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         "context"
   9         "encoding/json"
  10         "errors"
  11         "fmt"
  12         "html"
  13         "html/template"
  14         "io"
  15         "net"
  16         "net/http"
  17         "net/url"
  18         "os"
  19         "path"
  20         "slices"
  21         "sort"
  22         "strconv"
  23         "strings"
  24         "sync"
  25         "time"
  26
  27         "git.arvados.org/arvados.git/lib/cmd"
  28         "git.arvados.org/arvados.git/lib/ctrlctx"
  29         "git.arvados.org/arvados.git/lib/webdavfs"
  30         "git.arvados.org/arvados.git/sdk/go/arvados"
  31         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  32         "git.arvados.org/arvados.git/sdk/go/auth"
  33         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  34         "git.arvados.org/arvados.git/sdk/go/httpserver"
  35         "github.com/gotd/contrib/http_range"
  36         "github.com/sirupsen/logrus"
  37         "golang.org/x/net/webdav"
  38 )
  39
  40 type handler struct {
  41         Cache   cache
  42         Cluster *arvados.Cluster
  43         metrics *metrics
  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         dbConnector    *ctrlctx.DBConnector
  54         dbConnectorMtx sync.Mutex
  55 }
  56
  57 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  58
  59 var notFoundMessage = "Not Found"
  60 var unauthorizedMessage = "401 Unauthorized\n\nA valid Arvados token must be provided to access this resource."
  61
  62 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  63 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  64 // otherwise "".
  65 func parseCollectionIDFromURL(s string) string {
  66         if arvadosclient.UUIDMatch(s) {
  67                 return s
  68         }
  69         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  70                 return pdh
  71         }
  72         return ""
  73 }
  74
  75 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  76         json.NewEncoder(w).Encode(struct{ Version string }{cmd.Version.String()})
  77 }
  78
  79 type errorWithHTTPStatus interface {
  80         HTTPStatus() int
  81 }
  82
  83 // updateOnSuccess wraps httpserver.ResponseWriter. If the handler
  84 // sends an HTTP header indicating success, updateOnSuccess first
  85 // calls the provided update func. If the update func fails, an error
  86 // response is sent (using the error's HTTP status or 500 if none),
  87 // and the status code and body sent by the handler are ignored (all
  88 // response writes return the update error).
  89 type updateOnSuccess struct {
  90         httpserver.ResponseWriter
  91         logger     logrus.FieldLogger
  92         update     func() error
  93         sentHeader bool
  94         err        error
  95 }
  96
  97 func (uos *updateOnSuccess) Write(p []byte) (int, error) {
  98         if !uos.sentHeader {
  99                 uos.WriteHeader(http.StatusOK)
 100         }
 101         if uos.err != nil {
 102                 return 0, uos.err
 103         }
 104         return uos.ResponseWriter.Write(p)
 105 }
 106
 107 func (uos *updateOnSuccess) WriteHeader(code int) {
 108         if !uos.sentHeader {
 109                 uos.sentHeader = true
 110                 if code >= 200 && code < 400 {
 111                         if uos.err = uos.update(); uos.err != nil {
 112                                 code := http.StatusInternalServerError
 113                                 if he := errorWithHTTPStatus(nil); errors.As(uos.err, &he) {
 114                                         code = he.HTTPStatus()
 115                                 }
 116                                 uos.logger.WithError(uos.err).Errorf("update() returned %T error, changing response to HTTP %d", uos.err, code)
 117                                 http.Error(uos.ResponseWriter, uos.err.Error(), code)
 118                                 return
 119                         }
 120                 }
 121         }
 122         uos.ResponseWriter.WriteHeader(code)
 123 }
 124
 125 var (
 126         corsAllowHeadersHeader = strings.Join([]string{
 127                 "Authorization", "Content-Type", "Range",
 128                 // WebDAV request headers:
 129                 "Depth", "Destination", "If", "Lock-Token", "Overwrite", "Timeout", "Cache-Control",
 130         }, ", ")
 131         writeMethod = map[string]bool{
 132                 "COPY":      true,
 133                 "DELETE":    true,
 134                 "LOCK":      true,
 135                 "MKCOL":     true,
 136                 "MOVE":      true,
 137                 "PROPPATCH": true,
 138                 "PUT":       true,
 139                 "UNLOCK":    true,
 140         }
 141         webdavMethod = map[string]bool{
 142                 "COPY":      true,
 143                 "DELETE":    true,
 144                 "LOCK":      true,
 145                 "MKCOL":     true,
 146                 "MOVE":      true,
 147                 "OPTIONS":   true,
 148                 "PROPFIND":  true,
 149                 "PROPPATCH": true,
 150                 "PUT":       true,
 151                 "RMCOL":     true,
 152                 "UNLOCK":    true,
 153         }
 154         browserMethod = map[string]bool{
 155                 "GET":  true,
 156                 "HEAD": true,
 157                 "POST": true,
 158         }
 159         // top-level dirs to serve with siteFS
 160         siteFSDir = map[string]bool{
 161                 "":      true, // root directory
 162                 "by_id": true,
 163                 "users": true,
 164         }
 165 )
 166
 167 func stripDefaultPort(host string) string {
 168         // Will consider port 80 and port 443 to be the same vhost.  I think that's fine.
 169         u := &url.URL{Host: host}
 170         if p := u.Port(); p == "80" || p == "443" {
 171                 return strings.ToLower(u.Hostname())
 172         } else {
 173                 return strings.ToLower(host)
 174         }
 175 }
 176
 177 // CheckHealth implements service.Handler.
 178 func (h *handler) CheckHealth() error {
 179         return nil
 180 }
 181
 182 // Done implements service.Handler.
 183 func (h *handler) Done() <-chan struct{} {
 184         return nil
 185 }
 186
 187 func (h *handler) getDBConnector() *ctrlctx.DBConnector {
 188         h.dbConnectorMtx.Lock()
 189         defer h.dbConnectorMtx.Unlock()
 190         if h.dbConnector == nil {
 191                 h.dbConnector = &ctrlctx.DBConnector{PostgreSQL: h.Cluster.PostgreSQL}
 192         }
 193         return h.dbConnector
 194 }
 195
 196 // ServeHTTP implements http.Handler.
