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         "git.arvados.org/arvados.git/sdk/go/keepclient"
  31         "github.com/sirupsen/logrus"
  32         "golang.org/x/net/webdav"
  33 )
  34
  35 type handler struct {
  36         Cache     cache
  37         Cluster   *arvados.Cluster
  38         setupOnce sync.Once
  39
  40         lockMtx    sync.Mutex
  41         lock       map[string]*sync.RWMutex
  42         lockTidied time.Time
  43 }
  44
  45 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  46
  47 var notFoundMessage = "Not Found"
  48 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  49
  50 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  51 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  52 // otherwise "".
  53 func parseCollectionIDFromURL(s string) string {
  54         if arvadosclient.UUIDMatch(s) {
  55                 return s
  56         }
  57         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  58                 return pdh
  59         }
  60         return ""
  61 }
  62
  63 func (h *handler) setup() {
  64         keepclient.DefaultBlockCache.MaxBlocks = h.Cluster.Collections.WebDAVCache.MaxBlockEntries
  65 }
  66
  67 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  68         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  69 }
  70
  71 type errorWithHTTPStatus interface {
  72         HTTPStatus() int
  73 }
  74
  75 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  76 // sends an HTTP header indicating success, updateOnSuccess first
  77 // calls the provided update func. If the update func fails, an error
  78 // response is sent (using the error's HTTP status or 500 if none),
  79 // and the status code and body sent by the handler are ignored (all
  80 // response writes return the update error).
  81 type updateOnSuccess struct {
  82         httpserver.ResponseWriter
  83         logger     logrus.FieldLogger
  84         update     func() error
  85         sentHeader bool
  86         err        error
  87 }
  88
  89 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  90         if !uos.sentHeader {
  91                 uos.WriteHeader(http.StatusOK)
  92         }
  93         if uos.err != nil {
  94                 return 0, uos.err
  95         }
  96         return uos.ResponseWriter.Write(p)
  97 }
  98
  99 func (uos *updateOnSuccess) WriteHeader(code int) {
 100         if !uos.sentHeader {
 101                 uos.sentHeader = true
 102                 if code >= 200 && code < 400 {
 103                         if uos.err = uos.update(); uos.err != nil {
 104                                 code := http.StatusInternalServerError
 105                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 106                                         code = he.HTTPStatus()
 107                                 }
 108                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 109                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 110                                 return
 111                         }
 112                 }
 113         }
 114         uos.ResponseWriter.WriteHeader(code)
 115 }
 116
 117 var (
 118         corsAllowHeadersHeader = strings.Join([]string{
 119                 "Authorization", "Content-Type", "Range",
 120                 // WebDAV request headers:
 121                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 122         }, ", ")
 123         writeMethod = map[string]bool{
 124                 "COPY":      true,
 125                 "DELETE":    true,
 126                 "LOCK":      true,
 127                 "MKCOL":     true,
 128                 "MOVE":      true,
 129                 "PROPPATCH": true,
 130                 "PUT":       true,
 131                 "RMCOL":     true,
 132                 "UNLOCK":    true,
 133         }
 134         webdavMethod = map[string]bool{
 135                 "COPY":      true,
 136                 "DELETE":    true,
 137                 "LOCK":      true,
 138                 "MKCOL":     true,
 139                 "MOVE":      true,
 140                 "OPTIONS":   true,
 141                 "PROPFIND":  true,
 142                 "PROPPATCH": true,
 143                 "PUT":       true,
 144                 "RMCOL":     true,
 145                 "UNLOCK":    true,
 146         }
 147         browserMethod = map[string]bool{
 148                 "GET":  true,
 149                 "HEAD": true,
 150                 "POST": true,
 151         }
 152         // top-level dirs to serve with siteFS
 153         siteFSDir = map[string]bool{
 154                 "":      true, // root directory
 155                 "by_id": true,
 156                 "users": true,
 157         }
 158 )
 159
 160 func stripDefaultPort(host string) string {
 161         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 162         u := &url.URL{Host: host}
 163         if p := u.Port(); p == "80" || p == "443" {
 164                 return strings.ToLower(u.Hostname())
 165         } else {
 166                 return strings.ToLower(host)
 167         }
 168 }
 169
 170 // CheckHealth implements service.Handler.
