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"
  15         "net/http"
  16         "net/url"
  17         "os"
  18         "slices"
  19         "sort"
  20         "strconv"
  21         "strings"
  22         "sync"
  23         "time"
  24
  25         "git.arvados.org/arvados.git/lib/cmd"
  26         "git.arvados.org/arvados.git/lib/webdavfs"
  27         "git.arvados.org/arvados.git/sdk/go/arvados"
  28         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  29         "git.arvados.org/arvados.git/sdk/go/auth"
  30         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  31         "git.arvados.org/arvados.git/sdk/go/httpserver"
  32         "github.com/sirupsen/logrus"
  33         "golang.org/x/net/webdav"
  34 )
  35
  36 type handler struct {
  37         Cache   cache
  38         Cluster *arvados.Cluster
  39         metrics *metrics
  40
  41         lockMtx    sync.Mutex
  42         lock       map[string]*sync.RWMutex
  43         lockTidied time.Time
  44
  45         fileEventLogs         map[fileEventLog]time.Time
  46         fileEventLogsMtx      sync.Mutex
  47         fileEventLogsNextTidy time.Time
  48
  49         s3SecretCache         map[string]*cachedS3Secret
  50         s3SecretCacheMtx      sync.Mutex
  51         s3SecretCacheNextTidy time.Time
  52 }
  53
  54 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  55
  56 var notFoundMessage = "Not Found"
  57 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  58
  59 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  60 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  61 // otherwise "".
  62 func parseCollectionIDFromURL(s string) string {
  63         if arvadosclient.UUIDMatch(s) {
  64                 return s
  65         }
  66         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  67                 return pdh
  68         }
  69         return ""
  70 }
  71
  72 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  73         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  74 }
  75
  76 type errorWithHTTPStatus interface {
  77         HTTPStatus() int
  78 }
  79
  80 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  81 // sends an HTTP header indicating success, updateOnSuccess first
  82 // calls the provided update func. If the update func fails, an error
  83 // response is sent (using the error's HTTP status or 500 if none),
  84 // and the status code and body sent by the handler are ignored (all
  85 // response writes return the update error).
  86 type updateOnSuccess struct {
  87         httpserver.ResponseWriter
  88         logger     logrus.FieldLogger
  89         update     func() error
  90         sentHeader bool
  91         err        error
  92 }
  93
  94 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  95         if !uos.sentHeader {
  96                 uos.WriteHeader(http.StatusOK)
  97         }
  98         if uos.err != nil {
  99                 return 0, uos.err
 100         }
 101         return uos.ResponseWriter.Write(p)
 102 }
 103
 104 func (uos *updateOnSuccess) WriteHeader(code int) {
 105         if !uos.sentHeader {
 106                 uos.sentHeader = true
 107                 if code >= 200 && code < 400 {
 108                         if uos.err = uos.update(); uos.err != nil {
 109                                 code := http.StatusInternalServerError
 110                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 111                                         code = he.HTTPStatus()
 112                                 }
 113                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 114                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 115                                 return
 116                         }
 117                 }
 118         }
 119         uos.ResponseWriter.WriteHeader(code)
 120 }
 121
 122 var (
 123         corsAllowHeadersHeader = strings.Join([]string{
 124                 "Authorization", "Content-Type", "Range",
 125                 // WebDAV request headers:
 126                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 127         }, ", ")
 128         writeMethod = map[string]bool{
 129                 "COPY":      true,
 130                 "DELETE":    true,
 131                 "LOCK":      true,
 132                 "MKCOL":     true,
 133                 "MOVE":      true,
 134                 "PROPPATCH": true,
 135                 "PUT":       true,
 136                 "RMCOL":     true,
 137                 "UNLOCK":    true,
 138         }
 139         webdavMethod = map[string]bool{
 140                 "COPY":      true,
 141                 "DELETE":    true,
 142                 "LOCK":      true,
 143                 "MKCOL":     true,
 144                 "MOVE":      true,
 145                 "OPTIONS":   true,
 146                 "PROPFIND":  true,
 147                 "PROPPATCH": true,
 148                 "PUT":       true,
 149                 "RMCOL":     true,
 150                 "UNLOCK":    true,
 151         }
 152         browserMethod = map[string]bool{
 153                 "GET":  true,
 154                 "HEAD": true,
 155                 "POST": true,
 156         }
 157         // top-level dirs to serve with siteFS
 158         siteFSDir = map[string]bool{
 159                 "":      true, // root directory
 160                 "by_id": true,
 161                 "users": true,
 162         }
 163 )
 164
 165 func stripDefaultPort(host string) string {
 166         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 167         u := &url.URL{Host: host}
 168         if p := u.Port(); p == "80" || p == "443" {
 169                 return strings.ToLower(u.Hostname())
 170         } else {
 171                 return strings.ToLower(host)
 172         }
 173 }
 174
 175 // CheckHealth implements service.Handler.
 176 func (h *handler) CheckHealth() error {
 177         return nil
 178 }
 179
 180 // Done implements service.Handler.