 197 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 198         if xfp := r.Header.Get("X-Forwarded-Proto"); xfp != "" && xfp != "http" {
 199                 r.URL.Scheme = xfp
 200         }
 201
 202         httpserver.SetResponseLogFields(r.Context(), logrus.Fields{
 203                 "webdavDepth":       r.Header.Get("Depth"),
 204                 "webdavDestination": r.Header.Get("Destination"),
 205                 "webdavOverwrite":   r.Header.Get("Overwrite"),
 206         })
 207
 208         wbuffer := newWriteBuffer(wOrig, int(h.Cluster.Collections.WebDAVOutputBuffer))
 209         defer wbuffer.Close()
 210         w := httpserver.WrapResponseWriter(responseWriter{
 211                 Writer:         wbuffer,
 212                 ResponseWriter: wOrig,
 213         })
 214
 215         if r.Method == "OPTIONS" && ServeCORSPreflight(w, r.Header) {
 216                 return
 217         }
 218
 219         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 220                 w.WriteHeader(http.StatusMethodNotAllowed)
 221                 return
 222         }
 223
 224         if r.Header.Get("Origin") != "" {
 225                 // Allow simple cross-origin requests without user
 226                 // credentials ("user credentials" as defined by CORS,
 227                 // i.e., cookies, HTTP authentication, and client-side
 228                 // SSL certificates. See
 229                 // http://www.w3.org/TR/cors/#user-credentials).
 230                 w.Header().Set("Access-Control-Allow-Origin", "*")
 231                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 232         }
 233
 234         if h.serveS3(w, r) {
 235                 return
 236         }
 237
 238         // webdavPrefix is the leading portion of r.URL.Path that
 239         // should be ignored by the webdav handler, if any.
 240         //
 241         // req "/c={id}/..." -> webdavPrefix "/c={id}"
 242         // req "/by_id/..." -> webdavPrefix ""
 243         //
 244         // Note: in the code immediately below, we set webdavPrefix
 245         // only if it was explicitly set by the client. Otherwise, it
 246         // gets set later, after checking the request path for cases
 247         // like "/c={id}/...".
 248         webdavPrefix := ""
 249         arvPath := r.URL.Path
 250         if prefix := r.Header.Get("X-Webdav-Prefix"); prefix != "" {
 251                 // Enable a proxy (e.g., container log handler in
 252                 // controller) to satisfy a request for path
 253                 // "/foo/bar/baz.txt" using content from
 254                 // "//abc123-4.internal/bar/baz.txt", by adding a
 255                 // request header "X-Webdav-Prefix: /foo"
 256                 if !strings.HasPrefix(arvPath, prefix) {
 257                         http.Error(w, "X-Webdav-Prefix header is not a prefix of the requested path", http.StatusBadRequest)
 258                         return
 259                 }
 260                 arvPath = r.URL.Path[len(prefix):]
 261                 if arvPath == "" {
 262                         arvPath = "/"
 263                 }
 264                 w.Header().Set("Vary", "X-Webdav-Prefix, "+w.Header().Get("Vary"))
 265                 webdavPrefix = prefix
 266         }
 267         pathParts := strings.Split(arvPath[1:], "/")
 268
 269         var stripParts int
 270         var collectionID string
 271         var tokens []string
 272         var reqTokens []string
 273         var pathToken bool
 274         var attachment bool
 275         var useSiteFS bool
 276         credentialsOK := h.Cluster.Collections.TrustAllContent
 277         reasonNotAcceptingCredentials := ""
 278
 279         if r.Host != "" && stripDefaultPort(r.Host) == stripDefaultPort(h.Cluster.Services.WebDAVDownload.ExternalURL.Host) {
 280                 credentialsOK = true
 281                 attachment = true
 282         } else if r.FormValue("disposition") == "attachment" {
 283                 attachment = true
 284         }
 285
 286         if !credentialsOK {
 287                 reasonNotAcceptingCredentials = fmt.Sprintf("vhost %q does not specify a single collection ID or match Services.WebDAVDownload.ExternalURL %q, and Collections.TrustAllContent is false",
 288                         r.Host, h.Cluster.Services.WebDAVDownload.ExternalURL)
 289         }
 290
 291         if collectionID = arvados.CollectionIDFromDNSName(r.Host); collectionID != "" {
 292                 // http://ID.collections.example/PATH...
 293                 credentialsOK = true
 294         } else if r.URL.Path == "/status.json" {
 295                 h.serveStatus(w, r)
 296                 return
 297         } else if siteFSDir[pathParts[0]] {
 298                 useSiteFS = true
 299         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 300                 // /c=ID[/PATH...]
 301                 collectionID = parseCollectionIDFromURL(pathParts[0][2:])
 302                 stripParts = 1
 303         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 304                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 305                         // /collections/download/ID/TOKEN/PATH...
 306                         collectionID = parseCollectionIDFromURL(pathParts[2])
 307                         tokens = []string{pathParts[3]}
 308                         stripParts = 4
 309                         pathToken = true
 310                 } else {
 311                         // /collections/ID/PATH...
 312                         collectionID = parseCollectionIDFromURL(pathParts[1])
 313                         stripParts = 2
 314                         // This path is only meant to work for public
 315                         // data. Tokens provided with the request are
 316                         // ignored.
 317                         credentialsOK = false
 318                         reasonNotAcceptingCredentials = "the '/collections/UUID/PATH' form only works for public data"
 319                 }
 320         }
 321
 322         forceReload := false
 323         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 324                 forceReload = true
 325         }
 326
 327         if credentialsOK {
 328                 reqTokens = auth.CredentialsFromRequest(r).Tokens
 329         }
 330
 331         r.ParseForm()
 332         origin := r.Header.Get("Origin")
 333         cors := origin != "" && !strings.HasSuffix(origin, "://"+r.Host)
 334         safeAjax := cors && (r.Method == http.MethodGet || r.Method == http.MethodHead)
 335         // Important distinction: safeAttachment checks whether api_token exists
 336         // as a query parameter. haveFormTokens checks whether api_token exists
 337         // as request form data *or* a query parameter. Different checks are
 338         // necessary because both the request disposition and the location of
 339         // the API token affect whether or not the request needs to be
 340         // redirected. The different branch comments below explain further.
 341         safeAttachment := attachment && !r.URL.Query().Has("api_token")
 342         if formTokens, haveFormTokens := r.Form["api_token"]; !haveFormTokens {
 343                 // No token to use or redact.
 344         } else if safeAjax || safeAttachment {
 345                 // If this is a cross-origin request, the URL won't
 346                 // appear in the browser's address bar, so
 347                 // substituting a clipboard-safe URL is pointless.
 348                 // Redirect-with-cookie wouldn't work anyway, because
 349                 // it's not safe to allow third-party use of our
 350                 // cookie.
 351                 //
 352                 // If we're supplying an attachment, we don't need to
 353                 // convert POST to GET to avoid the "really resubmit
 354                 // form?" problem, so provided the token isn't
 355                 // embedded in the URL, there's no reason to do
 356                 // redirect-with-cookie in this case either.
 357                 for _, tok := range formTokens {
 358                         reqTokens = append(reqTokens, tok)
 359                 }
 360         } else if browserMethod[r.Method] {
 361                 // If this is a page view, and the client provided a
 362                 // token via query string or POST body, we must put
 363                 // the token in an HttpOnly cookie, and redirect to an
 364                 // equivalent URL with the query param redacted and
 365                 // method = GET.