 171 func (h *handler) CheckHealth() error {
 172         return nil
 173 }
 174
 175 // Done implements service.Handler.
 176 func (h *handler) Done() <-chan struct{} {
 177         return nil
 178 }
 179
 180 // ServeHTTP implements http.Handler.
 181 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 182         h.setupOnce.Do(h.setup)
 183
 184         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 185                 r.URL.Scheme = xfp
 186         }
 187
 188         w := httpserver.WrapResponseWriter(wOrig)
 189
 190         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 191                 return
 192         }
 193
 194         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 195                 w.WriteHeader(http.StatusMethodNotAllowed)
 196                 return
 197         }
 198
 199         if r.Header.Get("Origin") != "" {
 200                 // Allow simple cross-origin requests without user
 201                 // credentials ("user credentials" as defined by CORS,
 202                 // i.e., cookies, HTTP authentication, and client-side
 203                 // SSL certificates. See
 204                 // http://www.w3.org/TR/cors/#user-credentials).
 205                 w.Header().Set("Access-Control-Allow-Origin", "*")
 206                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 207         }
 208
 209         if h.serveS3(w, r) {
 210                 return
 211         }
 212
 213         webdavPrefix := ""
 214         arvPath := r.URL.Path
 215         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 216                 // Enable a proxy (e.g., container log handler in
 217                 // controller) to satisfy a request for path
 218                 // "/foo/bar/baz.txt" using content from
 219                 // "//abc123-4.internal/bar/baz.txt", by adding a
 220                 // request header "X-Webdav-Prefix: /foo"
 221                 if !strings.HasPrefix(arvPath, prefix) {
 222                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 223                         return
 224                 }
 225                 arvPath = r.URL.Path[len(prefix):]
 226                 if arvPath == "" {
 227                         arvPath = "/"
 228                 }
 229                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 230                 webdavPrefix = prefix
 231         }
 232         pathParts := strings.Split(arvPath[1:], "/")
 233
 234         var stripParts int
 235         var collectionID string
 236         var tokens []string
 237         var reqTokens []string
 238         var pathToken bool
 239         var attachment bool
 240         var useSiteFS bool
 241         credentialsOK := h.Cluster.Collections.TrustAllContent
 242         reasonNotAcceptingCredentials := ""
 243
 244         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 245                 credentialsOK = true
 246                 attachment = true
 247         } else if r.FormValue("disposition") == "attachment" {
 248                 attachment = true
 249         }
 250
 251         if !credentialsOK {
 252                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 253                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 254         }
 255
 256         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 257                 // http://ID.collections.example/PATH...
 258                 credentialsOK = true
 259         } else if r.URL.Path == "/status.json" {
 260                 h.serveStatus(w, r)
 261                 return
 262         } else if siteFSDir[pathParts[0]] {
 263                 useSiteFS = true
 264         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 265                 // /c=ID[/PATH...]
 266                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 267                 stripParts = 1
 268         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 269                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 270                         // /collections/download/ID/TOKEN/PATH...
 271                         collectionID = parseCollectionIDFromURL(pathParts[2])
 272                         tokens = []string{pathParts[3]}
 273                         stripParts = 4
 274                         pathToken = true
 275                 } else {
 276                         // /collections/ID/PATH...
 277                         collectionID = parseCollectionIDFromURL(pathParts[1])
 278                         stripParts = 2
 279                         // This path is only meant to work for public
 280                         // data. Tokens provided with the request are
 281                         // ignored.
