services/keep-web/handler.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package keepweb
   6
   7 import (
   8         "encoding/json"
   9         "errors"
  10         "fmt"
  11         "html"
  12         "html/template"
  13         "io"
  14         "net/http"
  15         "net/url"
  16         "os"
  17         "sort"
  18         "strconv"
  19         "strings"
  20         "sync"
  21         "time"
  22
  23         "git.arvados.org/arvados.git/lib/cmd"
  24         "git.arvados.org/arvados.git/lib/webdavfs"
  25         "git.arvados.org/arvados.git/sdk/go/arvados"
  26         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  27         "git.arvados.org/arvados.git/sdk/go/auth"
  28         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  29         "git.arvados.org/arvados.git/sdk/go/httpserver"
  30         "github.com/sirupsen/logrus"
  31         "golang.org/x/net/webdav"
  32 )
  33
  34 type handler struct {
  35         Cache   cache
  36         Cluster *arvados.Cluster
  37         metrics *metrics
  38
  39         lockMtx    sync.Mutex
  40         lock       map[string]*sync.RWMutex
  41         lockTidied time.Time
  42 }
  43
  44 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  45
  46 var notFoundMessage = "Not Found"
  47 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  48
  49 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  50 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  51 // otherwise "".
  52 func parseCollectionIDFromURL(s string) string {
  53         if arvadosclient.UUIDMatch(s) {
  54                 return s
  55         }
  56         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  57                 return pdh
  58         }
  59         return ""
  60 }
  61
  62 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  63         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  64 }
  65
  66 type errorWithHTTPStatus interface {
  67         HTTPStatus() int
  68 }
  69
  70 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  71 // sends an HTTP header indicating success, updateOnSuccess first
  72 // calls the provided update func. If the update func fails, an error
  73 // response is sent (using the error's HTTP status or 500 if none),
  74 // and the status code and body sent by the handler are ignored (all
  75 // response writes return the update error).
  76 type updateOnSuccess struct {
  77         httpserver.ResponseWriter
  78         logger     logrus.FieldLogger
  79         update     func() error
  80         sentHeader bool
  81         err        error
  82 }
  83
  84 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  85         if !uos.sentHeader {
  86                 uos.WriteHeader(http.StatusOK)
  87         }
  88         if uos.err != nil {
  89                 return 0, uos.err
  90         }
  91         return uos.ResponseWriter.Write(p)
  92 }
  93
  94 func (uos *updateOnSuccess) WriteHeader(code int) {
  95         if !uos.sentHeader {
  96                 uos.sentHeader = true
  97                 if code >= 200 && code < 400 {
  98                         if uos.err = uos.update(); uos.err != nil {
  99                                 code := http.StatusInternalServerError
 100                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 101                                         code = he.HTTPStatus()
 102                                 }
 103                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 104                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 105                                 return
 106                         }
 107                 }
 108         }
 109         uos.ResponseWriter.WriteHeader(code)
 110 }
 111
 112 var (
 113         corsAllowHeadersHeader = strings.Join([]string{
 114                 "Authorization", "Content-Type", "Range",
 115                 // WebDAV request headers:
 116                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 117         }, ", ")
 118         writeMethod = map[string]bool{
 119                 "COPY":      true,
 120                 "DELETE":    true,
 121                 "LOCK":      true,
 122                 "MKCOL":     true,
 123                 "MOVE":      true,
 124                 "PROPPATCH": true,
 125                 "PUT":       true,
 126                 "RMCOL":     true,
 127                 "UNLOCK":    true,
 128         }
 129         webdavMethod = map[string]bool{
 130                 "COPY":      true,
 131                 "DELETE":    true,
 132                 "LOCK":      true,
 133                 "MKCOL":     true,
 134                 "MOVE":      true,
 135                 "OPTIONS":   true,
 136                 "PROPFIND":  true,
 137                 "PROPPATCH": true,
 138                 "PUT":       true,
 139                 "RMCOL":     true,
 140                 "UNLOCK":    true,
 141         }
 142         browserMethod = map[string]bool{
 143                 "GET":  true,
 144                 "HEAD": true,
 145                 "POST": true,
 146         }
 147         // top-level dirs to serve with siteFS
 148         siteFSDir = map[string]bool{
 149                 "":      true, // root directory
 150                 "by_id": true,
 151                 "users": true,
 152         }
 153 )
 154
 155 func stripDefaultPort(host string) string {
 156         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 157         u := &url.URL{Host: host}
 158         if p := u.Port(); p == "80" || p == "443" {
 159                 return strings.ToLower(u.Hostname())
 160         } else {
 161                 return strings.ToLower(host)
 162         }
 163 }
 164
 165 // CheckHealth implements service.Handler.
