool
 389         var tokenScopeProblem bool
 390         var token string
 391         var tokenUser *arvados.User
 392         var sessionFS arvados.CustomFileSystem
 393         var session *cachedSession
 394         var collectionDir arvados.File
 395         for _, token = range tokens {
 396                 var statusErr errorWithHTTPStatus
 397                 fs, sess, user, err := h.Cache.GetSession(token)
 398                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 399                         // bad token
 400                         continue
 401                 } else if err != nil {
 402                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 403                         return
 404                 }
 405                 if token != h.Cluster.Users.AnonymousUserToken {
 406                         tokenValid = true
 407                 }
 408                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 409                 if errors.As(err, &statusErr) &&
 410                         statusErr.HTTPStatus() == http.StatusForbidden &&
 411                         token != h.Cluster.Users.AnonymousUserToken {
 412                         // collection id is outside scope of supplied
 413                         // token
 414                         tokenScopeProblem = true
 415                         sess.Release()
 416                         continue
 417                 } else if os.IsNotExist(err) {
 418                         // collection does not exist or is not
 419                         // readable using this token
 420                         sess.Release()
 421                         continue
 422                 } else if err != nil {
 423                         http.Error(w, err.Error(), http.StatusInternalServerError)
 424                         sess.Release()
 425                         return
 426                 }
 427                 defer f.Close()
 428                 defer sess.Release()
 429
 430                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 431                 break
 432         }
 433         if forceReload && collectionDir != nil {
 434                 err := collectionDir.Sync()
 435                 if err != nil {
 436                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 437                                 http.Error(w, err.Error(), he.HTTPStatus())
 438                         } else {
 439                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 440                         }
 441                         return
 442                 }
 443         }
 444         if session == nil {
 445                 if pathToken {
 446                         // The URL is a "secret sharing link" that
 447                         // didn't work out.  Asking the client for
 448                         // additional credentials would just be
 449                         // confusing.
 450                         http.Error(w, notFoundMessage, http.StatusNotFound)
 451                         return
 452                 }
 453                 if tokenValid {
 454                         // The client provided valid token(s), but the
 455                         // collection was not found.
 456                         http.Error(w, notFoundMessage, http.StatusNotFound)
 457                         return
 458                 }
 459                 if tokenScopeProblem {
 460                         // The client provided a valid token but
 461                         // fetching a collection returned 401, which
 462                         // means the token scope doesn't permit
 463                         // fetching that collection.
 464                         http.Error(w, notFoundMessage, http.StatusForbidden)
 465                         return
 466                 }
 467                 // The client's token was invalid (e.g., expired), or
 468                 // the client didn't even provide one.  Redirect to
 469                 // workbench2's login-and-redirect-to-download url if
 470                 // this is a browser navigation request. (The redirect
 471                 // flow can't preserve the original method if it's not
 472                 // GET, and doesn't make sense if the UA is a
 473                 // command-line tool, is trying to load an inline
 474                 // image, etc.; in these cases, there's nothing we can
 475                 // do, so return 401 unauthorized.)
 476                 //
 477                 // Note Sec-Fetch-Mode is sent by all non-EOL
 478                 // browsers, except Safari.
 479                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 480                 //
 481                 // TODO(TC): This response would be confusing to
 482                 // someone trying (anonymously) to download public
 483                 // data that has been deleted.  Allow a referrer to
 484                 // provide this context somehow?
 485                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 486                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 487                         redirkey := "redirectToPreview"
 488                         if attachment {
 489                                 redirkey = "redirectToDownload"
 490                         }
 491                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 492                         // target.RawQuery = url.Values{redirkey:
 493                         // {target}}.Encode() would be the obvious
 494                         // thing to do here, but wb2 doesn't decode
 495                         // this as a query param -- it takes
 496                         // everything after "${redirkey}=" as the
 497                         // target URL. If we encode "/" as "%2F" etc.,
 498                         // the redirect won't work.
