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