 282                         credentialsOK = false
 283                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 284                 }
 285         }
 286
 287         forceReload := false
 288         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 289                 forceReload = true
 290         }
 291
 292         if credentialsOK {
 293                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 294         }
 295
 296         formToken := r.FormValue("api_token")
 297         origin := r.Header.Get("Origin")
 298         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 299         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 300         safeAttachment := attachment && r.URL.Query().Get("api_token") == ""
 301         if formToken == "" {
 302                 // No token to use or redact.
 303         } else if safeAjax || safeAttachment {
 304                 // If this is a cross-origin request, the URL won't
 305                 // appear in the browser's address bar, so
 306                 // substituting a clipboard-safe URL is pointless.
 307                 // Redirect-with-cookie wouldn't work anyway, because
 308                 // it's not safe to allow third-party use of our
 309                 // cookie.
 310                 //
 311                 // If we're supplying an attachment, we don't need to
 312                 // convert POST to GET to avoid the "really resubmit
 313                 // form?" problem, so provided the token isn't
 314                 // embedded in the URL, there's no reason to do
 315                 // redirect-with-cookie in this case either.
 316                 reqTokens = append(reqTokens, formToken)
 317         } else if browserMethod[r.Method] {
 318                 // If this is a page view, and the client provided a
 319                 // token via query string or POST body, we must put
 320                 // the token in an HttpOnly cookie, and redirect to an
 321                 // equivalent URL with the query param redacted and
 322                 // method = GET.
 323                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 324                 return
 325         }
 326
 327         targetPath := pathParts[stripParts:]
 328         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 329                 // http://ID.example/t=TOKEN/PATH...
 330                 // /c=ID/t=TOKEN/PATH...
 331                 //
 332                 // This form must only be used to pass scoped tokens
 333                 // that give permission for a single collection. See
 334                 // FormValue case above.
 335                 tokens = []string{targetPath[0][2:]}
 336                 pathToken = true
 337                 targetPath = targetPath[1:]
 338                 stripParts++
 339         }
 340
 341         fsprefix := ""
 342         if useSiteFS {
 343                 if writeMethod[r.Method] {
 344                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 345                         return
 346                 }
 347                 if len(reqTokens) == 0 {
 348                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 349                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 350                         return
 351                 }
 352                 tokens = reqTokens
 353         } else if collectionID == "" {
 354                 http.Error(w, notFoundMessage, http.StatusNotFound)
 355                 return
 356         } else {
 357                 fsprefix = "by_id/" + collectionID + "/"
 358         }
 359
 360         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 361                 fsprefix += src[1:]
 362         }
 363
 364         if tokens == nil {
 365                 tokens = reqTokens
 366                 if h.Cluster.Users.AnonymousUserToken != "" {
 367                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 368                 }
 369         }
 370
 371         if len(targetPath) > 0 && targetPath[0] == "_" {
 372                 // If a collection has a directory called "t=foo" or
 373                 // "_", it can be served at
 374                 // //collections.example/_/t=foo/ or
 375                 // //collections.example/_/_/ respectively:
 376                 // //collections.example/t=foo/ won't work because
 377                 // t=foo will be interpreted as a token "foo".