 499                         target.RawQuery = redirkey + "=" + callback
 500                         w.Header().Add("Location", target.String())
 501                         w.WriteHeader(http.StatusSeeOther)
 502                         return
 503                 }
 504                 if !credentialsOK {
 505                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 506                         return
 507                 }
 508                 // If none of the above cases apply, suggest the
 509                 // user-agent (which is either a non-browser agent
 510                 // like wget, or a browser that can't redirect through
 511                 // a login flow) prompt the user for credentials.
 512                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 513                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 514                 return
 515         }
 516
 517         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 518                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 519                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 520                         if !strings.HasSuffix(r.URL.Path, "/") {
 521                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 522                         } else {
 523                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 524                         }
 525                         return
 526                 }
 527         }
 528
 529         var basename string
 530         if len(targetPath) > 0 {
 531                 basename = targetPath[len(targetPath)-1]
 532         }
 533         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 534                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 535                 return
 536         }
 537         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 538                 http.Error(w, "Not permitted", http.StatusForbidden)
 539                 return
 540         }
 541         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 542
 543         writing := writeMethod[r.Method]
 544         locker := h.collectionLock(collectionID, writing)
 545         defer locker.Unlock()
 546
 547         if writing {
 548                 // Save the collection only if/when all
 549                 // webdav->filesystem operations succeed --
 550                 // and send a 500 error if the modified
 551                 // collection can't be saved.
 552                 //
 553                 // Perform the write in a separate sitefs, so
 554                 // concurrent read operations on the same
 555                 // collection see the previous saved
 556                 // state. After the write succeeds and the
 557                 // collection record is updated, we reset the
 558                 // session so the updates are visible in
 559                 // subsequent read requests.
 560                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 561                 sessionFS = client.SiteFileSystem(session.keepclient)
 562                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 563                 if err != nil {
 564                         http.Error(w, err.Error(), http.StatusInternalServerError)
 565                         return
 566                 }
 567                 defer writingDir.Close()
 568                 w = &updateOnSuccess{
 569                         ResponseWriter: w,
 570                         logger:         ctxlog.FromContext(r.Context()),
 571                         update: func() error {
 572                                 err := writingDir.Sync()
 573                                 var te arvados.TransactionError
 574                                 if errors.As(err, &te) {
 575                                         err = te
 576                                 }
 577                                 if err != nil {
 578                                         return err
 579                                 }
 580                                 // Sync the changes to the persistent
 581                                 // sessionfs for this token.
 582                                 snap, err := writingDir.Snapshot()
 583                                 if err != nil {
 584                                         return err
 585                                 }
 586                                 collectionDir.Splice(snap)
 587                                 return nil
 588                         }}
 589         }
 590         if r.Method == http.MethodGet {
 591                 applyContentDispositionHdr(w, r, basename, attachment)
 592         }
 593         if webdavPrefix == "" {
 594                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 595         }
 596         wh := &webdav.Handler{
 597                 Prefix: webdavPrefix,
 598                 FileSystem: &webdavfs.FS{
 599                         FileSystem:    sessionFS,
 600                         Prefix:        fsprefix,
 601                         Writing:       writeMethod[r.Method],
 602                         AlwaysReadEOF: r.Method == "PROPFIND",
 603                 },
 604                 LockSystem: webdavfs.NoLockSystem,
 605                 Logger: func(r *http.Request, err error) {
 606                         if err != nil && !os.IsNotExist(err) {
 607                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 608                         }
 609                 },
 610         }
