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/lib/webdavfs"
  24         "git.arvados.org/arvados.git/sdk/go/arvados"
  25         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  26         "git.arvados.org/arvados.git/sdk/go/auth"
  27         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  28         "git.arvados.org/arvados.git/sdk/go/httpserver"
  29         "git.arvados.org/arvados.git/sdk/go/keepclient"
  30         "github.com/sirupsen/logrus"
  31         "golang.org/x/net/webdav"
  32 )
  33
  34 type handler struct {
  35         Cache     cache
  36         Cluster   *arvados.Cluster
  37         setupOnce sync.Once
  38 }
  39
  40 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  41
  42 var notFoundMessage = "Not Found"
  43 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  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
  62 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  63         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  64 }
  65
  66 type errorWithHTTPStatus interface {
  67         HTTPStatus() int
  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                                 if he := errorWithHTTPStatus(nil); 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", "Cache-Control",
 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, webdavfs.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 len(targetPath) > 0 && targetPath[0] == "_" {
 352                 // If a collection has a directory called "t=foo" or
 353                 // "_", it can be served at
 354                 // //collections.example/_/t=foo/ or
 355                 // //collections.example/_/_/ respectively:
 356                 // //collections.example/t=foo/ won't work because
 357                 // t=foo will be interpreted as a token "foo".
 358                 targetPath = targetPath[1:]
 359                 stripParts++
 360         }
 361
 362         dirOpenMode := os.O_RDONLY
 363         if writeMethod[r.Method] {
 364                 dirOpenMode = os.O_RDWR
 365         }
 366
 367         var tokenValid bool
 368         var tokenScopeProblem bool
 369         var token string
 370         var tokenUser *arvados.User
 371         var sessionFS arvados.CustomFileSystem
 372         var session *cachedSession
 373         var collectionDir arvados.File
 374         for _, token = range tokens {
 375                 var statusErr errorWithHTTPStatus
 376                 fs, sess, user, err := h.Cache.GetSession(token)
 377                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 378                         // bad token
 379                         continue
 380                 } else if err != nil {
 381                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 382                         return
 383                 }
 384                 if token != h.Cluster.Users.AnonymousUserToken {
 385                         tokenValid = true
 386                 }
 387                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 388                 if errors.As(err, &statusErr) &&
 389                         statusErr.HTTPStatus() == http.StatusForbidden &&
 390                         token != h.Cluster.Users.AnonymousUserToken {
 391                         // collection id is outside scope of supplied
 392                         // token
 393                         tokenScopeProblem = true
 394                         continue
 395                 } else if os.IsNotExist(err) {
 396                         // collection does not exist or is not
 397                         // readable using this token
 398                         continue
 399                 } else if err != nil {
 400                         http.Error(w, err.Error(), http.StatusInternalServerError)
 401                         return
 402                 }
 403                 defer f.Close()
 404
 405                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 406                 break
 407         }
 408         if forceReload && collectionDir != nil {
 409                 err := collectionDir.Sync()
 410                 if err != nil {
 411                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 412                                 http.Error(w, err.Error(), he.HTTPStatus())
 413                         } else {
 414                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 415                         }
 416                         return
 417                 }
 418         }
 419         if session == nil {
 420                 if pathToken {
 421                         // The URL is a "secret sharing link" that
 422                         // didn't work out.  Asking the client for
 423                         // additional credentials would just be
 424                         // confusing.
 425                         http.Error(w, notFoundMessage, http.StatusNotFound)
 426                         return
 427                 }
 428                 if tokenValid {
 429                         // The client provided valid token(s), but the
 430                         // collection was not found.
 431                         http.Error(w, notFoundMessage, http.StatusNotFound)
 432                         return
 433                 }
 434                 if tokenScopeProblem {
 435                         // The client provided a valid token but
 436                         // fetching a collection returned 401, which
 437                         // means the token scope doesn't permit
 438                         // fetching that collection.
 439                         http.Error(w, notFoundMessage, http.StatusForbidden)
 440                         return
 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                         return
 478                 }
 479                 if !credentialsOK {
 480                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 481                         return
 482                 }
 483                 // If none of the above cases apply, suggest the
 484                 // user-agent (which is either a non-browser agent
 485                 // like wget, or a browser that can't redirect through
 486                 // a login flow) prompt the user for credentials.
