services/keep-web/handler.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package keepweb
   6
   7 import (
   8         "encoding/json"
   9         "errors"
  10         "fmt"
  11         "html"
  12         "html/template"
  13         "io"
  14         "net/http"
  15         "net/url"
  16         "os"
  17         "sort"
  18         "strconv"
  19         "strings"
  20         "sync"
  21
  22         "git.arvados.org/arvados.git/lib/cmd"
  23         "git.arvados.org/arvados.git/sdk/go/arvados"
  24         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  25         "git.arvados.org/arvados.git/sdk/go/auth"
  26         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  27         "git.arvados.org/arvados.git/sdk/go/httpserver"
  28         "git.arvados.org/arvados.git/sdk/go/keepclient"
  29         "github.com/sirupsen/logrus"
  30         "golang.org/x/net/webdav"
  31 )
  32
  33 type handler struct {
  34         Cache     cache
  35         Cluster   *arvados.Cluster
  36         setupOnce sync.Once
  37         webdavLS  webdav.LockSystem
  38 }
  39
  40 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  41
  42 var notFoundMessage = "Not Found"
  43 var unauthorizedMessage = "401 Unauthorized\r\n\r\nA valid Arvados token must be provided to access this resource.\r\n"
  44
  45 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  46 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  47 // otherwise "".
  48 func parseCollectionIDFromURL(s string) string {
  49         if arvadosclient.UUIDMatch(s) {
  50                 return s
  51         }
  52         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  53                 return pdh
  54         }
  55         return ""
  56 }
  57
  58 func (h *handler) setup() {
  59         keepclient.DefaultBlockCache.MaxBlocks = h.Cluster.Collections.WebDAVCache.MaxBlockEntries
  60
  61         // Even though we don't accept LOCK requests, every webdav
  62         // handler must have a non-nil LockSystem.
  63         h.webdavLS = &noLockSystem{}
  64 }
  65
  66 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  67         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  68 }
  69
  70 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  71 // sends an HTTP header indicating success, updateOnSuccess first
  72 // calls the provided update func. If the update func fails, an error
  73 // response is sent (using the error's HTTP status or 500 if none),
  74 // and the status code and body sent by the handler are ignored (all
  75 // response writes return the update error).
  76 type updateOnSuccess struct {
  77         httpserver.ResponseWriter
  78         logger     logrus.FieldLogger
  79         update     func() error
  80         sentHeader bool
  81         err        error
  82 }
  83
  84 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  85         if !uos.sentHeader {
  86                 uos.WriteHeader(http.StatusOK)
  87         }
  88         if uos.err != nil {
  89                 return 0, uos.err
  90         }
  91         return uos.ResponseWriter.Write(p)
  92 }
  93
  94 func (uos *updateOnSuccess) WriteHeader(code int) {
  95         if !uos.sentHeader {
  96                 uos.sentHeader = true
  97                 if code >= 200 && code < 400 {
  98                         if uos.err = uos.update(); uos.err != nil {
  99                                 code := http.StatusInternalServerError
 100                                 var he interface{ HTTPStatus() int }
 101                                 if errors.As(uos.err, &he) {
 102                                         code = he.HTTPStatus()
 103                                 }
 104                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 105                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 106                                 return
 107                         }
 108                 }
 109         }
 110         uos.ResponseWriter.WriteHeader(code)
 111 }
 112
 113 var (
 114         corsAllowHeadersHeader = strings.Join([]string{
 115                 "Authorization", "Content-Type", "Range",
 116                 // WebDAV request headers:
 117                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout",
 118         }, ", ")
 119         writeMethod = map[string]bool{
 120                 "COPY":      true,
 121                 "DELETE":    true,
 122                 "LOCK":      true,
 123                 "MKCOL":     true,
 124                 "MOVE":      true,
 125                 "PROPPATCH": true,
 126                 "PUT":       true,
 127                 "RMCOL":     true,
 128                 "UNLOCK":    true,
 129         }
 130         webdavMethod = map[string]bool{
 131                 "COPY":      true,
 132                 "DELETE":    true,
 133                 "LOCK":      true,
 134                 "MKCOL":     true,
 135                 "MOVE":      true,
 136                 "OPTIONS":   true,
 137                 "PROPFIND":  true,
 138                 "PROPPATCH": true,
 139                 "PUT":       true,
 140                 "RMCOL":     true,
 141                 "UNLOCK":    true,
 142         }
 143         browserMethod = map[string]bool{
 144                 "GET":  true,
 145                 "HEAD": true,
 146                 "POST": true,
 147         }
 148         // top-level dirs to serve with siteFS
 149         siteFSDir = map[string]bool{
 150                 "":      true, // root directory
 151                 "by_id": true,
 152                 "users": true,
 153         }
 154 )
 155
 156 func stripDefaultPort(host string) string {
 157         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 158         u := &url.URL{Host: host}
 159         if p := u.Port(); p == "80" || p == "443" {
 160                 return strings.ToLower(u.Hostname())
 161         } else {
 162                 return strings.ToLower(host)
 163         }
 164 }
 165
 166 // CheckHealth implements service.Handler.
