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         "path/filepath"
  18         "sort"
  19         "strconv"
  20         "strings"
  21         "sync"
  22
  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         clientPool *arvadosclient.ClientPool
  37         setupOnce  sync.Once
  38         webdavLS   webdav.LockSystem
  39 }
  40
  41 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  42
  43 var notFoundMessage = "Not Found"
  44 var unauthorizedMessage = "401 Unauthorized\r\n\r\nA valid Arvados token must be provided to access this resource.\r\n"
  45
  46 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  47 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  48 // otherwise "".
  49 func parseCollectionIDFromURL(s string) string {
  50         if arvadosclient.UUIDMatch(s) {
  51                 return s
  52         }
  53         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  54                 return pdh
  55         }
  56         return ""
  57 }
  58
  59 func (h *handler) setup() {
  60         // Errors will be handled at the client pool.
  61         arv, _ := arvados.NewClientFromConfig(h.Cluster)
  62         h.clientPool = arvadosclient.MakeClientPoolWith(arv)
  63
  64         keepclient.DefaultBlockCache.MaxBlocks = h.Cluster.Collections.WebDAVCache.MaxBlockEntries
  65
  66         // Even though we don't accept LOCK requests, every webdav
  67         // handler must have a non-nil LockSystem.
  68         h.webdavLS = &noLockSystem{}
  69 }
  70
  71 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  72         json.NewEncoder(w).Encode(struct{ Version string }{version})
  73 }
  74
  75 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  76 // sends an HTTP header indicating success, updateOnSuccess first
  77 // calls the provided update func. If the update func fails, an error
  78 // response is sent (using the error's HTTP status or 500 if none),
  79 // and the status code and body sent by the handler are ignored (all
  80 // response writes return the update error).
  81 type updateOnSuccess struct {
  82         httpserver.ResponseWriter
  83         logger     logrus.FieldLogger
  84         update     func() error
  85         sentHeader bool
  86         err        error
  87 }
  88
  89 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  90         if !uos.sentHeader {
  91                 uos.WriteHeader(http.StatusOK)
  92         }
  93         if uos.err != nil {
  94                 return 0, uos.err
  95         }
  96         return uos.ResponseWriter.Write(p)
  97 }
  98
  99 func (uos *updateOnSuccess) WriteHeader(code int) {
 100         if !uos.sentHeader {
 101                 uos.sentHeader = true
 102                 if code >= 200 && code < 400 {
 103                         if uos.err = uos.update(); uos.err != nil {
 104                                 code := http.StatusInternalServerError
 105                                 var he interface{ HTTPStatus() int }
 106                                 if errors.As(uos.err, &he) {
 107                                         code = he.HTTPStatus()
 108                                 }
 109                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 110                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 111                                 return
 112                         }
 113                 }
 114         }
 115         uos.ResponseWriter.WriteHeader(code)
 116 }
 117
 118 var (
 119         corsAllowHeadersHeader = strings.Join([]string{
 120                 "Authorization", "Content-Type", "Range",
 121                 // WebDAV request headers:
 122                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout",
 123         }, ", ")
 124         writeMethod = map[string]bool{
 125                 "COPY":      true,
 126                 "DELETE":    true,
 127                 "LOCK":      true,
 128                 "MKCOL":     true,
 129                 "MOVE":      true,
 130                 "PROPPATCH": true,
 131                 "PUT":       true,
 132                 "RMCOL":     true,
 133                 "UNLOCK":    true,
 134         }
 135         webdavMethod = map[string]bool{
 136                 "COPY":      true,
 137                 "DELETE":    true,
 138                 "LOCK":      true,
 139                 "MKCOL":     true,
 140                 "MOVE":      true,
 141                 "OPTIONS":   true,
 142                 "PROPFIND":  true,
 143                 "PROPPATCH": true,
 144                 "PUT":       true,
 145                 "RMCOL":     true,
 146                 "UNLOCK":    true,
 147         }
 148         browserMethod = map[string]bool{
 149                 "GET":  true,
 150                 "HEAD": true,
 151                 "POST": true,
 152         }
 153         // top-level dirs to serve with siteFS
 154         siteFSDir = map[string]bool{
 155                 "":      true, // root directory
 156                 "by_id": true,
 157                 "users": true,
 158         }
 159 )
 160
 161 func stripDefaultPort(host string) string {
 162         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 163         u := &url.URL{Host: host}
 164         if p := u.Port(); p == "80" || p == "443" {
 165                 return strings.ToLower(u.Hostname())
 166         } else {
 167                 return strings.ToLower(host)
 168         }
 169 }
 170
 171 // CheckHealth implements service.Handler.