 487                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 488                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 489                 return
 490         }
 491
 492         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 493                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 494                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 495                         if !strings.HasSuffix(r.URL.Path, "/") {
 496                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 497                         } else {
 498                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 499                         }
 500                         return
 501                 }
 502         }
 503
 504         var basename string
 505         if len(targetPath) > 0 {
 506                 basename = targetPath[len(targetPath)-1]
 507         }
 508         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 509                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 510                 return
 511         }
 512         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 513                 http.Error(w, "Not permitted", http.StatusForbidden)
 514                 return
 515         }
 516         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 517
 518         if writeMethod[r.Method] {
 519                 // Save the collection only if/when all
 520                 // webdav->filesystem operations succeed --
 521                 // and send a 500 error if the modified
 522                 // collection can't be saved.
 523                 //
 524                 // Perform the write in a separate sitefs, so
 525                 // concurrent read operations on the same
 526                 // collection see the previous saved
 527                 // state. After the write succeeds and the
 528                 // collection record is updated, we reset the
 529                 // session so the updates are visible in
 530                 // subsequent read requests.
 531                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 532                 sessionFS = client.SiteFileSystem(session.keepclient)
 533                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 534                 if err != nil {
 535                         http.Error(w, err.Error(), http.StatusInternalServerError)
 536                         return
 537                 }
 538                 defer writingDir.Close()
 539                 w = &updateOnSuccess{
 540                         ResponseWriter: w,
 541                         logger:         ctxlog.FromContext(r.Context()),
 542                         update: func() error {
 543                                 err := writingDir.Sync()
 544                                 var te arvados.TransactionError
 545                                 if errors.As(err, &te) {
 546                                         err = te
 547                                 }
 548                                 if err != nil {
 549                                         return err
 550                                 }
 551                                 // Sync the changes to the persistent
 552                                 // sessionfs for this token.
 553                                 snap, err := writingDir.Snapshot()
 554                                 if err != nil {
 555                                         return err
 556                                 }
 557                                 collectionDir.Splice(snap)
 558                                 return nil
 559                         }}
 560         }
 561         if r.Method == http.MethodGet {
 562                 applyContentDispositionHdr(w, r, basename, attachment)
 563         }
 564         wh := webdav.Handler{
 565                 Prefix: "/" + strings.Join(pathParts[:stripParts], "/"),
 566                 FileSystem: &webdavfs.FS{
 567                         FileSystem:    sessionFS,
 568                         Prefix:        fsprefix,
 569                         Writing:       writeMethod[r.Method],
 570                         AlwaysReadEOF: r.Method == "PROPFIND",
 571                 },
 572                 LockSystem: webdavfs.NoLockSystem,
 573                 Logger: func(r *http.Request, err error) {
 574                         if err != nil {
 575                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 576                         }
 577                 },
 578         }
 579         wh.ServeHTTP(w, r)
 580         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 581                 wrote := int64(w.WroteBodyBytes())
 582                 fnm := strings.Join(pathParts[stripParts:], "/")
 583                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 584                 if err == nil && fi.Size() != wrote {
 585                         var n int
 586                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 587                         if err == nil {
 588                                 n, err = f.Read(make([]byte, 1024))
 589                                 f.Close()
 590                         }
 591                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 592                 }
 593         }
 594 }
 595
 596 var dirListingTemplate = `<!DOCTYPE HTML>
 597 <HTML><HEAD>
 598   <META name="robots" content="NOINDEX">
 599   <TITLE>{{ .CollectionName }}</TITLE>
 600   <STYLE type="text/css">
 601     body {
 602       margin: 1.5em;
 603     }
 604     pre {
 605       background-color: #D9EDF7;
 606       border-radius: .25em;
 607       padding: .75em;
 608       overflow: auto;
 609     }
 610     .footer p {
 611       font-size: 82%;
 612     }
 613     ul {
 614       padding: 0;
 615     }
 616     ul li {
 617       font-family: monospace;
 618       list-style: none;
 619     }
 620   </STYLE>
 621 </HEAD>
 622 <BODY>
 623
 624 <H1>{{ .CollectionName }}</H1>
 625
 626 <P>This collection of data files is being shared with you through
 627 Arvados.  You can download individual files listed below.  To download
 628 the entire directory tree with wget, try:</P>
 629
 630 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL.Path }}</PRE>
 631
 632 <H2>File Listing</H2>
 633
 634 {{if .Files}}
 635 <UL>
 636 {{range .Files}}
 637 {{if .IsDir }}
 638   <LI>{{" " | printf "%15s  " | nbsp}}<A href="{{print "./" .Name}}/">{{.Name}}/</A></LI>
 639 {{else}}
 640   <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{print "./" .Name}}">{{.Name}}</A></LI>