 166 func (h *handler) CheckHealth() error {
 167         return nil
 168 }
 169
 170 // Done implements service.Handler.
 171 func (h *handler) Done() <-chan struct{} {
 172         return nil
 173 }
 174
 175 // ServeHTTP implements http.Handler.
 176 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 177         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 178                 r.URL.Scheme = xfp
 179         }
 180
 181         wbuffer := newWriteBuffer(wOrig, int(h.Cluster.Collections.WebDAVOutputBuffer))
 182         defer wbuffer.Close()
 183         w := httpserver.WrapResponseWriter(responseWriter{
 184                 Writer:         wbuffer,
 185                 ResponseWriter: wOrig,
 186         })
 187
 188         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 189                 return
 190         }
 191
 192         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 193                 w.WriteHeader(http.StatusMethodNotAllowed)
 194                 return
 195         }
 196
 197         if r.Header.Get("Origin") != "" {
 198                 // Allow simple cross-origin requests without user
 199                 // credentials ("user credentials" as defined by CORS,
 200                 // i.e., cookies, HTTP authentication, and client-side
 201                 // SSL certificates. See
 202                 // http://www.w3.org/TR/cors/#user-credentials).
 203                 w.Header().Set("Access-Control-Allow-Origin", "*")
 204                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 205         }
 206
 207         if h.serveS3(w, r) {
 208                 return
 209         }
 210
 211         webdavPrefix := ""
 212         arvPath := r.URL.Path
 213         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 214                 // Enable a proxy (e.g., container log handler in
 215                 // controller) to satisfy a request for path
 216                 // "/foo/bar/baz.txt" using content from
 217                 // "//abc123-4.internal/bar/baz.txt", by adding a
 218                 // request header "X-Webdav-Prefix: /foo"
 219                 if !strings.HasPrefix(arvPath, prefix) {
 220                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 221                         return
 222                 }
 223                 arvPath = r.URL.Path[len(prefix):]
 224                 if arvPath == "" {
 225                         arvPath = "/"
 226                 }
 227                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 228                 webdavPrefix = prefix
 229         }
 230         pathParts := strings.Split(arvPath[1:], "/")
 231
 232         var stripParts int
 233         var collectionID string
 234         var tokens []string
 235         var reqTokens []string
 236         var pathToken bool
 237         var attachment bool
 238         var useSiteFS bool
 239         credentialsOK := h.Cluster.Collections.TrustAllContent
 240         reasonNotAcceptingCredentials := ""
 241
 242         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 243                 credentialsOK = true
 244                 attachment = true
 245         } else if r.FormValue("disposition") == "attachment" {
 246                 attachment = true
 247         }
 248
 249         if !credentialsOK {
 250                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 251                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 252         }
 253
 254         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 255                 // http://ID.collections.example/PATH...
 256                 credentialsOK = true
 257         } else if r.URL.Path == "/status.json" {
 258                 h.serveStatus(w, r)
 259                 return
 260         } else if siteFSDir[pathParts[0]] {
 261                 useSiteFS = true
 262         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 263                 // /c=ID[/PATH...]
 264                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 265                 stripParts = 1
 266         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 267                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 268                         // /collections/download/ID/TOKEN/PATH...
