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         "time"
  22
  23         "git.arvados.org/arvados.git/lib/cmd"
  24         "git.arvados.org/arvados.git/lib/webdavfs"
  25         "git.arvados.org/arvados.git/sdk/go/arvados"
  26         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  27         "git.arvados.org/arvados.git/sdk/go/auth"
  28         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  29         "git.arvados.org/arvados.git/sdk/go/httpserver"
  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         metrics *metrics
  38
  39         lockMtx    sync.Mutex
  40         lock       map[string]*sync.RWMutex
  41         lockTidied time.Time
  42
  43         s3SecretCache         map[string]*cachedS3Secret
  44         s3SecretCacheMtx      sync.Mutex
  45         s3SecretCacheNextTidy time.Time
  46 }
  47
  48 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  49
  50 var notFoundMessage = "Not Found"
  51 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  52
  53 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  54 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  55 // otherwise "".
  56 func parseCollectionIDFromURL(s string) string {
  57         if arvadosclient.UUIDMatch(s) {
  58                 return s
  59         }
  60         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  61                 return pdh
  62         }
  63         return ""
  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 type errorWithHTTPStatus interface {
  71         HTTPStatus() int
  72 }
  73
  74 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  75 // sends an HTTP header indicating success, updateOnSuccess first
  76 // calls the provided update func. If the update func fails, an error
  77 // response is sent (using the error's HTTP status or 500 if none),
  78 // and the status code and body sent by the handler are ignored (all
  79 // response writes return the update error).
  80 type updateOnSuccess struct {
  81         httpserver.ResponseWriter
  82         logger     logrus.FieldLogger
  83         update     func() error
  84         sentHeader bool
  85         err        error
  86 }
  87
  88 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  89         if !uos.sentHeader {
  90                 uos.WriteHeader(http.StatusOK)
  91         }
  92         if uos.err != nil {
  93                 return 0, uos.err
  94         }
  95         return uos.ResponseWriter.Write(p)
  96 }
  97
  98 func (uos *updateOnSuccess) WriteHeader(code int) {
  99         if !uos.sentHeader {
 100                 uos.sentHeader = true
 101                 if code >= 200 && code < 400 {
 102                         if uos.err = uos.update(); uos.err != nil {
 103                                 code := http.StatusInternalServerError
 104                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 105                                         code = he.HTTPStatus()
 106                                 }
 107                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 108                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 109                                 return
 110                         }
 111                 }
 112         }
 113         uos.ResponseWriter.WriteHeader(code)
 114 }
 115
 116 var (
 117         corsAllowHeadersHeader = strings.Join([]string{
 118                 "Authorization", "Content-Type", "Range",
 119                 // WebDAV request headers:
 120                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 121         }, ", ")
 122         writeMethod = map[string]bool{
 123                 "COPY":      true,
 124                 "DELETE":    true,
 125                 "LOCK":      true,
 126                 "MKCOL":     true,
 127                 "MOVE":      true,
 128                 "PROPPATCH": true,
 129                 "PUT":       true,
 130                 "RMCOL":     true,
 131                 "UNLOCK":    true,
 132         }
 133         webdavMethod = map[string]bool{
 134                 "COPY":      true,
 135                 "DELETE":    true,
 136                 "LOCK":      true,
 137                 "MKCOL":     true,
 138                 "MOVE":      true,
 139                 "OPTIONS":   true,
 140                 "PROPFIND":  true,
 141                 "PROPPATCH": true,
 142                 "PUT":       true,
 143                 "RMCOL":     true,
 144                 "UNLOCK":    true,
 145         }
 146         browserMethod = map[string]bool{
 147                 "GET":  true,
 148                 "HEAD": true,
 149                 "POST": true,
 150         }
 151         // top-level dirs to serve with siteFS
 152         siteFSDir = map[string]bool{
 153                 "":      true, // root directory
 154                 "by_id": true,
 155                 "users": true,
 156         }
 157 )
 158
 159 func stripDefaultPort(host string) string {
 160         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 161         u := &url.URL{Host: host}
 162         if p := u.Port(); p == "80" || p == "443" {
 163                 return strings.ToLower(u.Hostname())
 164         } else {
 165                 return strings.ToLower(host)
 166         }
 167 }
 168
 169 // CheckHealth implements service.Handler.
 170 func (h *handler) CheckHealth() error {
 171         return nil
 172 }
 173
 174 // Done implements service.Handler.