 641 {{end}}
 642 {{end}}
 643 </UL>
 644 {{else}}
 645 <P>(No files; this collection is empty.)</P>
 646 {{end}}
 647
 648 <HR noshade>
 649 <DIV class="footer">
 650   <P>
 651     About Arvados:
 652     Arvados is a free and open source software bioinformatics platform.
 653     To learn more, visit arvados.org.
 654     Arvados is not responsible for the files listed on this page.
 655   </P>
 656 </DIV>
 657
 658 </BODY>
 659 `
 660
 661 type fileListEnt struct {
 662         Name  string
 663         Size  int64
 664         IsDir bool
 665 }
 666
 667 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 668         var files []fileListEnt
 669         var walk func(string) error
 670         if !strings.HasSuffix(base, "/") {
 671                 base = base + "/"
 672         }
 673         walk = func(path string) error {
 674                 dirname := base + path
 675                 if dirname != "/" {
 676                         dirname = strings.TrimSuffix(dirname, "/")
 677                 }
 678                 d, err := fs.Open(dirname)
 679                 if err != nil {
 680                         return err
 681                 }
 682                 ents, err := d.Readdir(-1)
 683                 if err != nil {
 684                         return err
 685                 }
 686                 for _, ent := range ents {
 687                         if recurse && ent.IsDir() {
 688                                 err = walk(path + ent.Name() + "/")
 689                                 if err != nil {
 690                                         return err
 691                                 }
 692                         } else {
 693                                 files = append(files, fileListEnt{
 694                                         Name:  path + ent.Name(),
 695                                         Size:  ent.Size(),
 696                                         IsDir: ent.IsDir(),
 697                                 })
 698                         }
 699                 }
 700                 return nil
 701         }
 702         if err := walk(""); err != nil {
 703                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 704                 return
 705         }
 706
 707         funcs := template.FuncMap{
 708                 "nbsp": func(s string) template.HTML {
 709                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 710                 },
 711         }
 712         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 713         if err != nil {
 714                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 715                 return
 716         }
 717         sort.Slice(files, func(i, j int) bool {
 718                 return files[i].Name < files[j].Name
 719         })
 720         w.WriteHeader(http.StatusOK)
 721         tmpl.Execute(w, map[string]interface{}{
 722                 "CollectionName": collectionName,
 723                 "Files":          files,
 724                 "Request":        r,
 725                 "StripParts":     strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 726         })
 727 }
 728
 729 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 730         disposition := "inline"
 731         if isAttachment {
 732                 disposition = "attachment"
 733         }
 734         if strings.ContainsRune(r.RequestURI, '?') {
 735                 // Help the UA realize that the filename is just
 736                 // "filename.txt", not
 737                 // "filename.txt?disposition=attachment".
 738                 //
 739                 // TODO(TC): Follow advice at RFC 6266 appendix D
 740                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 741         }
 742         if disposition != "inline" {
 743                 w.Header().Set("Content-Disposition", disposition)
 744         }
 745 }
 746
 747 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 748         if formToken := r.FormValue("api_token"); formToken != "" {
 749                 if !credentialsOK {
 750                         // It is not safe to copy the provided token
 751                         // into a cookie unless the current vhost
 752                         // (origin) serves only a single collection or
 753                         // we are in TrustAllContent mode.
 754                         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)
 755                         return
 756                 }
 757
 758                 // The HttpOnly flag is necessary to prevent
 759                 // JavaScript code (included in, or loaded by, a page
 760                 // in the collection being served) from employing the
 761                 // user's token beyond reading other files in the same
 762                 // domain, i.e., same collection.
 763                 //
 764                 // The 303 redirect is necessary in the case of a GET
 765                 // request to avoid exposing the token in the Location
 766                 // bar, and in the case of a POST request to avoid
 767                 // raising warnings when the user refreshes the
 768                 // resulting page.
 769                 http.SetCookie(w, &http.Cookie{
 770                         Name:     "arvados_api_token",
 771                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 772                         Path:     "/",
 773                         HttpOnly: true,
 774                         SameSite: http.SameSiteLaxMode,
 775                 })
 776         }
 777
 778         // Propagate query parameters (except api_token) from
 779         // the original request.