 181 func (h *handler) Done() <-chan struct{} {
 182         return nil
 183 }
 184
 185 // ServeHTTP implements http.Handler.
 186 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 187         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 188                 r.URL.Scheme = xfp
 189         }
 190
 191         wbuffer := newWriteBuffer(wOrig, int(h.Cluster.Collections.WebDAVOutputBuffer))
 192         defer wbuffer.Close()
 193         w := httpserver.WrapResponseWriter(responseWriter{
 194                 Writer:         wbuffer,
 195                 ResponseWriter: wOrig,
 196         })
 197
 198         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 199                 return
 200         }
 201
 202         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 203                 w.WriteHeader(http.StatusMethodNotAllowed)
 204                 return
 205         }
 206
 207         if r.Header.Get("Origin") != "" {
 208                 // Allow simple cross-origin requests without user
 209                 // credentials ("user credentials" as defined by CORS,
 210                 // i.e., cookies, HTTP authentication, and client-side
 211                 // SSL certificates. See
 212                 // http://www.w3.org/TR/cors/#user-credentials).
 213                 w.Header().Set("Access-Control-Allow-Origin", "*")
 214                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 215         }
 216
 217         if h.serveS3(w, r) {
 218                 return
 219         }
 220
 221         webdavPrefix := ""
 222         arvPath := r.URL.Path
 223         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 224                 // Enable a proxy (e.g., container log handler in
 225                 // controller) to satisfy a request for path
 226                 // "/foo/bar/baz.txt" using content from
 227                 // "//abc123-4.internal/bar/baz.txt", by adding a
 228                 // request header "X-Webdav-Prefix: /foo"
 229                 if !strings.HasPrefix(arvPath, prefix) {
 230                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 231                         return
 232                 }
 233                 arvPath = r.URL.Path[len(prefix):]
 234                 if arvPath == "" {
 235                         arvPath = "/"
 236                 }
 237                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 238                 webdavPrefix = prefix
 239         }
 240         pathParts := strings.Split(arvPath[1:], "/")
 241
 242         var stripParts int
 243         var collectionID string
 244         var tokens []string
 245         var reqTokens []string
 246         var pathToken bool
 247         var attachment bool
 248         var useSiteFS bool
 249         credentialsOK := h.Cluster.Collections.TrustAllContent
 250         reasonNotAcceptingCredentials := ""
 251
 252         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 253                 credentialsOK = true
 254                 attachment = true
 255         } else if r.FormValue("disposition") == "attachment" {
 256                 attachment = true
 257         }
 258
 259         if !credentialsOK {
 260                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 261                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 262         }
 263
 264         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 265                 // http://ID.collections.example/PATH...
 266                 credentialsOK = true
 267         } else if r.URL.Path == "/status.json" {
 268                 h.serveStatus(w, r)
 269                 return
 270         } else if siteFSDir[pathParts[0]] {
 271                 useSiteFS = true
 272         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 273                 // /c=ID[/PATH...]
 274                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 275                 stripParts = 1
 276         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 277                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 278                         // /collections/download/ID/TOKEN/PATH...
 279                         collectionID = parseCollectionIDFromURL(pathParts[2])
 280                         tokens = []string{pathParts[3]}
 281                         stripParts = 4
 282                         pathToken = true
 283                 } else {
 284                         // /collections/ID/PATH...
 285                         collectionID = parseCollectionIDFromURL(pathParts[1])
 286                         stripParts = 2
 287                         // This path is only meant to work for public
 288                         // data. Tokens provided with the request are
 289                         // ignored.
 290                         credentialsOK = false
 291                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 292                 }
 293         }
 294
 295         forceReload := false
 296         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 297                 forceReload = true
 298         }
 299
 300         if credentialsOK {
 301                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 302         }
 303
 304         r.ParseForm()
 305         origin := r.Header.Get("Origin")
 306         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 307         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 308         // Important distinction: safeAttachment checks whether api_token exists
 309         // as a query parameter. haveFormTokens checks whether api_token exists
 310         // as request form data *or* a query parameter. Different checks are
 311         // necessary because both the request disposition and the location of
 312         // the API token affect whether or not the request needs to be
 313         // redirected. The different branch comments below explain further.
 314         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 315         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 316                 // No token to use or redact.
 317         } else if safeAjax || safeAttachment {
 318                 // If this is a cross-origin request, the URL won't
 319                 // appear in the browser's address bar, so
 320                 // substituting a clipboard-safe URL is pointless.
 321                 // Redirect-with-cookie wouldn't work anyway, because
 322                 // it's not safe to allow third-party use of our
 323                 // cookie.
 324                 //
 325                 // If we're supplying an attachment, we don't need to
 326                 // convert POST to GET to avoid the "really resubmit
 327                 // form?" problem, so provided the token isn't
 328                 // embedded in the URL, there's no reason to do
 329                 // redirect-with-cookie in this case either.
 330                 for _, tok := range formTokens {
 331                         reqTokens = append(reqTokens, tok)
 332                 }
 333         } else if browserMethod[r.Method] {
 334                 // If this is a page view, and the client provided a
 335                 // token via query string or POST body, we must put
 336                 // the token in an HttpOnly cookie, and redirect to an
 337                 // equivalent URL with the query param redacted and
 338                 // method = GET.