 172 func (h *handler) CheckHealth() error {
 173         return nil
 174 }
 175
 176 // Done implements service.Handler.
 177 func (h *handler) Done() <-chan struct{} {
 178         return nil
 179 }
 180
 181 // ServeHTTP implements http.Handler.
 182 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 183         h.setupOnce.Do(h.setup)
 184
 185         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 186                 r.URL.Scheme = xfp
 187         }
 188
 189         w := httpserver.WrapResponseWriter(wOrig)
 190
 191         if method := r.Header.Get("Access-Control-Request-Method"); method != "" && r.Method == "OPTIONS" {
 192                 if !browserMethod[method] && !webdavMethod[method] {
 193                         w.WriteHeader(http.StatusMethodNotAllowed)
 194                         return
 195                 }
 196                 w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
 197                 w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
 198                 w.Header().Set("Access-Control-Allow-Origin", "*")
 199                 w.Header().Set("Access-Control-Max-Age", "86400")
 200                 return
 201         }
 202
 203         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 204                 w.WriteHeader(http.StatusMethodNotAllowed)
 205                 return
 206         }
 207
 208         if r.Header.Get("Origin") != "" {
 209                 // Allow simple cross-origin requests without user
 210                 // credentials ("user credentials" as defined by CORS,
 211                 // i.e., cookies, HTTP authentication, and client-side
 212                 // SSL certificates. See
 213                 // http://www.w3.org/TR/cors/#user-credentials).
 214                 w.Header().Set("Access-Control-Allow-Origin", "*")
 215                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 216         }
 217
 218         if h.serveS3(w, r) {
 219                 return
 220         }
 221
 222         pathParts := strings.Split(r.URL.Path[1:], "/")
 223
 224         var stripParts int
 225         var collectionID string
 226         var tokens []string
 227         var reqTokens []string
 228         var pathToken bool
 229         var attachment bool
 230         var useSiteFS bool
 231         credentialsOK := h.Cluster.Collections.TrustAllContent
 232         reasonNotAcceptingCredentials := ""
 233
 234         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 235                 credentialsOK = true
 236                 attachment = true
 237         } else if r.FormValue("disposition") == "attachment" {
 238                 attachment = true
 239         }
 240
 241         if !credentialsOK {
 242                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 243                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 244         }
 245
 246         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 247                 // http://ID.collections.example/PATH...
 248                 credentialsOK = true
 249         } else if r.URL.Path == "/status.json" {
 250                 h.serveStatus(w, r)
 251                 return
 252         } else if siteFSDir[pathParts[0]] {
 253                 useSiteFS = true
 254         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 255                 // /c=ID[/PATH...]
 256                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 257                 stripParts = 1
 258         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 259                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 260                         // /collections/download/ID/TOKEN/PATH...
 261                         collectionID = parseCollectionIDFromURL(pathParts[2])
 262                         tokens = []string{pathParts[3]}
 263                         stripParts = 4
 264                         pathToken = true
 265                 } else {
 266                         // /collections/ID/PATH...
 267                         collectionID = parseCollectionIDFromURL(pathParts[1])
 268                         stripParts = 2
 269                         // This path is only meant to work for public
 270                         // data. Tokens provided with the request are
 271                         // ignored.