 780         redirQuery := r.URL.Query()
 781         redirQuery.Del("api_token")
 782
 783         u := r.URL
 784         if location != "" {
 785                 newu, err := u.Parse(location)
 786                 if err != nil {
 787                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 788                         return
 789                 }
 790                 u = newu
 791         }
 792         redir := (&url.URL{
 793                 Scheme:   r.URL.Scheme,
 794                 Host:     r.Host,
 795                 Path:     u.Path,
 796                 RawQuery: redirQuery.Encode(),
 797         }).String()
 798
 799         w.Header().Add("Location", redir)
 800         w.WriteHeader(http.StatusSeeOther)
 801         io.WriteString(w, `<A href="`)
 802         io.WriteString(w, html.EscapeString(redir))
 803         io.WriteString(w, `">Continue</A>`)
 804 }
 805
 806 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 807         var permitDownload bool
 808         var permitUpload bool
 809         if tokenUser != nil && tokenUser.IsAdmin {
 810                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 811                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 812         } else {
 813                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 814                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 815         }
 816         if (method == "PUT" || method == "POST") && !permitUpload {
 817                 // Disallow operations that upload new files.
 818                 // Permit webdav operations that move existing files around.
 819                 return false
 820         } else if method == "GET" && !permitDownload {
 821                 // Disallow downloading file contents.
 822                 // Permit webdav operations like PROPFIND that retrieve metadata
 823                 // but not file contents.
 824                 return false
 825         }
 826         return true
 827 }
 828
 829 func (h *handler) logUploadOrDownload(
 830         r *http.Request,
 831         client *arvadosclient.ArvadosClient,
 832         fs arvados.CustomFileSystem,
 833         filepath string,
 834         collection *arvados.Collection,
 835         user *arvados.User) {
 836
 837         log := ctxlog.FromContext(r.Context())
 838         props := make(map[string]string)
 839         props["reqPath"] = r.URL.Path
 840         var useruuid string
 841         if user != nil {
 842                 log = log.WithField("user_uuid", user.UUID).
 843                         WithField("user_full_name", user.FullName)
 844                 useruuid = user.UUID
 845         } else {
 846                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 847         }
 848         if collection == nil && fs != nil {
 849                 collection, filepath = h.determineCollection(fs, filepath)
 850         }
 851         if collection != nil {
 852                 log = log.WithField("collection_file_path", filepath)
 853                 props["collection_file_path"] = filepath
 854                 // h.determineCollection populates the collection_uuid
 855                 // prop with the PDH, if this collection is being
 856                 // accessed via PDH. For logging, we use a different
 857                 // field depending on whether it's a UUID or PDH.
 858                 if len(collection.UUID) > 32 {
 859                         log = log.WithField("portable_data_hash", collection.UUID)
 860                         props["portable_data_hash"] = collection.UUID
 861                 } else {
 862                         log = log.WithField("collection_uuid", collection.UUID)
 863                         props["collection_uuid"] = collection.UUID
 864                 }
 865         }
 866         if r.Method == "PUT" || r.Method == "POST" {
 867                 log.Info("File upload")
 868                 if h.Cluster.Collections.WebDAVLogEvents {
 869                         go func() {
 870                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 871                                         "object_uuid": useruuid,
 872                                         "event_type":  "file_upload",
 873                                         "properties":  props}}
 874                                 err := client.Create("logs", lr, nil)
 875                                 if err != nil {
 876                                         log.WithError(err).Error("Failed to create upload log event on API server")
 877                                 }
 878                         }()
 879                 }
 880         } else if r.Method == "GET" {
 881                 if collection != nil && collection.PortableDataHash != "" {
 882                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
 883                         props["portable_data_hash"] = collection.PortableDataHash
 884                 }
 885                 log.Info("File download")
 886                 if h.Cluster.Collections.WebDAVLogEvents {
 887                         go func() {
 888                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 889                                         "object_uuid": useruuid,
 890                                         "event_type":  "file_download",
 891                                         "properties":  props}}
 892                                 err := client.Create("logs", lr, nil)
 893                                 if err != nil {
 894                                         log.WithError(err).Error("Failed to create download log event on API server")
 895                                 }
 896                         }()
 897                 }
 898         }
 899 }
 900
 901 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
 902         target := strings.TrimSuffix(path, "/")
 903         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
 904                 target = target[:cut]
 905                 fi, err := fs.Stat(target)
 906                 if os.IsNotExist(err) {
 907                         // creating a new file/dir, or download
 908                         // destined to fail
 909                         continue
 910                 } else if err != nil {
 911                         return nil, ""
 912                 }
 913                 switch src := fi.Sys().(type) {
 914                 case *arvados.Collection:
 915                         return src, strings.TrimPrefix(path[len(target):], "/")
 916                 case *arvados.Group:
 917                         return nil, ""
 918                 default:
 919                         if _, ok := src.(error); ok {
 920                                 return nil, ""
 921                         }
 922                 }
 923         }
 924         return nil, ""
 925 }