 339                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 340                 return
 341         }
 342
 343         targetPath := pathParts[stripParts:]
 344         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 345                 // http://ID.example/t=TOKEN/PATH...
 346                 // /c=ID/t=TOKEN/PATH...
 347                 //
 348                 // This form must only be used to pass scoped tokens
 349                 // that give permission for a single collection. See
 350                 // FormValue case above.
 351                 tokens = []string{targetPath[0][2:]}
 352                 pathToken = true
 353                 targetPath = targetPath[1:]
 354                 stripParts++
 355         }
 356
 357         fsprefix := ""
 358         if useSiteFS {
 359                 if writeMethod[r.Method] {
 360                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 361                         return
 362                 }
 363                 if len(reqTokens) == 0 {
 364                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 365                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 366                         return
 367                 }
 368                 tokens = reqTokens
 369         } else if collectionID == "" {
 370                 http.Error(w, notFoundMessage, http.StatusNotFound)
 371                 return
 372         } else {
 373                 fsprefix = "by_id/" + collectionID + "/"
 374         }
 375
 376         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 377                 fsprefix += src[1:]
 378         }
 379
 380         if tokens == nil {
 381                 tokens = reqTokens
 382                 if h.Cluster.Users.AnonymousUserToken != "" {
 383                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 384                 }
 385         }
 386
 387         if len(targetPath) > 0 && targetPath[0] == "_" {
 388                 // If a collection has a directory called "t=foo" or
 389                 // "_", it can be served at
 390                 // //collections.example/_/t=foo/ or
 391                 // //collections.example/_/_/ respectively:
 392                 // //collections.example/t=foo/ won't work because
 393                 // t=foo will be interpreted as a token "foo".
 394                 targetPath = targetPath[1:]
 395                 stripParts++
 396         }
 397
 398         dirOpenMode := os.O_RDONLY
 399         if writeMethod[r.Method] {
 400                 dirOpenMode = os.O_RDWR
 401         }
 402
 403         var tokenValid bool
 404         var tokenScopeProblem bool
 405         var token string
 406         var tokenUser *arvados.User
 407         var sessionFS arvados.CustomFileSystem
 408         var session *cachedSession
 409         var collectionDir arvados.File
 410         for _, token = range tokens {
 411                 var statusErr errorWithHTTPStatus
 412                 fs, sess, user, err := h.Cache.GetSession(token)
 413                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 414                         // bad token
 415                         continue
 416                 } else if err != nil {
 417                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 418                         return
 419                 }
 420                 if token != h.Cluster.Users.AnonymousUserToken {
 421                         tokenValid = true
 422                 }
 423                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 424                 if errors.As(err, &statusErr) &&
 425                         statusErr.HTTPStatus() == http.StatusForbidden &&
 426                         token != h.Cluster.Users.AnonymousUserToken {
 427                         // collection id is outside scope of supplied
 428                         // token
 429                         tokenScopeProblem = true
 430                         sess.Release()
 431                         continue
 432                 } else if os.IsNotExist(err) {
 433                         // collection does not exist or is not
 434                         // readable using this token
 435                         sess.Release()
 436                         continue
 437                 } else if err != nil {
 438                         http.Error(w, err.Error(), http.StatusInternalServerError)
 439                         sess.Release()
 440                         return
 441                 }
 442                 defer f.Close()
 443
 444                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 445                 break
 446         }
 447
 448         // releaseSession() is equivalent to session.Release() except
 449         // that it's a no-op if (1) session is nil, or (2) it has
 450         // already been called.
 451         //
 452         // This way, we can do a defer call here to ensure it gets
 453         // called in all code paths, and also call it inline (see
 454         // below) in the cases where we want to release the lock
 455         // before returning.
 456         releaseSession := func() {}
 457         if session != nil {
 458                 var releaseSessionOnce sync.Once
 459                 releaseSession = func() { releaseSessionOnce.Do(func() { session.Release() }) }
 460         }
 461         defer releaseSession()
 462
 463         if forceReload && collectionDir != nil {
 464                 err := collectionDir.Sync()
 465                 if err != nil {
 466                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 467                                 http.Error(w, err.Error(), he.HTTPStatus())
 468                         } else {
 469                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 470                         }
 471                         return
 472                 }
 473         }
 474         if session == nil {
 475                 if pathToken {
 476                         // The URL is a "secret sharing link" that
 477                         // didn't work out.  Asking the client for
 478                         // additional credentials would just be
 479                         // confusing.
 480                         http.Error(w, notFoundMessage, http.StatusNotFound)
 481                         return
 482                 }
 483                 if tokenValid {
 484                         // The client provided valid token(s), but the
 485                         // collection was not found.
 486                         http.Error(w, notFoundMessage, http.StatusNotFound)
 487                         return
 488                 }
 489                 if tokenScopeProblem {
 490                         // The client provided a valid token but
 491                         // fetching a collection returned 401, which
 492                         // means the token scope doesn't permit
 493                         // fetching that collection.
