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 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  67 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  68 // otherwise "".
  69 func parseCollectionIDFromURL(s string) string {
  70         if arvadosclient.UUIDMatch(s) {
  71                 return s
  72         }
  73         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  74                 return pdh
  75         }
  76         return ""
  77 }
  78
  79 func (h *handler) setup() {
  80         // Errors will be handled at the client pool.
  81         arv, _ := arvados.NewClientFromConfig(h.Config.cluster)
  82         h.clientPool = arvadosclient.MakeClientPoolWith(arv)
  83
  84         keepclient.RefreshServiceDiscoveryOnSIGHUP()
  85         keepclient.DefaultBlockCache.MaxBlocks = h.Config.cluster.Collections.WebDAVCache.MaxBlockEntries
  86
  87         h.healthHandler = &health.Handler{
  88                 Token:  h.Config.cluster.ManagementToken,
  89                 Prefix: "/_health/",
  90         }
  91
  92         // Even though we don't accept LOCK requests, every webdav
  93         // handler must have a non-nil LockSystem.
  94         h.webdavLS = &noLockSystem{}
  95 }
  96
  97 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  98         json.NewEncoder(w).Encode(struct{ Version string }{version})
  99 }
 100
 101 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
 102 // sends an HTTP header indicating success, updateOnSuccess first
 103 // calls the provided update func. If the update func fails, a 500
 104 // response is sent, and the status code and body sent by the handler
 105 // are ignored (all response writes return the update error).
 106 type updateOnSuccess struct {
 107         httpserver.ResponseWriter
 108         logger     logrus.FieldLogger
 109         update     func() error
 110         sentHeader bool
 111         err        error
 112 }
 113
 114 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
 115         if !uos.sentHeader {
 116                 uos.WriteHeader(http.StatusOK)
 117         }
 118         if uos.err != nil {
 119                 return 0, uos.err
 120         }
 121         return uos.ResponseWriter.Write(p)
 122 }
 123
 124 func (uos *updateOnSuccess) WriteHeader(code int) {
 125         if !uos.sentHeader {
 126                 uos.sentHeader = true
 127                 if code >= 200 && code < 400 {
 128                         if uos.err = uos.update(); uos.err != nil {
 129                                 code := http.StatusInternalServerError
 130                                 if err, ok := uos.err.(*arvados.TransactionError); ok {
 131                                         code = err.StatusCode
 132                                 }
 133                                 uos.logger.WithError(uos.err).Errorf("update() returned error type %T, changing response to HTTP %d", uos.err, code)
 134                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 135                                 return
 136                         }
 137                 }
 138         }
 139         uos.ResponseWriter.WriteHeader(code)
 140 }
 141
 142 var (
 143         corsAllowHeadersHeader = strings.Join([]string{
 144                 "Authorization", "Content-Type", "Range",
 145                 // WebDAV request headers:
 146                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout",
 147         }, ", ")
 148         writeMethod = map[string]bool{
 149                 "COPY":      true,
 150                 "DELETE":    true,
 151                 "LOCK":      true,
 152                 "MKCOL":     true,
 153                 "MOVE":      true,
 154                 "PROPPATCH": true,
 155                 "PUT":       true,
 156                 "RMCOL":     true,
 157                 "UNLOCK":    true,
 158         }
 159         webdavMethod = map[string]bool{
 160                 "COPY":      true,
 161                 "DELETE":    true,
 162                 "LOCK":      true,
 163                 "MKCOL":     true,
 164                 "MOVE":      true,
 165                 "OPTIONS":   true,
 166                 "PROPFIND":  true,
 167                 "PROPPATCH": true,
 168                 "PUT":       true,
 169                 "RMCOL":     true,
 170                 "UNLOCK":    true,
 171         }
 172         browserMethod = map[string]bool{
 173                 "GET":  true,
 174                 "HEAD": true,
 175                 "POST": true,
 176         }
 177         // top-level dirs to serve with siteFS
 178         siteFSDir = map[string]bool{
 179                 "":      true, // root directory
 180                 "by_id": true,
 181                 "users": true,
 182         }
 183 )
 184
 185 // ServeHTTP implements http.Handler.