 175 func (h *handler) Done() <-chan struct{} {
 176         return nil
 177 }
 178
 179 // ServeHTTP implements http.Handler.
 180 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 181         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 182                 r.URL.Scheme = xfp
 183         }
 184
 185         wbuffer := newWriteBuffer(wOrig, int(h.Cluster.Collections.WebDAVOutputBuffer))
 186         defer wbuffer.Close()
 187         w := httpserver.WrapResponseWriter(responseWriter{
 188                 Writer:         wbuffer,
 189                 ResponseWriter: wOrig,
 190         })
 191
 192         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 193                 return
 194         }
 195
 196         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 197                 w.WriteHeader(http.StatusMethodNotAllowed)
 198                 return
 199         }
 200
 201         if r.Header.Get("Origin") != "" {
 202                 // Allow simple cross-origin requests without user
 203                 // credentials ("user credentials" as defined by CORS,
 204                 // i.e., cookies, HTTP authentication, and client-side
 205                 // SSL certificates. See
 206                 // http://www.w3.org/TR/cors/#user-credentials).
 207                 w.Header().Set("Access-Control-Allow-Origin", "*")
 208                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 209         }
 210
 211         if h.serveS3(w, r) {
 212                 return
 213         }
 214
 215         webdavPrefix := ""
 216         arvPath := r.URL.Path
 217         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 218                 // Enable a proxy (e.g., container log handler in
 219                 // controller) to satisfy a request for path
 220                 // "/foo/bar/baz.txt" using content from
 221                 // "//abc123-4.internal/bar/baz.txt", by adding a
 222                 // request header "X-Webdav-Prefix: /foo"
 223                 if !strings.HasPrefix(arvPath, prefix) {
 224                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 225                         return
 226                 }
 227                 arvPath = r.URL.Path[len(prefix):]
 228                 if arvPath == "" {
 229                         arvPath = "/"
 230                 }
 231                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 232                 webdavPrefix = prefix
 233         }
 234         pathParts := strings.Split(arvPath[1:], "/")
 235
 236         var stripParts int
 237         var collectionID string
 238         var tokens []string
 239         var reqTokens []string
 240         var pathToken bool
 241         var attachment bool
 242         var useSiteFS bool
 243         credentialsOK := h.Cluster.Collections.TrustAllContent
 244         reasonNotAcceptingCredentials := ""
 245
 246         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 247                 credentialsOK = true
 248                 attachment = true
 249         } else if r.FormValue("disposition") == "attachment" {
 250                 attachment = true
 251         }
 252
 253         if !credentialsOK {
 254                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 255                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 256         }
 257
 258         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 259                 // http://ID.collections.example/PATH...
 260                 credentialsOK = true
 261         } else if r.URL.Path == "/status.json" {
 262                 h.serveStatus(w, r)
 263                 return
 264         } else if siteFSDir[pathParts[0]] {
 265                 useSiteFS = true
 266         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 267                 // /c=ID[/PATH...]
 268                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 269                 stripParts = 1
 270         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 271                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 272                         // /collections/download/ID/TOKEN/PATH...
 273                         collectionID = parseCollectionIDFromURL(pathParts[2])
 274                         tokens = []string{pathParts[3]}
 275                         stripParts = 4
 276                         pathToken = true
 277                 } else {
 278                         // /collections/ID/PATH...
 279                         collectionID = parseCollectionIDFromURL(pathParts[1])
 280                         stripParts = 2
 281                         // This path is only meant to work for public
 282                         // data. Tokens provided with the request are
 283                         // ignored.
 284                         credentialsOK = false
 285                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 286                 }
 287         }
 288
 289         forceReload := false
 290         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 291                 forceReload = true
 292         }
 293
 294         if credentialsOK {
 295                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 296         }
 297
 298         r.ParseForm()
 299         origin := r.Header.Get("Origin")
 300         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 301         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 302         // Important distinction: safeAttachment checks whether api_token exists
 303         // as a query parameter. haveFormTokens checks whether api_token exists
 304         // as request form data *or* a query parameter. Different checks are
 305         // necessary because both the request disposition and the location of
 306         // the API token affect whether or not the request needs to be
 307         // redirected. The different branch comments below explain further.
 308         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 309         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 310                 // No token to use or redact.
 311         } else if safeAjax || safeAttachment {
 312                 // If this is a cross-origin request, the URL won't
 313                 // appear in the browser's address bar, so
 314                 // substituting a clipboard-safe URL is pointless.