 167 func (h *handler) CheckHealth() error {
 168         return nil
 169 }
 170
 171 // Done implements service.Handler.
 172 func (h *handler) Done() <-chan struct{} {
 173         return nil
 174 }
 175
 176 // ServeHTTP implements http.Handler.
 177 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 178         h.setupOnce.Do(h.setup)
 179
 180         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 181                 r.URL.Scheme = xfp
 182         }
 183
 184         w := httpserver.WrapResponseWriter(wOrig)
 185
 186         if method := r.Header.Get("Access-Control-Request-Method"); method != "" && r.Method == "OPTIONS" {
 187                 if !browserMethod[method] && !webdavMethod[method] {
 188                         w.WriteHeader(http.StatusMethodNotAllowed)
 189                         return
 190                 }
 191                 w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
 192                 w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
 193                 w.Header().Set("Access-Control-Allow-Origin", "*")
 194                 w.Header().Set("Access-Control-Max-Age", "86400")
 195                 return
 196         }
 197
 198         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 199                 w.WriteHeader(http.StatusMethodNotAllowed)
 200                 return
 201         }
 202
 203         if r.Header.Get("Origin") != "" {
 204                 // Allow simple cross-origin requests without user
 205                 // credentials ("user credentials" as defined by CORS,
 206                 // i.e., cookies, HTTP authentication, and client-side
 207                 // SSL certificates. See
 208                 // http://www.w3.org/TR/cors/#user-credentials).
 209                 w.Header().Set("Access-Control-Allow-Origin", "*")
 210                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 211         }
 212
 213         if h.serveS3(w, r) {
 214                 return
 215         }
 216
 217         pathParts := strings.Split(r.URL.Path[1:], "/")
 218
 219         var stripParts int
 220         var collectionID string
 221         var tokens []string
 222         var reqTokens []string
 223         var pathToken bool
 224         var attachment bool
 225         var useSiteFS bool
 226         credentialsOK := h.Cluster.Collections.TrustAllContent
 227         reasonNotAcceptingCredentials := ""
 228
 229         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 230                 credentialsOK = true
 231                 attachment = true
 232         } else if r.FormValue("disposition") == "attachment" {
 233                 attachment = true
 234         }
 235
 236         if !credentialsOK {
 237                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 238                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 239         }
 240
 241         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 242                 // http://ID.collections.example/PATH...
 243                 credentialsOK = true
 244         } else if r.URL.Path == "/status.json" {
 245                 h.serveStatus(w, r)
 246                 return
 247         } else if siteFSDir[pathParts[0]] {
 248                 useSiteFS = true
 249         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 250                 // /c=ID[/PATH...]
 251                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 252                 stripParts = 1
 253         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 254                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 255                         // /collections/download/ID/TOKEN/PATH...
 256                         collectionID = parseCollectionIDFromURL(pathParts[2])
 257                         tokens = []string{pathParts[3]}
 258                         stripParts = 4
 259                         pathToken = true
 260                 } else {
 261                         // /collections/ID/PATH...
 262                         collectionID = parseCollectionIDFromURL(pathParts[1])
 263                         stripParts = 2
 264                         // This path is only meant to work for public
 265                         // data. Tokens provided with the request are
 266                         // ignored.
