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         r.ParseForm()
 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         // Important distinction: safeAttachment checks whether api_token exists
 301         // as a query parameter. haveFormTokens checks whether api_token exists
 302         // as request form data *or* a query parameter. Different checks are
 303         // necessary because both the request disposition and the location of
 304         // the API token affect whether or not the request needs to be
 305         // redirected. The different branch comments below explain further.
 306         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 307         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 308                 // No token to use or redact.
 309         } else if safeAjax || safeAttachment {
 310                 // If this is a cross-origin request, the URL won't
 311                 // appear in the browser's address bar, so
 312                 // substituting a clipboard-safe URL is pointless.
 313                 // Redirect-with-cookie wouldn't work anyway, because
 314                 // it's not safe to allow third-party use of our
 315                 // cookie.
 316                 //
 317                 // If we're supplying an attachment, we don't need to
 318                 // convert POST to GET to avoid the "really resubmit
 319                 // form?" problem, so provided the token isn't
 320                 // embedded in the URL, there's no reason to do
 321                 // redirect-with-cookie in this case either.
 322                 for _, tok := range formTokens {
 323                         reqTokens = append(reqTokens, tok)
 324                 }
 325         } else if browserMethod[r.Method] {
 326                 // If this is a page view, and the client provided a
 327                 // token via query string or POST body, we must put
 328                 // the token in an HttpOnly cookie, and redirect to an
 329                 // equivalent URL with the query param redacted and
 330                 // method = GET.
 331                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 332                 return
 333         }
 334
 335         targetPath := pathParts[stripParts:]
 336         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 337                 // http://ID.example/t=TOKEN/PATH...
 338                 // /c=ID/t=TOKEN/PATH...
 339                 //
 340                 // This form must only be used to pass scoped tokens
 341                 // that give permission for a single collection. See
 342                 // FormValue case above.
 343                 tokens = []string{targetPath[0][2:]}
 344                 pathToken = true
 345                 targetPath = targetPath[1:]
 346                 stripParts++
 347         }
 348
 349         fsprefix := ""
 350         if useSiteFS {
 351                 if writeMethod[r.Method] {
 352                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 353                         return
 354                 }
 355                 if len(reqTokens) == 0 {
 356                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 357                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 358                         return
 359                 }
 360                 tokens = reqTokens
 361         } else if collectionID == "" {
 362                 http.Error(w, notFoundMessage, http.StatusNotFound)
 363                 return
 364         } else {
 365                 fsprefix = "by_id/" + collectionID + "/"
 366         }
 367
 368         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 369                 fsprefix += src[1:]
 370         }
 371
 372         if tokens == nil {
 373                 tokens = reqTokens
 374                 if h.Cluster.Users.AnonymousUserToken != "" {
 375                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 376                 }
 377         }
 378
 379         if len(targetPath) > 0 && targetPath[0] == "_" {
 380                 // If a collection has a directory called "t=foo" or
 381                 // "_", it can be served at
 382                 // //collections.example/_/t=foo/ or
 383                 // //collections.example/_/_/ respectively:
 384                 // //collections.example/t=foo/ won't work because
 385                 // t=foo will be interpreted as a token "foo".