 186 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 187         h.setupOnce.Do(h.setup)
 188
 189         remoteAddr := r.RemoteAddr
 190         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
 191                 remoteAddr = xff + "," + remoteAddr
 192         }
 193         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 194                 r.URL.Scheme = xfp
 195         }
 196
 197         w := httpserver.WrapResponseWriter(wOrig)
 198
 199         if strings.HasPrefix(r.URL.Path, "/_health/") && r.Method == "GET" {
 200                 h.healthHandler.ServeHTTP(w, r)
 201                 return
 202         }
 203
 204         if method := r.Header.Get("Access-Control-Request-Method"); method != "" && r.Method == "OPTIONS" {
 205                 if !browserMethod[method] && !webdavMethod[method] {
 206                         w.WriteHeader(http.StatusMethodNotAllowed)
 207                         return
 208                 }
 209                 w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
 210                 w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
 211                 w.Header().Set("Access-Control-Allow-Origin", "*")
 212                 w.Header().Set("Access-Control-Max-Age", "86400")
 213                 return
 214         }
 215
 216         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 217                 w.WriteHeader(http.StatusMethodNotAllowed)
 218                 return
 219         }
 220
 221         if r.Header.Get("Origin") != "" {
 222                 // Allow simple cross-origin requests without user
 223                 // credentials ("user credentials" as defined by CORS,
 224                 // i.e., cookies, HTTP authentication, and client-side
 225                 // SSL certificates. See
 226                 // http://www.w3.org/TR/cors/#user-credentials).
 227                 w.Header().Set("Access-Control-Allow-Origin", "*")
 228                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 229         }
 230
 231         if auth := r.Header.Get("Authorization"); strings.HasPrefix(auth, "AWS ") {
 232                 split := strings.SplitN(auth[4:], ":", 2)
 233                 if len(split) < 2 {
 234                         w.WriteHeader(http.StatusUnauthorized)
 235                         return
 236                 }
 237                 h.serveS3(w, r, split[0])
 238                 return
 239         } else if strings.HasPrefix(auth, "AWS4-HMAC-SHA256 ") {
 240                 w.WriteHeader(http.StatusBadRequest)
 241                 fmt.Println(w, "V4 signature is not supported")
 242                 return
 243         }
 244
 245         pathParts := strings.Split(r.URL.Path[1:], "/")
 246
 247         var stripParts int
 248         var collectionID string
 249         var tokens []string
 250         var reqTokens []string
 251         var pathToken bool
 252         var attachment bool
 253         var useSiteFS bool
 254         credentialsOK := h.Config.cluster.Collections.TrustAllContent
 255
 256         if r.Host != "" && r.Host == 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 collectionID = parseCollectionIDFromDNSName(r.Host); collectionID != "" {
 264                 // http://ID.collections.example/PATH...
 265                 credentialsOK = true
 266         } else if r.URL.Path == "/status.json" {
 267                 h.serveStatus(w, r)
 268                 return
 269         } else if strings.HasPrefix(r.URL.Path, "/metrics") {
 270                 h.MetricsAPI.ServeHTTP(w, r)
 271                 return
 272         } else if siteFSDir[pathParts[0]] {
 273                 useSiteFS = true
 274         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 275                 // /c=ID[/PATH...]
 276                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 277                 stripParts = 1
 278         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 279                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 280                         // /collections/download/ID/TOKEN/PATH...
 281                         collectionID = parseCollectionIDFromURL(pathParts[2])
 282                         tokens = []string{pathParts[3]}
 283                         stripParts = 4
 284                         pathToken = true
 285                 } else {
 286                         // /collections/ID/PATH...
 287                         collectionID = parseCollectionIDFromURL(pathParts[1])
 288                         stripParts = 2
 289                         // This path is only meant to work for public
 290                         // data. Tokens provided with the request are
 291                         // ignored.
