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