 269                         collectionID = parseCollectionIDFromURL(pathParts[2])
 270                         tokens = []string{pathParts[3]}
 271                         stripParts = 4
 272                         pathToken = true
 273                 } else {
 274                         // /collections/ID/PATH...
 275                         collectionID = parseCollectionIDFromURL(pathParts[1])
 276                         stripParts = 2
 277                         // This path is only meant to work for public
 278                         // data. Tokens provided with the request are
 279                         // ignored.
 280                         credentialsOK = false
 281                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 282                 }
 283         }
 284
 285         forceReload := false
 286         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 287                 forceReload = true
 288         }
 289
 290         if credentialsOK {
 291                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 292         }
 293
 294         r.ParseForm()
 295         origin := r.Header.Get("Origin")
 296         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 297         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 298         // Important distinction: safeAttachment checks whether api_token exists
 299         // as a query parameter. haveFormTokens checks whether api_token exists
 300         // as request form data *or* a query parameter. Different checks are
 301         // necessary because both the request disposition and the location of
 302         // the API token affect whether or not the request needs to be
 303         // redirected. The different branch comments below explain further.
 304         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 305         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 306                 // No token to use or redact.
 307         } else if safeAjax || safeAttachment {
 308                 // If this is a cross-origin request, the URL won't
 309                 // appear in the browser's address bar, so
 310                 // substituting a clipboard-safe URL is pointless.
 311                 // Redirect-with-cookie wouldn't work anyway, because
 312                 // it's not safe to allow third-party use of our
 313                 // cookie.
 314                 //
 315                 // If we're supplying an attachment, we don't need to
 316                 // convert POST to GET to avoid the "really resubmit
 317                 // form?" problem, so provided the token isn't
 318                 // embedded in the URL, there's no reason to do
 319                 // redirect-with-cookie in this case either.
 320                 for _, tok := range formTokens {
 321                         reqTokens = append(reqTokens, tok)
 322                 }
 323         } else if browserMethod[r.Method] {
 324                 // If this is a page view, and the client provided a
 325                 // token via query string or POST body, we must put
 326                 // the token in an HttpOnly cookie, and redirect to an
 327                 // equivalent URL with the query param redacted and
 328                 // method = GET.
 329                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 330                 return
 331         }
 332
 333         targetPath := pathParts[stripParts:]
 334         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 335                 // http://ID.example/t=TOKEN/PATH...
 336                 // /c=ID/t=TOKEN/PATH...
 337                 //
 338                 // This form must only be used to pass scoped tokens
 339                 // that give permission for a single collection. See
 340                 // FormValue case above.
 341                 tokens = []string{targetPath[0][2:]}
 342                 pathToken = true
 343                 targetPath = targetPath[1:]
 344                 stripParts++
 345         }
 346
 347         fsprefix := ""
 348         if useSiteFS {
 349                 if writeMethod[r.Method] {
 350                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 351                         return
 352                 }
 353                 if len(reqTokens) == 0 {
 354                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 355                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 356                         return
 357                 }
 358                 tokens = reqTokens
 359         } else if collectionID == "" {
 360                 http.Error(w, notFoundMessage, http.StatusNotFound)
 361                 return
 362         } else {
 363                 fsprefix = "by_id/" + collectionID + "/"
 364         }
 365
 366         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 367                 fsprefix += src[1:]
 368         }
 369
 370         if tokens == nil {
 371                 tokens = reqTokens
 372                 if h.Cluster.Users.AnonymousUserToken != "" {
 373                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 374                 }
 375         }
 376
 377         if len(targetPath) > 0 && targetPath[0] == "_" {
 378                 // If a collection has a directory called "t=foo" or
 379                 // "_", it can be served at
 380                 // //collections.example/_/t=foo/ or
 381                 // //collections.example/_/_/ respectively:
 382                 // //collections.example/t=foo/ won't work because
 383                 // t=foo will be interpreted as a token "foo".