 611         h.metrics.track(wh, w, r)
 612         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 613                 wrote := int64(w.WroteBodyBytes())
 614                 fnm := strings.Join(pathParts[stripParts:], "/")
 615                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 616                 if err == nil && fi.Size() != wrote {
 617                         var n int
 618                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 619                         if err == nil {
 620                                 n, err = f.Read(make([]byte, 1024))
 621                                 f.Close()
 622                         }
 623                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 624                 }
 625         }
 626 }
 627
 628 var dirListingTemplate = `<!DOCTYPE HTML>
 629 <HTML><HEAD>
 630   <META name="robots" content="NOINDEX">
 631   <TITLE>{{ .CollectionName }}</TITLE>
 632   <STYLE type="text/css">
 633     body {
 634       margin: 1.5em;
 635     }
 636     pre {
 637       background-color: #D9EDF7;
 638       border-radius: .25em;
 639       padding: .75em;
 640       overflow: auto;
 641     }
 642     .footer p {
 643       font-size: 82%;
 644     }
 645     ul {
 646       padding: 0;
 647     }
 648     ul li {
 649       font-family: monospace;
 650       list-style: none;
 651     }
 652   </STYLE>
 653 </HEAD>
 654 <BODY>
 655
 656 <H1>{{ .CollectionName }}</H1>
 657
 658 <P>This collection of data files is being shared with you through
 659 Arvados.  You can download individual files listed below.  To download
 660 the entire directory tree with wget, try:</P>
 661
 662 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 663
 664 <H2>File Listing</H2>
 665
 666 {{if .Files}}
 667 <UL>
 668 {{range .Files}}
 669 {{if .IsDir }}
 670   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 671 {{else}}
 672   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 673 {{end}}
 674 {{end}}
 675 </UL>
 676 {{else}}
 677 <P>(No files; this collection is empty.)</P>
 678 {{end}}
 679
 680 <HR noshade>
 681 <DIV class="footer">
 682   <P>
 683     About Arvados:
 684     Arvados is a free and open source software bioinformatics platform.
 685     To learn more, visit arvados.org.
 686     Arvados is not responsible for the files listed on this page.
 687   </P>
 688 </DIV>
 689
 690 </BODY>
 691 `
 692
 693 type fileListEnt struct {
 694         Name  string
 695         Size  int64
 696         IsDir bool
 697 }
 698
 699 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 700         var files []fileListEnt
 701         var walk func(string) error
 702         if !strings.HasSuffix(base, "/") {
 703                 base = base + "/"
 704         }
 705         walk = func(path string) error {
 706                 dirname := base + path
 707                 if dirname != "/" {
 708                         dirname = strings.TrimSuffix(dirname, "/")
 709                 }
 710                 d, err := fs.Open(dirname)
 711                 if err != nil {
 712                         return err
 713                 }
 714                 ents, err := d.Readdir(-1)
 715                 if err != nil {
 716                         return err
 717                 }
 718                 for _, ent := range ents {
 719                         if recurse && ent.IsDir() {
 720                                 err = walk(path + ent.Name() + "/")
 721                                 if err != nil {
 722                                         return err
 723                                 }
 724                         } else {
 725                                 files = append(files, fileListEnt{
 726                                         Name:  path + ent.Name(),
 727                                         Size:  ent.Size(),
 728                                         IsDir: ent.IsDir(),
 729                                 })
 730                         }
 731                 }
 732                 return nil
 733         }
 734         if err := walk(""); err != nil {
 735                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 736                 return
 737         }
 738
 739         funcs := template.FuncMap{
 740                 "nbsp": func(s string) template.HTML {
 741                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 742                 },
 743         }
 744         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 745         if err != nil {
 746                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 747                 return
 748         }
 749         sort.Slice(files, func(i, j int) bool {
 750                 return files[i].Name < files[j].Name
 751         })
 752         w.WriteHeader(http.StatusOK)
 753         tmpl.Execute(w, map[string]interface{}{
 754                 "CollectionName": collectionName,
 755                 "Files":          files,
 756                 "Request":        r,
 757                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 758         })
 759 }
 760
 761 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 762         disposition := "inline"
 763         if isAttachment {
 764                 disposition = "attachment"
 765         }
 766         if strings.ContainsRune(r.RequestURI, '?') {
 767                 // Help the UA realize that the filename is just
 768                 // "filename.txt", not
 769                 // "filename.txt?disposition=attachment".