 366                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 367                 return
 368         }
 369
 370         targetPath := pathParts[stripParts:]
 371         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 372                 // http://ID.example/t=TOKEN/PATH...
 373                 // /c=ID/t=TOKEN/PATH...
 374                 //
 375                 // This form must only be used to pass scoped tokens
 376                 // that give permission for a single collection. See
 377                 // FormValue case above.
 378                 tokens = []string{targetPath[0][2:]}
 379                 pathToken = true
 380                 targetPath = targetPath[1:]
 381                 stripParts++
 382         }
 383
 384         // fsprefix is the path from sitefs root to the sitefs
 385         // directory (implicitly or explicitly) indicated by the
 386         // leading / in the request path.
 387         //
 388         // Request "/by_id/..." -> fsprefix ""
 389         // Request "/c={id}/..." -> fsprefix "/by_id/{id}/"
 390         fsprefix := ""
 391         if useSiteFS {
 392                 if writeMethod[r.Method] {
 393                         http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 394                         return
 395                 }
 396                 if len(reqTokens) == 0 {
 397                         w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 398                         http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 399                         return
 400                 }
 401                 tokens = reqTokens
 402         } else if collectionID == "" {
 403                 http.Error(w, notFoundMessage, http.StatusNotFound)
 404                 return
 405         } else {
 406                 fsprefix = "by_id/" + collectionID + "/"
 407         }
 408
 409         if src := r.Header.Get("X-Webdav-Source"); strings.HasPrefix(src, "/") && !strings.Contains(src, "//") && !strings.Contains(src, "/../") {
 410                 // Clients (specifically, the container log gateway)
 411                 // use X-Webdav-Source to specify that although the
 412                 // request path (and other webdav fields in the
 413                 // request) refer to target "/abc", the intended
 414                 // target is actually
 415                 // "{x-webdav-source-value}/abc".
 416                 //
 417                 // This, combined with X-Webdav-Prefix, enables the
 418                 // container log gateway to effectively alter the
 419                 // target path when proxying a request, without
 420                 // needing to rewrite all the other webdav
 421                 // request/response fields that might mention the
 422                 // target path.
 423                 fsprefix += src[1:]
 424         }
 425
 426         if tokens == nil {
 427                 tokens = reqTokens
 428                 if h.Cluster.Users.AnonymousUserToken != "" {
 429                         tokens = append(tokens, h.Cluster.Users.AnonymousUserToken)
 430                 }
 431         }
 432
 433         if len(targetPath) > 0 && targetPath[0] == "_" {
 434                 // If a collection has a directory called "t=foo" or
 435                 // "_", it can be served at
 436                 // //collections.example/_/t=foo/ or
 437                 // //collections.example/_/_/ respectively:
 438                 // //collections.example/t=foo/ won't work because
 439                 // t=foo will be interpreted as a token "foo".
 440                 targetPath = targetPath[1:]
 441                 stripParts++
 442         }
 443
 444         dirOpenMode := os.O_RDONLY
 445         if writeMethod[r.Method] {
 446                 dirOpenMode = os.O_RDWR
 447         }
 448
 449         var tokenValid bool
 450         var tokenScopeProblem bool
 451         var token string
 452         var tokenUser *arvados.User
 453         var sessionFS arvados.CustomFileSystem
 454         var targetFS arvados.FileSystem
 455         var session *cachedSession
 456         var collectionDir arvados.File
 457         for _, token = range tokens {
 458                 var statusErr errorWithHTTPStatus
 459                 fs, sess, user, err := h.Cache.GetSession(token)
 460                 if errors.As(err, &statusErr) && statusErr.HTTPStatus() == http.StatusUnauthorized {
 461                         // bad token
 462                         continue
 463                 } else if err != nil {
 464                         http.Error(w, "cache error: "+err.Error(), http.StatusInternalServerError)
 465                         return
 466                 }
 467                 if token != h.Cluster.Users.AnonymousUserToken {
 468                         tokenValid = true
 469                 }
 470                 f, err := fs.OpenFile(fsprefix, dirOpenMode, 0)
 471                 if errors.As(err, &statusErr) &&
 472                         statusErr.HTTPStatus() == http.StatusForbidden &&
 473                         token != h.Cluster.Users.AnonymousUserToken {
 474                         // collection id is outside scope of supplied
 475                         // token
 476                         tokenScopeProblem = true
 477                         sess.Release()
 478                         continue
 479                 } else if os.IsNotExist(err) {
 480                         // collection does not exist or is not
 481                         // readable using this token
 482                         sess.Release()
 483                         continue
 484                 } else if err != nil {
 485                         http.Error(w, err.Error(), http.StatusInternalServerError)
 486                         sess.Release()
 487                         return
 488                 }
 489                 defer f.Close()
 490
 491                 collectionDir, sessionFS, session, tokenUser = f, fs, sess, user
 492                 break
 493         }
 494
 495         if session == nil {
 496                 if pathToken {
 497                         // The URL is a "secret sharing link" that
 498                         // didn't work out.  Asking the client for
 499                         // additional credentials would just be
 500                         // confusing.
 501                         http.Error(w, notFoundMessage, http.StatusNotFound)
 502                         return
 503                 }
 504                 if tokenValid {
 505                         // The client provided valid token(s), but the
 506                         // collection was not found.
 507                         http.Error(w, notFoundMessage, http.StatusNotFound)
 508                         return
 509                 }
 510                 if tokenScopeProblem {
 511                         // The client provided a valid token but
 512                         // fetching a collection returned 401, which
 513                         // means the token scope doesn't permit
 514                         // fetching that collection.
 515                         http.Error(w, notFoundMessage, http.StatusForbidden)
 516                         return
 517                 }
 518                 // The client's token was invalid (e.g., expired), or
 519                 // the client didn't even provide one.  Redirect to
 520                 // workbench2's login-and-redirect-to-download url if
 521                 // this is a browser navigation request. (The redirect
 522                 // flow can't preserve the original method if it's not
 523                 // GET, and doesn't make sense if the UA is a
 524                 // command-line tool, is trying to load an inline
 525                 // image, etc.; in these cases, there's nothing we can
 526                 // do, so return 401 unauthorized.)
 527                 //
 528                 // Note Sec-Fetch-Mode is sent by all non-EOL
 529                 // browsers, except Safari.
 530                 // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Sec-Fetch-Mode
 531                 //
 532                 // TODO(TC): This response would be confusing to
 533                 // someone trying (anonymously) to download public
 534                 // data that has been deleted.  Allow a referrer to
 535                 // provide this context somehow?