 378                 targetPath = targetPath[1:]
 379                 stripParts++
 380         }
 381
 382         dirOpenMode := os.O_RDONLY
 383         if writeMethod[r.Method] {
 384                 dirOpenMode = os.O_RDWR
 385         }
 386
 387         var tokenValid bool
 388         var tokenScopeProblem bool
 389         var token string
 390         var tokenUser *arvados.User
 391         var sessionFS arvados.CustomFileSystem
 392         var session *cachedSession
 393         var collectionDir arvados.File
 394         for _, token = range tokens {
 395                 var statusErr errorWithHTTPStatus
 396                 fs, sess, user, err := h.Cache.GetSession(token)
 397                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 398                         // bad token
 399                         continue
 400                 } else if err != nil {
 401                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 402                         return
 403                 }
 404                 if token != h.Cluster.Users.AnonymousUserToken {
 405                         tokenValid = true
 406                 }
 407                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 408                 if errors.As(err, &statusErr) &&
 409                         statusErr.HTTPStatus() == http.StatusForbidden &&
 410                         token != h.Cluster.Users.AnonymousUserToken {
 411                         // collection id is outside scope of supplied
 412                         // token
 413                         tokenScopeProblem = true
 414                         sess.Release()
 415                         continue
 416                 } else if os.IsNotExist(err) {
 417                         // collection does not exist or is not
 418                         // readable using this token
 419                         sess.Release()
 420                         continue
 421                 } else if err != nil {
 422                         http.Error(w, err.Error(), http.StatusInternalServerError)
 423                         sess.Release()
 424                         return
 425                 }
 426                 defer f.Close()
 427                 defer sess.Release()
 428
 429                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 430                 break
 431         }
 432         if forceReload && collectionDir != nil {
 433                 err := collectionDir.Sync()
 434                 if err != nil {
 435                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 436                                 http.Error(w, err.Error(), he.HTTPStatus())
 437                         } else {
 438                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 439                         }
 440                         return
 441                 }
 442         }
 443         if session == nil {
 444                 if pathToken {
 445                         // The URL is a "secret sharing link" that
 446                         // didn't work out.  Asking the client for
 447                         // additional credentials would just be
 448                         // confusing.
 449                         http.Error(w, notFoundMessage, http.StatusNotFound)
 450                         return
 451                 }
 452                 if tokenValid {
 453                         // The client provided valid token(s), but the
 454                         // collection was not found.
 455                         http.Error(w, notFoundMessage, http.StatusNotFound)
 456                         return
 457                 }
 458                 if tokenScopeProblem {
 459                         // The client provided a valid token but
 460                         // fetching a collection returned 401, which
 461                         // means the token scope doesn't permit
 462                         // fetching that collection.
 463                         http.Error(w, notFoundMessage, http.StatusForbidden)
 464                         return
 465                 }
 466                 // The client's token was invalid (e.g., expired), or
 467                 // the client didn't even provide one.  Redirect to
 468                 // workbench2's login-and-redirect-to-download url if
 469                 // this is a browser navigation request. (The redirect
 470                 // flow can't preserve the original method if it's not
 471                 // GET, and doesn't make sense if the UA is a
 472                 // command-line tool, is trying to load an inline
 473                 // image, etc.; in these cases, there's nothing we can
 474                 // do, so return 401 unauthorized.)
 475                 //
 476                 // Note Sec-Fetch-Mode is sent by all non-EOL
 477                 // browsers, except Safari.
 478                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 479                 //
 480                 // TODO(TC): This response would be confusing to
 481                 // someone trying (anonymously) to download public
 482                 // data that has been deleted.  Allow a referrer to
 483                 // provide this context somehow?
 484                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 485                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 486                         redirkey := "redirectToPreview"
 487                         if attachment {
 488                                 redirkey = "redirectToDownload"
 489                         }
 490                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 491                         // target.RawQuery = url.Values{redirkey:
 492                         // {target}}.Encode() would be the obvious
 493                         // thing to do here, but wb2 doesn't decode
 494                         // this as a query param -- it takes
 495                         // everything after "${redirkey}=" as the
 496                         // target URL. If we encode "/" as "%2F" etc.,
 497                         // the redirect won't work.
 498                         target.RawQuery = redirkey + "=" + callback
 499                         w.Header().Add("Location", target.String())
 500                         w.WriteHeader(http.StatusSeeOther)
 501                         return
 502                 }
 503                 if !credentialsOK {
 504                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 505                         return
 506                 }
 507                 // If none of the above cases apply, suggest the
 508                 // user-agent (which is either a non-browser agent
 509                 // like wget, or a browser that can't redirect through
 510                 // a login flow) prompt the user for credentials.