 384                 targetPath = targetPath[1:]
 385                 stripParts++
 386         }
 387
 388         dirOpenMode := os.O_RDONLY
 389         if writeMethod[r.Method] {
 390                 dirOpenMode = os.O_RDWR
 391         }
 392
 393         var tokenValid bool
 394         var tokenScopeProblem bool
 395         var token string
 396         var tokenUser *arvados.User
 397         var sessionFS arvados.CustomFileSystem
 398         var session *cachedSession
 399         var collectionDir arvados.File
 400         for _, token = range tokens {
 401                 var statusErr errorWithHTTPStatus
 402                 fs, sess, user, err := h.Cache.GetSession(token)
 403                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 404                         // bad token
 405                         continue
 406                 } else if err != nil {
 407                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 408                         return
 409                 }
 410                 if token != h.Cluster.Users.AnonymousUserToken {
 411                         tokenValid = true
 412                 }
 413                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 414                 if errors.As(err, &statusErr) &&
 415                         statusErr.HTTPStatus() == http.StatusForbidden &&
 416                         token != h.Cluster.Users.AnonymousUserToken {
 417                         // collection id is outside scope of supplied
 418                         // token
 419                         tokenScopeProblem = true
 420                         sess.Release()
 421                         continue
 422                 } else if os.IsNotExist(err) {
 423                         // collection does not exist or is not
 424                         // readable using this token
 425                         sess.Release()
 426                         continue
 427                 } else if err != nil {
 428                         http.Error(w, err.Error(), http.StatusInternalServerError)
 429                         sess.Release()
 430                         return
 431                 }
 432                 defer f.Close()
 433                 defer sess.Release()
 434
 435                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 436                 break
 437         }
 438         if forceReload && collectionDir != nil {
 439                 err := collectionDir.Sync()
 440                 if err != nil {
 441                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 442                                 http.Error(w, err.Error(), he.HTTPStatus())
 443                         } else {
 444                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 445                         }
 446                         return
 447                 }
 448         }
 449         if session == nil {
 450                 if pathToken {
 451                         // The URL is a "secret sharing link" that
 452                         // didn't work out.  Asking the client for
 453                         // additional credentials would just be
 454                         // confusing.
 455                         http.Error(w, notFoundMessage, http.StatusNotFound)
 456                         return
 457                 }
 458                 if tokenValid {
 459                         // The client provided valid token(s), but the
 460                         // collection was not found.
 461                         http.Error(w, notFoundMessage, http.StatusNotFound)
 462                         return
 463                 }
 464                 if tokenScopeProblem {
 465                         // The client provided a valid token but
 466                         // fetching a collection returned 401, which
 467                         // means the token scope doesn't permit
 468                         // fetching that collection.
 469                         http.Error(w, notFoundMessage, http.StatusForbidden)
 470                         return
 471                 }
 472                 // The client's token was invalid (e.g., expired), or
 473                 // the client didn't even provide one.  Redirect to
 474                 // workbench2's login-and-redirect-to-download url if
 475                 // this is a browser navigation request. (The redirect
 476                 // flow can't preserve the original method if it's not
 477                 // GET, and doesn't make sense if the UA is a
 478                 // command-line tool, is trying to load an inline
 479                 // image, etc.; in these cases, there's nothing we can
 480                 // do, so return 401 unauthorized.)
 481                 //
 482                 // Note Sec-Fetch-Mode is sent by all non-EOL
 483                 // browsers, except Safari.
 484                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 485                 //
 486                 // TODO(TC): This response would be confusing to
 487                 // someone trying (anonymously) to download public
 488                 // data that has been deleted.  Allow a referrer to
 489                 // provide this context somehow?
 490                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 491                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 492                         redirkey := "redirectToPreview"
 493                         if attachment {
 494                                 redirkey = "redirectToDownload"
 495                         }
 496                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 497                         // target.RawQuery = url.Values{redirkey:
 498                         // {target}}.Encode() would be the obvious
 499                         // thing to do here, but wb2 doesn't decode
 500                         // this as a query param -- it takes
 501                         // everything after "${redirkey}=" as the
 502                         // target URL. If we encode "/" as "%2F" etc.,
 503                         // the redirect won't work.