 272                         credentialsOK = false
 273                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 274                 }
 275         }
 276
 277         if collectionID == "" && !useSiteFS {
 278                 http.Error(w, notFoundMessage, http.StatusNotFound)
 279                 return
 280         }
 281
 282         forceReload := false
 283         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 284                 forceReload = true
 285         }
 286
 287         if credentialsOK {
 288                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 289         }
 290
 291         formToken := r.FormValue("api_token")
 292         origin := r.Header.Get("Origin")
 293         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 294         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 295         safeAttachment := attachment && r.URL.Query().Get("api_token") == ""
 296         if formToken == "" {
 297                 // No token to use or redact.
 298         } else if safeAjax || safeAttachment {
 299                 // If this is a cross-origin request, the URL won't
 300                 // appear in the browser's address bar, so
 301                 // substituting a clipboard-safe URL is pointless.
 302                 // Redirect-with-cookie wouldn't work anyway, because
 303                 // it's not safe to allow third-party use of our
 304                 // cookie.
 305                 //
 306                 // If we're supplying an attachment, we don't need to
 307                 // convert POST to GET to avoid the "really resubmit
 308                 // form?" problem, so provided the token isn't
 309                 // embedded in the URL, there's no reason to do
 310                 // redirect-with-cookie in this case either.
 311                 reqTokens = append(reqTokens, formToken)
 312         } else if browserMethod[r.Method] {
 313                 // If this is a page view, and the client provided a
 314                 // token via query string or POST body, we must put
 315                 // the token in an HttpOnly cookie, and redirect to an
 316                 // equivalent URL with the query param redacted and
 317                 // method = GET.
 318                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 319                 return
 320         }
 321
 322         if useSiteFS {
 323                 h.serveSiteFS(w, r, reqTokens, credentialsOK, attachment)
 324                 return
 325         }
 326
 327         targetPath := pathParts[stripParts:]
 328         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 329                 // http://ID.example/t=TOKEN/PATH...
 330                 // /c=ID/t=TOKEN/PATH...
 331                 //
 332                 // This form must only be used to pass scoped tokens
 333                 // that give permission for a single collection. See
 334                 // FormValue case above.
 335                 tokens = []string{targetPath[0][2:]}
 336                 pathToken = true
 337                 targetPath = targetPath[1:]
 338                 stripParts++
 339         }
 340
 341         if tokens == nil {
 342                 tokens = reqTokens
 343                 if h.Cluster.Users.AnonymousUserToken != "" {
 344                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 345                 }
 346         }
 347
 348         if tokens == nil {
 349                 if !credentialsOK {
 350                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 351                 } else {
 352                         http.Error(w, fmt.Sprintf("No authorization token in request, and no anonymous user token is configured."), http.StatusUnauthorized)
 353                 }
 354                 return
 355         }
 356
 357         if len(targetPath) > 0 && targetPath[0] == "_" {
 358                 // If a collection has a directory called "t=foo" or
 359                 // "_", it can be served at
 360                 // //collections.example/_/t=foo/ or
 361                 // //collections.example/_/_/ respectively:
 362                 // //collections.example/t=foo/ won't work because
 363                 // t=foo will be interpreted as a token "foo".
 364                 targetPath = targetPath[1:]
 365                 stripParts++
 366         }
 367
 368         arv := h.clientPool.Get()
 369         if arv == nil {
 370                 http.Error(w, "client pool error: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
 371                 return
 372         }
 373         defer h.clientPool.Put(arv)
 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 interface{ HTTPStatus() int }
 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("by_id/"+collectionID, 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 {
 417                 err := collectionDir.Sync()
 418                 if err != nil {
 419                         var statusErr interface{ HTTPStatus() int }
 420                         if errors.As(err, &statusErr) {
 421                                 http.Error(w, err.Error(), statusErr.HTTPStatus())
 422                         } else {
 423                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 424                         }
 425                         return
 426                 }
 427         }
 428         if session == nil {
 429                 if pathToken || !credentialsOK {
 430                         // Either the URL is a "secret sharing link"
 431                         // that didn't work out (and asking the client
 432                         // for additional credentials would just be
 433                         // confusing), or we don't even accept
 434                         // credentials at this path.