 511                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 512                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 513                 return
 514         }
 515
 516         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 517                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 518                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 519                         if !strings.HasSuffix(r.URL.Path, "/") {
 520                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 521                         } else {
 522                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 523                         }
 524                         return
 525                 }
 526         }
 527
 528         var basename string
 529         if len(targetPath) > 0 {
 530                 basename = targetPath[len(targetPath)-1]
 531         }
 532         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 533                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 534                 return
 535         }
 536         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 537                 http.Error(w, "Not permitted", http.StatusForbidden)
 538                 return
 539         }
 540         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 541
 542         writing := writeMethod[r.Method]
 543         locker := h.collectionLock(collectionID, writing)
 544         defer locker.Unlock()
 545
 546         if writing {
 547                 // Save the collection only if/when all
 548                 // webdav->filesystem operations succeed --
 549                 // and send a 500 error if the modified
 550                 // collection can't be saved.
 551                 //
 552                 // Perform the write in a separate sitefs, so
 553                 // concurrent read operations on the same
 554                 // collection see the previous saved
 555                 // state. After the write succeeds and the
 556                 // collection record is updated, we reset the
 557                 // session so the updates are visible in
 558                 // subsequent read requests.
 559                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 560                 sessionFS = client.SiteFileSystem(session.keepclient)
 561                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 562                 if err != nil {
 563                         http.Error(w, err.Error(), http.StatusInternalServerError)
 564                         return
 565                 }
 566                 defer writingDir.Close()
 567                 w = &updateOnSuccess{
 568                         ResponseWriter: w,
 569                         logger:         ctxlog.FromContext(r.Context()),
 570                         update: func() error {
 571                                 err := writingDir.Sync()
 572                                 var te arvados.TransactionError
 573                                 if errors.As(err, &te) {
 574                                         err = te
 575                                 }
 576                                 if err != nil {
 577                                         return err
 578                                 }
 579                                 // Sync the changes to the persistent
 580                                 // sessionfs for this token.
 581                                 snap, err := writingDir.Snapshot()
 582                                 if err != nil {
 583                                         return err
 584                                 }
 585                                 collectionDir.Splice(snap)
 586                                 return nil
 587                         }}
 588         }
 589         if r.Method == http.MethodGet {
 590                 applyContentDispositionHdr(w, r, basename, attachment)
 591         }
 592         if webdavPrefix == "" {
 593                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 594         }
 595         wh := webdav.Handler{
 596                 Prefix: webdavPrefix,
 597                 FileSystem: &webdavfs.FS{
 598                         FileSystem:    sessionFS,
 599                         Prefix:        fsprefix,
 600                         Writing:       writeMethod[r.Method],
 601                         AlwaysReadEOF: r.Method == "PROPFIND",
 602                 },
 603                 LockSystem: webdavfs.NoLockSystem,
 604                 Logger: func(r *http.Request, err error) {
 605                         if err != nil && !os.IsNotExist(err) {
 606                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 607                         }
 608                 },
 609         }
 610         wh.ServeHTTP(w, r)
 611         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 612                 wrote := int64(w.WroteBodyBytes())
 613                 fnm := strings.Join(pathParts[stripParts:], "/")
 614                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 615                 if err == nil && fi.Size() != wrote {
 616                         var n int
 617                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 618                         if err == nil {
 619                                 n, err = f.Read(make([]byte, 1024))
 620                                 f.Close()
 621                         }
 622                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 623                 }