 267                         credentialsOK = false
 268                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 269                 }
 270         }
 271
 272         forceReload := false
 273         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 274                 forceReload = true
 275         }
 276
 277         if credentialsOK {
 278                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 279         }
 280
 281         formToken := r.FormValue("api_token")
 282         origin := r.Header.Get("Origin")
 283         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 284         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 285         safeAttachment := attachment && r.URL.Query().Get("api_token") == ""
 286         if formToken == "" {
 287                 // No token to use or redact.
 288         } else if safeAjax || safeAttachment {
 289                 // If this is a cross-origin request, the URL won't
 290                 // appear in the browser's address bar, so
 291                 // substituting a clipboard-safe URL is pointless.
 292                 // Redirect-with-cookie wouldn't work anyway, because
 293                 // it's not safe to allow third-party use of our
 294                 // cookie.
 295                 //
 296                 // If we're supplying an attachment, we don't need to
 297                 // convert POST to GET to avoid the "really resubmit
 298                 // form?" problem, so provided the token isn't
 299                 // embedded in the URL, there's no reason to do
 300                 // redirect-with-cookie in this case either.
 301                 reqTokens = append(reqTokens, formToken)
 302         } else if browserMethod[r.Method] {
 303                 // If this is a page view, and the client provided a
 304                 // token via query string or POST body, we must put
 305                 // the token in an HttpOnly cookie, and redirect to an
 306                 // equivalent URL with the query param redacted and
 307                 // method = GET.
 308                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 309                 return
 310         }
 311
 312         targetPath := pathParts[stripParts:]
 313         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 314                 // http://ID.example/t=TOKEN/PATH...
 315                 // /c=ID/t=TOKEN/PATH...
 316                 //
 317                 // This form must only be used to pass scoped tokens
 318                 // that give permission for a single collection. See
 319                 // FormValue case above.
 320                 tokens = []string{targetPath[0][2:]}
 321                 pathToken = true
 322                 targetPath = targetPath[1:]
 323                 stripParts++
 324         }
 325
 326         fsprefix := ""
 327         if useSiteFS {
 328                 if writeMethod[r.Method] {
 329                         http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 330                         return
 331                 }
 332                 if len(reqTokens) == 0 {
 333                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 334                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 335                         return
 336                 }
 337                 tokens = reqTokens
 338         } else if collectionID == "" {
 339                 http.Error(w, notFoundMessage, http.StatusNotFound)
 340                 return
 341         } else {
 342                 fsprefix = "by_id/" + collectionID + "/"
 343         }
 344
 345         if tokens == nil {
 346                 tokens = reqTokens
 347                 if h.Cluster.Users.AnonymousUserToken != "" {
 348                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 349                 }
 350         }
 351
 352         if tokens == nil {
 353                 if !credentialsOK {
 354                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 355                 } else {
 356                         http.Error(w, fmt.Sprintf("No authorization token in request, and no anonymous user token is configured."), http.StatusUnauthorized)
 357                 }
 358                 return
 359         }
 360
 361         if len(targetPath) > 0 && targetPath[0] == "_" {
 362                 // If a collection has a directory called "t=foo" or
 363                 // "_", it can be served at
 364                 // //collections.example/_/t=foo/ or
 365                 // //collections.example/_/_/ respectively:
 366                 // //collections.example/t=foo/ won't work because
 367                 // t=foo will be interpreted as a token "foo".