 536                 if r.Method == http.MethodGet && r.Header.Get("Sec-Fetch-Mode") == "navigate" {
 537                         target := url.URL(h.Cluster.Services.Workbench2.ExternalURL)
 538                         redirkey := "redirectToPreview"
 539                         if attachment {
 540                                 redirkey = "redirectToDownload"
 541                         }
 542                         callback := "/c=" + collectionID + "/" + strings.Join(targetPath, "/")
 543                         query := url.Values{redirkey: {callback}}
 544                         queryString := query.Encode()
 545                         // Note: Encode (and QueryEscape function) turns space
 546                         // into plus sign (+) rather than %20 (the plus sign
 547                         // becomes %2B); that is the rule for web forms data
 548                         // sent in URL query part via GET, but we're not
 549                         // emulating forms here. Client JS APIs
 550                         // (URLSearchParam#get, decodeURIComponent) will
 551                         // decode %20, but while the former also expects the
 552                         // form-specific encoding, the latter doesn't.
 553                         // Encode() almost encodes everything; RFC 3986 3.4
 554                         // says "it is sometimes better for usability" to not
 555                         // encode / and ? when passing URI reference in query.
 556                         // This is also legal according to WHATWG URL spec and
 557                         // can be desirable for debugging webapp.
 558                         // We can let slash / appear in the encoded query, and
 559                         // equality-sign = too, but exempting ? is not very
 560                         // useful.
 561                         // Plus-sign, hash, and ampersand are never exempt.
 562                         r := strings.NewReplacer("+", "%20", "%2F", "/", "%3D", "=")
 563                         target.RawQuery = r.Replace(queryString)
 564                         w.Header().Add("Location", target.String())
 565                         w.WriteHeader(http.StatusSeeOther)
 566                         return
 567                 }
 568                 if !credentialsOK {
 569                         http.Error(w, fmt.Sprintf("Authorization tokens are not accepted here: %v, and no anonymous user token is configured.", reasonNotAcceptingCredentials), http.StatusUnauthorized)
 570                         return
 571                 }
 572                 // If none of the above cases apply, suggest the
 573                 // user-agent (which is either a non-browser agent
 574                 // like wget, or a browser that can't redirect through
 575                 // a login flow) prompt the user for credentials.
 576                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 577                 http.Error(w, unauthorizedMessage, http.StatusUnauthorized)
 578                 return
 579         }
 580
 581         // The first call to releaseSession() calls session.Release(),
 582         // then subsequent calls are no-ops.  This lets us use a defer
 583         // call here to ensure it gets called in all code paths, and
 584         // also call it inline (see below) in the cases where we want
 585         // to release the lock before returning.
 586         var releaseSessionOnce sync.Once
 587         releaseSession := func() { releaseSessionOnce.Do(func() { session.Release() }) }
 588         defer releaseSession()
 589
 590         colltarget := strings.Join(targetPath, "/")
 591         colltarget = strings.TrimSuffix(colltarget, "/")
 592         fstarget := fsprefix + colltarget
 593         if !forceReload {
 594                 need, err := h.needSync(r.Context(), sessionFS, fstarget)
 595                 if err != nil {
 596                         http.Error(w, err.Error(), http.StatusBadGateway)
 597                         return
 598                 }
 599                 forceReload = need
 600         }
 601         if forceReload {
 602                 err := collectionDir.Sync()
 603                 if err != nil {
 604                         if he := errorWithHTTPStatus(nil); errors.As(err, &he) {
 605                                 http.Error(w, err.Error(), he.HTTPStatus())
 606                         } else {
 607                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 608                         }
 609                         return
 610                 }
 611         }
 612
 613         if r.Method == http.MethodGet || r.Method == http.MethodHead {
 614                 if fi, err := sessionFS.Stat(fstarget); err == nil && fi.IsDir() {
 615                         releaseSession() // because we won't be writing anything
 616                         if !strings.HasSuffix(r.URL.Path, "/") {
 617                                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 618                         } else {
 619                                 h.serveDirectory(w, r, fi.Name(), sessionFS, fstarget, !useSiteFS)
 620                         }
 621                         return
 622                 }
 623         }
 624
 625         var basename string
 626         if len(targetPath) > 0 {
 627                 basename = targetPath[len(targetPath)-1]
 628         }
 629         if arvadosclient.PDHMatch(collectionID) && writeMethod[r.Method] {
 630                 http.Error(w, webdavfs.ErrReadOnly.Error(), http.StatusMethodNotAllowed)
 631                 return
 632         }
 633         if !h.userPermittedToUploadOrDownload(r.Method, tokenUser) {
 634                 http.Error(w, "Not permitted", http.StatusForbidden)
 635                 return
 636         }
 637         h.logUploadOrDownload(r, session.arvadosclient, sessionFS, fstarget, nil, tokenUser)
 638
 639         if webdavPrefix == "" && stripParts > 0 {
 640                 webdavPrefix = "/" + strings.Join(pathParts[:stripParts], "/")
 641         }
 642
 643         writing := writeMethod[r.Method]
 644         if writing {
 645                 // We implement write operations by writing to a
 646                 // temporary collection, then applying the change to
 647                 // the real collection using the replace_files option
 648                 // in a collection update request.  This lets us do
 649                 // the slow part (i.e., receive the file data from the
 650                 // client and write it to Keep) without worrying about
 651                 // side effects of other read/write operations.
 652                 //
 653                 // Collection update requests for a given collection
 654                 // are serialized by the controller, so we don't need
 655                 // to do any locking for that part either.
 656
 657                 // collprefix is the subdirectory in the target
 658                 // collection which (according to X-Webdav-Source) we
 659                 // should pretend is "/" for this request.
 660                 collprefix := strings.TrimPrefix(fsprefix, "by_id/"+collectionID+"/")
 661                 if len(collprefix) == len(fsprefix) {
 662                         http.Error(w, "internal error: writing to anything other than /by_id/{collectionID}", http.StatusInternalServerError)
 663                         return
 664                 }
 665
 666                 // Create a temporary collection filesystem for webdav
 667                 // to operate on.
 668                 var tmpcoll arvados.Collection
 669                 client := session.client.WithRequestID(r.Header.Get("X-Request-Id"))
 670                 tmpfs, err := tmpcoll.FileSystem(client, session.keepclient)
 671                 if err != nil {
 672                         http.Error(w, err.Error(), http.StatusInternalServerError)
 673                         return
 674                 }
 675                 snap, err := arvados.Snapshot(sessionFS, "by_id/"+collectionID+"/")
 676                 if err != nil {
 677                         http.Error(w, "snapshot: "+err.Error(), http.StatusInternalServerError)
 678                         return
 679                 }
 680                 err = arvados.Splice(tmpfs, "/", snap)
 681                 if err != nil {
 682                         http.Error(w, "splice: "+err.Error(), http.StatusInternalServerError)
 683                         return
 684                 }
 685
 686                 targetFS = tmpfs
 687                 fsprefix = collprefix
 688                 replace := make(map[string]string)
 689
 690                 switch r.Method {
 691                 case "COPY", "MOVE":
 692                         dsttarget, err := copyMoveDestination(r, webdavPrefix)
 693                         if err != nil {
 694                                 http.Error(w, err.Error(), http.StatusBadRequest)
 695                                 return
 696                         }
 697
 698                         srcspec := "current/" + colltarget
 699                         // RFC 4918 9.8.3: A COPY of "Depth: 0" only
 700                         // instructs that the collection and its
 701                         // properties, but not resources identified by
 702                         // its internal member URLs, are to be copied.
