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