 624         }
 625 }
 626
 627 var dirListingTemplate = `<!DOCTYPE HTML>
 628 <HTML><HEAD>
 629   <META name="robots" content="NOINDEX">
 630   <TITLE>{{ .CollectionName }}</TITLE>
 631   <STYLE type="text/css">
 632     body {
 633       margin: 1.5em;
 634     }
 635     pre {
 636       background-color: #D9EDF7;
 637       border-radius: .25em;
 638       padding: .75em;
 639       overflow: auto;
 640     }
 641     .footer p {
 642       font-size: 82%;
 643     }
 644     ul {
 645       padding: 0;
 646     }
 647     ul li {
 648       font-family: monospace;
 649       list-style: none;
 650     }
 651   </STYLE>
 652 </HEAD>
 653 <BODY>
 654
 655 <H1>{{ .CollectionName }}</H1>
 656
 657 <P>This collection of data files is being shared with you through
 658 Arvados.  You can download individual files listed below.  To download
 659 the entire directory tree with wget, try:</P>
 660
 661 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 662
 663 <H2>File Listing</H2>
 664
 665 {{if .Files}}
 666 <UL>
 667 {{range .Files}}
 668 {{if .IsDir }}
 669   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 670 {{else}}
 671   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 672 {{end}}
 673 {{end}}
 674 </UL>
 675 {{else}}
 676 <P>(No files; this collection is empty.)</P>
 677 {{end}}
 678
 679 <HR noshade>
 680 <DIV class="footer">
 681   <P>
 682     About Arvados:
 683     Arvados is a free and open source software bioinformatics platform.
 684     To learn more, visit arvados.org.
 685     Arvados is not responsible for the files listed on this page.
 686   </P>
 687 </DIV>
 688
 689 </BODY>
 690 `
 691
 692 type fileListEnt struct {
 693         Name  string
 694         Size  int64
 695         IsDir bool
 696 }
 697
 698 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 699         var files []fileListEnt
 700         var walk func(string) error
 701         if !strings.HasSuffix(base, "/") {
 702                 base = base + "/"
 703         }
 704         walk = func(path string) error {
 705                 dirname := base + path
 706                 if dirname != "/" {
 707                         dirname = strings.TrimSuffix(dirname, "/")
 708                 }
 709                 d, err := fs.Open(dirname)
 710                 if err != nil {
 711                         return err
 712                 }
 713                 ents, err := d.Readdir(-1)
 714                 if err != nil {
 715                         return err
 716                 }
 717                 for _, ent := range ents {
 718                         if recurse && ent.IsDir() {
 719                                 err = walk(path + ent.Name() + "/")
 720                                 if err != nil {
 721                                         return err
 722                                 }
 723                         } else {
 724                                 files = append(files, fileListEnt{
 725                                         Name:  path + ent.Name(),
 726                                         Size:  ent.Size(),
 727                                         IsDir: ent.IsDir(),
 728                                 })
 729                         }
 730                 }
 731                 return nil
 732         }
 733         if err := walk(""); err != nil {
 734                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 735                 return
 736         }
 737
 738         funcs := template.FuncMap{
 739                 "nbsp": func(s string) template.HTML {
 740                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 741                 },
 742         }
 743         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 744         if err != nil {
 745                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 746                 return
 747         }
 748         sort.Slice(files, func(i, j int) bool {
 749                 return files[i].Name < files[j].Name
 750         })
 751         w.WriteHeader(http.StatusOK)
 752         tmpl.Execute(w, map[string]interface{}{
 753                 "CollectionName": collectionName,
 754                 "Files":          files,
 755                 "Request":        r,
 756                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 757         })
 758 }
 759
 760 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 761         disposition := "inline"
 762         if isAttachment {
 763                 disposition = "attachment"
 764         }
 765         if strings.ContainsRune(r.RequestURI, '?') {
 766                 // Help the UA realize that the filename is just
 767                 // "filename.txt", not
 768                 // "filename.txt?disposition=attachment".
 769                 //
 770                 // TODO(TC): Follow advice at RFC 6266 appendix D
 771                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 772         }
 773         if disposition != "inline" {
 774                 w.Header().Set("Content-Disposition", disposition)
 775         }
 776 }
 777
 778 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 779         if formToken := r.FormValue("api_token"); formToken != "" {
 780                 if !credentialsOK {
 781                         // It is not safe to copy the provided token
 782                         // into a cookie unless the current vhost
 783                         // (origin) serves only a single collection or
 784                         // we are in TrustAllContent mode.