 435                         http.Error(w, notFoundMessage, http.StatusNotFound)
 436                         return
 437                 }
 438                 for _, t := range reqTokens {
 439                         if validToken[t] {
 440                                 // The client provided valid token(s),
 441                                 // but the collection was not found.
 442                                 http.Error(w, notFoundMessage, http.StatusNotFound)
 443                                 return
 444                         }
 445                 }
 446                 // The client's token was invalid (e.g., expired), or
 447                 // the client didn't even provide one.  Redirect to
 448                 // workbench2's login-and-redirect-to-download url if
 449                 // this is a browser navigation request. (The redirect
 450                 // flow can't preserve the original method if it's not
 451                 // GET, and doesn't make sense if the UA is a
 452                 // command-line tool, is trying to load an inline
 453                 // image, etc.; in these cases, there's nothing we can
 454                 // do, so return 401 unauthorized.)
 455                 //
 456                 // Note Sec-Fetch-Mode is sent by all non-EOL
 457                 // browsers, except Safari.
 458                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 459                 //
 460                 // TODO(TC): This response would be confusing to
 461                 // someone trying (anonymously) to download public
 462                 // data that has been deleted.  Allow a referrer to
 463                 // provide this context somehow?
 464                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 465                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 466                         redirkey := "redirectToPreview"
 467                         if attachment {
 468                                 redirkey = "redirectToDownload"
 469                         }
 470                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 471                         // target.RawQuery = url.Values{redirkey:
 472                         // {target}}.Encode() would be the obvious
 473                         // thing to do here, but wb2 doesn't decode
 474                         // this as a query param -- it takes
 475                         // everything after "${redirkey}=" as the
 476                         // target URL. If we encode "/" as "%2F" etc.,
 477                         // the redirect won't work.
 478                         target.RawQuery = redirkey + "=" + callback
 479                         w.Header().Add("Location", target.String())
 480                         w.WriteHeader(http.StatusSeeOther)
 481                 } else {
 482                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 483                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 484                 }
 485                 return
 486         }
 487
 488         var basename string
 489         if len(targetPath) > 0 {
 490                 basename = targetPath[len(targetPath)-1]
 491         }
 492         applyContentDispositionHdr(w, r, basename, attachment)
 493
 494         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 495                 http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 496                 return
 497         }
 498         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 499                 http.Error(w, "Not permitted", http.StatusForbidden)
 500                 return
 501         }
 502         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, "by_id/"+collectionID+"/"+strings.Join(targetPath, "/"), nil, tokenUser)
 503
 504         if writeMethod[r.Method] {
 505                 // Save the collection only if/when all
 506                 // webdav->filesystem operations succeed --
 507                 // and send a 500 error if the modified
 508                 // collection can't be saved.
 509                 //
 510                 // Perform the write in a separate sitefs, so
 511                 // concurrent read operations on the same
 512                 // collection see the previous saved
 513                 // state. After the write succeeds and the
 514                 // collection record is updated, we reset the
 515                 // session so the updates are visible in
 516                 // subsequent read requests.