 494                         http.Error(w, notFoundMessage, http.StatusForbidden)
 495                         return
 496                 }
 497                 // The client's token was invalid (e.g., expired), or
 498                 // the client didn't even provide one.  Redirect to
 499                 // workbench2's login-and-redirect-to-download url if
 500                 // this is a browser navigation request. (The redirect
 501                 // flow can't preserve the original method if it's not
 502                 // GET, and doesn't make sense if the UA is a
 503                 // command-line tool, is trying to load an inline
 504                 // image, etc.; in these cases, there's nothing we can
 505                 // do, so return 401 unauthorized.)
 506                 //
 507                 // Note Sec-Fetch-Mode is sent by all non-EOL
 508                 // browsers, except Safari.
 509                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 510                 //
 511                 // TODO(TC): This response would be confusing to
 512                 // someone trying (anonymously) to download public
 513                 // data that has been deleted.  Allow a referrer to
 514                 // provide this context somehow?
 515                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 516                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 517                         redirkey := "redirectToPreview"
 518                         if attachment {
 519                                 redirkey = "redirectToDownload"
 520                         }
 521                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 522                         query := url.Values{redirkey: {callback}}
 523                         queryString := query.Encode()
 524                         // Note: Encode (and QueryEscape function) turns space
 525                         // into plus sign (+) rather than %20 (the plus sign
 526                         // becomes %2B); that is the rule for web forms data
 527                         // sent in URL query part via GET, but we're not
 528                         // emulating forms here. Client JS APIs
 529                         // (URLSearchParam#get, decodeURIComponent) will
 530                         // decode %20, but while the former also expects the
 531                         // form-specific encoding, the latter doesn't.
 532                         // Encode() almost encodes everything; RFC3986 sec. 3.4
 533                         // says "it is sometimes better for usability" to not
 534                         // encode / and ? when passing URI reference in query.
 535                         // This is also legal according to WHATWG URL spec and
 536                         // can be desirable for debugging webapp.
 537                         // We can let slash / appear in the encoded query, and
 538                         // equality-sign = too, but exempting ? is not very
 539                         // useful.
 540                         // Plus-sign, hash, and ampersand are never exempt.
 541                         r := strings.NewReplacer("+", "%20", "%2F", "/", "%3D", "=")
 542                         target.RawQuery = r.Replace(queryString)
 543                         w.Header().Add("Location", target.String())
 544                         w.WriteHeader(http.StatusSeeOther)
 545                         return
 546                 }
 547                 if !credentialsOK {
 548                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 549                         return
 550                 }
 551                 // If none of the above cases apply, suggest the
 552                 // user-agent (which is either a non-browser agent
 553                 // like wget, or a browser that can't redirect through
 554                 // a login flow) prompt the user for credentials.
 555                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 556                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 557                 return
 558         }
 559
 560         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 561                 targetfnm := fsprefix + strings.Join(pathParts[stripParts:], "/")
 562                 if fi, err := sessionFS.Stat(targetfnm); err == nil && fi.IsDir() {
 563                         releaseSession() // because we won't be writing anything
 564                         if !strings.HasSuffix(r.URL.Path, "/") {
 565                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 566                         } else {
 567                                 h.serveDirectory(w, r, fi.Name(), sessionFS, targetfnm, !useSiteFS)
 568                         }
 569                         return
 570                 }
 571         }
 572
 573         var basename string
 574         if len(targetPath) > 0 {
 575                 basename = targetPath[len(targetPath)-1]
 576         }
 577         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 578                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 579                 return
 580         }
 581         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 582                 http.Error(w, "Not permitted", http.StatusForbidden)
 583                 return
 584         }
 585         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fsprefix+strings.Join(targetPath, "/"), nil, tokenUser)
 586
 587         writing := writeMethod[r.Method]
 588         locker := h.collectionLock(collectionID, writing)
 589         defer locker.Unlock()
 590
 591         if writing {
 592                 // Save the collection only if/when all
 593                 // webdav->filesystem operations succeed --
 594                 // and send a 500 error if the modified
 595                 // collection can't be saved.
 596                 //
 597                 // Perform the write in a separate sitefs, so
 598                 // concurrent read operations on the same
 599                 // collection see the previous saved
 600                 // state. After the write succeeds and the
 601                 // collection record is updated, we reset the
 602                 // session so the updates are visible in
 603                 // subsequent read requests.
 604                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 605                 sessionFS = client.SiteFileSystem(session.keepclient)
 606                 writingDir, err := sessionFS.OpenFile(fsprefix, os.O_RDONLY, 0)
 607                 if err != nil {
 608                         http.Error(w, err.Error(), http.StatusInternalServerError)
 609                         return
 610                 }
 611                 defer writingDir.Close()
 612                 w = &updateOnSuccess{
 613                         ResponseWriter: w,
 614                         logger:         ctxlog.FromContext(r.Context()),
 615                         update: func() error {
 616                                 err := writingDir.Sync()
 617                                 var te arvados.TransactionError
 618                                 if errors.As(err, &te) {
 619                                         err = te
 620                                 }
 621                                 if err != nil {
 622                                         return err
 623                                 }
 624                                 // Sync the changes to the persistent
 625                                 // sessionfs for this token.