 368                 targetPath = targetPath[1:]
 369                 stripParts++
 370         }
 371
 372         dirOpenMode := os.O_RDONLY
 373         if writeMethod[r.Method] {
 374                 dirOpenMode = os.O_RDWR
 375         }
 376
 377         validToken := make(map[string]bool)
 378         var token string
 379         var tokenUser *arvados.User
 380         var sessionFS arvados.CustomFileSystem
 381         var session *cachedSession
 382         var collectionDir arvados.File
 383         for _, token = range tokens {
 384                 var statusErr interface{ HTTPStatus() int }
 385                 fs, sess, user, err := h.Cache.GetSession(token)
 386                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 387                         // bad token
 388                         continue
 389                 } else if err != nil {
 390                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 391                         return
 392                 }
 393                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 394                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusForbidden {
 395                         // collection id is outside token scope
 396                         validToken[token] = true
 397                         continue
 398                 }
 399                 validToken[token] = true
 400                 if os.IsNotExist(err) {
 401                         // collection does not exist or is not
 402                         // readable using this token
 403                         continue
 404                 } else if err != nil {
 405                         http.Error(w, err.Error(), http.StatusInternalServerError)
 406                         return
 407                 }
 408                 defer f.Close()
 409
 410                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 411                 break
 412         }
 413         if forceReload {
 414                 err := collectionDir.Sync()
 415                 if err != nil {
 416                         var statusErr interface{ HTTPStatus() int }
 417                         if errors.As(err, &statusErr) {
 418                                 http.Error(w, err.Error(), statusErr.HTTPStatus())
 419                         } else {
 420                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 421                         }
 422                         return
 423                 }
 424         }
 425         if session == nil {
 426                 if pathToken || !credentialsOK {
 427                         // Either the URL is a "secret sharing link"
 428                         // that didn't work out (and asking the client
 429                         // for additional credentials would just be
 430                         // confusing), or we don't even accept
 431                         // credentials at this path.
 432                         http.Error(w, notFoundMessage, http.StatusNotFound)
 433                         return
 434                 }
 435                 for _, t := range reqTokens {
 436                         if validToken[t] {
 437                                 // The client provided valid token(s),
 438                                 // but the collection was not found.
 439                                 http.Error(w, notFoundMessage, http.StatusNotFound)
 440                                 return
 441                         }
 442                 }
 443                 // The client's token was invalid (e.g., expired), or
 444                 // the client didn't even provide one.  Redirect to
 445                 // workbench2's login-and-redirect-to-download url if
 446                 // this is a browser navigation request. (The redirect
 447                 // flow can't preserve the original method if it's not
 448                 // GET, and doesn't make sense if the UA is a
 449                 // command-line tool, is trying to load an inline
 450                 // image, etc.; in these cases, there's nothing we can
 451                 // do, so return 401 unauthorized.)
 452                 //
 453                 // Note Sec-Fetch-Mode is sent by all non-EOL
 454                 // browsers, except Safari.
 455                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 456                 //
 457                 // TODO(TC): This response would be confusing to
 458                 // someone trying (anonymously) to download public
 459                 // data that has been deleted.  Allow a referrer to
 460                 // provide this context somehow?
 461                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 462                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 463                         redirkey := "redirectToPreview"
 464                         if attachment {
 465                                 redirkey = "redirectToDownload"
 466                         }
 467                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 468                         // target.RawQuery = url.Values{redirkey:
 469                         // {target}}.Encode() would be the obvious
 470                         // thing to do here, but wb2 doesn't decode
 471                         // this as a query param -- it takes
 472                         // everything after "${redirkey}=" as the
 473                         // target URL. If we encode "/" as "%2F" etc.,
 474                         // the redirect won't work.
 475                         target.RawQuery = redirkey + "=" + callback
 476                         w.Header().Add("Location", target.String())
 477                         w.WriteHeader(http.StatusSeeOther)
 478                 } else {
 479                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 480                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 481                 }
 482                 return
 483         }
 484
 485         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 486                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 487                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 488                         if !strings.HasSuffix(r.URL.Path, "/") {
 489                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 490                         } else {
 491                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 492                         }
 493                         return
 494                 }
 495         }
 496
 497         var basename string
 498         if len(targetPath) > 0 {
 499                 basename = targetPath[len(targetPath)-1]
 500         }
 501         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 502                 http.Error(w, errReadOnly.Error(), http.StatusMethodNotAllowed)
 503                 return
 504         }
 505         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 506                 http.Error(w, "Not permitted", http.StatusForbidden)
 507                 return
 508         }
 509         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 510
 511         if writeMethod[r.Method] {
 512                 // Save the collection only if/when all
 513                 // webdav->filesystem operations succeed --
 514                 // and send a 500 error if the modified
 515                 // collection can't be saved.
 516                 //
 517                 // Perform the write in a separate sitefs, so
 518                 // concurrent read operations on the same
 519                 // collection see the previous saved
 520                 // state. After the write succeeds and the
 521                 // collection record is updated, we reset the
 522                 // session so the updates are visible in
 523                 // subsequent read requests.