 386                 targetPath = targetPath[1:]
 387                 stripParts++
 388         }
 389
 390         dirOpenMode := os.O_RDONLY
 391         if writeMethod[r.Method] {
 392                 dirOpenMode = os.O_RDWR
 393         }
 394
 395         var tokenValid bool
 396         var tokenScopeProblem bool
 397         var token string
 398         var tokenUser *arvados.User
 399         var sessionFS arvados.CustomFileSystem
 400         var session *cachedSession
 401         var collectionDir arvados.File
 402         for _, token = range tokens {
 403                 var statusErr errorWithHTTPStatus
 404                 fs, sess, user, err := h.Cache.GetSession(token)
 405                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 406                         // bad token
 407                         continue
 408                 } else if err != nil {
 409                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 410                         return
 411                 }
 412                 if token != h.Cluster.Users.AnonymousUserToken {
 413                         tokenValid = true
 414                 }
 415                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 416                 if errors.As(err, &statusErr) &&
 417                         statusErr.HTTPStatus() == http.StatusForbidden &&
 418                         token != h.Cluster.Users.AnonymousUserToken {
 419                         // collection id is outside scope of supplied
 420                         // token
 421                         tokenScopeProblem = true
 422                         sess.Release()
 423                         continue
 424                 } else if os.IsNotExist(err) {
 425                         // collection does not exist or is not
 426                         // readable using this token
 427                         sess.Release()
 428                         continue
 429                 } else if err != nil {
 430                         http.Error(w, err.Error(), http.StatusInternalServerError)
 431                         sess.Release()
 432                         return
 433                 }
 434                 defer f.Close()
 435                 defer sess.Release()
 436
 437                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 438                 break
 439         }
 440         if forceReload && collectionDir != nil {
 441                 err := collectionDir.Sync()
 442                 if err != nil {
 443                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 444                                 http.Error(w, err.Error(), he.HTTPStatus())
 445                         } else {
 446                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 447                         }
 448                         return
 449                 }
 450         }
 451         if session == nil {
 452                 if pathToken {
 453                         // The URL is a "secret sharing link" that
 454                         // didn't work out.  Asking the client for
 455                         // additional credentials would just be
 456                         // confusing.
 457                         http.Error(w, notFoundMessage, http.StatusNotFound)
 458                         return
 459                 }
 460                 if tokenValid {
 461                         // The client provided valid token(s), but the
 462                         // collection was not found.
 463                         http.Error(w, notFoundMessage, http.StatusNotFound)
 464                         return
 465                 }
 466                 if tokenScopeProblem {
 467                         // The client provided a valid token but
 468                         // fetching a collection returned 401, which
 469                         // means the token scope doesn't permit
 470                         // fetching that collection.
 471                         http.Error(w, notFoundMessage, http.StatusForbidden)
 472                         return
 473                 }
 474                 // The client's token was invalid (e.g., expired), or
 475                 // the client didn't even provide one.  Redirect to
 476                 // workbench2's login-and-redirect-to-download url if
 477                 // this is a browser navigation request. (The redirect
 478                 // flow can't preserve the original method if it's not
 479                 // GET, and doesn't make sense if the UA is a
 480                 // command-line tool, is trying to load an inline
 481                 // image, etc.; in these cases, there's nothing we can
 482                 // do, so return 401 unauthorized.)
 483                 //
 484                 // Note Sec-Fetch-Mode is sent by all non-EOL
 485                 // browsers, except Safari.
 486                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 487                 //
 488                 // TODO(TC): This response would be confusing to
 489                 // someone trying (anonymously) to download public
 490                 // data that has been deleted.  Allow a referrer to
 491                 // provide this context somehow?
 492                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 493                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 494                         redirkey := "redirectToPreview"
 495                         if attachment {
 496                                 redirkey = "redirectToDownload"
 497                         }
 498                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 499                         // target.RawQuery = url.Values{redirkey:
 500                         // {target}}.Encode() would be the obvious
 501                         // thing to do here, but wb2 doesn't decode
 502                         // this as a query param -- it takes
 503                         // everything after "${redirkey}=" as the
 504                         // target URL. If we encode "/" as "%2F" etc.,
 505                         // the redirect won't work.
 506                         target.RawQuery = redirkey + "=" + callback
 507                         w.Header().Add("Location", target.String())
 508                         w.WriteHeader(http.StatusSeeOther)
 509                         return
 510                 }
 511                 if !credentialsOK {
 512                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 513                         return
 514                 }
 515                 // If none of the above cases apply, suggest the
 516                 // user-agent (which is either a non-browser agent
 517                 // like wget, or a browser that can't redirect through
 518                 // a login flow) prompt the user for credentials.