 315                 // Redirect-with-cookie wouldn't work anyway, because
 316                 // it's not safe to allow third-party use of our
 317                 // cookie.
 318                 //
 319                 // If we're supplying an attachment, we don't need to
 320                 // convert POST to GET to avoid the "really resubmit
 321                 // form?" problem, so provided the token isn't
 322                 // embedded in the URL, there's no reason to do
 323                 // redirect-with-cookie in this case either.
 324                 for _, tok := range formTokens {
 325                         reqTokens = append(reqTokens, tok)
 326                 }
 327         } else if browserMethod[r.Method] {
 328                 // If this is a page view, and the client provided a
 329                 // token via query string or POST body, we must put
 330                 // the token in an HttpOnly cookie, and redirect to an
 331                 // equivalent URL with the query param redacted and
 332                 // method = GET.
 333                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 334                 return
 335         }
 336
 337         targetPath := pathParts[stripParts:]
 338         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 339                 // http://ID.example/t=TOKEN/PATH...
 340                 // /c=ID/t=TOKEN/PATH...
 341                 //
 342                 // This form must only be used to pass scoped tokens
 343                 // that give permission for a single collection. See
 344                 // FormValue case above.
 345                 tokens = []string{targetPath[0][2:]}
 346                 pathToken = true
 347                 targetPath = targetPath[1:]
 348                 stripParts++
 349         }
 350
 351         fsprefix := ""
 352         if useSiteFS {
 353                 if writeMethod[r.Method] {
 354                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 355                         return
 356                 }
 357                 if len(reqTokens) == 0 {
 358                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 359                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 360                         return
 361                 }
 362                 tokens = reqTokens
 363         } else if collectionID == "" {
 364                 http.Error(w, notFoundMessage, http.StatusNotFound)
 365                 return
 366         } else {
 367                 fsprefix = "by_id/" + collectionID + "/"
 368         }
 369
 370         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 371                 fsprefix += src[1:]
 372         }
 373
 374         if tokens == nil {
 375                 tokens = reqTokens
 376                 if h.Cluster.Users.AnonymousUserToken != "" {
 377                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 378                 }
 379         }
 380
 381         if len(targetPath) > 0 && targetPath[0] == "_" {
 382                 // If a collection has a directory called "t=foo" or
 383                 // "_", it can be served at
 384                 // //collections.example/_/t=foo/ or
 385                 // //collections.example/_/_/ respectively:
 386                 // //collections.example/t=foo/ won't work because
 387                 // t=foo will be interpreted as a token "foo".
 388                 targetPath = targetPath[1:]
 389                 stripParts++
 390         }
 391
 392         dirOpenMode := os.O_RDONLY
 393         if writeMethod[r.Method] {
 394                 dirOpenMode = os.O_RDWR
 395         }
 396
 397         var tokenValid bool
 398         var tokenScopeProblem bool
 399         var token string
 400         var tokenUser *arvados.User
 401         var sessionFS arvados.CustomFileSystem
 402         var session *cachedSession
 403         var collectionDir arvados.File
 404         for _, token = range tokens {
 405                 var statusErr errorWithHTTPStatus
 406                 fs, sess, user, err := h.Cache.GetSession(token)
 407                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 408                         // bad token
 409                         continue
 410                 } else if err != nil {
 411                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 412                         return
 413                 }
 414                 if token != h.Cluster.Users.AnonymousUserToken {
 415                         tokenValid = true
 416                 }
 417                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 418                 if errors.As(err, &statusErr) &&
 419                         statusErr.HTTPStatus() == http.StatusForbidden &&
 420                         token != h.Cluster.Users.AnonymousUserToken {
 421                         // collection id is outside scope of supplied
 422                         // token
 423                         tokenScopeProblem = true
 424                         sess.Release()
 425                         continue
 426                 } else if os.IsNotExist(err) {
 427                         // collection does not exist or is not
 428                         // readable using this token
 429                         sess.Release()
 430                         continue
 431                 } else if err != nil {
 432                         http.Error(w, err.Error(), http.StatusInternalServerError)
 433                         sess.Release()
 434                         return
 435                 }
 436                 defer f.Close()
 437
 438                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 439                 break
 440         }
 441
 442         // releaseSession() is equivalent to session.Release() except
 443         // that it's a no-op if (1) session is nil, or (2) it has
 444         // already been called.
 445         //
 446         // This way, we can do a defer call here to ensure it gets
 447         // called in all code paths, and also call it inline (see
 448         // below) in the cases where we want to release the lock
 449         // before returning.