 517                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 518                 sessionFS = client.SiteFileSystem(session.keepclient)
 519                 writingDir, err := sessionFS.OpenFile("by_id/"+collectionID, os.O_RDONLY, 0)
 520                 if err != nil {
 521                         http.Error(w, err.Error(), http.StatusInternalServerError)
 522                         return
 523                 }
 524                 defer writingDir.Close()
 525                 w = &updateOnSuccess{
 526                         ResponseWriter: w,
 527                         logger:         ctxlog.FromContext(r.Context()),
 528                         update: func() error {
 529                                 err := writingDir.Sync()
 530                                 var te arvados.TransactionError
 531                                 if errors.As(err, &te) {
 532                                         err = te
 533                                 }
 534                                 if err != nil {
 535                                         return err
 536                                 }
 537                                 // Sync the changes to the persistent
 538                                 // sessionfs for this token.
 539                                 snap, err := writingDir.Snapshot()
 540                                 if err != nil {
 541                                         return err
 542                                 }
 543                                 collectionDir.Splice(snap)
 544                                 return nil
 545                         }}
 546         }
 547         wh := webdav.Handler{
 548                 Prefix: "/" + strings.Join(pathParts[:stripParts], "/"),
 549                 FileSystem: &webdavFS{
 550                         collfs:        sessionFS,
 551                         prefix:        "by_id/" + collectionID + "/",
 552                         writing:       writeMethod[r.Method],
 553                         alwaysReadEOF: r.Method == "PROPFIND",
 554                 },
 555                 LockSystem: h.webdavLS,
 556                 Logger: func(r *http.Request, err error) {
 557                         if err != nil {
 558                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 559                         }
 560                 },
 561         }
 562         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 563                 targetfnm := "by_id/" + collectionID + "/" + strings.Join(pathParts[stripParts:], "/")
 564                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 565                         if !strings.HasSuffix(r.URL.Path, "/") {
 566                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 567                         } else {
 568                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, true)
 569                         }
 570                         return
 571                 }
 572         }
 573         wh.ServeHTTP(w, r)
 574         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 575                 wrote := int64(w.WroteBodyBytes())
 576                 fnm := strings.Join(pathParts[stripParts:], "/")
 577                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 578                 if err == nil && fi.Size() != wrote {
 579                         var n int
 580                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 581                         if err == nil {
 582                                 n, err = f.Read(make([]byte, 1024))
 583                                 f.Close()
 584                         }
 585                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 586                 }
 587         }
 588 }
 589
 590 func (h *handler) getClients(reqID, token string) (arv *arvadosclient.ArvadosClient, kc *keepclient.KeepClient, client *arvados.Client, release func(), err error) {