 703                         //
 704                         // ...meaning we will be creating an empty
 705                         // directory.
 706                         //
 707                         // RFC 4918 9.9.2: A client MUST NOT submit a
 708                         // Depth header on a MOVE on a collection with
 709                         // any value but "infinity".
 710                         //
 711                         // ...meaning we only need to consider this
 712                         // case for COPY, not for MOVE.
 713                         if fi, err := tmpfs.Stat(colltarget); err == nil && fi.IsDir() && r.Method == "COPY" && r.Header.Get("Depth") == "0" {
 714                                 srcspec = "manifest_text/"
 715                         }
 716
 717                         replace[strings.TrimSuffix(dsttarget, "/")] = srcspec
 718                         if r.Method == "MOVE" {
 719                                 replace["/"+colltarget] = ""
 720                         }
 721                 case "MKCOL":
 722                         replace["/"+colltarget] = "manifest_text/"
 723                 case "DELETE":
 724                         if depth := r.Header.Get("Depth"); depth != "" && depth != "infinity" {
 725                                 http.Error(w, "invalid depth header, see RFC 4918 9.6.1", http.StatusBadRequest)
 726                                 return
 727                         }
 728                         replace["/"+colltarget] = ""
 729                 case "PUT":
 730                         // changes will be applied by updateOnSuccess
 731                         // update func below
 732                 case "LOCK", "UNLOCK", "PROPPATCH":
 733                         // no changes
 734                 default:
 735                         http.Error(w, "method missing", http.StatusInternalServerError)
 736                         return
 737                 }
 738
 739                 // Save the collection only if/when all
 740                 // webdav->filesystem operations succeed using our
 741                 // temporary collection -- and send a 500 error if the
 742                 // updates can't be saved.
 743                 logger := ctxlog.FromContext(r.Context())
 744                 w = &updateOnSuccess{
 745                         ResponseWriter: w,
 746                         logger:         logger,
 747                         update: func() error {
 748                                 var manifest string
 749                                 var snap *arvados.Subtree
 750                                 var err error
 751                                 if r.Method == "PUT" {
 752                                         snap, err = arvados.Snapshot(tmpfs, colltarget)
 753                                         if err != nil {
 754                                                 return fmt.Errorf("snapshot tmpfs: %w", err)
 755                                         }
 756                                         tmpfs, err = (&arvados.Collection{}).FileSystem(client, session.keepclient)
 757                                         err = arvados.Splice(tmpfs, "file", snap)
 758                                         if err != nil {
 759                                                 return fmt.Errorf("splice tmpfs: %w", err)
 760                                         }
 761                                         manifest, err = tmpfs.MarshalManifest(".")
 762                                         if err != nil {
 763                                                 return fmt.Errorf("marshal tmpfs: %w", err)
 764                                         }
 765                                         replace["/"+colltarget] = "manifest_text/file"
 766                                 } else if len(replace) == 0 {
 767                                         return nil
 768                                 }
 769                                 err = client.RequestAndDecode(nil, "PATCH", "arvados/v1/collections/"+collectionID, nil, map[string]interface{}{
 770                                         "replace_files": replace,
 771                                         "collection":    map[string]interface{}{"manifest_text": manifest}})
 772                                 var te arvados.TransactionError
 773                                 if errors.As(err, &te) {
 774                                         err = te
 775                                 }
 776                                 if err != nil {
 777                                         return err
 778                                 }
 779                                 return nil
 780                         }}
 781         } else {
 782                 // When writing, we need to block session renewal
 783                 // until we're finished, in order to guarantee the
 784                 // effect of the write is visible in future responses.
 785                 // But if we're not writing, we can release the lock
 786                 // early.  This enables us to keep renewing sessions
 787                 // and processing more requests even if a slow client
 788                 // takes a long time to download a large file.
 789                 releaseSession()
 790                 targetFS = sessionFS
 791         }
 792         if r.Method == http.MethodGet {
 793                 applyContentDispositionHdr(w, r, basename, attachment)
 794         }
 795         wh := &webdav.Handler{
 796                 Prefix: webdavPrefix,
 797                 FileSystem: &webdavfs.FS{
 798                         FileSystem:    targetFS,
 799                         Prefix:        fsprefix,
 800                         Writing:       writeMethod[r.Method],
 801                         AlwaysReadEOF: r.Method == "PROPFIND",
 802                 },
 803                 LockSystem: webdavfs.NoLockSystem,
 804                 Logger: func(r *http.Request, err error) {
 805                         if err != nil && !os.IsNotExist(err) {
 806                                 ctxlog.FromContext(r.Context()).WithError(err).Error("error reported by webdav handler")
 807                         }
 808                 },
 809         }
 810         h.metrics.track(wh, w, r)
 811         if r.Method == http.MethodGet && w.WroteStatus() == http.StatusOK {
 812                 wrote := int64(w.WroteBodyBytes())
 813                 fi, err := wh.FileSystem.Stat(r.Context(), colltarget)
 814                 if err == nil && fi.Size() != wrote {
 815                         var n int
 816                         f, err := wh.FileSystem.OpenFile(r.Context(), colltarget, os.O_RDONLY, 0)
 817                         if err == nil {
 818                                 n, err = f.Read(make([]byte, 1024))
 819                                 f.Close()
 820                         }
 821                         ctxlog.FromContext(r.Context()).Errorf("stat.Size()==%d but only wrote %d bytes; read(1024) returns %d, %v", fi.Size(), wrote, n, err)
 822                 }
 823         }
 824 }
 825
 826 var dirListingTemplate = `<!DOCTYPE HTML>
 827 <HTML><HEAD>
 828   <META name="robots" content="NOINDEX">
 829   <TITLE>{{ .CollectionName }}</TITLE>
 830   <STYLE type="text/css">
 831     body {
 832       margin: 1.5em;
 833     }
 834     pre {
 835       background-color: #D9EDF7;
 836       border-radius: .25em;
 837       padding: .75em;
 838       overflow: auto;
 839     }
 840     .footer p {
 841       font-size: 82%;
 842     }
 843     hr {
 844       border: 1px solid #808080;
 845     }
 846     ul {
 847       padding: 0;
 848     }
 849     ul li {
 850       font-family: monospace;
 851       list-style: none;
 852     }
 853   </STYLE>
 854 </HEAD>
 855 <BODY>
 856
 857 <H1>{{ .CollectionName }}</H1>
 858
 859 <P>This collection of data files is being shared with you through
 860 Arvados.  You can download individual files listed below.  To download
 861 the entire directory tree with <CODE>wget</CODE>, try:</P>
 862
 863 <PRE id="wget-example">$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} {{ .QuotedUrlForWget }}</PRE>
 864
 865 <H2>File Listing</H2>
 866
 867 {{if .Files}}
 868 <UL>
 869 {{range .Files}}
 870 {{if .IsDir }}
 871   <LI>{{" " | printf "%15s  " | nbsp}}<A class="item" href="{{ .Href }}/">{{ .Name }}/</A></LI>
 872 {{else}}
 873   <LI>{{.Size | printf "%15d  " | nbsp}}<A class="item" href="{{ .Href }}">{{ .Name }}</A></LI>