 292                         credentialsOK = false
 293                 }
 294         }
 295
 296         if collectionID == "" && !useSiteFS {
 297                 w.WriteHeader(http.StatusNotFound)
 298                 return
 299         }
 300
 301         forceReload := false
 302         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 303                 forceReload = true
 304         }
 305
 306         if credentialsOK {
 307                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 308         }
 309
 310         formToken := r.FormValue("api_token")
 311         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 312                 // The client provided an explicit token in the POST
 313                 // body. The Origin header indicates this *might* be
 314                 // an AJAX request, in which case redirect-with-cookie
 315                 // won't work: we should just serve the content in the
 316                 // POST response. This is safe because:
 317                 //
 318                 // * We're supplying an attachment, not inline
 319                 //   content, so we don't need to convert the POST to
 320                 //   a GET and avoid the "really resubmit form?"
 321                 //   problem.
 322                 //
 323                 // * The token isn't embedded in the URL, so we don't
 324                 //   need to worry about bookmarks and copy/paste.
 325                 reqTokens = append(reqTokens, formToken)
 326         } else if formToken != "" && browserMethod[r.Method] {
 327                 // The client provided an explicit token in the query
 328                 // string, or a form in POST body. We must put the
 329                 // token in an HttpOnly cookie, and redirect to the
 330                 // same URL with the query param redacted and method =
 331                 // GET.
 332                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 333                 return
 334         }
 335
 336         if useSiteFS {
 337                 h.serveSiteFS(w, r, reqTokens, credentialsOK, attachment)
 338                 return
 339         }
 340
 341         targetPath := pathParts[stripParts:]
 342         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 343                 // http://ID.example/t=TOKEN/PATH...
 344                 // /c=ID/t=TOKEN/PATH...
 345                 //
 346                 // This form must only be used to pass scoped tokens
 347                 // that give permission for a single collection. See
 348                 // FormValue case above.
 349                 tokens = []string{targetPath[0][2:]}
 350                 pathToken = true
 351                 targetPath = targetPath[1:]
 352                 stripParts++
 353         }
 354
 355         if tokens == nil {
 356                 tokens = append(reqTokens, h.Config.cluster.Users.AnonymousUserToken)
 357         }
 358
 359         if len(targetPath) > 0 && targetPath[0] == "_" {
 360                 // If a collection has a directory called "t=foo" or
 361                 // "_", it can be served at
 362                 // //collections.example/_/t=foo/ or
 363                 // //collections.example/_/_/ respectively:
 364                 // //collections.example/t=foo/ won't work because
 365                 // t=foo will be interpreted as a token "foo".
 366                 targetPath = targetPath[1:]
 367                 stripParts++
 368         }
 369
 370         arv := h.clientPool.Get()
 371         if arv == nil {
 372                 http.Error(w, "client pool error: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
 373                 return
 374         }
 375         defer h.clientPool.Put(arv)
 376
 377         var collection *arvados.Collection
 378         tokenResult := make(map[string]int)
 379         for _, arv.ApiToken = range tokens {
 380                 var err error
 381                 collection, err = h.Config.Cache.Get(arv, collectionID, forceReload)
 382                 if err == nil {
 383                         // Success
 384                         break
 385                 }
 386                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 387                         switch srvErr.HttpStatusCode {
 388                         case 404, 401:
 389                                 // Token broken or insufficient to
 390                                 // retrieve collection
 391                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 392                                 continue
 393                         }
 394                 }
 395                 // Something more serious is wrong
 396                 http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 397                 return
 398         }
 399         if collection == nil {
 400                 if pathToken || !credentialsOK {
 401                         // Either the URL is a "secret sharing link"
 402                         // that didn't work out (and asking the client
 403                         // for additional credentials would just be
 404                         // confusing), or we don't even accept
 405                         // credentials at this path.
 406                         w.WriteHeader(http.StatusNotFound)
 407                         return
 408                 }
 409                 for _, t := range reqTokens {
 410                         if tokenResult[t] == 404 {
 411                                 // The client provided valid token(s), but the
 412                                 // collection was not found.
 413                                 w.WriteHeader(http.StatusNotFound)
 414                                 return
 415                         }
 416                 }
 417                 // The client's token was invalid (e.g., expired), or
 418                 // the client didn't even provide one.  Propagate the
 419                 // 401 to encourage the client to use a [different]
 420                 // token.