 519                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 520                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 521                 return
 522         }
 523
 524         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 525                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 526                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 527                         if !strings.HasSuffix(r.URL.Path, "/") {
 528                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 529                         } else {
 530                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 531                         }
 532                         return
 533                 }
 534         }
 535
 536         var basename string
 537         if len(targetPath) > 0 {
 538                 basename = targetPath[len(targetPath)-1]
 539         }
 540         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 541                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 542                 return
 543         }
 544         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 545                 http.Error(w, "Not permitted", http.StatusForbidden)
 546                 return
 547         }
 548         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 549
 550         writing := writeMethod[r.Method]
 551         locker := h.collectionLock(collectionID, writing)
 552         defer locker.Unlock()
 553
 554         if writing {
 555                 // Save the collection only if/when all
 556                 // webdav->filesystem operations succeed --
 557                 // and send a 500 error if the modified
 558                 // collection can't be saved.
 559                 //
 560                 // Perform the write in a separate sitefs, so
 561                 // concurrent read operations on the same
 562                 // collection see the previous saved
 563                 // state. After the write succeeds and the
 564                 // collection record is updated, we reset the
 565                 // session so the updates are visible in
 566                 // subsequent read requests.
 567                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 568                 sessionFS = client.SiteFileSystem(session.keepclient)
 569                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 570                 if err != nil {
 571                         http.Error(w, err.Error(), http.StatusInternalServerError)
 572                         return
 573                 }
 574                 defer writingDir.Close()
 575                 w = &updateOnSuccess{
 576                         ResponseWriter: w,
 577                         logger:         ctxlog.FromContext(r.Context()),
 578                         update: func() error {
 579                                 err := writingDir.Sync()
 580                                 var te arvados.TransactionError
 581                                 if errors.As(err, &te) {
 582                                         err = te
 583                                 }
 584                                 if err != nil {
 585                                         return err
 586                                 }
 587                                 // Sync the changes to the persistent
 588                                 // sessionfs for this token.
 589                                 snap, err := writingDir.Snapshot()
 590                                 if err != nil {
 591                                         return err
 592                                 }
 593                                 collectionDir.Splice(snap)
 594                                 return nil
 595                         }}
 596         }
 597         if r.Method == http.MethodGet {
 598                 applyContentDispositionHdr(w, r, basename, attachment)
 599         }
 600         if webdavPrefix == "" {
 601                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 602         }
 603         wh := webdav.Handler{
 604                 Prefix: webdavPrefix,
 605                 FileSystem: &webdavfs.FS{
 606                         FileSystem:    sessionFS,
 607                         Prefix:        fsprefix,
 608                         Writing:       writeMethod[r.Method],
 609                         AlwaysReadEOF: r.Method == "PROPFIND",
 610                 },
 611                 LockSystem: webdavfs.NoLockSystem,
 612                 Logger: func(r *http.Request, err error) {
 613                         if err != nil && !os.IsNotExist(err) {
 614                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 615                         }
 616                 },
 617         }
 618         wh.ServeHTTP(w, r)
 619         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 620                 wrote := int64(w.WroteBodyBytes())
 621                 fnm := strings.Join(pathParts[stripParts:], "/")
 622                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 623                 if err == nil && fi.Size() != wrote {
 624                         var n int
 625                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 626                         if err == nil {
 627                                 n, err = f.Read(make([]byte, 1024))
 628                                 f.Close()
 629                         }
 630                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 631                 }
 632         }
 633 }
 634
 635 var dirListingTemplate = `<!DOCTYPE HTML>
 636 <HTML><HEAD>
 637   <META name="robots" content="NOINDEX">
 638   <TITLE>{{ .CollectionName }}</TITLE>
 639   <STYLE type="text/css">
 640     body {
 641       margin: 1.5em;
 642     }
 643     pre {
 644       background-color: #D9EDF7;
 645       border-radius: .25em;
 646       padding: .75em;
 647       overflow: auto;
 648     }
 649     .footer p {
 650       font-size: 82%;
 651     }
 652     ul {
 653       padding: 0;
 654     }
 655     ul li {
 656       font-family: monospace;
 657       list-style: none;