 785                         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)
 786                         return
 787                 }
 788
 789                 // The HttpOnly flag is necessary to prevent
 790                 // JavaScript code (included in, or loaded by, a page
 791                 // in the collection being served) from employing the
 792                 // user's token beyond reading other files in the same
 793                 // domain, i.e., same collection.
 794                 //
 795                 // The 303 redirect is necessary in the case of a GET
 796                 // request to avoid exposing the token in the Location
 797                 // bar, and in the case of a POST request to avoid
 798                 // raising warnings when the user refreshes the
 799                 // resulting page.
 800                 http.SetCookie(w, &http.Cookie{
 801                         Name:     "arvados_api_token",
 802                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 803                         Path:     "/",
 804                         HttpOnly: true,
 805                         SameSite: http.SameSiteLaxMode,
 806                 })
 807         }
 808
 809         // Propagate query parameters (except api_token) from
 810         // the original request.
 811         redirQuery := r.URL.Query()
 812         redirQuery.Del("api_token")
 813
 814         u := r.URL
 815         if location != "" {
 816                 newu, err := u.Parse(location)
 817                 if err != nil {
 818                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 819                         return
 820                 }
 821                 u = newu
 822         }
 823         redir := (&url.URL{
 824                 Scheme:   r.URL.Scheme,
 825                 Host:     r.Host,
 826                 Path:     u.Path,
 827                 RawQuery: redirQuery.Encode(),
 828         }).String()
 829
 830         w.Header().Add("Location", redir)
 831         w.WriteHeader(http.StatusSeeOther)
 832         io.WriteString(w, `<A href="`)
 833         io.WriteString(w, html.EscapeString(redir))
 834         io.WriteString(w, `">Continue</A>`)
 835 }
 836
 837 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 838         var permitDownload bool
 839         var permitUpload bool
 840         if tokenUser != nil && tokenUser.IsAdmin {
 841                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 842                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 843         } else {
 844                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 845                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 846         }
 847         if (method == "PUT" || method == "POST") && !permitUpload {
 848                 // Disallow operations that upload new files.
 849                 // Permit webdav operations that move existing files around.
 850                 return false
 851         } else if method == "GET" && !permitDownload {
 852                 // Disallow downloading file contents.
 853                 // Permit webdav operations like PROPFIND that retrieve metadata
 854                 // but not file contents.
 855                 return false
 856         }
 857         return true
 858 }
 859
 860 func (h *handler) logUploadOrDownload(
 861         r *http.Request,
 862         client *arvadosclient.ArvadosClient,
 863         fs arvados.CustomFileSystem,
 864         filepath string,
 865         collection *arvados.Collection,
 866         user *arvados.User) {
 867
 868         log := ctxlog.FromContext(r.Context())
 869         props := make(map[string]string)
 870         props["reqPath"] = r.URL.Path
 871         var useruuid string
 872         if user != nil {
 873                 log = log.WithField("user_uuid", user.UUID).
 874                         WithField("user_full_name", user.FullName)
 875                 useruuid = user.UUID
 876         } else {
 877                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 878         }
 879         if collection == nil && fs != nil {
 880                 collection, filepath = h.determineCollection(fs, filepath)
 881         }
 882         if collection != nil {
 883                 log = log.WithField("collection_file_path", filepath)
 884                 props["collection_file_path"] = filepath
 885                 // h.determineCollection populates the collection_uuid
 886                 // prop with the PDH, if this collection is being
 887                 // accessed via PDH. For logging, we use a different
 888                 // field depending on whether it's a UUID or PDH.