 626                                 snap, err := writingDir.Snapshot()
 627                                 if err != nil {
 628                                         return err
 629                                 }
 630                                 collectionDir.Splice(snap)
 631                                 return nil
 632                         }}
 633         } else {
 634                 // When writing, we need to block session renewal
 635                 // until we're finished, in order to guarantee the
 636                 // effect of the write is visible in future responses.
 637                 // But if we're not writing, we can release the lock
 638                 // early.  This enables us to keep renewing sessions
 639                 // and processing more requests even if a slow client
 640                 // takes a long time to download a large file.
 641                 releaseSession()
 642         }
 643         if r.Method == http.MethodGet {
 644                 applyContentDispositionHdr(w, r, basename, attachment)
 645         }
 646         if webdavPrefix == "" {
 647                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 648         }
 649         wh := &webdav.Handler{
 650                 Prefix: webdavPrefix,
 651                 FileSystem: &webdavfs.FS{
 652                         FileSystem:    sessionFS,
 653                         Prefix:        fsprefix,
 654                         Writing:       writeMethod[r.Method],
 655                         AlwaysReadEOF: r.Method == "PROPFIND",
 656                 },
 657                 LockSystem: webdavfs.NoLockSystem,
 658                 Logger: func(r *http.Request, err error) {
 659                         if err != nil && !os.IsNotExist(err) {
 660                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 661                         }
 662                 },
 663         }
 664         h.metrics.track(wh, w, r)
 665         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 666                 wrote := int64(w.WroteBodyBytes())
 667                 fnm := strings.Join(pathParts[stripParts:], "/")
 668                 fi, err := wh.FileSystem.Stat(r.Context(), fnm)
 669                 if err == nil && fi.Size() != wrote {
 670                         var n int
 671                         f, err := wh.FileSystem.OpenFile(r.Context(), fnm, os.O_RDONLY, 0)
 672                         if err == nil {
 673                                 n, err = f.Read(make([]byte, 1024))
 674                                 f.Close()
 675                         }
 676                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 677                 }
 678         }
 679 }
 680
 681 var dirListingTemplate = `<!DOCTYPE HTML>
 682 <HTML><HEAD>
 683   <META name="robots" content="NOINDEX">
 684   <TITLE>{{ .CollectionName }}</TITLE>
 685   <STYLE type="text/css">
 686     body {
 687       margin: 1.5em;
 688     }
 689     pre {
 690       background-color: #D9EDF7;
 691       border-radius: .25em;
 692       padding: .75em;
 693       overflow: auto;
 694     }
 695     .footer p {
 696       font-size: 82%;
 697     }
 698     hr {
 699       border: 1px solid #808080;
 700     }
 701     ul {
 702       padding: 0;
 703     }
 704     ul li {
 705       font-family: monospace;
 706       list-style: none;
 707     }
 708   </STYLE>
 709 </HEAD>
 710 <BODY>
 711
 712 <H1>{{ .CollectionName }}</H1>
 713
 714 <P>This collection of data files is being shared with you through
 715 Arvados.  You can download individual files listed below.  To download
 716 the entire directory tree with <CODE>wget</CODE>, try:</P>
 717
 718 <PRE id="wget-example">$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} {{ .QuotedUrlForWget }}</PRE>
 719
 720 <H2>File Listing</H2>
 721
 722 {{if .Files}}
 723 <UL>
 724 {{range .Files}}
 725 {{if .IsDir }}
 726   <LI>{{" " | printf "%15s  " | nbsp}}<A class="item" href="{{ .Href }}/">{{ .Name }}/</A></LI>
 727 {{else}}
 728   <LI>{{.Size | printf "%15d  " | nbsp}}<A class="item" href="{{ .Href }}">{{ .Name }}</A></LI>
 729 {{end}}
 730 {{end}}
 731 </UL>
 732 {{else}}
 733 <P>(No files; this collection is empty.)</P>
 734 {{end}}
 735
 736 <HR>
 737 <DIV class="footer">
 738   <P>
 739     About Arvados:
 740     Arvados is a free and open source software bioinformatics platform.
 741     To learn more, visit arvados.org.
 742     Arvados is not responsible for the files listed on this page.
 743   </P>
 744 </DIV>
 745
 746 </BODY>
 747 </HTML>
 748 `
 749
 750 type fileListEnt struct {
 751         Name  string
 752         Href  string
 753         Size  int64
 754         IsDir bool
 755 }
 756
 757 // Given a filesystem path like `foo/"bar baz"`, return an escaped
 758 // (percent-encoded) relative path like `./foo/%22bar%20%baz%22`.
 759 //
 760 // Note the result may contain html-unsafe characters like '&'. These
 761 // will be handled separately by the HTML templating engine as needed.