 658     }
 659   </STYLE>
 660 </HEAD>
 661 <BODY>
 662
 663 <H1>{{ .CollectionName }}</H1>
 664
 665 <P>This collection of data files is being shared with you through
 666 Arvados.  You can download individual files listed below.  To download
 667 the entire directory tree with wget, try:</P>
 668
 669 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 670
 671 <H2>File Listing</H2>
 672
 673 {{if .Files}}
 674 <UL>
 675 {{range .Files}}
 676 {{if .IsDir }}
 677   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 678 {{else}}
 679   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 680 {{end}}
 681 {{end}}
 682 </UL>
 683 {{else}}
 684 <P>(No files; this collection is empty.)</P>
 685 {{end}}
 686
 687 <HR noshade>
 688 <DIV class="footer">
 689   <P>
 690     About Arvados:
 691     Arvados is a free and open source software bioinformatics platform.
 692     To learn more, visit arvados.org.
 693     Arvados is not responsible for the files listed on this page.
 694   </P>
 695 </DIV>
 696
 697 </BODY>
 698 `
 699
 700 type fileListEnt struct {
 701         Name  string
 702         Size  int64
 703         IsDir bool
 704 }
 705
 706 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 707         var files []fileListEnt
 708         var walk func(string) error
 709         if !strings.HasSuffix(base, "/") {
 710                 base = base + "/"
 711         }
 712         walk = func(path string) error {
 713                 dirname := base + path
 714                 if dirname != "/" {
 715                         dirname = strings.TrimSuffix(dirname, "/")
 716                 }
 717                 d, err := fs.Open(dirname)
 718                 if err != nil {
 719                         return err
 720                 }
 721                 ents, err := d.Readdir(-1)
 722                 if err != nil {
 723                         return err
 724                 }
 725                 for _, ent := range ents {
 726                         if recurse && ent.IsDir() {
 727                                 err = walk(path + ent.Name() + "/")
 728                                 if err != nil {
 729                                         return err
 730                                 }
 731                         } else {
 732                                 files = append(files, fileListEnt{
 733                                         Name:  path + ent.Name(),
 734                                         Size:  ent.Size(),
 735                                         IsDir: ent.IsDir(),
 736                                 })
 737                         }
 738                 }
 739                 return nil
 740         }
 741         if err := walk(""); err != nil {
 742                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 743                 return
 744         }
 745
 746         funcs := template.FuncMap{
 747                 "nbsp": func(s string) template.HTML {
 748                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 749                 },
 750         }
 751         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 752         if err != nil {
 753                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 754                 return
 755         }
 756         sort.Slice(files, func(i, j int) bool {
 757                 return files[i].Name < files[j].Name
 758         })
 759         w.WriteHeader(http.StatusOK)
 760         tmpl.Execute(w, map[string]interface{}{
 761                 "CollectionName": collectionName,
 762                 "Files":          files,
 763                 "Request":        r,
 764                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 765         })
 766 }
 767
 768 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 769         disposition := "inline"
 770         if isAttachment {
 771                 disposition = "attachment"
 772         }
 773         if strings.ContainsRune(r.RequestURI, '?') {
 774                 // Help the UA realize that the filename is just
 775                 // "filename.txt", not
 776                 // "filename.txt?disposition=attachment".
 777                 //
 778                 // TODO(TC): Follow advice at RFC 6266 appendix D
 779                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 780         }
 781         if disposition != "inline" {
 782                 w.Header().Set("Content-Disposition", disposition)
 783         }
 784 }
 785
 786 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 787         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
 788                 if !credentialsOK {
 789                         // It is not safe to copy the provided token
 790                         // into a cookie unless the current vhost
 791                         // (origin) serves only a single collection or
 792                         // we are in TrustAllContent mode.
