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 main
   6
   7 import (
   8         "encoding/json"
   9         "fmt"
  10         "html"
  11         "html/template"
  12         "io"
  13         "net/http"
  14         "net/url"
  15         "os"
  16         "sort"
  17         "strconv"
  18         "strings"
  19         "sync"
  20
  21         "git.curoverse.com/arvados.git/sdk/go/arvados"
  22         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  23         "git.curoverse.com/arvados.git/sdk/go/auth"
  24         "git.curoverse.com/arvados.git/sdk/go/health"
  25         "git.curoverse.com/arvados.git/sdk/go/httpserver"
  26         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  27         "golang.org/x/net/webdav"
  28 )
  29
  30 type handler struct {
  31         Config        *Config
  32         clientPool    *arvadosclient.ClientPool
  33         setupOnce     sync.Once
  34         healthHandler http.Handler
  35         webdavLS      webdav.LockSystem
  36 }
  37
  38 // parseCollectionIDFromDNSName returns a UUID or PDH if s begins with
  39 // a UUID or URL-encoded PDH; otherwise "".
  40 func parseCollectionIDFromDNSName(s string) string {
  41         // Strip domain.
  42         if i := strings.IndexRune(s, '.'); i >= 0 {
  43                 s = s[:i]
  44         }
  45         // Names like {uuid}--collections.example.com serve the same
  46         // purpose as {uuid}.collections.example.com but can reduce
  47         // cost/effort of using [additional] wildcard certificates.
  48         if i := strings.Index(s, "--"); i >= 0 {
  49                 s = s[:i]
  50         }
  51         if arvadosclient.UUIDMatch(s) {
  52                 return s
  53         }
  54         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  55                 return pdh
  56         }
  57         return ""
  58 }
  59
  60 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  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) setup() {
  76         h.clientPool = arvadosclient.MakeClientPool()
  77
  78         keepclient.RefreshServiceDiscoveryOnSIGHUP()
  79
  80         h.healthHandler = &health.Handler{
  81                 Token:  h.Config.ManagementToken,
  82                 Prefix: "/_health/",
  83         }
  84
  85         // Even though we don't accept LOCK requests, every webdav
  86         // handler must have a non-nil LockSystem.
  87         h.webdavLS = &noLockSystem{}
  88 }
  89
  90 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  91         status := struct {
  92                 cacheStats
  93                 Version string
  94         }{
  95                 cacheStats: h.Config.Cache.Stats(),
  96                 Version:    version,
  97         }
  98         json.NewEncoder(w).Encode(status)
  99 }
 100
 101 var (
 102         webdavMethod = map[string]bool{
 103                 "OPTIONS":  true,
 104                 "PROPFIND": true,
 105         }
 106         browserMethod = map[string]bool{
 107                 "GET":  true,
 108                 "HEAD": true,
 109                 "POST": true,
 110         }
 111 )
 112
 113 // ServeHTTP implements http.Handler.
 114 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 115         h.setupOnce.Do(h.setup)
 116
 117         var statusCode = 0
 118         var statusText string
 119
 120         remoteAddr := r.RemoteAddr
 121         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
 122                 remoteAddr = xff + "," + remoteAddr
 123         }
 124
 125         w := httpserver.WrapResponseWriter(wOrig)
 126         defer func() {
 127                 if statusCode == 0 {
 128                         statusCode = w.WroteStatus()
 129                 } else if w.WroteStatus() == 0 {
 130                         w.WriteHeader(statusCode)
 131                 } else if w.WroteStatus() != statusCode {
 132                         httpserver.Log(r.RemoteAddr, "WARNING",
 133                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
 134                 }
 135                 if statusText == "" {
 136                         statusText = http.StatusText(statusCode)
 137                 }
 138                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
 139         }()
 140
 141         if strings.HasPrefix(r.URL.Path, "/_health/") && r.Method == "GET" {
 142                 h.healthHandler.ServeHTTP(w, r)
 143                 return
 144         }
 145
 146         if method := r.Header.Get("Access-Control-Request-Method"); method != "" && r.Method == "OPTIONS" {
 147                 if !browserMethod[method] && !webdavMethod[method] {
 148                         statusCode = http.StatusMethodNotAllowed
 149                         return
 150                 }
 151                 w.Header().Set("Access-Control-Allow-Headers", "Authorization, Content-Type, Range")
 152                 w.Header().Set("Access-Control-Allow-Methods", "GET, POST, OPTIONS, PROPFIND")
 153                 w.Header().Set("Access-Control-Allow-Origin", "*")
 154                 w.Header().Set("Access-Control-Max-Age", "86400")
 155                 statusCode = http.StatusOK
 156                 return
 157         }
 158
 159         if !browserMethod[r.Method] && !webdavMethod[r.Method] {
 160                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
 161                 return
 162         }
 163
 164         if r.Header.Get("Origin") != "" {
 165                 // Allow simple cross-origin requests without user
 166                 // credentials ("user credentials" as defined by CORS,
 167                 // i.e., cookies, HTTP authentication, and client-side
 168                 // SSL certificates. See
 169                 // http://www.w3.org/TR/cors/#user-credentials).