 762 func relativeHref(path string) string {
 763         u := &url.URL{Path: path}
 764         return "./" + u.EscapedPath()
 765 }
 766
 767 // Return a shell-quoted URL suitable for pasting to a command line
 768 // ("wget ...") to repeat the given HTTP request.
 769 func makeQuotedUrlForWget(r *http.Request) string {
 770         scheme := r.Header.Get("X-Forwarded-Proto")
 771         if scheme == "http" || scheme == "https" {
 772                 // use protocol reported by load balancer / proxy
 773         } else if r.TLS != nil {
 774                 scheme = "https"
 775         } else {
 776                 scheme = "http"
 777         }
 778         p := r.URL.EscapedPath()
 779         // An escaped path may still contain single quote chars, which
 780         // would interfere with our shell quoting. Avoid this by
 781         // escaping them as %27.
 782         return fmt.Sprintf("'%s://%s%s'", scheme, r.Host, strings.Replace(p, "'", "%27", -1))
 783 }
 784
 785 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 786         var files []fileListEnt
 787         var walk func(string) error
 788         if !strings.HasSuffix(base, "/") {
 789                 base = base + "/"
 790         }
 791         walk = func(path string) error {
 792                 dirname := base + path
 793                 if dirname != "/" {
 794                         dirname = strings.TrimSuffix(dirname, "/")
 795                 }
 796                 d, err := fs.Open(dirname)
 797                 if err != nil {
 798                         return err
 799                 }
 800                 ents, err := d.Readdir(-1)
 801                 if err != nil {
 802                         return err
 803                 }
 804                 for _, ent := range ents {
 805                         if recurse && ent.IsDir() {
 806                                 err = walk(path + ent.Name() + "/")
 807                                 if err != nil {
 808                                         return err
 809                                 }
 810                         } else {
 811                                 listingName := path + ent.Name()
 812                                 files = append(files, fileListEnt{
 813                                         Name:  listingName,
 814                                         Href:  relativeHref(listingName),
 815                                         Size:  ent.Size(),
 816                                         IsDir: ent.IsDir(),
 817                                 })
 818                         }
 819                 }
 820                 return nil
 821         }
 822         if err := walk(""); err != nil {
 823                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 824                 return
 825         }
 826
 827         funcs := template.FuncMap{
 828                 "nbsp": func(s string) template.HTML {
 829                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 830                 },
 831         }
 832         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 833         if err != nil {
 834                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 835                 return
 836         }
 837         sort.Slice(files, func(i, j int) bool {
 838                 return files[i].Name < files[j].Name
 839         })
 840         w.WriteHeader(http.StatusOK)
 841         tmpl.Execute(w, map[string]interface{}{
 842                 "CollectionName":   collectionName,
 843                 "Files":            files,
 844                 "Request":          r,
 845                 "StripParts":       strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 846                 "QuotedUrlForWget": makeQuotedUrlForWget(r),
 847         })
 848 }
 849
 850 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 851         disposition := "inline"
 852         if isAttachment {
 853                 disposition = "attachment"
 854         }
 855         if strings.ContainsRune(r.RequestURI, '?') {
 856                 // Help the UA realize that the filename is just
 857                 // "filename.txt", not
 858                 // "filename.txt?disposition=attachment".
 859                 //
 860                 // TODO(TC): Follow advice at RFC 6266 appendix D
 861                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 862         }
 863         if disposition != "inline" {
 864                 w.Header().Set("Content-Disposition", disposition)
 865         }
 866 }
 867
 868 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 869         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
 870                 if !credentialsOK {
 871                         // It is not safe to copy the provided token
 872                         // into a cookie unless the current vhost
 873                         // (origin) serves only a single collection or
 874                         // we are in TrustAllContent mode.
 875                         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)
 876                         return
 877                 }
 878
 879                 // The HttpOnly flag is necessary to prevent
 880                 // JavaScript code (included in, or loaded by, a page
 881                 // in the collection being served) from employing the
 882                 // user's token beyond reading other files in the same
 883                 // domain, i.e., same collection.
 884                 //
 885                 // The 303 redirect is necessary in the case of a GET
 886                 // request to avoid exposing the token in the Location
 887                 // bar, and in the case of a POST request to avoid
 888                 // raising warnings when the user refreshes the
 889                 // resulting page.
 890                 for _, tok := range formTokens {
 891                         if tok == "" {
 892                                 continue
 893                         }
 894                         http.SetCookie(w, &http.Cookie{
 895                                 Name:     "arvados_api_token",
 896                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
 897                                 Path:     "/",
 898                                 HttpOnly: true,
 899                                 SameSite: http.SameSiteLaxMode,
 900                         })
 901                         break
 902                 }
 903         }
 904
 905         // Propagate query parameters (except api_token) from
 906         // the original request.