 770                 //
 771                 // TODO(TC): Follow advice at RFC 6266 appendix D
 772                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 773         }
 774         if disposition != "inline" {
 775                 w.Header().Set("Content-Disposition", disposition)
 776         }
 777 }
 778
 779 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 780         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
 781                 if !credentialsOK {
 782                         // It is not safe to copy the provided token
 783                         // into a cookie unless the current vhost
 784                         // (origin) serves only a single collection or
 785                         // we are in TrustAllContent mode.
 786                         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)
 787                         return
 788                 }
 789
 790                 // The HttpOnly flag is necessary to prevent
 791                 // JavaScript code (included in, or loaded by, a page
 792                 // in the collection being served) from employing the
 793                 // user's token beyond reading other files in the same
 794                 // domain, i.e., same collection.
 795                 //
 796                 // The 303 redirect is necessary in the case of a GET
 797                 // request to avoid exposing the token in the Location
 798                 // bar, and in the case of a POST request to avoid
 799                 // raising warnings when the user refreshes the
 800                 // resulting page.
 801                 for _, tok := range formTokens {
 802                         if tok == "" {
 803                                 continue
 804                         }
 805                         http.SetCookie(w, &http.Cookie{
 806                                 Name:     "arvados_api_token",
 807                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
 808                                 Path:     "/",
 809                                 HttpOnly: true,
 810                                 SameSite: http.SameSiteLaxMode,
 811                         })
 812                         break
 813                 }
 814         }
 815
 816         // Propagate query parameters (except api_token) from
 817         // the original request.
 818         redirQuery := r.URL.Query()
 819         redirQuery.Del("api_token")
 820
 821         u := r.URL
 822         if location != "" {
 823                 newu, err := u.Parse(location)
 824                 if err != nil {
 825                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 826                         return
 827                 }
 828                 u = newu
 829         }
 830         redir := (&url.URL{
 831                 Scheme:   r.URL.Scheme,
 832                 Host:     r.Host,
 833                 Path:     u.Path,
 834                 RawQuery: redirQuery.Encode(),
 835         }).String()
 836
 837         w.Header().Add("Location", redir)
 838         w.WriteHeader(http.StatusSeeOther)
 839         io.WriteString(w, `<A href="`)
 840         io.WriteString(w, html.EscapeString(redir))
 841         io.WriteString(w, `">Continue</A>`)
 842 }
 843
 844 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 845         var permitDownload bool
 846         var permitUpload bool
 847         if tokenUser != nil && tokenUser.IsAdmin {
 848                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 849                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 850         } else {
 851                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 852                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 853         }
 854         if (method == "PUT" || method == "POST") && !permitUpload {
 855                 // Disallow operations that upload new files.
 856                 // Permit webdav operations that move existing files around.
 857                 return false
 858         } else if method == "GET" && !permitDownload {
 859                 // Disallow downloading file contents.
 860                 // Permit webdav operations like PROPFIND that retrieve metadata
 861                 // but not file contents.
 862                 return false
 863         }
 864         return true
 865 }
 866
 867 func (h *handler) logUploadOrDownload(
 868         r *http.Request,
 869         client *arvadosclient.ArvadosClient,
 870         fs arvados.CustomFileSystem,
 871         filepath string,
 872         collection *arvados.Collection,
 873         user *arvados.User) {
 874
 875         log := ctxlog.FromContext(r.Context())
 876         props := make(map[string]string)
 877         props["reqPath"] = r.URL.Path
 878         var useruuid string
 879         if user != nil {
 880                 log = log.WithField("user_uuid", user.UUID).
 881                         WithField("user_full_name", user.FullName)
 882                 useruuid = user.UUID
 883         } else {
 884                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 885         }
 886         if collection == nil && fs != nil {
 887                 collection, filepath = h.determineCollection(fs, filepath)
 888         }
 889         if collection != nil {
 890                 log = log.WithField("collection_file_path", filepath)
 891                 props["collection_file_path"] = filepath
 892                 // h.determineCollection populates the collection_uuid
 893                 // prop with the PDH, if this collection is being
 894                 // accessed via PDH. For logging, we use a different
 895                 // field depending on whether it's a UUID or PDH.