 450         releaseSession := func() {}
 451         if session != nil {
 452                 var releaseSessionOnce sync.Once
 453                 releaseSession = func() { releaseSessionOnce.Do(func() { session.Release() }) }
 454         }
 455         defer releaseSession()
 456
 457         if forceReload && collectionDir != nil {
 458                 err := collectionDir.Sync()
 459                 if err != nil {
 460                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 461                                 http.Error(w, err.Error(), he.HTTPStatus())
 462                         } else {
 463                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 464                         }
 465                         return
 466                 }
 467         }
 468         if session == nil {
 469                 if pathToken {
 470                         // The URL is a "secret sharing link" that
 471                         // didn't work out.  Asking the client for
 472                         // additional credentials would just be
 473                         // confusing.
 474                         http.Error(w, notFoundMessage, http.StatusNotFound)
 475                         return
 476                 }
 477                 if tokenValid {
 478                         // The client provided valid token(s), but the
 479                         // collection was not found.
 480                         http.Error(w, notFoundMessage, http.StatusNotFound)
 481                         return
 482                 }
 483                 if tokenScopeProblem {
 484                         // The client provided a valid token but
 485                         // fetching a collection returned 401, which
 486                         // means the token scope doesn't permit
 487                         // fetching that collection.
 488                         http.Error(w, notFoundMessage, http.StatusForbidden)
 489                         return
 490                 }
 491                 // The client's token was invalid (e.g., expired), or
 492                 // the client didn't even provide one.  Redirect to
 493                 // workbench2's login-and-redirect-to-download url if
 494                 // this is a browser navigation request. (The redirect
 495                 // flow can't preserve the original method if it's not
 496                 // GET, and doesn't make sense if the UA is a
 497                 // command-line tool, is trying to load an inline
 498                 // image, etc.; in these cases, there's nothing we can
 499                 // do, so return 401 unauthorized.)
 500                 //
 501                 // Note Sec-Fetch-Mode is sent by all non-EOL
 502                 // browsers, except Safari.
 503                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 504                 //
 505                 // TODO(TC): This response would be confusing to
 506                 // someone trying (anonymously) to download public
 507                 // data that has been deleted.  Allow a referrer to
 508                 // provide this context somehow?
 509                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 510                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 511                         redirkey := "redirectToPreview"
 512                         if attachment {
 513                                 redirkey = "redirectToDownload"
 514                         }
 515                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 516                         query := url.Values{redirkey: {callback}}
 517                         queryString := query.Encode()
 518                         // Note: Encode (and QueryEscape function) turns space
 519                         // into plus sign (+) rather than %20 (the plus sign
 520                         // becomes %2B); that is the rule for web forms data
 521                         // sent in URL query part via GET, but we're not
 522                         // emulating forms here. Client JS APIs
 523                         // (URLSearchParam#get, decodeURIComponent) will
 524                         // decode %20, but while the former also expects the
 525                         // form-specific encoding, the latter doesn't.
 526                         // Encode() almost encodes everything; RFC3986 sec. 3.4
 527                         // says "it is sometimes better for usability" to not
 528                         // encode / and ? when passing URI reference in query.
 529                         // This is also legal according to WHATWG URL spec and
 530                         // can be desirable for debugging webapp.
 531                         // We can let slash / appear in the encoded query, and
 532                         // equality-sign = too, but exempting ? is not very
 533                         // useful.
 534                         // Plus-sign, hash, and ampersand are never exempt.
 535                         r := strings.NewReplacer("+", "%20", "%2F", "/", "%3D", "=")
 536                         target.RawQuery = r.Replace(queryString)
 537                         w.Header().Add("Location", target.String())
 538                         w.WriteHeader(http.StatusSeeOther)
 539                         return
 540                 }
 541                 if !credentialsOK {
 542                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 543                         return
 544                 }
 545                 // If none of the above cases apply, suggest the
 546                 // user-agent (which is either a non-browser agent
 547                 // like wget, or a browser that can't redirect through
 548                 // a login flow) prompt the user for credentials.