 504                         target.RawQuery = redirkey + "=" + callback
 505                         w.Header().Add("Location", target.String())
 506                         w.WriteHeader(http.StatusSeeOther)
 507                         return
 508                 }
 509                 if !credentialsOK {
 510                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 511                         return
 512                 }
 513                 // If none of the above cases apply, suggest the
 514                 // user-agent (which is either a non-browser agent
 515                 // like wget, or a browser that can't redirect through
 516                 // a login flow) prompt the user for credentials.
 517                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 518                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 519                 return
 520         }
 521
 522         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 523                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 524                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 525                         if !strings.HasSuffix(r.URL.Path, "/") {
 526                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 527                         } else {
 528                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 529                         }
 530                         return
 531                 }
 532         }
 533
 534         var basename string
 535         if len(targetPath) > 0 {
 536                 basename = targetPath[len(targetPath)-1]
 537         }
 538         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 539                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 540                 return
 541         }
 542         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 543                 http.Error(w, "Not permitted", http.StatusForbidden)
 544                 return
 545         }
 546         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 547
 548         writing := writeMethod[r.Method]
 549         locker := h.collectionLock(collectionID, writing)
 550         defer locker.Unlock()
 551
 552         if writing {
 553                 // Save the collection only if/when all
 554                 // webdav->filesystem operations succeed --
 555                 // and send a 500 error if the modified
 556                 // collection can't be saved.
 557                 //
 558                 // Perform the write in a separate sitefs, so
 559                 // concurrent read operations on the same
 560                 // collection see the previous saved
 561                 // state. After the write succeeds and the
 562                 // collection record is updated, we reset the
 563                 // session so the updates are visible in
 564                 // subsequent read requests.
 565                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 566                 sessionFS = client.SiteFileSystem(session.keepclient)
 567                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 568                 if err != nil {
 569                         http.Error(w, err.Error(), http.StatusInternalServerError)
 570                         return
 571                 }
 572                 defer writingDir.Close()
 573                 w = &updateOnSuccess{
 574                         ResponseWriter: w,
 575                         logger:         ctxlog.FromContext(r.Context()),
 576                         update: func() error {
 577                                 err := writingDir.Sync()
 578                                 var te arvados.TransactionError
 579                                 if errors.As(err, &te) {
 580                                         err = te
 581                                 }
 582                                 if err != nil {
 583                                         return err
 584                                 }
 585                                 // Sync the changes to the persistent
 586                                 // sessionfs for this token.
 587                                 snap, err := writingDir.Snapshot()
 588                                 if err != nil {
 589                                         return err
 590                                 }
 591                                 collectionDir.Splice(snap)
 592                                 return nil
 593                         }}
 594         }
 595         if r.Method == http.MethodGet {
 596                 applyContentDispositionHdr(w, r, basename, attachment)
 597         }
 598         if webdavPrefix == "" {
 599                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 600         }
 601         wh := &webdav.Handler{
 602                 Prefix: webdavPrefix,
 603                 FileSystem: &webdavfs.FS{
 604                         FileSystem:    sessionFS,
 605                         Prefix:        fsprefix,
 606                         Writing:       writeMethod[r.Method],
 607                         AlwaysReadEOF: r.Method == "PROPFIND",
 608                 },
 609                 LockSystem: webdavfs.NoLockSystem,
 610                 Logger: func(r *http.Request, err error) {
 611                         if err != nil && !os.IsNotExist(err) {
 612                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 613                         }
 614                 },
 615         }