 874 {{end}}
 875 {{end}}
 876 </UL>
 877 {{else}}
 878 <P>(No files; this collection is empty.)</P>
 879 {{end}}
 880
 881 <HR>
 882 <DIV class="footer">
 883   <P>
 884     About Arvados:
 885     Arvados is a free and open source software bioinformatics platform.
 886     To learn more, visit arvados.org.
 887     Arvados is not responsible for the files listed on this page.
 888   </P>
 889 </DIV>
 890
 891 </BODY>
 892 </HTML>
 893 `
 894
 895 type fileListEnt struct {
 896         Name  string
 897         Href  string
 898         Size  int64
 899         IsDir bool
 900 }
 901
 902 // Given a filesystem path like `foo/"bar baz"`, return an escaped
 903 // (percent-encoded) relative path like `./foo/%22bar%20%baz%22`.
 904 //
 905 // Note the result may contain html-unsafe characters like '&'. These
 906 // will be handled separately by the HTML templating engine as needed.
 907 func relativeHref(path string) string {
 908         u := &url.URL{Path: path}
 909         return "./" + u.EscapedPath()
 910 }
 911
 912 // Return a shell-quoted URL suitable for pasting to a command line
 913 // ("wget ...") to repeat the given HTTP request.
 914 func makeQuotedUrlForWget(r *http.Request) string {
 915         scheme := r.Header.Get("X-Forwarded-Proto")
 916         if scheme == "http" || scheme == "https" {
 917                 // use protocol reported by load balancer / proxy
 918         } else if r.TLS != nil {
 919                 scheme = "https"
 920         } else {
 921                 scheme = "http"
 922         }
 923         p := r.URL.EscapedPath()
 924         // An escaped path may still contain single quote chars, which
 925         // would interfere with our shell quoting. Avoid this by
 926         // escaping them as %27.
 927         return fmt.Sprintf("'%s://%s%s'", scheme, r.Host, strings.Replace(p, "'", "%27", -1))
 928 }
 929
 930 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, recurse bool) {
 931         var files []fileListEnt
 932         var walk func(string) error
 933         if !strings.HasSuffix(base, "/") {
 934                 base = base + "/"
 935         }
 936         walk = func(path string) error {
 937                 dirname := base + path
 938                 if dirname != "/" {
 939                         dirname = strings.TrimSuffix(dirname, "/")
 940                 }
 941                 d, err := fs.Open(dirname)
 942                 if err != nil {
 943                         return err
 944                 }
 945                 ents, err := d.Readdir(-1)
 946                 if err != nil {
 947                         return err
 948                 }
 949                 for _, ent := range ents {
 950                         if recurse && ent.IsDir() {
 951                                 err = walk(path + ent.Name() + "/")
 952                                 if err != nil {
 953                                         return err
 954                                 }
 955                         } else {
 956                                 listingName := path + ent.Name()
 957                                 files = append(files, fileListEnt{
 958                                         Name:  listingName,
 959                                         Href:  relativeHref(listingName),
 960                                         Size:  ent.Size(),
 961                                         IsDir: ent.IsDir(),
 962                                 })
 963                         }
 964                 }
 965                 return nil
 966         }
 967         if err := walk(""); err != nil {
 968                 http.Error(w, "error getting directory listing: "+err.Error(), http.StatusInternalServerError)
 969                 return
 970         }
 971
 972         funcs := template.FuncMap{
 973                 "nbsp": func(s string) template.HTML {
 974                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 975                 },
 976         }
 977         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 978         if err != nil {
 979                 http.Error(w, "error parsing template: "+err.Error(), http.StatusInternalServerError)
 980                 return
 981         }
 982         sort.Slice(files, func(i, j int) bool {
 983                 return files[i].Name < files[j].Name
 984         })
 985         w.WriteHeader(http.StatusOK)
 986         tmpl.Execute(w, map[string]interface{}{
 987                 "CollectionName":   collectionName,
 988                 "Files":            files,
 989                 "Request":          r,
 990                 "StripParts":       strings.Count(strings.TrimRight(r.URL.Path, "/"), "/"),
 991                 "QuotedUrlForWget": makeQuotedUrlForWget(r),
 992         })
 993 }
 994
 995 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 996         disposition := "inline"
 997         if isAttachment {
 998                 disposition = "attachment"
 999         }
1000         if strings.ContainsRune(r.RequestURI, '?') {
1001                 // Help the UA realize that the filename is just
1002                 // "filename.txt", not
1003                 // "filename.txt?disposition=attachment".
1004                 //
1005                 // TODO(TC): Follow advice at RFC 6266 appendix D
1006                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
1007         }
1008         if disposition != "inline" {
1009                 w.Header().Set("Content-Disposition", disposition)
1010         }
1011 }
1012
1013 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
1014         if formTokens, haveFormTokens := r.Form["api_token"]; haveFormTokens {
1015                 if !credentialsOK {
1016                         // It is not safe to copy the provided token
1017                         // into a cookie unless the current vhost
1018                         // (origin) serves only a single collection or
1019                         // we are in TrustAllContent mode.
1020                         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)
1021                         return
1022                 }
1023
1024                 // The HttpOnly flag is necessary to prevent
1025                 // JavaScript code (included in, or loaded by, a page
1026                 // in the collection being served) from employing the
1027                 // user's token beyond reading other files in the same
1028                 // domain, i.e., same collection.
1029                 //
1030                 // The 303 redirect is necessary in the case of a GET
1031                 // request to avoid exposing the token in the Location
1032                 // bar, and in the case of a POST request to avoid
1033                 // raising warnings when the user refreshes the
1034                 // resulting page.
1035                 for _, tok := range formTokens {
1036                         if tok == "" {
1037                                 continue
1038                         }
1039                         http.SetCookie(w, &http.Cookie{
1040                                 Name:     "arvados_api_token",
1041                                 Value:    auth.EncodeTokenCookie([]byte(tok)),
1042                                 Path:     "/",
1043                                 HttpOnly: true,
1044                                 SameSite: http.SameSiteLaxMode,
1045                         })
1046                         break
1047                 }
1048         }
1049
1050         // Propagate query parameters (except api_token) from
1051         // the original request.