 549                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 550                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 551                 return
 552         }
 553
 554         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 555                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 556                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 557                         releaseSession() // because we won't be writing anything
 558                         if !strings.HasSuffix(r.URL.Path, "/") {
 559                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 560                         } else {
 561                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 562                         }
 563                         return
 564                 }
 565         }
 566
 567         var basename string
 568         if len(targetPath) > 0 {
 569                 basename = targetPath[len(targetPath)-1]
 570         }
 571         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 572                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 573                 return
 574         }
 575         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 576                 http.Error(w, "Not permitted", http.StatusForbidden)
 577                 return
 578         }
 579         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 580
 581         writing := writeMethod[r.Method]
 582         locker := h.collectionLock(collectionID, writing)
 583         defer locker.Unlock()
 584
 585         if writing {
 586                 // Save the collection only if/when all
 587                 // webdav->filesystem operations succeed --
 588                 // and send a 500 error if the modified
 589                 // collection can't be saved.
 590                 //
 591                 // Perform the write in a separate sitefs, so
 592                 // concurrent read operations on the same
 593                 // collection see the previous saved
 594                 // state. After the write succeeds and the
 595                 // collection record is updated, we reset the
 596                 // session so the updates are visible in
 597                 // subsequent read requests.
 598                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 599                 sessionFS = client.SiteFileSystem(session.keepclient)
 600                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 601                 if err != nil {
 602                         http.Error(w, err.Error(), http.StatusInternalServerError)
 603                         return
 604                 }
 605                 defer writingDir.Close()
 606                 w = &updateOnSuccess{
 607                         ResponseWriter: w,
 608                         logger:         ctxlog.FromContext(r.Context()),
 609                         update: func() error {
 610                                 err := writingDir.Sync()
 611                                 var te arvados.TransactionError
 612                                 if errors.As(err, &te) {
 613                                         err = te
 614                                 }
 615                                 if err != nil {
 616                                         return err
 617                                 }
 618                                 // Sync the changes to the persistent
 619                                 // sessionfs for this token.
 620                                 snap, err := writingDir.Snapshot()
 621                                 if err != nil {
 622                                         return err
 623                                 }
 624                                 collectionDir.Splice(snap)
 625                                 return nil
 626                         }}
 627         } else {
 628                 // When writing, we need to block session renewal
 629                 // until we're finished, in order to guarantee the
 630                 // effect of the write is visible in future responses.
 631                 // But if we're not writing, we can release the lock
 632                 // early.  This enables us to keep renewing sessions
 633                 // and processing more requests even if a slow client
 634                 // takes a long time to download a large file.
 635                 releaseSession()
 636         }
 637         if r.Method == http.MethodGet {
 638                 applyContentDispositionHdr(w, r, basename, attachment)
 639         }
 640         if webdavPrefix == "" {
 641                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 642         }
 643         wh := &webdav.Handler{
 644                 Prefix: webdavPrefix,
 645                 FileSystem: &webdavfs.FS{
 646                         FileSystem:    sessionFS,
 647                         Prefix:        fsprefix,
 648                         Writing:       writeMethod[r.Method],
 649                         AlwaysReadEOF: r.Method == "PROPFIND",
 650                 },
 651                 LockSystem: webdavfs.NoLockSystem,
 652                 Logger: func(r *http.Request, err error) {
 653                         if err != nil && !os.IsNotExist(err) {
 654                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 655                         }
 656                 },
 657         }
 658         h.metrics.track(wh, w, r)
 659         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 660                 wrote := int64(w.WroteBodyBytes())
 661                 fnm := strings.Join(pathParts[stripParts:], "/")
 662                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 663                 if err == nil && fi.Size() != wrote {
 664                         var n int
 665                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 666                         if err == nil {
 667                                 n, err = f.Read(make([]byte, 1024))
 668                                 f.Close()
 669                         }
 670                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 671                 }
 672         }
 673 }
 674
 675 var dirListingTemplate = `<!DOCTYPE HTML>
 676 <HTML><HEAD>
 677   <META name="robots" content="NOINDEX">
 678   <TITLE>{{ .CollectionName }}</TITLE>
 679   <STYLE type="text/css">
 680     body {
 681       margin: 1.5em;
 682     }
 683     pre {
 684       background-color: #D9EDF7;
 685       border-radius: .25em;
 686       padding: .75em;
 687       overflow: auto;
 688     }
 689     .footer p {
 690       font-size: 82%;
 691     }
 692     hr {
 693       border: 1px solid #808080;
 694     }
 695     ul {
 696       padding: 0;
 697     }
 698     ul li {
 699       font-family: monospace;
 700       list-style: none;
 701     }
 702   </STYLE>
 703 </HEAD>
 704 <BODY>
 705
 706 <H1>{{ .CollectionName }}</H1>
 707
 708 <P>This collection of data files is being shared with you through
 709 Arvados.  You can download individual files listed below.  To download
 710 the entire directory tree with <CODE>wget</CODE>, try:</P>
 711
 712 <PRE id="wget-example">$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} {{ .QuotedUrlForWget }}</PRE>
 713
 714 <H2>File Listing</H2>
 715
 716 {{if .Files}}
 717 <UL>
 718 {{range .Files}}
 719 {{if .IsDir }}
 720   <LI>{{" " | printf "%15s  " | nbsp}}<A class="item" href="{{ .Href }}/">{{ .Name }}/</A></LI>
 721 {{else}}
 722   <LI>{{.Size | printf "%15d  " | nbsp}}<A class="item" href="{{ .Href }}">{{ .Name }}</A></LI>