 524                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 525                 sessionFS = client.SiteFileSystem(session.keepclient)
 526                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 527                 if err != nil {
 528                         http.Error(w, err.Error(), http.StatusInternalServerError)
 529                         return
 530                 }
 531                 defer writingDir.Close()
 532                 w = &updateOnSuccess{
 533                         ResponseWriter: w,
 534                         logger:         ctxlog.FromContext(r.Context()),
 535                         update: func() error {
 536                                 err := writingDir.Sync()
 537                                 var te arvados.TransactionError
 538                                 if errors.As(err, &te) {
 539                                         err = te
 540                                 }
 541                                 if err != nil {
 542                                         return err
 543                                 }
 544                                 // Sync the changes to the persistent
 545                                 // sessionfs for this token.
 546                                 snap, err := writingDir.Snapshot()
 547                                 if err != nil {
 548                                         return err
 549                                 }
 550                                 collectionDir.Splice(snap)
 551                                 return nil
 552                         }}
 553         }
 554         if r.Method == http.MethodGet {
 555                 applyContentDispositionHdr(w, r, basename, attachment)
 556         }
 557         wh := webdav.Handler{
 558                 Prefix: "/" + strings.Join(pathParts[:stripParts], "/"),
 559                 FileSystem: &webdavFS{
 560                         collfs:        sessionFS,
 561                         prefix:        fsprefix,
 562                         writing:       writeMethod[r.Method],
 563                         alwaysReadEOF: r.Method == "PROPFIND",
 564                 },
 565                 LockSystem: h.webdavLS,
 566                 Logger: func(r *http.Request, err error) {
 567                         if err != nil {
 568                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 569                         }
 570                 },
 571         }
 572         wh.ServeHTTP(w, r)
 573         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 574                 wrote := int64(w.WroteBodyBytes())
 575                 fnm := strings.Join(pathParts[stripParts:], "/")
 576                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 577                 if err == nil && fi.Size() != wrote {
 578                         var n int
 579                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 580                         if err == nil {
 581                                 n, err = f.Read(make([]byte, 1024))
 582                                 f.Close()
 583                         }
 584                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 585                 }
 586         }
 587 }
 588
 589 var dirListingTemplate = `<!DOCTYPE HTML>
 590 <HTML><HEAD>
 591   <META name="robots" content="NOINDEX">
 592   <TITLE>{{ .CollectionName }}</TITLE>
 593   <STYLE type="text/css">
 594     body {
 595       margin: 1.5em;
 596     }
 597     pre {
 598       background-color: #D9EDF7;
 599       border-radius: .25em;
 600       padding: .75em;
 601       overflow: auto;
 602     }
 603     .footer p {
 604       font-size: 82%;
 605     }
 606     ul {
 607       padding: 0;
 608     }
 609     ul li {
 610       font-family: monospace;
 611       list-style: none;