 591         arv = h.clientPool.Get()
 592         if arv == nil {
 593                 err = h.clientPool.Err()
 594                 return
 595         }
 596         release = func() { h.clientPool.Put(arv) }
 597         arv.ApiToken = token
 598         kc, err = keepclient.MakeKeepClient(arv)
 599         if err != nil {
 600                 release()
 601                 return
 602         }
 603         kc.RequestID = reqID
 604         client = (&arvados.Client{
 605                 APIHost:   arv.ApiServer,
 606                 AuthToken: arv.ApiToken,
 607                 Insecure:  arv.ApiInsecure,
 608         }).WithRequestID(reqID)
 609         return
 610 }
 611
 612 func (h *handler) serveSiteFS(w http.ResponseWriter, r *http.Request, tokens []string, credentialsOK, attachment bool) {
 613         if len(tokens) == 0 {
 614                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 615                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 616                 return
 617         }
 618         if writeMethod[r.Method] {
 619                 http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 620                 return
 621         }
 622
 623         fs, sess, user, err := h.Cache.GetSession(tokens[0])
 624         if err != nil {
 625                 http.Error(w, err.Error(), http.StatusInternalServerError)
 626                 return
 627         }
 628         f, err := fs.Open(r.URL.Path)
 629         if os.IsNotExist(err) {
 630                 http.Error(w, err.Error(), http.StatusNotFound)
 631                 return
 632         } else if err != nil {
 633                 http.Error(w, err.Error(), http.StatusInternalServerError)
 634                 return
 635         }
 636         defer f.Close()
 637         if fi, err := f.Stat(); err == nil && fi.IsDir() && r.Method == "GET" {
 638                 if !strings.HasSuffix(r.URL.Path, "/") {
 639                         h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 640                 } else {
 641                         h.serveDirectory(w, r, fi.Name(), fs, r.URL.Path, false)
 642                 }
 643                 return
 644         }
 645
 646         if !h.userPermittedToUploadOrDownload(r.Method, user) {
 647                 http.Error(w, "Not permitted", http.StatusForbidden)
 648                 return
 649         }
 650         h.logUploadOrDownload(r, sess.arvadosclient, fs, r.URL.Path, nil, user)
 651
 652         if r.Method == "GET" {
 653                 _, basename := filepath.Split(r.URL.Path)
 654                 applyContentDispositionHdr(w, r, basename, attachment)
 655         }
 656         wh := webdav.Handler{
 657                 FileSystem: &webdavFS{
 658                         collfs:        fs,
 659                         writing:       writeMethod[r.Method],
 660                         alwaysReadEOF: r.Method == "PROPFIND",
 661                 },
 662                 LockSystem: h.webdavLS,
 663                 Logger: func(_ *http.Request, err error) {
 664                         if err != nil {
 665                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 666                         }
 667                 },
 668         }
 669         wh.ServeHTTP(w, r)
 670 }
 671
 672 var dirListingTemplate = `<!DOCTYPE HTML>
 673 <HTML><HEAD>
 674   <META name="robots" content="NOINDEX">
 675   <TITLE>{{ .CollectionName }}</TITLE>
 676   <STYLE type="text/css">
 677     body {
 678       margin: 1.5em;
 679     }
 680     pre {
 681       background-color: #D9EDF7;
 682       border-radius: .25em;
 683       padding: .75em;
 684       overflow: auto;
 685     }
 686     .footer p {
 687       font-size: 82%;
 688     }
 689     ul {
 690       padding: 0;
 691     }
 692     ul li {
 693       font-family: monospace;
 694       list-style: none;