1052         redirQuery := r.URL.Query()
1053         redirQuery.Del("api_token")
1054
1055         u := r.URL
1056         if location != "" {
1057                 newu, err := u.Parse(location)
1058                 if err != nil {
1059                         http.Error(w, "error resolving redirect target: "+err.Error(), http.StatusInternalServerError)
1060                         return
1061                 }
1062                 u = newu
1063         }
1064         redir := (&url.URL{
1065                 Scheme:   r.URL.Scheme,
1066                 Host:     r.Host,
1067                 Path:     u.Path,
1068                 RawQuery: redirQuery.Encode(),
1069         }).String()
1070
1071         w.Header().Add("Location", redir)
1072         w.WriteHeader(http.StatusSeeOther)
1073         io.WriteString(w, `<A href="`)
1074         io.WriteString(w, html.EscapeString(redir))
1075         io.WriteString(w, `">Continue</A>`)
1076 }
1077
1078 func (h *handler) userPermittedToUploadOrDownload(method string, tokenUser *arvados.User) bool {
1079         var permitDownload bool
1080         var permitUpload bool
1081         if tokenUser != nil && tokenUser.IsAdmin {
1082                 permitUpload = h.Cluster.Collections.WebDAVPermission.Admin.Upload
1083                 permitDownload = h.Cluster.Collections.WebDAVPermission.Admin.Download
1084         } else {
1085                 permitUpload = h.Cluster.Collections.WebDAVPermission.User.Upload
1086                 permitDownload = h.Cluster.Collections.WebDAVPermission.User.Download
1087         }
1088         if (method == "PUT" || method == "POST") && !permitUpload {
1089                 // Disallow operations that upload new files.
1090                 // Permit webdav operations that move existing files around.
1091                 return false
1092         } else if method == "GET" && !permitDownload {
1093                 // Disallow downloading file contents.
1094                 // Permit webdav operations like PROPFIND that retrieve metadata
1095                 // but not file contents.
1096                 return false
1097         }
1098         return true
1099 }
1100
1101 // Parse the request's Destination header and return the destination
1102 // path relative to the current collection, i.e., with webdavPrefix
1103 // stripped off.
1104 func copyMoveDestination(r *http.Request, webdavPrefix string) (string, error) {
1105         dsturl, err := url.Parse(r.Header.Get("Destination"))
1106         if err != nil {
1107                 return "", err
1108         }
1109         if dsturl.Host != "" && dsturl.Host != r.Host {
1110                 return "", errors.New("destination host mismatch")
1111         }
1112         if webdavPrefix == "" {
1113                 return dsturl.Path, nil
1114         }
1115         dsttarget := strings.TrimPrefix(dsturl.Path, webdavPrefix)
1116         if len(dsttarget) == len(dsturl.Path) {
1117                 return "", errors.New("destination path not supported")
1118         }
1119         return dsttarget, nil
1120 }
1121
1122 // Check whether fstarget is in a collection whose PDH has changed
1123 // since it was last Sync()ed in sessionFS.
1124 //
1125 // If fstarget doesn't exist, but would be in such a collection if it
1126 // did exist, return true.
1127 func (h *handler) needSync(ctx context.Context, sessionFS arvados.CustomFileSystem, fstarget string) (bool, error) {
1128         collection, _ := h.determineCollection(sessionFS, fstarget)
1129         if collection == nil || len(collection.UUID) != 27 || !strings.HasPrefix(collection.UUID, h.Cluster.ClusterID) {
1130                 return false, nil
1131         }
1132         db, err := h.getDBConnector().GetDB(ctx)
1133         if err != nil {
1134                 return false, err
1135         }
1136         var currentPDH string
1137         err = db.QueryRowContext(ctx, `select portable_data_hash from collections where uuid=$1`, collection.UUID).Scan(&currentPDH)
1138         if err != nil {
1139                 return false, err
1140         }
1141         if currentPDH != collection.PortableDataHash {
1142                 return true, nil
1143         }
1144         return false, nil
1145 }
1146
1147 type fileEventLog struct {
1148         requestPath  string
1149         eventType    string
1150         userUUID     string
1151         userFullName string
1152         collUUID     string
1153         collPDH      string
1154         collFilePath string
1155         clientAddr   string
1156         clientToken  string
1157 }
1158
1159 func newFileEventLog(
1160         h *handler,
1161         r *http.Request,
1162         filepath string,
1163         collection *arvados.Collection,
1164         user *arvados.User,
1165         token string,
1166 ) *fileEventLog {
1167         var eventType string
1168         switch r.Method {
1169         case "POST", "PUT":
1170                 eventType = "file_upload"
1171         case "GET":
1172                 eventType = "file_download"
1173         default:
1174                 return nil
1175         }
1176
1177         // We want to log the address of the proxy closest to keep-web—the last
1178         // value in the X-Forwarded-For list—or the client address if there is no
1179         // valid proxy.
1180         var clientAddr string
1181         // 1. Build a slice of proxy addresses from X-Forwarded-For.
1182         xff := strings.Join(r.Header.Values("X-Forwarded-For"), ",")
1183         addrs := strings.Split(xff, ",")
1184         // 2. Reverse the slice so it's in our most preferred order for logging.
1185         slices.Reverse(addrs)
1186         // 3. Append the client address to that slice.
1187         if addr, _, err := net.SplitHostPort(r.RemoteAddr); err == nil {
1188                 addrs = append(addrs, addr)
1189         }
1190         // 4. Use the first valid address in the slice.
1191         for _, addr := range addrs {
1192                 if ip := net.ParseIP(strings.TrimSpace(addr)); ip != nil {
1193                         clientAddr = ip.String()
1194                         break
1195                 }
1196         }
1197
1198         ev := &fileEventLog{
1199                 requestPath: r.URL.Path,
1200                 eventType:   eventType,
1201                 clientAddr:  clientAddr,
1202                 clientToken: token,
1203         }
1204
1205         if user != nil {
1206                 ev.userUUID = user.UUID
1207                 ev.userFullName = user.FullName
1208         } else {
1209                 ev.userUUID = fmt.Sprintf("%s-tpzed-anonymouspublic", h.Cluster.ClusterID)
1210         }
1211
1212         if collection != nil {
1213                 ev.collFilePath = filepath
1214                 // h.determineCollection populates the collection_uuid
1215                 // prop with the PDH, if this collection is being
1216                 // accessed via PDH. For logging, we use a different
1217                 // field depending on whether it's a UUID or PDH.