 896                 if len(collection.UUID) > 32 {
 897                         log = log.WithField("portable_data_hash", collection.UUID)
 898                         props["portable_data_hash"] = collection.UUID
 899                 } else {
 900                         log = log.WithField("collection_uuid", collection.UUID)
 901                         props["collection_uuid"] = collection.UUID
 902                 }
 903         }
 904         if r.Method == "PUT" || r.Method == "POST" {
 905                 log.Info("File upload")
 906                 if h.Cluster.Collections.WebDAVLogEvents {
 907                         go func() {
 908                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 909                                         "object_uuid": useruuid,
 910                                         "event_type":  "file_upload",
 911                                         "properties":  props}}
 912                                 err := client.Create("logs", lr, nil)
 913                                 if err != nil {
 914                                         log.WithError(err).Error("Failed to create upload log event on API server")
 915                                 }
 916                         }()
 917                 }
 918         } else if r.Method == "GET" {
 919                 if collection != nil && collection.PortableDataHash != "" {
 920                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 921                         props["portable_data_hash"] = collection.PortableDataHash
 922                 }
 923                 log.Info("File download")
 924                 if h.Cluster.Collections.WebDAVLogEvents {
 925                         go func() {
 926                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 927                                         "object_uuid": useruuid,
 928                                         "event_type":  "file_download",
 929                                         "properties":  props}}
 930                                 err := client.Create("logs", lr, nil)
 931                                 if err != nil {
 932                                         log.WithError(err).Error("Failed to create download log event on API server")
 933                                 }
 934                         }()
 935                 }
 936         }
 937 }
 938
 939 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 940         target := strings.TrimSuffix(path, "/")
 941         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 942                 target = target[:cut]
 943                 fi, err := fs.Stat(target)
 944                 if os.IsNotExist(err) {
 945                         // creating a new file/dir, or download
 946                         // destined to fail
 947                         continue
 948                 } else if err != nil {
 949                         return nil, ""
 950                 }
 951                 switch src := fi.Sys().(type) {
 952                 case *arvados.Collection:
 953                         return src, strings.TrimPrefix(path[len(target):], "/")
 954                 case *arvados.Group:
 955                         return nil, ""
 956                 default:
 957                         if _, ok := src.(error); ok {
 958                                 return nil, ""
 959                         }
 960                 }
 961         }
 962         return nil, ""
 963 }
 964
 965 var lockTidyInterval = time.Minute * 10
 966
 967 // Lock the specified collection for reading or writing. Caller must
 968 // call Unlock() on the returned Locker when the operation is
 969 // finished.
 970 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
 971         h.lockMtx.Lock()
 972         defer h.lockMtx.Unlock()
 973         if time.Since(h.lockTidied) > lockTidyInterval {
 974                 // Periodically delete all locks that aren't in use.
 975                 h.lockTidied = time.Now()
 976                 for id, locker := range h.lock {
 977                         if locker.TryLock() {
 978                                 locker.Unlock()
 979                                 delete(h.lock, id)
 980                         }
 981                 }
 982         }
 983         locker := h.lock[collectionID]
 984         if locker == nil {
 985                 locker = new(sync.RWMutex)
 986                 if h.lock == nil {
 987                         h.lock = map[string]*sync.RWMutex{}
 988                 }
 989                 h.lock[collectionID] = locker
 990         }
 991         if writing {
 992                 locker.Lock()
 993                 return locker
 994         } else {
 995                 locker.RLock()
 996                 return locker.RLocker()
 997         }
 998 }
 999
1000 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1001         method := header.Get("Access-Control-Request-Method")
1002         if method == "" {
1003                 return false
1004         }
1005         if !browserMethod[method] && !webdavMethod[method] {
1006                 w.WriteHeader(http.StatusMethodNotAllowed)
1007                 return true
1008         }
1009         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1010         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1011         w.Header().Set("Access-Control-Allow-Origin", "*")
1012         w.Header().Set("Access-Control-Max-Age", "86400")
1013         return true
1014 }