 695     }
 696   </STYLE>
 697 </HEAD>
 698 <BODY>
 699
 700 <H1>{{ .CollectionName }}</H1>
 701
 702 <P>This collection of data files is being shared with you through
 703 Arvados.  You can download individual files listed below.  To download
 704 the entire directory tree with wget, try:</P>
 705
 706 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 707
 708 <H2>File Listing</H2>
 709
 710 {{if .Files}}
 711 <UL>
 712 {{range .Files}}
 713 {{if .IsDir }}
 714   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 715 {{else}}
 716   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 717 {{end}}
 718 {{end}}
 719 </UL>
 720 {{else}}
 721 <P>(No files; this collection is empty.)</P>
 722 {{end}}
 723
 724 <HR noshade>
 725 <DIV class="footer">
 726   <P>
 727     About Arvados:
 728     Arvados is a free and open source software bioinformatics platform.
 729     To learn more, visit arvados.org.
 730     Arvados is not responsible for the files listed on this page.
 731   </P>
 732 </DIV>
 733
 734 </BODY>
 735 `
 736
 737 type fileListEnt struct {
 738         Name  string
 739         Size  int64
 740         IsDir bool
 741 }
 742
 743 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 744         var files []fileListEnt
 745         var walk func(string) error
 746         if !strings.HasSuffix(base, "/") {
 747                 base = base + "/"
 748         }
 749         walk = func(path string) error {
 750                 dirname := base + path
 751                 if dirname != "/" {
 752                         dirname = strings.TrimSuffix(dirname, "/")
 753                 }
 754                 d, err := fs.Open(dirname)
 755                 if err != nil {
 756                         return err
 757                 }
 758                 ents, err := d.Readdir(-1)
 759                 if err != nil {
 760                         return err
 761                 }
 762                 for _, ent := range ents {
 763                         if recurse && ent.IsDir() {
 764                                 err = walk(path + ent.Name() + "/")
 765                                 if err != nil {
 766                                         return err
 767                                 }
 768                         } else {
 769                                 files = append(files, fileListEnt{
 770                                         Name:  path + ent.Name(),
 771                                         Size:  ent.Size(),
 772                                         IsDir: ent.IsDir(),
 773                                 })
 774                         }
 775                 }
 776                 return nil
 777         }
 778         if err := walk(""); err != nil {
 779                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 780                 return
 781         }
 782
 783         funcs := template.FuncMap{
 784                 "nbsp": func(s string) template.HTML {
 785                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 786                 },
 787         }
 788         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 789         if err != nil {
 790                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 791                 return
 792         }
 793         sort.Slice(files, func(i, j int) bool {
 794                 return files[i].Name < files[j].Name
 795         })
 796         w.WriteHeader(http.StatusOK)
 797         tmpl.Execute(w, map[string]interface{}{
 798                 "CollectionName": collectionName,
 799                 "Files":          files,
 800                 "Request":        r,
 801                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 802         })
 803 }
 804
 805 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 806         disposition := "inline"
 807         if isAttachment {
 808                 disposition = "attachment"
 809         }
 810         if strings.ContainsRune(r.RequestURI, '?') {
 811                 // Help the UA realize that the filename is just
 812                 // "filename.txt", not
 813                 // "filename.txt?disposition=attachment".
 814                 //
 815                 // TODO(TC): Follow advice at RFC 6266 appendix D
 816                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 817         }
 818         if disposition != "inline" {
 819                 w.Header().Set("Content-Disposition", disposition)
 820         }
 821 }
 822
 823 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 824         if formToken := r.FormValue("api_token"); formToken != "" {
 825                 if !credentialsOK {
 826                         // It is not safe to copy the provided token
 827                         // into a cookie unless the current vhost
 828                         // (origin) serves only a single collection or
 829                         // we are in TrustAllContent mode.
 830                         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)
 831                         return
 832                 }
 833
 834                 // The HttpOnly flag is necessary to prevent
 835                 // JavaScript code (included in, or loaded by, a page
 836                 // in the collection being served) from employing the
 837                 // user's token beyond reading other files in the same
 838                 // domain, i.e., same collection.
 839                 //
 840                 // The 303 redirect is necessary in the case of a GET
 841                 // request to avoid exposing the token in the Location
 842                 // bar, and in the case of a POST request to avoid
 843                 // raising warnings when the user refreshes the
 844                 // resulting page.
 845                 http.SetCookie(w, &http.Cookie{
 846                         Name:     "arvados_api_token",
 847                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 848                         Path:     "/",
 849                         HttpOnly: true,
 850                         SameSite: http.SameSiteLaxMode,
 851                 })
 852         }
 853
 854         // Propagate query parameters (except api_token) from
 855         // the original request.
 856         redirQuery := r.URL.Query()
 857         redirQuery.Del("api_token")
 858
 859         u := r.URL
 860         if location != "" {
 861                 newu, err := u.Parse(location)
 862                 if err != nil {
 863                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 864                         return
 865                 }
 866                 u = newu
 867         }
 868         redir := (&url.URL{
 869                 Scheme:   r.URL.Scheme,
 870                 Host:     r.Host,
 871                 Path:     u.Path,
 872                 RawQuery: redirQuery.Encode(),
 873         }).String()
 874
 875         w.Header().Add("Location", redir)
 876         w.WriteHeader(http.StatusSeeOther)
 877         io.WriteString(w, `<A href="`)
 878         io.WriteString(w, html.EscapeString(redir))
 879         io.WriteString(w, `">Continue</A>`)
 880 }
 881
 882 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 883         var permitDownload bool
 884         var permitUpload bool
 885         if tokenUser != nil && tokenUser.IsAdmin {
 886                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 887                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 888         } else {
 889                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 890                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 891         }
 892         if (method == "PUT" || method == "POST") && !permitUpload {
 893                 // Disallow operations that upload new files.