 421                 //
 422                 // TODO(TC): This response would be confusing to
 423                 // someone trying (anonymously) to download public
 424                 // data that has been deleted.  Allow a referrer to
 425                 // provide this context somehow?
 426                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 427                 w.WriteHeader(http.StatusUnauthorized)
 428                 return
 429         }
 430
 431         kc, err := keepclient.MakeKeepClient(arv)
 432         if err != nil {
 433                 http.Error(w, "error setting up keep client: "+err.Error(), http.StatusInternalServerError)
 434                 return
 435         }
 436         kc.RequestID = r.Header.Get("X-Request-Id")
 437
 438         var basename string
 439         if len(targetPath) > 0 {
 440                 basename = targetPath[len(targetPath)-1]
 441         }
 442         applyContentDispositionHdr(w, r, basename, attachment)
 443
 444         client := (&arvados.Client{
 445                 APIHost:   arv.ApiServer,
 446                 AuthToken: arv.ApiToken,
 447                 Insecure:  arv.ApiInsecure,
 448         }).WithRequestID(r.Header.Get("X-Request-Id"))
 449
 450         fs, err := collection.FileSystem(client, kc)
 451         if err != nil {
 452                 http.Error(w, "error creating collection filesystem: "+err.Error(), http.StatusInternalServerError)
 453                 return
 454         }
 455
 456         writefs, writeOK := fs.(arvados.CollectionFileSystem)
 457         targetIsPDH := arvadosclient.PDHMatch(collectionID)
 458         if (targetIsPDH || !writeOK) && writeMethod[r.Method] {
 459                 http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 460                 return
 461         }
 462
 463         if webdavMethod[r.Method] {
 464                 if writeMethod[r.Method] {
 465                         // Save the collection only if/when all
 466                         // webdav->filesystem operations succeed --
 467                         // and send a 500 error if the modified
 468                         // collection can't be saved.
 469                         w = &updateOnSuccess{
 470                                 ResponseWriter: w,
 471                                 logger:         ctxlog.FromContext(r.Context()),
 472                                 update: func() error {
 473                                         return h.Config.Cache.Update(client, *collection, writefs)
 474                                 }}
 475                 }
 476                 h := webdav.Handler{
 477                         Prefix: "/" + strings.Join(pathParts[:stripParts], "/"),
 478                         FileSystem: &webdavFS{
 479                                 collfs:        fs,
 480                                 writing:       writeMethod[r.Method],
 481                                 alwaysReadEOF: r.Method == "PROPFIND",
 482                         },
 483                         LockSystem: h.webdavLS,
 484                         Logger: func(_ *http.Request, err error) {
 485                                 if err != nil {
 486                                         ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 487                                 }
 488                         },
 489                 }
 490                 h.ServeHTTP(w, r)
 491                 return
 492         }
 493
 494         openPath := "/" + strings.Join(targetPath, "/")
 495         if f, err := fs.Open(openPath); os.IsNotExist(err) {
 496                 // Requested non-existent path
 497                 w.WriteHeader(http.StatusNotFound)
 498         } else if err != nil {
 499                 // Some other (unexpected) error
 500                 http.Error(w, "open: "+err.Error(), http.StatusInternalServerError)
 501         } else if stat, err := f.Stat(); err != nil {
 502                 // Can't get Size/IsDir (shouldn't happen with a collectionFS!)
 503                 http.Error(w, "stat: "+err.Error(), http.StatusInternalServerError)
 504         } else if stat.IsDir() && !strings.HasSuffix(r.URL.Path, "/") {
 505                 // If client requests ".../dirname", redirect to
 506                 // ".../dirname/". This way, relative links in the
 507                 // listing for "dirname" can always be "fnm", never
 508                 // "dirname/fnm".
 509                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 510         } else if stat.IsDir() {
 511                 h.serveDirectory(w, r, collection.Name, fs, openPath, true)
 512         } else {
 513                 http.ServeContent(w, r, basename, stat.ModTime(), f)
 514                 if wrote := int64(w.WroteBodyBytes()); wrote != stat.Size() && r.Header.Get("Range") == "" {
 515                         // If we wrote fewer bytes than expected, it's
 516                         // too late to change the real response code
 517                         // or send an error message to the client, but
 518                         // at least we can try to put some useful
 519                         // debugging info in the logs.