 616         h.metrics.track(wh, w, r)
 617         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 618                 wrote := int64(w.WroteBodyBytes())
 619                 fnm := strings.Join(pathParts[stripParts:], "/")
 620                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 621                 if err == nil && fi.Size() != wrote {
 622                         var n int
 623                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 624                         if err == nil {
 625                                 n, err = f.Read(make([]byte, 1024))
 626                                 f.Close()
 627                         }
 628                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 629                 }
 630         }
 631 }
 632
 633 var dirListingTemplate = `<!DOCTYPE HTML>
 634 <HTML><HEAD>
 635   <META name="robots" content="NOINDEX">
 636   <TITLE>{{ .CollectionName }}</TITLE>
 637   <STYLE type="text/css">
 638     body {
 639       margin: 1.5em;
 640     }
 641     pre {
 642       background-color: #D9EDF7;
 643       border-radius: .25em;
 644       padding: .75em;
 645       overflow: auto;
 646     }
 647     .footer p {
 648       font-size: 82%;
 649     }
 650     ul {
 651       padding: 0;
 652     }
 653     ul li {
 654       font-family: monospace;
 655       list-style: none;
 656     }
 657   </STYLE>
 658 </HEAD>
 659 <BODY>
 660
 661 <H1>{{ .CollectionName }}</H1>
 662
 663 <P>This collection of data files is being shared with you through
 664 Arvados.  You can download individual files listed below.  To download
 665 the entire directory tree with wget, try:</P>
 666
 667 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 668
 669 <H2>File Listing</H2>
 670
 671 {{if .Files}}
 672 <UL>
 673 {{range .Files}}
 674 {{if .IsDir }}
 675   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 676 {{else}}
 677   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 678 {{end}}
 679 {{end}}
 680 </UL>
 681 {{else}}
 682 <P>(No files; this collection is empty.)</P>
 683 {{end}}
 684
 685 <HR noshade>
 686 <DIV class="footer">
 687   <P>
 688     About Arvados:
 689     Arvados is a free and open source software bioinformatics platform.
 690     To learn more, visit arvados.org.
 691     Arvados is not responsible for the files listed on this page.
 692   </P>
 693 </DIV>
 694
 695 </BODY>
 696 `
 697
 698 type fileListEnt struct {
 699         Name  string
 700         Size  int64
 701         IsDir bool
 702 }
 703
 704 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 705         var files []fileListEnt
 706         var walk func(string) error
 707         if !strings.HasSuffix(base, "/") {
 708                 base = base + "/"
 709         }
 710         walk = func(path string) error {
 711                 dirname := base + path
 712                 if dirname != "/" {
 713                         dirname = strings.TrimSuffix(dirname, "/")
 714                 }
 715                 d, err := fs.Open(dirname)
 716                 if err != nil {
 717                         return err
 718                 }
 719                 ents, err := d.Readdir(-1)
 720                 if err != nil {
 721                         return err
 722                 }
 723                 for _, ent := range ents {
 724                         if recurse && ent.IsDir() {
 725                                 err = walk(path + ent.Name() + "/")
 726                                 if err != nil {
 727                                         return err
 728                                 }
 729                         } else {
 730                                 files = append(files, fileListEnt{
 731                                         Name:  path + ent.Name(),
 732                                         Size:  ent.Size(),
 733                                         IsDir: ent.IsDir(),
 734                                 })
 735                         }
 736                 }
 737                 return nil
 738         }
 739         if err := walk(""); err != nil {
 740                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 741                 return
 742         }
 743
 744         funcs := template.FuncMap{
 745                 "nbsp": func(s string) template.HTML {
 746                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 747                 },
 748         }
 749         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 750         if err != nil {
 751                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 752                 return
 753         }
 754         sort.Slice(files, func(i, j int) bool {
 755                 return files[i].Name < files[j].Name
 756         })
 757         w.WriteHeader(http.StatusOK)
 758         tmpl.Execute(w, map[string]interface{}{
 759                 "CollectionName": collectionName,
 760                 "Files":          files,
 761                 "Request":        r,
 762                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 763         })
 764 }
 765
 766 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 767         disposition := "inline"
 768         if isAttachment {
 769                 disposition = "attachment"
 770         }
 771         if strings.ContainsRune(r.RequestURI, '?') {
 772                 // Help the UA realize that the filename is just
 773                 // "filename.txt", not
 774                 // "filename.txt?disposition=attachment".
