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