 793                         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)
 794                         return
 795                 }
 796
 797                 // The HttpOnly flag is necessary to prevent
 798                 // JavaScript code (included in, or loaded by, a page
 799                 // in the collection being served) from employing the
 800                 // user's token beyond reading other files in the same
 801                 // domain, i.e., same collection.
 802                 //
 803                 // The 303 redirect is necessary in the case of a GET
 804                 // request to avoid exposing the token in the Location
 805                 // bar, and in the case of a POST request to avoid
 806                 // raising warnings when the user refreshes the
 807                 // resulting page.
 808                 for _, tok := range formTokens {
 809                         if tok == "" {
 810                                 continue
 811                         }
 812                         http.SetCookie(w, &http.Cookie{
 813                                 Name:     "arvados_api_token",
 814                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
 815                                 Path:     "/",
 816                                 HttpOnly: true,
 817                                 SameSite: http.SameSiteLaxMode,
 818                         })
 819                         break
 820                 }
 821         }
 822
 823         // Propagate query parameters (except api_token) from
 824         // the original request.
 825         redirQuery := r.URL.Query()
 826         redirQuery.Del("api_token")
 827
 828         u := r.URL
 829         if location != "" {
 830                 newu, err := u.Parse(location)
 831                 if err != nil {
 832                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 833                         return
 834                 }
 835                 u = newu
 836         }
 837         redir := (&url.URL{
 838                 Scheme:   r.URL.Scheme,
 839                 Host:     r.Host,
 840                 Path:     u.Path,
 841                 RawQuery: redirQuery.Encode(),
 842         }).String()
 843
 844         w.Header().Add("Location", redir)
 845         w.WriteHeader(http.StatusSeeOther)
 846         io.WriteString(w, `<A href="`)
 847         io.WriteString(w, html.EscapeString(redir))
 848         io.WriteString(w, `">Continue</A>`)
 849 }
 850
 851 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 852         var permitDownload bool
 853         var permitUpload bool
 854         if tokenUser != nil && tokenUser.IsAdmin {
 855                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 856                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 857         } else {
 858                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 859                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 860         }
 861         if (method == "PUT" || method == "POST") && !permitUpload {
 862                 // Disallow operations that upload new files.
 863                 // Permit webdav operations that move existing files around.
 864                 return false
 865         } else if method == "GET" && !permitDownload {
 866                 // Disallow downloading file contents.
 867                 // Permit webdav operations like PROPFIND that retrieve metadata
 868                 // but not file contents.
 869                 return false
 870         }
 871         return true
 872 }
 873
 874 func (h *handler) logUploadOrDownload(
 875         r *http.Request,
 876         client *arvadosclient.ArvadosClient,
 877         fs arvados.CustomFileSystem,
 878         filepath string,
 879         collection *arvados.Collection,
 880         user *arvados.User) {
 881
 882         log := ctxlog.FromContext(r.Context())
 883         props := make(map[string]string)
 884         props["reqPath"] = r.URL.Path
 885         var useruuid string
 886         if user != nil {
 887                 log = log.WithField("user_uuid", user.UUID).
 888                         WithField("user_full_name", user.FullName)
 889                 useruuid = user.UUID
 890         } else {
 891                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 892         }
 893         if collection == nil && fs != nil {
 894                 collection, filepath = h.determineCollection(fs, filepath)
 895         }
 896         if collection != nil {
 897                 log = log.WithField("collection_file_path", filepath)
 898                 props["collection_file_path"] = filepath
 899                 // h.determineCollection populates the collection_uuid
 900                 // prop with the PDH, if this collection is being
 901                 // accessed via PDH. For logging, we use a different
 902                 // field depending on whether it's a UUID or PDH.