 775                 //
 776                 // TODO(TC): Follow advice at RFC 6266 appendix D
 777                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 778         }
 779         if disposition != "inline" {
 780                 w.Header().Set("Content-Disposition", disposition)
 781         }
 782 }
 783
 784 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 785         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
 786                 if !credentialsOK {
 787                         // It is not safe to copy the provided token
 788                         // into a cookie unless the current vhost
 789                         // (origin) serves only a single collection or
 790                         // we are in TrustAllContent mode.
 791                         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)
 792                         return
 793                 }
 794
 795                 // The HttpOnly flag is necessary to prevent
 796                 // JavaScript code (included in, or loaded by, a page
 797                 // in the collection being served) from employing the
 798                 // user's token beyond reading other files in the same
 799                 // domain, i.e., same collection.
 800                 //
 801                 // The 303 redirect is necessary in the case of a GET
 802                 // request to avoid exposing the token in the Location
 803                 // bar, and in the case of a POST request to avoid
 804                 // raising warnings when the user refreshes the
 805                 // resulting page.
 806                 for _, tok := range formTokens {
 807                         if tok == "" {
 808                                 continue
 809                         }
 810                         http.SetCookie(w, &http.Cookie{
 811                                 Name:     "arvados_api_token",
 812                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
 813                                 Path:     "/",
 814                                 HttpOnly: true,
 815                                 SameSite: http.SameSiteLaxMode,
 816                         })
 817                         break
 818                 }
 819         }
 820
 821         // Propagate query parameters (except api_token) from
 822         // the original request.
 823         redirQuery := r.URL.Query()
 824         redirQuery.Del("api_token")
 825
 826         u := r.URL
 827         if location != "" {
 828                 newu, err := u.Parse(location)
 829                 if err != nil {
 830                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 831                         return
 832                 }
 833                 u = newu
 834         }
 835         redir := (&url.URL{
 836                 Scheme:   r.URL.Scheme,
 837                 Host:     r.Host,
 838                 Path:     u.Path,
 839                 RawQuery: redirQuery.Encode(),
 840         }).String()
 841
 842         w.Header().Add("Location", redir)
 843         w.WriteHeader(http.StatusSeeOther)
 844         io.WriteString(w, `<A href="`)
 845         io.WriteString(w, html.EscapeString(redir))
 846         io.WriteString(w, `">Continue</A>`)
 847 }
 848
 849 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 850         var permitDownload bool
 851         var permitUpload bool
 852         if tokenUser != nil && tokenUser.IsAdmin {
 853                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 854                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 855         } else {
 856                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 857                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 858         }
 859         if (method == "PUT" || method == "POST") && !permitUpload {
 860                 // Disallow operations that upload new files.
 861                 // Permit webdav operations that move existing files around.
 862                 return false
 863         } else if method == "GET" && !permitDownload {
 864                 // Disallow downloading file contents.
 865                 // Permit webdav operations like PROPFIND that retrieve metadata
 866                 // but not file contents.
 867                 return false
 868         }
 869         return true
 870 }
 871
 872 func (h *handler) logUploadOrDownload(
 873         r *http.Request,
 874         client *arvadosclient.ArvadosClient,
 875         fs arvados.CustomFileSystem,
 876         filepath string,
 877         collection *arvados.Collection,
 878         user *arvados.User) {
 879
 880         log := ctxlog.FromContext(r.Context())
 881         props := make(map[string]string)
 882         props["reqPath"] = r.URL.Path
 883         var useruuid string
 884         if user != nil {
 885                 log = log.WithField("user_uuid", user.UUID).
 886                         WithField("user_full_name", user.FullName)
 887                 useruuid = user.UUID
 888         } else {
 889                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 890         }
 891         if collection == nil && fs != nil {
 892                 collection, filepath = h.determineCollection(fs, filepath)
 893         }
 894         if collection != nil {
 895                 log = log.WithField("collection_file_path", filepath)
 896                 props["collection_file_path"] = filepath
 897                 // h.determineCollection populates the collection_uuid
 898                 // prop with the PDH, if this collection is being
 899                 // accessed via PDH. For logging, we use a different
 900                 // field depending on whether it's a UUID or PDH.