 612     }
 613   </STYLE>
 614 </HEAD>
 615 <BODY>
 616
 617 <H1>{{ .CollectionName }}</H1>
 618
 619 <P>This collection of data files is being shared with you through
 620 Arvados.  You can download individual files listed below.  To download
 621 the entire directory tree with wget, try:</P>
 622
 623 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 624
 625 <H2>File Listing</H2>
 626
 627 {{if .Files}}
 628 <UL>
 629 {{range .Files}}
 630 {{if .IsDir }}
 631   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 632 {{else}}
 633   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 634 {{end}}
 635 {{end}}
 636 </UL>
 637 {{else}}
 638 <P>(No files; this collection is empty.)</P>
 639 {{end}}
 640
 641 <HR noshade>
 642 <DIV class="footer">
 643   <P>
 644     About Arvados:
 645     Arvados is a free and open source software bioinformatics platform.
 646     To learn more, visit arvados.org.
 647     Arvados is not responsible for the files listed on this page.
 648   </P>
 649 </DIV>
 650
 651 </BODY>
 652 `
 653
 654 type fileListEnt struct {
 655         Name  string
 656         Size  int64
 657         IsDir bool
 658 }
 659
 660 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 661         var files []fileListEnt
 662         var walk func(string) error
 663         if !strings.HasSuffix(base, "/") {
 664                 base = base + "/"
 665         }
 666         walk = func(path string) error {
 667                 dirname := base + path
 668                 if dirname != "/" {
 669                         dirname = strings.TrimSuffix(dirname, "/")
 670                 }
 671                 d, err := fs.Open(dirname)
 672                 if err != nil {
 673                         return err
 674                 }
 675                 ents, err := d.Readdir(-1)
 676                 if err != nil {
 677                         return err
 678                 }
 679                 for _, ent := range ents {
 680                         if recurse && ent.IsDir() {
 681                                 err = walk(path + ent.Name() + "/")
 682                                 if err != nil {
 683                                         return err
 684                                 }
 685                         } else {
 686                                 files = append(files, fileListEnt{
 687                                         Name:  path + ent.Name(),
 688                                         Size:  ent.Size(),
 689                                         IsDir: ent.IsDir(),
 690                                 })
 691                         }
 692                 }
 693                 return nil
 694         }
 695         if err := walk(""); err != nil {
 696                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 697                 return
 698         }
 699
 700         funcs := template.FuncMap{
 701                 "nbsp": func(s string) template.HTML {
 702                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 703                 },
 704         }
 705         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 706         if err != nil {
 707                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 708                 return
 709         }
 710         sort.Slice(files, func(i, j int) bool {
 711                 return files[i].Name < files[j].Name
 712         })
 713         w.WriteHeader(http.StatusOK)
 714         tmpl.Execute(w, map[string]interface{}{
 715                 "CollectionName": collectionName,
 716                 "Files":          files,
 717                 "Request":        r,
 718                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 719         })
 720 }
 721
 722 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 723         disposition := "inline"
 724         if isAttachment {
 725                 disposition = "attachment"
 726         }
 727         if strings.ContainsRune(r.RequestURI, '?') {
 728                 // Help the UA realize that the filename is just
 729                 // "filename.txt", not
 730                 // "filename.txt?disposition=attachment".
 731                 //
 732                 // TODO(TC): Follow advice at RFC 6266 appendix D
 733                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 734         }
 735         if disposition != "inline" {
 736                 w.Header().Set("Content-Disposition", disposition)
 737         }
 738 }
 739
 740 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 741         if formToken := r.FormValue("api_token"); formToken != "" {
 742                 if !credentialsOK {
 743                         // It is not safe to copy the provided token
 744                         // into a cookie unless the current vhost
 745                         // (origin) serves only a single collection or
 746                         // we are in TrustAllContent mode.
 747                         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)
 748                         return
 749                 }
 750
 751                 // The HttpOnly flag is necessary to prevent
 752                 // JavaScript code (included in, or loaded by, a page
 753                 // in the collection being served) from employing the
 754                 // user's token beyond reading other files in the same
 755                 // domain, i.e., same collection.
 756                 //
 757                 // The 303 redirect is necessary in the case of a GET
 758                 // request to avoid exposing the token in the Location
 759                 // bar, and in the case of a POST request to avoid
 760                 // raising warnings when the user refreshes the
 761                 // resulting page.
 762                 http.SetCookie(w, &http.Cookie{
 763                         Name:     "arvados_api_token",
 764                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 765                         Path:     "/",
 766                         HttpOnly: true,
 767                         SameSite: http.SameSiteLaxMode,
 768                 })
 769         }
 770
 771         // Propagate query parameters (except api_token) from
 772         // the original request.
 773         redirQuery := r.URL.Query()
 774         redirQuery.Del("api_token")
 775
 776         u := r.URL
 777         if location != "" {
 778                 newu, err := u.Parse(location)
 779                 if err != nil {
 780                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 781                         return
 782                 }
 783                 u = newu
 784         }
 785         redir := (&url.URL{
 786                 Scheme:   r.URL.Scheme,
 787                 Host:     r.Host,
 788                 Path:     u.Path,
 789                 RawQuery: redirQuery.Encode(),
 790         }).String()
 791
 792         w.Header().Add("Location", redir)
 793         w.WriteHeader(http.StatusSeeOther)
 794         io.WriteString(w, `<A href="`)
 795         io.WriteString(w, html.EscapeString(redir))
 796         io.WriteString(w, `">Continue</A>`)
 797 }
 798
 799 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 800         var permitDownload bool
 801         var permitUpload bool
 802         if tokenUser != nil && tokenUser.IsAdmin {
 803                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 804                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 805         } else {
 806                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 807                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 808         }
 809         if (method == "PUT" || method == "POST") && !permitUpload {
 810                 // Disallow operations that upload new files.