 723 {{end}}
 724 {{end}}
 725 </UL>
 726 {{else}}
 727 <P>(No files; this collection is empty.)</P>
 728 {{end}}
 729
 730 <HR>
 731 <DIV class="footer">
 732   <P>
 733     About Arvados:
 734     Arvados is a free and open source software bioinformatics platform.
 735     To learn more, visit arvados.org.
 736     Arvados is not responsible for the files listed on this page.
 737   </P>
 738 </DIV>
 739
 740 </BODY>
 741 </HTML>
 742 `
 743
 744 type fileListEnt struct {
 745         Name  string
 746         Href  string
 747         Size  int64
 748         IsDir bool
 749 }
 750
 751 // Given a filesystem path like `foo/"bar baz"`, return an escaped
 752 // (percent-encoded) relative path like `./foo/%22bar%20%baz%22`.
 753 //
 754 // Note the result may contain html-unsafe characters like '&'. These
 755 // will be handled separately by the HTML templating engine as needed.
 756 func relativeHref(path string) string {
 757         u := &url.URL{Path: path}
 758         return "./" + u.EscapedPath()
 759 }
 760
 761 // Return a shell-quoted URL suitable for pasting to a command line
 762 // ("wget ...") to repeat the given HTTP request.
 763 func makeQuotedUrlForWget(r *http.Request) string {
 764         scheme := r.Header.Get("X-Forwarded-Proto")
 765         if scheme == "http" || scheme == "https" {
 766                 // use protocol reported by load balancer / proxy
 767         } else if r.TLS != nil {
 768                 scheme = "https"
 769         } else {
 770                 scheme = "http"
 771         }
 772         p := r.URL.EscapedPath()
 773         // An escaped path may still contain single quote chars, which
 774         // would interfere with our shell quoting. Avoid this by
 775         // escaping them as %27.
 776         return fmt.Sprintf("'%s://%s%s'", scheme, r.Host, strings.Replace(p, "'", "%27", -1))
 777 }
 778
 779 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 780         var files []fileListEnt
 781         var walk func(string) error
 782         if !strings.HasSuffix(base, "/") {
 783                 base = base + "/"
 784         }
 785         walk = func(path string) error {
 786                 dirname := base + path
 787                 if dirname != "/" {
 788                         dirname = strings.TrimSuffix(dirname, "/")
 789                 }
 790                 d, err := fs.Open(dirname)
 791                 if err != nil {
 792                         return err
 793                 }
 794                 ents, err := d.Readdir(-1)
 795                 if err != nil {
 796                         return err
 797                 }
 798                 for _, ent := range ents {
 799                         if recurse && ent.IsDir() {
 800                                 err = walk(path + ent.Name() + "/")
 801                                 if err != nil {
 802                                         return err
 803                                 }
 804                         } else {
 805                                 listingName := path + ent.Name()
 806                                 files = append(files, fileListEnt{
 807                                         Name:  listingName,
 808                                         Href:  relativeHref(listingName),
 809                                         Size:  ent.Size(),
 810                                         IsDir: ent.IsDir(),
 811                                 })
 812                         }
 813                 }
 814                 return nil
 815         }
 816         if err := walk(""); err != nil {
 817                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 818                 return
 819         }
 820
 821         funcs := template.FuncMap{
 822                 "nbsp": func(s string) template.HTML {
 823                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 824                 },
 825         }
 826         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 827         if err != nil {
 828                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 829                 return
 830         }
 831         sort.Slice(files, func(i, j int) bool {
 832                 return files[i].Name < files[j].Name
 833         })
 834         w.WriteHeader(http.StatusOK)
 835         tmpl.Execute(w, map[string]interface{}{
 836                 "CollectionName":   collectionName,
 837                 "Files":            files,
 838                 "Request":          r,
 839                 "StripParts":       strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 840                 "QuotedUrlForWget": makeQuotedUrlForWget(r),
 841         })
 842 }
 843
 844 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 845         disposition := "inline"
 846         if isAttachment {
 847                 disposition = "attachment"
 848         }
 849         if strings.ContainsRune(r.RequestURI, '?') {
 850                 // Help the UA realize that the filename is just
 851                 // "filename.txt", not
 852                 // "filename.txt?disposition=attachment".
