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