 907         redirQuery := r.URL.Query()
 908         redirQuery.Del("api_token")
 909
 910         u := r.URL
 911         if location != "" {
 912                 newu, err := u.Parse(location)
 913                 if err != nil {
 914                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
 915                         return
 916                 }
 917                 u = newu
 918         }
 919         redir := (&url.URL{
 920                 Scheme:   r.URL.Scheme,
 921                 Host:     r.Host,
 922                 Path:     u.Path,
 923                 RawQuery: redirQuery.Encode(),
 924         }).String()
 925
 926         w.Header().Add("Location", redir)
 927         w.WriteHeader(http.StatusSeeOther)
 928         io.WriteString(w, `<A href="`)
 929         io.WriteString(w, html.EscapeString(redir))
 930         io.WriteString(w, `">Continue</A>`)
 931 }
 932
 933 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
 934         var permitDownload bool
 935         var permitUpload bool
 936         if tokenUser != nil && tokenUser.IsAdmin {
 937                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
 938                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
 939         } else {
 940                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
 941                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
 942         }
 943         if (method == "PUT" || method == "POST") && !permitUpload {
 944                 // Disallow operations that upload new files.
 945                 // Permit webdav operations that move existing files around.
 946                 return false
 947         } else if method == "GET" && !permitDownload {
 948                 // Disallow downloading file contents.
 949                 // Permit webdav operations like PROPFIND that retrieve metadata
 950                 // but not file contents.
 951                 return false
 952         }
 953         return true
 954 }
 955
 956 type fileEventLog struct {
 957         requestPath  string
 958         eventType    string
 959         userUUID     string
 960         userFullName string
 961         collUUID     string
 962         collPDH      string
 963         collFilePath string
 964         clientAddr   string
 965         clientToken  string
 966 }
 967
 968 func newFileEventLog(
 969         h *handler,
 970         r *http.Request,
 971         filepath string,
 972         collection *arvados.Collection,
 973         user *arvados.User,
 974         token string,
 975 ) *fileEventLog {
 976         var eventType string
 977         switch r.Method {
 978         case "POST", "PUT":
 979                 eventType = "file_upload"
 980         case "GET":
 981                 eventType = "file_download"
 982         default:
 983                 return nil
 984         }
 985
 986         // We want to log the address of the proxy closest to keep-web—the last
 987         // value in the X-Forwarded-For list—or the client address if there is no
 988         // valid proxy.
 989         var clientAddr string
 990         // 1. Build a slice of proxy addresses from X-Forwarded-For.
 991         xff := strings.Join(r.Header.Values("X-Forwarded-For"), ",")
 992         addrs := strings.Split(xff, ",")
 993         // 2. Reverse the slice so it's in our most preferred order for logging.
 994         slices.Reverse(addrs)
 995         // 3. Append the client address to that slice.
 996         if addr, _, err := net.SplitHostPort(r.RemoteAddr); err == nil {
 997                 addrs = append(addrs, addr)
 998         }
 999         // 4. Use the first valid address in the slice.
1000         for _, addr := range addrs {
1001                 if ip := net.ParseIP(strings.TrimSpace(addr)); ip != nil {
1002                         clientAddr = ip.String()
1003                         break
1004                 }
1005         }
1006
1007         ev := &fileEventLog{
1008                 requestPath: r.URL.Path,
1009                 eventType:   eventType,
1010                 clientAddr:  clientAddr,
1011                 clientToken: token,
1012         }
1013
1014         if user != nil {
1015                 ev.userUUID = user.UUID
1016                 ev.userFullName = user.FullName
1017         } else {
1018                 ev.userUUID = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
1019         }
1020
1021         if collection != nil {
1022                 ev.collFilePath = filepath
1023                 // h.determineCollection populates the collection_uuid
1024                 // prop with the PDH, if this collection is being
1025                 // accessed via PDH. For logging, we use a different
1026                 // field depending on whether it's a UUID or PDH.