 853                 //
 854                 // TODO(TC): Follow advice at RFC 6266 appendix D
 855                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 856         }
 857         if disposition != "inline" {
 858                 w.Header().Set("Content-Disposition", disposition)
 859         }
 860 }
 861
 862 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 863         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
 864                 if !credentialsOK {
 865                         // It is not safe to copy the provided token
 866                         // into a cookie unless the current vhost
 867                         // (origin) serves only a single collection or
 868                         // we are in TrustAllContent mode.
 869                         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)
 870                         return
 871                 }
 872
 873                 // The HttpOnly flag is necessary to prevent
 874                 // JavaScript code (included in, or loaded by, a page
 875                 // in the collection being served) from employing the
 876                 // user's token beyond reading other files in the same
 877                 // domain, i.e., same collection.
 878                 //
 879                 // The 303 redirect is necessary in the case of a GET
 880                 // request to avoid exposing the token in the Location
 881                 // bar, and in the case of a POST request to avoid
 882                 // raising warnings when the user refreshes the
 883                 // resulting page.
 884                 for _, tok := range formTokens {
 885                         if tok == "" {
 886                                 continue
 887                         }
 888                         http.SetCookie(w, &http.Cookie{
 889                                 Name:     "arvados_api_token",
 890                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
 891                                 Path:     "/",
 892                                 HttpOnly: true,
 893                                 SameSite: http.SameSiteLaxMode,
 894                         })
 895                         break
 896                 }
 897         }
 898
 899         // Propagate query parameters (except api_token) from
 900         // the original request.
 901         redirQuery := r.URL.Query()
 902         redirQuery.Del("api_token")
 903
 904         u := r.URL
 905         if location != "" {
 906                 newu, err := u.Parse(location)
 907                 if err != nil {
 908                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 909                         return
 910                 }
 911                 u = newu
 912         }
 913         redir := (&url.URL{
 914                 Scheme:   r.URL.Scheme,
 915                 Host:     r.Host,
 916                 Path:     u.Path,
 917                 RawQuery: redirQuery.Encode(),
 918         }).String()
 919
 920         w.Header().Add("Location", redir)
 921         w.WriteHeader(http.StatusSeeOther)
 922         io.WriteString(w, `<A href="`)
 923         io.WriteString(w, html.EscapeString(redir))
 924         io.WriteString(w, `">Continue</A>`)
 925 }
 926
 927 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 928         var permitDownload bool
 929         var permitUpload bool
 930         if tokenUser != nil && tokenUser.IsAdmin {
 931                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 932                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 933         } else {
 934                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 935                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 936         }
 937         if (method == "PUT" || method == "POST") && !permitUpload {
 938                 // Disallow operations that upload new files.
 939                 // Permit webdav operations that move existing files around.
 940                 return false
 941         } else if method == "GET" && !permitDownload {
 942                 // Disallow downloading file contents.
 943                 // Permit webdav operations like PROPFIND that retrieve metadata
 944                 // but not file contents.
 945                 return false
 946         }
 947         return true
 948 }
 949
 950 func (h *handler) logUploadOrDownload(
 951         r *http.Request,
 952         client *arvadosclient.ArvadosClient,
 953         fs arvados.CustomFileSystem,
 954         filepath string,
 955         collection *arvados.Collection,
 956         user *arvados.User) {
 957
 958         log := ctxlog.FromContext(r.Context())
 959         props := make(map[string]string)
 960         props["reqPath"] = r.URL.Path
 961         var useruuid string
 962         if user != nil {
 963                 log = log.WithField("user_uuid", user.UUID).
 964                         WithField("user_full_name", user.FullName)
 965                 useruuid = user.UUID
 966         } else {
 967                 useruuid = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
 968         }
 969         if collection == nil && fs != nil {
 970                 collection, filepath = h.determineCollection(fs, filepath)
 971         }
 972         if collection != nil {
 973                 log = log.WithField("collection_file_path", filepath)
 974                 props["collection_file_path"] = filepath
 975                 // h.determineCollection populates the collection_uuid
 976                 // prop with the PDH, if this collection is being
 977                 // accessed via PDH. For logging, we use a different
 978                 // field depending on whether it's a UUID or PDH.