 170                 w.Header().Set("Access-Control-Allow-Origin", "*")
 171                 w.Header().Set("Access-Control-Expose-Headers", "Content-Range")
 172         }
 173
 174         arv := h.clientPool.Get()
 175         if arv == nil {
 176                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+h.clientPool.Err().Error()
 177                 return
 178         }
 179         defer h.clientPool.Put(arv)
 180
 181         pathParts := strings.Split(r.URL.Path[1:], "/")
 182
 183         var stripParts int
 184         var targetID string
 185         var tokens []string
 186         var reqTokens []string
 187         var pathToken bool
 188         var attachment bool
 189         credentialsOK := h.Config.TrustAllContent
 190
 191         if r.Host != "" && r.Host == h.Config.AttachmentOnlyHost {
 192                 credentialsOK = true
 193                 attachment = true
 194         } else if r.FormValue("disposition") == "attachment" {
 195                 attachment = true
 196         }
 197
 198         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 199                 // http://ID.collections.example/PATH...
 200                 credentialsOK = true
 201         } else if r.URL.Path == "/status.json" {
 202                 h.serveStatus(w, r)
 203                 return
 204         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 205                 // /c=ID[/PATH...]
 206                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 207                 stripParts = 1
 208         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 209                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 210                         // /collections/download/ID/TOKEN/PATH...
 211                         targetID = parseCollectionIDFromURL(pathParts[2])
 212                         tokens = []string{pathParts[3]}
 213                         stripParts = 4
 214                         pathToken = true
 215                 } else {
 216                         // /collections/ID/PATH...
 217                         targetID = parseCollectionIDFromURL(pathParts[1])
 218                         tokens = h.Config.AnonymousTokens
 219                         stripParts = 2
 220                 }
 221         }
 222
 223         if targetID == "" {
 224                 statusCode = http.StatusNotFound
 225                 return
 226         }
 227
 228         formToken := r.FormValue("api_token")
 229         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 230                 // The client provided an explicit token in the POST
 231                 // body. The Origin header indicates this *might* be
 232                 // an AJAX request, in which case redirect-with-cookie
 233                 // won't work: we should just serve the content in the
 234                 // POST response. This is safe because:
 235                 //
 236                 // * We're supplying an attachment, not inline
 237                 //   content, so we don't need to convert the POST to
 238                 //   a GET and avoid the "really resubmit form?"
 239                 //   problem.
 240                 //
 241                 // * The token isn't embedded in the URL, so we don't
 242                 //   need to worry about bookmarks and copy/paste.
 243                 tokens = append(tokens, formToken)
 244         } else if formToken != "" && browserMethod[r.Method] {
 245                 // The client provided an explicit token in the query
 246                 // string, or a form in POST body. We must put the
 247                 // token in an HttpOnly cookie, and redirect to the
 248                 // same URL with the query param redacted and method =
 249                 // GET.
 250                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 251                 return
 252         }
 253
 254         targetPath := pathParts[stripParts:]
 255         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 256                 // http://ID.example/t=TOKEN/PATH...
 257                 // /c=ID/t=TOKEN/PATH...
 258                 //
 259                 // This form must only be used to pass scoped tokens
 260                 // that give permission for a single collection. See
 261                 // FormValue case above.