1218                 if len(collection.UUID) > 32 {
1219                         ev.collPDH = collection.UUID
1220                 } else {
1221                         ev.collPDH = collection.PortableDataHash
1222                         ev.collUUID = collection.UUID
1223                 }
1224         }
1225
1226         return ev
1227 }
1228
1229 func (ev *fileEventLog) shouldLogPDH() bool {
1230         return ev.eventType == "file_download" && ev.collPDH != ""
1231 }
1232
1233 func (ev *fileEventLog) asDict() arvadosclient.Dict {
1234         props := arvadosclient.Dict{
1235                 "reqPath":              ev.requestPath,
1236                 "collection_uuid":      ev.collUUID,
1237                 "collection_file_path": ev.collFilePath,
1238         }
1239         if ev.shouldLogPDH() {
1240                 props["portable_data_hash"] = ev.collPDH
1241         }
1242         return arvadosclient.Dict{
1243                 "object_uuid": ev.userUUID,
1244                 "event_type":  ev.eventType,
1245                 "properties":  props,
1246         }
1247 }
1248
1249 func (ev *fileEventLog) asFields() logrus.Fields {
1250         fields := logrus.Fields{
1251                 "collection_file_path": ev.collFilePath,
1252                 "collection_uuid":      ev.collUUID,
1253                 "user_uuid":            ev.userUUID,
1254         }
1255         if ev.shouldLogPDH() {
1256                 fields["portable_data_hash"] = ev.collPDH
1257         }
1258         if !strings.HasSuffix(ev.userUUID, "-tpzed-anonymouspublic") {
1259                 fields["user_full_name"] = ev.userFullName
1260         }
1261         return fields
1262 }
1263
1264 func (h *handler) shouldLogEvent(
1265         event *fileEventLog,
1266         req *http.Request,
1267         fileInfo os.FileInfo,
1268         t time.Time,
1269 ) bool {
1270         if event == nil {
1271                 return false
1272         } else if event.eventType != "file_download" ||
1273                 h.Cluster.Collections.WebDAVLogDownloadInterval == 0 ||
1274                 fileInfo == nil {
1275                 return true
1276         }
1277         td := h.Cluster.Collections.WebDAVLogDownloadInterval.Duration()
1278         cutoff := t.Add(-td)
1279         ev := *event
1280         h.fileEventLogsMtx.Lock()
1281         defer h.fileEventLogsMtx.Unlock()
1282         if h.fileEventLogs == nil {
1283                 h.fileEventLogs = make(map[fileEventLog]time.Time)
1284         }
1285         shouldLog := h.fileEventLogs[ev].Before(cutoff)
1286         if !shouldLog {
1287                 // Go's http fs server evaluates http.Request.Header.Get("Range")
1288                 // (as of Go 1.22) so we should do the same.
1289                 // Don't worry about merging multiple headers, etc.
1290                 ranges, err := http_range.ParseRange(req.Header.Get("Range"), fileInfo.Size())
1291                 if ranges == nil || err != nil {
1292                         // The Range header was either empty or malformed.
1293                         // Err on the side of logging.
1294                         shouldLog = true
1295                 } else {
1296                         // Log this request only if it requested the first byte
1297                         // (our heuristic for "starting a new download").
1298                         for _, reqRange := range ranges {
1299                                 if reqRange.Start == 0 {
1300                                         shouldLog = true
1301                                         break
1302                                 }
1303                         }
1304                 }
1305         }
1306         if shouldLog {
1307                 h.fileEventLogs[ev] = t
1308         }
1309         if t.After(h.fileEventLogsNextTidy) {
1310                 for key, logTime := range h.fileEventLogs {
1311                         if logTime.Before(cutoff) {
1312                                 delete(h.fileEventLogs, key)
1313                         }
1314                 }
1315                 h.fileEventLogsNextTidy = t.Add(td)
1316         }
1317         return shouldLog
1318 }
1319
1320 func (h *handler) logUploadOrDownload(
1321         r *http.Request,
1322         client *arvadosclient.ArvadosClient,
1323         fs arvados.CustomFileSystem,
1324         filepath string,
1325         collection *arvados.Collection,
1326         user *arvados.User,
1327 ) {
1328         var fileInfo os.FileInfo
1329         if fs != nil {
1330                 if collection == nil {
1331                         collection, filepath = h.determineCollection(fs, filepath)
1332                 }
1333                 if collection != nil {
1334                         // It's okay to ignore this error because shouldLogEvent will
1335                         // always return true if fileInfo == nil.
1336                         fileInfo, _ = fs.Stat(path.Join("by_id", collection.UUID, filepath))
1337                 }
1338         }
1339         event := newFileEventLog(h, r, filepath, collection, user, client.ApiToken)
1340         if !h.shouldLogEvent(event, r, fileInfo, time.Now()) {
1341                 return
1342         }
1343         log := ctxlog.FromContext(r.Context()).WithFields(event.asFields())
1344         log.Info(strings.Replace(event.eventType, "file_", "File ", 1))
1345         if h.Cluster.Collections.WebDAVLogEvents {
1346                 go func() {
1347                         logReq := arvadosclient.Dict{"log": event.asDict()}
1348                         err := client.Create("logs", logReq, nil)
1349                         if err != nil {
1350                                 log.WithError(err).Errorf("Failed to create %s log event on API server", event.eventType)
1351                         }
1352                 }()
1353         }
1354 }
1355
1356 func (h *handler) determineCollection(fs arvados.CustomFileSystem, path string) (*arvados.Collection, string) {
1357         target := strings.TrimSuffix(path, "/")
1358         for cut := len(target); cut >= 0; cut = strings.LastIndexByte(target, '/') {
1359                 target = target[:cut]
1360                 fi, err := fs.Stat(target)
1361                 if os.IsNotExist(err) {
1362                         // creating a new file/dir, or download
1363                         // destined to fail
1364                         continue
1365                 } else if err != nil {
1366                         return nil, ""
1367                 }
1368                 switch src := fi.Sys().(type) {
1369                 case *arvados.Collection:
1370                         return src, strings.TrimPrefix(path[len(target):], "/")
1371                 case *arvados.Group:
1372                         return nil, ""
1373                 default:
1374                         if _, ok := src.(error); ok {
1375                                 return nil, ""
1376                         }
1377                 }
1378         }
1379         return nil, ""
1380 }
1381
1382 func ServeCORSPreflight(w http.ResponseWriter, header http.Header) bool {
1383         method := header.Get("Access-Control-Request-Method")
1384         if method == "" {
1385                 return false
1386         }
1387         if !browserMethod[method] && !webdavMethod[method] {
1388                 w.WriteHeader(http.StatusMethodNotAllowed)
1389                 return true
1390         }
1391         w.Header().Set("Access-Control-Allow-Headers", corsAllowHeadersHeader)
1392         w.Header().Set("Access-Control-Allow-Methods", "COPY, DELETE, GET, LOCK, MKCOL, MOVE, OPTIONS, POST, PROPFIND, PROPPATCH, PUT, RMCOL, UNLOCK")
1393         w.Header().Set("Access-Control-Allow-Origin", "*")
1394         w.Header().Set("Access-Control-Max-Age", "86400")
1395         return true
1396 }