 520                         n, err := f.Read(make([]byte, 1024))
 521                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %s", stat.Size(), wrote, n, err)
 522
 523                 }
 524         }
 525 }
 526
 527 func (h *handler) getClients(reqID, token string) (arv *arvadosclient.ArvadosClient, kc *keepclient.KeepClient, client *arvados.Client, release func(), err error) {
 528         arv = h.clientPool.Get()
 529         if arv == nil {
 530                 return nil, nil, nil, nil, err
 531         }
 532         release = func() { h.clientPool.Put(arv) }
 533         arv.ApiToken = token
 534         kc, err = keepclient.MakeKeepClient(arv)
 535         if err != nil {
 536                 release()
 537                 return
 538         }
 539         kc.RequestID = reqID
 540         client = (&arvados.Client{
 541                 APIHost:   arv.ApiServer,
 542                 AuthToken: arv.ApiToken,
 543                 Insecure:  arv.ApiInsecure,
 544         }).WithRequestID(reqID)
 545         return
 546 }
 547
 548 func (h *handler) serveS3(w http.ResponseWriter, r *http.Request, token string) {
 549         _, kc, client, release, err := h.getClients(r.Header.Get("X-Request-Id"), token)
 550         if err != nil {
 551                 http.Error(w, "Pool failed: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
 552                 return
 553         }
 554         defer release()
 555
 556         r.URL.Path = "/by_id" + r.URL.Path
 557
 558         fs := client.SiteFileSystem(kc)
 559         fs.ForwardSlashNameSubstitution(h.Config.cluster.Collections.ForwardSlashNameSubstitution)
 560
 561         switch r.Method {
 562         case "GET":
 563                 fi, err := fs.Stat(r.URL.Path)
 564                 if os.IsNotExist(err) {
 565                         http.Error(w, err.Error(), http.StatusNotFound)
 566                         return
 567                 } else if err != nil {
 568                         http.Error(w, err.Error(), http.StatusInternalServerError)
 569                         return
 570                 } else if fi.IsDir() {
 571                         http.Error(w, "not found", http.StatusNotFound)
 572                 }
 573                 http.FileServer(fs).ServeHTTP(w, r)
 574                 return
 575         case "PUT":
 576                 f, err := fs.OpenFile(r.URL.Path, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 577                 if os.IsNotExist(err) {
 578                         // create missing intermediate directories, then try again
 579                         for i, c := range r.URL.Path {
 580                                 if i > 0 && c == '/' {
 581                                         dir := r.URL.Path[:i]
 582                                         err := fs.Mkdir(dir, 0755)
 583                                         if err != nil && err != os.ErrExist {
 584                                                 err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 585                                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 586                                                 return
 587                                         }
 588                                 }
 589                         }
 590                         f, err = fs.OpenFile(r.URL.Path, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 591                 }
 592                 if err != nil {
 593                         err = fmt.Errorf("open %q failed: %w", r.URL.Path, err)
 594                         http.Error(w, err.Error(), http.StatusBadRequest)
 595                         return
 596                 }
 597                 defer f.Close()
 598                 _, err = io.Copy(f, r.Body)
 599                 if err != nil {
 600                         err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 601                         http.Error(w, err.Error(), http.StatusBadGateway)
 602                         return