 903                 if len(collection.UUID) > 32 {
 904                         log = log.WithField("portable_data_hash", collection.UUID)
 905                         props["portable_data_hash"] = collection.UUID
 906                 } else {
 907                         log = log.WithField("collection_uuid", collection.UUID)
 908                         props["collection_uuid"] = collection.UUID
 909                 }
 910         }
 911         if r.Method == "PUT" || r.Method == "POST" {
 912                 log.Info("File upload")
 913                 if h.Cluster.Collections.WebDAVLogEvents {
 914                         go func() {
 915                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 916                                         "object_uuid": useruuid,
 917                                         "event_type":  "file_upload",
 918                                         "properties":  props}}
 919                                 err := client.Create("logs", lr, nil)
 920                                 if err != nil {
 921                                         log.WithError(err).Error("Failed to create upload log event on API server")
 922                                 }
 923                         }()
 924                 }
 925         } else if r.Method == "GET" {
 926                 if collection != nil && collection.PortableDataHash != "" {
 927                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 928                         props["portable_data_hash"] = collection.PortableDataHash
 929                 }
 930                 log.Info("File download")
 931                 if h.Cluster.Collections.WebDAVLogEvents {
 932                         go func() {
 933                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 934                                         "object_uuid": useruuid,
 935                                         "event_type":  "file_download",
 936                                         "properties":  props}}
 937                                 err := client.Create("logs", lr, nil)
 938                                 if err != nil {
 939                                         log.WithError(err).Error("Failed to create download log event on API server")
 940                                 }
 941                         }()
 942                 }
 943         }
 944 }
 945
 946 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 947         target := strings.TrimSuffix(path, "/")
 948         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 949                 target = target[:cut]
 950                 fi, err := fs.Stat(target)
 951                 if os.IsNotExist(err) {
 952                         // creating a new file/dir, or download
 953                         // destined to fail
 954                         continue
 955                 } else if err != nil {
 956                         return nil, ""
 957                 }
 958                 switch src := fi.Sys().(type) {
 959                 case *arvados.Collection:
 960                         return src, strings.TrimPrefix(path[len(target):], "/")
 961                 case *arvados.Group:
 962                         return nil, ""
 963                 default:
 964                         if _, ok := src.(error); ok {
 965                                 return nil, ""
 966                         }
 967                 }
 968         }
 969         return nil, ""
 970 }
 971
 972 var lockTidyInterval = time.Minute * 10
 973
 974 // Lock the specified collection for reading or writing. Caller must
 975 // call Unlock() on the returned Locker when the operation is
 976 // finished.
 977 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
 978         h.lockMtx.Lock()
 979         defer h.lockMtx.Unlock()
 980         if time.Since(h.lockTidied) > lockTidyInterval {
 981                 // Periodically delete all locks that aren't in use.
 982                 h.lockTidied = time.Now()
 983                 for id, locker := range h.lock {
 984                         if locker.TryLock() {
 985                                 locker.Unlock()
 986                                 delete(h.lock, id)
 987                         }
 988                 }
 989         }
 990         locker := h.lock[collectionID]
 991         if locker == nil {
 992                 locker = new(sync.RWMutex)
 993                 if h.lock == nil {
 994                         h.lock = map[string]*sync.RWMutex{}
 995                 }
 996                 h.lock[collectionID] = locker
 997         }
 998         if writing {
 999                 locker.Lock()
1000                 return locker
1001         } else {
1002                 locker.RLock()
1003                 return locker.RLocker()
1004         }
1005 }
1006
1007 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1008         method := header.Get("Access-Control-Request-Method")
1009         if method == "" {
1010                 return false
1011         }
1012         if !browserMethod[method] && !webdavMethod[method] {
1013                 w.WriteHeader(http.StatusMethodNotAllowed)
1014                 return true
1015         }
1016         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1017         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1018         w.Header().Set("Access-Control-Allow-Origin", "*")
1019         w.Header().Set("Access-Control-Max-Age", "86400")
1020         return true
1021 }