 889                 if len(collection.UUID) > 32 {
 890                         log = log.WithField("portable_data_hash", collection.UUID)
 891                         props["portable_data_hash"] = collection.UUID
 892                 } else {
 893                         log = log.WithField("collection_uuid", collection.UUID)
 894                         props["collection_uuid"] = collection.UUID
 895                 }
 896         }
 897         if r.Method == "PUT" || r.Method == "POST" {
 898                 log.Info("File upload")
 899                 if h.Cluster.Collections.WebDAVLogEvents {
 900                         go func() {
 901                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 902                                         "object_uuid": useruuid,
 903                                         "event_type":  "file_upload",
 904                                         "properties":  props}}
 905                                 err := client.Create("logs", lr, nil)
 906                                 if err != nil {
 907                                         log.WithError(err).Error("Failed to create upload log event on API server")
 908                                 }
 909                         }()
 910                 }
 911         } else if r.Method == "GET" {
 912                 if collection != nil && collection.PortableDataHash != "" {
 913                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 914                         props["portable_data_hash"] = collection.PortableDataHash
 915                 }
 916                 log.Info("File download")
 917                 if h.Cluster.Collections.WebDAVLogEvents {
 918                         go func() {
 919                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 920                                         "object_uuid": useruuid,
 921                                         "event_type":  "file_download",
 922                                         "properties":  props}}
 923                                 err := client.Create("logs", lr, nil)
 924                                 if err != nil {
 925                                         log.WithError(err).Error("Failed to create download log event on API server")
 926                                 }
 927                         }()
 928                 }
 929         }
 930 }
 931
 932 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 933         target := strings.TrimSuffix(path, "/")
 934         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 935                 target = target[:cut]
 936                 fi, err := fs.Stat(target)
 937                 if os.IsNotExist(err) {
 938                         // creating a new file/dir, or download
 939                         // destined to fail
 940                         continue
 941                 } else if err != nil {
 942                         return nil, ""
 943                 }
 944                 switch src := fi.Sys().(type) {
 945                 case *arvados.Collection:
 946                         return src, strings.TrimPrefix(path[len(target):], "/")
 947                 case *arvados.Group:
 948                         return nil, ""
 949                 default:
 950                         if _, ok := src.(error); ok {
 951                                 return nil, ""
 952                         }
 953                 }
 954         }
 955         return nil, ""
 956 }
 957
 958 var lockTidyInterval = time.Minute * 10
 959
 960 // Lock the specified collection for reading or writing. Caller must
 961 // call Unlock() on the returned Locker when the operation is
 962 // finished.
 963 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
 964         h.lockMtx.Lock()
 965         defer h.lockMtx.Unlock()
 966         if time.Since(h.lockTidied) > lockTidyInterval {
 967                 // Periodically delete all locks that aren't in use.
 968                 h.lockTidied = time.Now()
 969                 for id, locker := range h.lock {
 970                         if locker.TryLock() {
 971                                 locker.Unlock()
 972                                 delete(h.lock, id)
 973                         }
 974                 }
 975         }
 976         locker := h.lock[collectionID]
 977         if locker == nil {
 978                 locker = new(sync.RWMutex)
 979                 if h.lock == nil {
 980                         h.lock = map[string]*sync.RWMutex{}
 981                 }
 982                 h.lock[collectionID] = locker
 983         }
 984         if writing {
 985                 locker.Lock()
 986                 return locker
 987         } else {
 988                 locker.RLock()
 989                 return locker.RLocker()
 990         }
 991 }
 992
 993 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
 994         method := header.Get("Access-Control-Request-Method")
 995         if method == "" {
 996                 return false
 997         }
 998         if !browserMethod[method] && !webdavMethod[method] {
 999                 w.WriteHeader(http.StatusMethodNotAllowed)
1000                 return true
1001         }
1002         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1003         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1004         w.Header().Set("Access-Control-Allow-Origin", "*")
1005         w.Header().Set("Access-Control-Max-Age", "86400")
1006         return true
1007 }