 901                 if len(collection.UUID) > 32 {
 902                         log = log.WithField("portable_data_hash", collection.UUID)
 903                         props["portable_data_hash"] = collection.UUID
 904                 } else {
 905                         log = log.WithField("collection_uuid", collection.UUID)
 906                         props["collection_uuid"] = collection.UUID
 907                 }
 908         }
 909         if r.Method == "PUT" || r.Method == "POST" {
 910                 log.Info("File upload")
 911                 if h.Cluster.Collections.WebDAVLogEvents {
 912                         go func() {
 913                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 914                                         "object_uuid": useruuid,
 915                                         "event_type":  "file_upload",
 916                                         "properties":  props}}
 917                                 err := client.Create("logs", lr, nil)
 918                                 if err != nil {
 919                                         log.WithError(err).Error("Failed to create upload log event on API server")
 920                                 }
 921                         }()
 922                 }
 923         } else if r.Method == "GET" {
 924                 if collection != nil && collection.PortableDataHash != "" {
 925                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 926                         props["portable_data_hash"] = collection.PortableDataHash
 927                 }
 928                 log.Info("File download")
 929                 if h.Cluster.Collections.WebDAVLogEvents {
 930                         go func() {
 931                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 932                                         "object_uuid": useruuid,
 933                                         "event_type":  "file_download",
 934                                         "properties":  props}}
 935                                 err := client.Create("logs", lr, nil)
 936                                 if err != nil {
 937                                         log.WithError(err).Error("Failed to create download log event on API server")
 938                                 }
 939                         }()
 940                 }
 941         }
 942 }
 943
 944 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 945         target := strings.TrimSuffix(path, "/")
 946         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 947                 target = target[:cut]
 948                 fi, err := fs.Stat(target)
 949                 if os.IsNotExist(err) {
 950                         // creating a new file/dir, or download
 951                         // destined to fail
 952                         continue
 953                 } else if err != nil {
 954                         return nil, ""
 955                 }
 956                 switch src := fi.Sys().(type) {
 957                 case *arvados.Collection:
 958                         return src, strings.TrimPrefix(path[len(target):], "/")
 959                 case *arvados.Group:
 960                         return nil, ""
 961                 default:
 962                         if _, ok := src.(error); ok {
 963                                 return nil, ""
 964                         }
 965                 }
 966         }
 967         return nil, ""
 968 }
 969
 970 var lockTidyInterval = time.Minute * 10
 971
 972 // Lock the specified collection for reading or writing. Caller must
 973 // call Unlock() on the returned Locker when the operation is
 974 // finished.
 975 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
 976         h.lockMtx.Lock()
 977         defer h.lockMtx.Unlock()
 978         if time.Since(h.lockTidied) > lockTidyInterval {
 979                 // Periodically delete all locks that aren't in use.
 980                 h.lockTidied = time.Now()
 981                 for id, locker := range h.lock {
 982                         if locker.TryLock() {
 983                                 locker.Unlock()
 984                                 delete(h.lock, id)
 985                         }
 986                 }
 987         }
 988         locker := h.lock[collectionID]
 989         if locker == nil {
 990                 locker = new(sync.RWMutex)
 991                 if h.lock == nil {
 992                         h.lock = map[string]*sync.RWMutex{}
 993                 }
 994                 h.lock[collectionID] = locker
 995         }
 996         if writing {
 997                 locker.Lock()
 998                 return locker
 999         } else {
1000                 locker.RLock()
1001                 return locker.RLocker()
1002         }
1003 }
1004
1005 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1006         method := header.Get("Access-Control-Request-Method")
1007         if method == "" {
1008                 return false
1009         }
1010         if !browserMethod[method] && !webdavMethod[method] {
1011                 w.WriteHeader(http.StatusMethodNotAllowed)
1012                 return true
1013         }
1014         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1015         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1016         w.Header().Set("Access-Control-Allow-Origin", "*")
1017         w.Header().Set("Access-Control-Max-Age", "86400")
1018         return true
1019 }