 894                 // Permit webdav operations that move existing files around.
 895                 return false
 896         } else if method == "GET" && !permitDownload {
 897                 // Disallow downloading file contents.
 898                 // Permit webdav operations like PROPFIND that retrieve metadata
 899                 // but not file contents.
 900                 return false
 901         }
 902         return true
 903 }
 904
 905 func (h *handler) logUploadOrDownload(
 906         r *http.Request,
 907         client *arvadosclient.ArvadosClient,
 908         fs arvados.CustomFileSystem,
 909         filepath string,
 910         collection *arvados.Collection,
 911         user *arvados.User) {
 912
 913         log := ctxlog.FromContext(r.Context())
 914         props := make(map[string]string)
 915         props["reqPath"] = r.URL.Path
 916         var useruuid string
 917         if user != nil {
 918                 log = log.WithField("user_uuid", user.UUID).
 919                         WithField("user_full_name", user.FullName)
 920                 useruuid = user.UUID
 921         } else {
 922                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 923         }
 924         if collection == nil && fs != nil {
 925                 collection, filepath = h.determineCollection(fs, filepath)
 926         }
 927         if collection != nil {
 928                 log = log.WithField("collection_file_path", filepath)
 929                 props["collection_file_path"] = filepath
 930                 // h.determineCollection populates the collection_uuid
 931                 // prop with the PDH, if this collection is being
 932                 // accessed via PDH. For logging, we use a different
 933                 // field depending on whether it's a UUID or PDH.
 934                 if len(collection.UUID) > 32 {
 935                         log = log.WithField("portable_data_hash", collection.UUID)
 936                         props["portable_data_hash"] = collection.UUID
 937                 } else {
 938                         log = log.WithField("collection_uuid", collection.UUID)
 939                         props["collection_uuid"] = collection.UUID
 940                 }
 941         }
 942         if r.Method == "PUT" || r.Method == "POST" {
 943                 log.Info("File upload")
 944                 if h.Cluster.Collections.WebDAVLogEvents {
 945                         go func() {
 946                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 947                                         "object_uuid": useruuid,
 948                                         "event_type":  "file_upload",
 949                                         "properties":  props}}
 950                                 err := client.Create("logs", lr, nil)
 951                                 if err != nil {
 952                                         log.WithError(err).Error("Failed to create upload log event on API server")
 953                                 }
 954                         }()
 955                 }
 956         } else if r.Method == "GET" {
 957                 if collection != nil && collection.PortableDataHash != "" {
 958                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 959                         props["portable_data_hash"] = collection.PortableDataHash
 960                 }
 961                 log.Info("File download")
 962                 if h.Cluster.Collections.WebDAVLogEvents {
 963                         go func() {
 964                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 965                                         "object_uuid": useruuid,
 966                                         "event_type":  "file_download",
 967                                         "properties":  props}}
 968                                 err := client.Create("logs", lr, nil)
 969                                 if err != nil {
 970                                         log.WithError(err).Error("Failed to create download log event on API server")
 971                                 }
 972                         }()
 973                 }
 974         }
 975 }
 976
 977 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 978         target := strings.TrimSuffix(path, "/")
 979         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 980                 target = target[:cut]
 981                 fi, err := fs.Stat(target)
 982                 if os.IsNotExist(err) {
 983                         // creating a new file/dir, or download
 984                         // destined to fail
 985                         continue
 986                 } else if err != nil {
 987                         return nil, ""
 988                 }
 989                 switch src := fi.Sys().(type) {
 990                 case *arvados.Collection:
 991                         return src, strings.TrimPrefix(path[len(target):], "/")
 992                 case *arvados.Group:
 993                         return nil, ""
 994                 default:
 995                         if _, ok := src.(error); ok {
 996                                 return nil, ""
 997                         }
 998                 }
 999         }
1000         return nil, ""
1001 }