 603                 }
 604                 err = f.Close()
 605                 if err != nil {
 606                         err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 607                         http.Error(w, err.Error(), http.StatusBadGateway)
 608                         return
 609                 }
 610                 err = fs.Sync()
 611                 if err != nil {
 612                         err = fmt.Errorf("sync failed: %w", err)
 613                         http.Error(w, err.Error(), http.StatusInternalServerError)
 614                         return
 615                 }
 616                 w.WriteHeader(http.StatusOK)
 617                 return
 618         default:
 619                 http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
 620                 return
 621         }
 622 }
 623
 624 func (h *handler) serveSiteFS(w http.ResponseWriter, r *http.Request, tokens []string, credentialsOK, attachment bool) {
 625         if len(tokens) == 0 {
 626                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 627                 http.Error(w, http.StatusText(http.StatusUnauthorized), http.StatusUnauthorized)
 628                 return
 629         }
 630         if writeMethod[r.Method] {
 631                 http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 632                 return
 633         }
 634         _, kc, client, release, err := h.getClients(r.Header.Get("X-Request-Id"), tokens[0])
 635         if err != nil {
 636                 http.Error(w, "Pool failed: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
 637                 return
 638         }
 639         defer release()
 640
 641         fs := client.SiteFileSystem(kc)
 642         fs.ForwardSlashNameSubstitution(h.Config.cluster.Collections.ForwardSlashNameSubstitution)
 643         f, err := fs.Open(r.URL.Path)
 644         if os.IsNotExist(err) {
 645                 http.Error(w, err.Error(), http.StatusNotFound)
 646                 return
 647         } else if err != nil {
 648                 http.Error(w, err.Error(), http.StatusInternalServerError)
 649                 return
 650         }
 651         defer f.Close()
 652         if fi, err := f.Stat(); err == nil && fi.IsDir() && r.Method == "GET" {
 653                 if !strings.HasSuffix(r.URL.Path, "/") {
 654                         h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 655                 } else {
 656                         h.serveDirectory(w, r, fi.Name(), fs, r.URL.Path, false)
 657                 }
 658                 return
 659         }
 660         if r.Method == "GET" {
 661                 _, basename := filepath.Split(r.URL.Path)
 662                 applyContentDispositionHdr(w, r, basename, attachment)
 663         }
 664         wh := webdav.Handler{
 665                 Prefix: "/",
 666                 FileSystem: &webdavFS{
 667                         collfs:        fs,
 668                         writing:       writeMethod[r.Method],
 669                         alwaysReadEOF: r.Method == "PROPFIND",
 670                 },
 671                 LockSystem: h.webdavLS,
 672                 Logger: func(_ *http.Request, err error) {
 673                         if err != nil {
 674                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 675                         }
 676                 },
 677         }
 678         wh.ServeHTTP(w, r)
 679 }
 680
 681 var dirListingTemplate = `<!DOCTYPE HTML>
 682 <HTML><HEAD>
 683   <META name="robots" content="NOINDEX">
 684   <TITLE>{{ .CollectionName }}</TITLE>
 685   <STYLE type="text/css">
 686     body {
 687       margin: 1.5em;
 688     }
 689     pre {
 690       background-color: #D9EDF7;
 691       border-radius: .25em;
 692       padding: .75em;
 693       overflow: auto;
 694     }
 695     .footer p {
 696       font-size: 82%;
 697     }
 698     ul {
 699       padding: 0;
 700     }
 701     ul li {
 702       font-family: monospace;
 703       list-style: none;