 262                 tokens = []string{targetPath[0][2:]}
 263                 pathToken = true
 264                 targetPath = targetPath[1:]
 265                 stripParts++
 266         }
 267
 268         if tokens == nil {
 269                 if credentialsOK {
 270                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 271                 }
 272                 tokens = append(reqTokens, h.Config.AnonymousTokens...)
 273         }
 274
 275         if len(targetPath) > 0 && targetPath[0] == "_" {
 276                 // If a collection has a directory called "t=foo" or
 277                 // "_", it can be served at
 278                 // //collections.example/_/t=foo/ or
 279                 // //collections.example/_/_/ respectively:
 280                 // //collections.example/t=foo/ won't work because
 281                 // t=foo will be interpreted as a token "foo".
 282                 targetPath = targetPath[1:]
 283                 stripParts++
 284         }
 285
 286         forceReload := false
 287         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 288                 forceReload = true
 289         }
 290
 291         var collection *arvados.Collection
 292         tokenResult := make(map[string]int)
 293         for _, arv.ApiToken = range tokens {
 294                 var err error
 295                 collection, err = h.Config.Cache.Get(arv, targetID, forceReload)
 296                 if err == nil {
 297                         // Success
 298                         break
 299                 }
 300                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 301                         switch srvErr.HttpStatusCode {
 302                         case 404, 401:
 303                                 // Token broken or insufficient to
 304                                 // retrieve collection
 305                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 306                                 continue
 307                         }
 308                 }
 309                 // Something more serious is wrong
 310                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 311                 return
 312         }
 313         if collection == nil {
 314                 if pathToken || !credentialsOK {
 315                         // Either the URL is a "secret sharing link"
 316                         // that didn't work out (and asking the client
 317                         // for additional credentials would just be
 318                         // confusing), or we don't even accept
 319                         // credentials at this path.
 320                         statusCode = http.StatusNotFound
 321                         return
 322                 }
 323                 for _, t := range reqTokens {
 324                         if tokenResult[t] == 404 {
 325                                 // The client provided valid token(s), but the
 326                                 // collection was not found.
 327                                 statusCode = http.StatusNotFound
 328                                 return
 329                         }
 330                 }
 331                 // The client's token was invalid (e.g., expired), or
 332                 // the client didn't even provide one.  Propagate the
 333                 // 401 to encourage the client to use a [different]
 334                 // token.
 335                 //
 336                 // TODO(TC): This response would be confusing to
 337                 // someone trying (anonymously) to download public
 338                 // data that has been deleted.  Allow a referrer to
 339                 // provide this context somehow?
 340                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 341                 statusCode = http.StatusUnauthorized
 342                 return
 343         }
 344
 345         kc, err := keepclient.MakeKeepClient(arv)
 346         if err != nil {
 347                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 348                 return
 349         }
 350
 351         var basename string
 352         if len(targetPath) > 0 {
 353                 basename = targetPath[len(targetPath)-1]
 354         }
 355         applyContentDispositionHdr(w, r, basename, attachment)
 356
 357         fs := collection.FileSystem(&arvados.Client{
 358                 APIHost:   arv.ApiServer,
 359                 AuthToken: arv.ApiToken,
 360                 Insecure:  arv.ApiInsecure,
 361         }, kc)
 362         if webdavMethod[r.Method] {
 363                 h := webdav.Handler{
 364                         Prefix:     "/" + strings.Join(pathParts[:stripParts], "/"),
 365                         FileSystem: &webdavFS{collfs: fs},
 366                         LockSystem: h.webdavLS,
 367                         Logger: func(_ *http.Request, err error) {
 368                                 if os.IsNotExist(err) {
 369                                         statusCode, statusText = http.StatusNotFound, err.Error()
 370                                 } else if err != nil {
 371                                         statusCode, statusText = http.StatusInternalServerError, err.Error()
 372                                 }
 373                         },
 374                 }
 375                 h.ServeHTTP(w, r)
 376                 return
 377         }
 378
 379         openPath := "/" + strings.Join(targetPath, "/")
 380         if f, err := fs.Open(openPath); os.IsNotExist(err) {
 381                 // Requested non-existent path
 382                 statusCode = http.StatusNotFound
 383         } else if err != nil {
 384                 // Some other (unexpected) error
 385                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 386         } else if stat, err := f.Stat(); err != nil {
 387                 // Can't get Size/IsDir (shouldn't happen with a collectionFS!)