 979                 if len(collection.UUID) > 32 {
 980                         log = log.WithField("portable_data_hash", collection.UUID)
 981                         props["portable_data_hash"] = collection.UUID
 982                 } else {
 983                         log = log.WithField("collection_uuid", collection.UUID)
 984                         props["collection_uuid"] = collection.UUID
 985                 }
 986         }
 987         if r.Method == "PUT" || r.Method == "POST" {
 988                 log.Info("File upload")
 989                 if h.Cluster.Collections.WebDAVLogEvents {
 990                         go func() {
 991                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
 992                                         "object_uuid": useruuid,
 993                                         "event_type":  "file_upload",
 994                                         "properties":  props}}
 995                                 err := client.Create("logs", lr, nil)
 996                                 if err != nil {
 997                                         log.WithError(err).Error("Failed to create upload log event on API server")
 998                                 }
 999                         }()
1000                 }
1001         } else if r.Method == "GET" {
1002                 if collection != nil && collection.PortableDataHash != "" {
1003                         log = log.WithField("portable_data_hash", collection.PortableDataHash)
1004                         props["portable_data_hash"] = collection.PortableDataHash
1005                 }
1006                 log.Info("File download")
1007                 if h.Cluster.Collections.WebDAVLogEvents {
1008                         go func() {
1009                                 lr := arvadosclient.Dict{"log": arvadosclient.Dict{
1010                                         "object_uuid": useruuid,
1011                                         "event_type":  "file_download",
1012                                         "properties":  props}}
1013                                 err := client.Create("logs", lr, nil)
1014                                 if err != nil {
1015                                         log.WithError(err).Error("Failed to create download log event on API server")
1016                                 }
1017                         }()
1018                 }
1019         }
1020 }
1021
1022 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
1023         target := strings.TrimSuffix(path, "/")
1024         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
1025                 target = target[:cut]
1026                 fi, err := fs.Stat(target)
1027                 if os.IsNotExist(err) {
1028                         // creating a new file/dir, or download
1029                         // destined to fail
1030                         continue
1031                 } else if err != nil {
1032                         return nil, ""
1033                 }
1034                 switch src := fi.Sys().(type) {
1035                 case *arvados.Collection:
1036                         return src, strings.TrimPrefix(path[len(target):], "/")
1037                 case *arvados.Group:
1038                         return nil, ""
1039                 default:
1040                         if _, ok := src.(error); ok {
1041                                 return nil, ""
1042                         }
1043                 }
1044         }
1045         return nil, ""
1046 }
1047
1048 var lockTidyInterval = time.Minute * 10
1049
1050 // Lock the specified collection for reading or writing. Caller must
1051 // call Unlock() on the returned Locker when the operation is
1052 // finished.
1053 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
1054         h.lockMtx.Lock()
1055         defer h.lockMtx.Unlock()
1056         if time.Since(h.lockTidied) > lockTidyInterval {
1057                 // Periodically delete all locks that aren't in use.
1058                 h.lockTidied = time.Now()
1059                 for id, locker := range h.lock {
1060                         if locker.TryLock() {
1061                                 locker.Unlock()
1062                                 delete(h.lock, id)
1063                         }
1064                 }
1065         }
1066         locker := h.lock[collectionID]
1067         if locker == nil {
1068                 locker = new(sync.RWMutex)
1069                 if h.lock == nil {
1070                         h.lock = map[string]*sync.RWMutex{}
1071                 }
1072                 h.lock[collectionID] = locker
1073         }
1074         if writing {
1075                 locker.Lock()
1076                 return locker
1077         } else {
1078                 locker.RLock()
1079                 return locker.RLocker()
1080         }
1081 }
1082
1083 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1084         method := header.Get("Access-Control-Request-Method")
1085         if method == "" {
1086                 return false
1087         }
1088         if !browserMethod[method] && !webdavMethod[method] {
1089                 w.WriteHeader(http.StatusMethodNotAllowed)
1090                 return true
1091         }
1092         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1093         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1094         w.Header().Set("Access-Control-Allow-Origin", "*")
1095         w.Header().Set("Access-Control-Max-Age", "86400")
1096         return true
1097 }