1027                 if len(collection.UUID) > 32 {
1028                         ev.collPDH = collection.UUID
1029                 } else {
1030                         ev.collPDH = collection.PortableDataHash
1031                         ev.collUUID = collection.UUID
1032                 }
1033         }
1034
1035         return ev
1036 }
1037
1038 func (ev *fileEventLog) shouldLogPDH() bool {
1039         return ev.eventType == "file_download" && ev.collPDH != ""
1040 }
1041
1042 func (ev *fileEventLog) asDict() arvadosclient.Dict {
1043         props := arvadosclient.Dict{
1044                 "reqPath":              ev.requestPath,
1045                 "collection_uuid":      ev.collUUID,
1046                 "collection_file_path": ev.collFilePath,
1047         }
1048         if ev.shouldLogPDH() {
1049                 props["portable_data_hash"] = ev.collPDH
1050         }
1051         return arvadosclient.Dict{
1052                 "object_uuid": ev.userUUID,
1053                 "event_type":  ev.eventType,
1054                 "properties":  props,
1055         }
1056 }
1057
1058 func (ev *fileEventLog) asFields() logrus.Fields {
1059         fields := logrus.Fields{
1060                 "collection_file_path": ev.collFilePath,
1061                 "collection_uuid":      ev.collUUID,
1062                 "user_uuid":            ev.userUUID,
1063         }
1064         if ev.shouldLogPDH() {
1065                 fields["portable_data_hash"] = ev.collPDH
1066         }
1067         if !strings.HasSuffix(ev.userUUID, "-tpzed-anonymouspublic") {
1068                 fields["user_full_name"] = ev.userFullName
1069         }
1070         return fields
1071 }
1072
1073 func (h *handler) shouldLogEvent(event *fileEventLog, t time.Time) bool {
1074         if event == nil {
1075                 return false
1076         } else if event.eventType != "file_download" ||
1077                 h.Cluster.Collections.WebDAVLogDownloadInterval == 0 {
1078                 return true
1079         }
1080         td := h.Cluster.Collections.WebDAVLogDownloadInterval.Duration()
1081         cutoff := t.Add(-td)
1082         ev := *event
1083         h.fileEventLogsMtx.Lock()
1084         defer h.fileEventLogsMtx.Unlock()
1085         if h.fileEventLogs == nil {
1086                 h.fileEventLogs = make(map[fileEventLog]time.Time)
1087         }
1088         shouldLog := h.fileEventLogs[ev].Before(cutoff)
1089         if shouldLog {
1090                 h.fileEventLogs[ev] = t
1091         }
1092         if t.After(h.fileEventLogsNextTidy) {
1093                 for key, logTime := range h.fileEventLogs {
1094                         if logTime.Before(cutoff) {
1095                                 delete(h.fileEventLogs, key)
1096                         }
1097                 }
1098                 h.fileEventLogsNextTidy = t.Add(td)
1099         }
1100         return shouldLog
1101 }
1102
1103 func (h *handler) logUploadOrDownload(
1104         r *http.Request,
1105         client *arvadosclient.ArvadosClient,
1106         fs arvados.CustomFileSystem,
1107         filepath string,
1108         collection *arvados.Collection,
1109         user *arvados.User,
1110 ) {
1111         if collection == nil && fs != nil {
1112                 collection, filepath = h.determineCollection(fs, filepath)
1113         }
1114         event := newFileEventLog(h, r, filepath, collection, user, client.ApiToken)
1115         if !h.shouldLogEvent(event, time.Now()) {
1116                 return
1117         }
1118         log := ctxlog.FromContext(r.Context()).WithFields(event.asFields())
1119         log.Info(strings.Replace(event.eventType, "file_", "File ", 1))
1120         if h.Cluster.Collections.WebDAVLogEvents {
1121                 go func() {
1122                         logReq := arvadosclient.Dict{"log": event.asDict()}
1123                         err := client.Create("logs", logReq, nil)
1124                         if err != nil {
1125                                 log.WithError(err).Errorf("Failed to create %s log event on API server", event.eventType)
1126                         }
1127                 }()
1128         }
1129 }
1130
1131 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
1132         target := strings.TrimSuffix(path, "/")
1133         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
1134                 target = target[:cut]
1135                 fi, err := fs.Stat(target)
1136                 if os.IsNotExist(err) {
1137                         // creating a new file/dir, or download
1138                         // destined to fail
1139                         continue
1140                 } else if err != nil {
1141                         return nil, ""
1142                 }
1143                 switch src := fi.Sys().(type) {
1144                 case *arvados.Collection:
1145                         return src, strings.TrimPrefix(path[len(target):], "/")
1146                 case *arvados.Group:
1147                         return nil, ""
1148                 default:
1149                         if _, ok := src.(error); ok {
1150                                 return nil, ""
1151                         }
1152                 }
1153         }
1154         return nil, ""
1155 }
1156
1157 var lockTidyInterval = time.Minute * 10
1158
1159 // Lock the specified collection for reading or writing. Caller must
1160 // call Unlock() on the returned Locker when the operation is
1161 // finished.
1162 func (h *handler) collectionLock(collectionID string, writing bool) sync.Locker {
1163         h.lockMtx.Lock()
1164         defer h.lockMtx.Unlock()
1165         if time.Since(h.lockTidied) > lockTidyInterval {
1166                 // Periodically delete all locks that aren't in use.
1167                 h.lockTidied = time.Now()
1168                 for id, locker := range h.lock {
1169                         if locker.TryLock() {
1170                                 locker.Unlock()
1171                                 delete(h.lock, id)
1172                         }
1173                 }
1174         }
1175         locker := h.lock[collectionID]
1176         if locker == nil {
1177                 locker = new(sync.RWMutex)
1178                 if h.lock == nil {
1179                         h.lock = map[string]*sync.RWMutex{}
1180                 }
1181                 h.lock[collectionID] = locker
1182         }
1183         if writing {
1184                 locker.Lock()
1185                 return locker
1186         } else {
1187                 locker.RLock()
1188                 return locker.RLocker()
1189         }
1190 }
1191
1192 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1193         method := header.Get("Access-Control-Request-Method")
1194         if method == "" {
1195                 return false
1196         }
1197         if !browserMethod[method] && !webdavMethod[method] {
1198                 w.WriteHeader(http.StatusMethodNotAllowed)
1199                 return true
1200         }
1201         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1202         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1203         w.Header().Set("Access-Control-Allow-Origin", "*")
1204         w.Header().Set("Access-Control-Max-Age", "86400")
1205         return true
1206 }