 811                 // Permit webdav operations that move existing files around.
 812                 return false
 813         } else if method == "GET" && !permitDownload {
 814                 // Disallow downloading file contents.
 815                 // Permit webdav operations like PROPFIND that retrieve metadata
 816                 // but not file contents.
 817                 return false
 818         }
 819         return true
 820 }
 821
 822 func (h *handler) logUploadOrDownload(
 823         r *http.Request,
 824         client *arvadosclient.ArvadosClient,
 825         fs arvados.CustomFileSystem,
 826         filepath string,
 827         collection *arvados.Collection,
 828         user *arvados.User) {
 829
 830         log := ctxlog.FromContext(r.Context())
 831         props := make(map[string]string)
 832         props["reqPath"] = r.URL.Path
 833         var useruuid string
 834         if user != nil {
 835                 log = log.WithField("user_uuid", user.UUID).
 836                         WithField("user_full_name", user.FullName)
 837                 useruuid = user.UUID
 838         } else {
 839                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 840         }
 841         if collection == nil && fs != nil {
 842                 collection, filepath = h.determineCollection(fs, filepath)
 843         }
 844         if collection != nil {
 845                 log = log.WithField("collection_file_path", filepath)
 846                 props["collection_file_path"] = filepath
 847                 // h.determineCollection populates the collection_uuid
 848                 // prop with the PDH, if this collection is being
 849                 // accessed via PDH. For logging, we use a different
 850                 // field depending on whether it's a UUID or PDH.
 851                 if len(collection.UUID) > 32 {
 852                         log = log.WithField("portable_data_hash", collection.UUID)
 853                         props["portable_data_hash"] = collection.UUID
 854                 } else {
 855                         log = log.WithField("collection_uuid", collection.UUID)
 856                         props["collection_uuid"] = collection.UUID
 857                 }
 858         }
 859         if r.Method == "PUT" || r.Method == "POST" {
 860                 log.Info("File upload")
 861                 if h.Cluster.Collections.WebDAVLogEvents {
 862                         go func() {
 863                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 864                                         "object_uuid": useruuid,
 865                                         "event_type":  "file_upload",
 866                                         "properties":  props}}
 867                                 err := client.Create("logs", lr, nil)
 868                                 if err != nil {
 869                                         log.WithError(err).Error("Failed to create upload log event on API server")
 870                                 }
 871                         }()
 872                 }
 873         } else if r.Method == "GET" {
 874                 if collection != nil && collection.PortableDataHash != "" {
 875                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 876                         props["portable_data_hash"] = collection.PortableDataHash
 877                 }
 878                 log.Info("File download")
 879                 if h.Cluster.Collections.WebDAVLogEvents {
 880                         go func() {
 881                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 882                                         "object_uuid": useruuid,
 883                                         "event_type":  "file_download",
 884                                         "properties":  props}}
 885                                 err := client.Create("logs", lr, nil)
 886                                 if err != nil {
 887                                         log.WithError(err).Error("Failed to create download log event on API server")
 888                                 }
 889                         }()
 890                 }
 891         }
 892 }
 893
 894 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 895         target := strings.TrimSuffix(path, "/")
 896         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 897                 target = target[:cut]
 898                 fi, err := fs.Stat(target)
 899                 if os.IsNotExist(err) {
 900                         // creating a new file/dir, or download
 901                         // destined to fail
 902                         continue
 903                 } else if err != nil {
 904                         return nil, ""
 905                 }
 906                 switch src := fi.Sys().(type) {
 907                 case *arvados.Collection:
 908                         return src, strings.TrimPrefix(path[len(target):], "/")
 909                 case *arvados.Group:
 910                         return nil, ""
 911                 default:
 912                         if _, ok := src.(error); ok {
 913                                 return nil, ""
 914                         }
 915                 }
 916         }
 917         return nil, ""
 918 }