 704     }
 705   </STYLE>
 706 </HEAD>
 707 <BODY>
 708
 709 <H1>{{ .CollectionName }}</H1>
 710
 711 <P>This collection of data files is being shared with you through
 712 Arvados.  You can download individual files listed below.  To download
 713 the entire directory tree with wget, try:</P>
 714
 715 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 716
 717 <H2>File Listing</H2>
 718
 719 {{if .Files}}
 720 <UL>
 721 {{range .Files}}
 722 {{if .IsDir }}
 723   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 724 {{else}}
 725   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 726 {{end}}
 727 {{end}}
 728 </UL>
 729 {{else}}
 730 <P>(No files; this collection is empty.)</P>
 731 {{end}}
 732
 733 <HR noshade>
 734 <DIV class="footer">
 735   <P>
 736     About Arvados:
 737     Arvados is a free and open source software bioinformatics platform.
 738     To learn more, visit arvados.org.
 739     Arvados is not responsible for the files listed on this page.
 740   </P>
 741 </DIV>
 742
 743 </BODY>
 744 `
 745
 746 type fileListEnt struct {
 747         Name  string
 748         Size  int64
 749         IsDir bool
 750 }
 751
 752 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 753         var files []fileListEnt
 754         var walk func(string) error
 755         if !strings.HasSuffix(base, "/") {
 756                 base = base + "/"
 757         }
 758         walk = func(path string) error {
 759                 dirname := base + path
 760                 if dirname != "/" {
 761                         dirname = strings.TrimSuffix(dirname, "/")
 762                 }
 763                 d, err := fs.Open(dirname)
 764                 if err != nil {
 765                         return err
 766                 }
 767                 ents, err := d.Readdir(-1)
 768                 if err != nil {
 769                         return err
 770                 }
 771                 for _, ent := range ents {
 772                         if recurse && ent.IsDir() {
 773                                 err = walk(path + ent.Name() + "/")
 774                                 if err != nil {
 775                                         return err
 776                                 }
 777                         } else {
 778                                 files = append(files, fileListEnt{
 779                                         Name:  path + ent.Name(),
 780                                         Size:  ent.Size(),
 781                                         IsDir: ent.IsDir(),
 782                                 })
 783                         }
 784                 }
 785                 return nil
 786         }
 787         if err := walk(""); err != nil {
 788                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 789                 return
 790         }
 791
 792         funcs := template.FuncMap{
 793                 "nbsp": func(s string) template.HTML {
 794                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 795                 },
 796         }
 797         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 798         if err != nil {
 799                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 800                 return
 801         }
 802         sort.Slice(files, func(i, j int) bool {
 803                 return files[i].Name < files[j].Name
 804         })
 805         w.WriteHeader(http.StatusOK)
 806         tmpl.Execute(w, map[string]interface{}{
 807                 "CollectionName": collectionName,
 808                 "Files":          files,
 809                 "Request":        r,
 810                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 811         })
 812 }
 813
 814 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 815         disposition := "inline"
 816         if isAttachment {
 817                 disposition = "attachment"
 818         }
 819         if strings.ContainsRune(r.RequestURI, '?') {
 820                 // Help the UA realize that the filename is just
 821                 // "filename.txt", not
 822                 // "filename.txt?disposition=attachment".
 823                 //
 824                 // TODO(TC): Follow advice at RFC 6266 appendix D
 825                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 826         }
 827         if disposition != "inline" {
 828                 w.Header().Set("Content-Disposition", disposition)
 829         }
 830 }
 831
 832 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 833         if formToken := r.FormValue("api_token"); formToken != "" {
 834                 if !credentialsOK {
 835                         // It is not safe to copy the provided token
 836                         // into a cookie unless the current vhost
 837                         // (origin) serves only a single collection or
 838                         // we are in TrustAllContent mode.
 839                         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)
 840                         return
 841                 }
 842
 843                 // The HttpOnly flag is necessary to prevent
 844                 // JavaScript code (included in, or loaded by, a page
 845                 // in the collection being served) from employing the
 846                 // user's token beyond reading other files in the same
 847                 // domain, i.e., same collection.
 848                 //
 849                 // The 303 redirect is necessary in the case of a GET
 850                 // request to avoid exposing the token in the Location
 851                 // bar, and in the case of a POST request to avoid
 852                 // raising warnings when the user refreshes the
 853                 // resulting page.
 854                 http.SetCookie(w, &http.Cookie{
 855                         Name:     "arvados_api_token",
 856                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 857                         Path:     "/",
 858                         HttpOnly: true,
 859                 })
 860         }
 861
 862         // Propagate query parameters (except api_token) from
 863         // the original request.
 864         redirQuery := r.URL.Query()
 865         redirQuery.Del("api_token")
 866
 867         u := r.URL
 868         if location != "" {
 869                 newu, err := u.Parse(location)
 870                 if err != nil {
 871                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 872                         return
 873                 }
 874                 u = newu
 875         }
 876         redir := (&url.URL{
 877                 Scheme:   r.URL.Scheme,
 878                 Host:     r.Host,
 879                 Path:     u.Path,
 880                 RawQuery: redirQuery.Encode(),
 881         }).String()
 882
 883         w.Header().Add("Location", redir)
 884         w.WriteHeader(http.StatusSeeOther)
 885         io.WriteString(w, `<A href="`)
 886         io.WriteString(w, html.EscapeString(redir))
 887         io.WriteString(w, `">Continue</A>`)
 888 }