 388                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 389         } else if stat.IsDir() && !strings.HasSuffix(r.URL.Path, "/") {
 390                 // If client requests ".../dirname", redirect to
 391                 // ".../dirname/". This way, relative links in the
 392                 // listing for "dirname" can always be "fnm", never
 393                 // "dirname/fnm".
 394                 h.seeOtherWithCookie(w, r, r.URL.Path+"/", credentialsOK)
 395         } else if stat.IsDir() {
 396                 h.serveDirectory(w, r, collection.Name, fs, openPath, stripParts)
 397         } else {
 398                 http.ServeContent(w, r, basename, stat.ModTime(), f)
 399                 if r.Header.Get("Range") == "" && int64(w.WroteBodyBytes()) != stat.Size() {
 400                         // If we wrote fewer bytes than expected, it's
 401                         // too late to change the real response code
 402                         // or send an error message to the client, but
 403                         // at least we can try to put some useful
 404                         // debugging info in the logs.
 405                         n, err := f.Read(make([]byte, 1024))
 406                         statusCode, statusText = http.StatusInternalServerError, fmt.Sprintf("f.Size()==%d but only wrote %d bytes; read(1024) returns %d, %s", stat.Size(), w.WroteBodyBytes(), n, err)
 407
 408                 }
 409         }
 410 }
 411
 412 var dirListingTemplate = `<!DOCTYPE HTML>
 413 <HTML><HEAD>
 414   <META name="robots" content="NOINDEX">
 415   <TITLE>{{ .Collection.Name }}</TITLE>
 416   <STYLE type="text/css">
 417     body {
 418       margin: 1.5em;
 419     }
 420     pre {
 421       background-color: #D9EDF7;
 422       border-radius: .25em;
 423       padding: .75em;
 424       overflow: auto;
 425     }
 426     .footer p {
 427       font-size: 82%;
 428     }
 429     ul {
 430       padding: 0;
 431     }
 432     ul li {
 433       font-family: monospace;
 434       list-style: none;
 435     }
 436   </STYLE>
 437 </HEAD>
 438 <BODY>
 439 <H1>{{ .CollectionName }}</H1>
 440
 441 <P>This collection of data files is being shared with you through
 442 Arvados.  You can download individual files listed below.  To download
 443 the entire collection with wget, try:</P>
 444
 445 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL }}</PRE>
 446
 447 <H2>File Listing</H2>
 448
 449 {{if .Files}}
 450 <UL>
 451 {{range .Files}}  <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{.Name}}">{{.Name}}</A></LI>{{end}}
 452 </UL>
 453 {{else}}
 454 <P>(No files; this collection is empty.)</P>
 455 {{end}}
 456
 457 <HR noshade>
 458 <DIV class="footer">
 459   <P>
 460     About Arvados:
 461     Arvados is a free and open source software bioinformatics platform.
 462     To learn more, visit arvados.org.
 463     Arvados is not responsible for the files listed on this page.
 464   </P>
 465 </DIV>
 466
 467 </BODY>
 468 `
 469
 470 type fileListEnt struct {
 471         Name string
 472         Size int64
 473 }
 474
 475 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, stripParts int) {
 476         var files []fileListEnt
 477         var walk func(string) error
 478         if !strings.HasSuffix(base, "/") {
 479                 base = base + "/"
 480         }
 481         walk = func(path string) error {
 482                 dirname := base + path
 483                 if dirname != "/" {
 484                         dirname = strings.TrimSuffix(dirname, "/")
 485                 }
 486                 d, err := fs.Open(dirname)
 487                 if err != nil {
 488                         return err
 489                 }
 490                 ents, err := d.Readdir(-1)
 491                 if err != nil {
 492                         return err
 493                 }
 494                 for _, ent := range ents {
 495                         if ent.IsDir() {
 496                                 err = walk(path + ent.Name() + "/")
 497                                 if err != nil {
 498                                         return err
 499                                 }
 500                         } else {
 501                                 files = append(files, fileListEnt{
 502                                         Name: path + ent.Name(),
 503                                         Size: ent.Size(),
 504                                 })
 505                         }
 506                 }
 507                 return nil
 508         }
 509         if err := walk(""); err != nil {
 510                 http.Error(w, err.Error(), http.StatusInternalServerError)
 511                 return
 512         }
 513
 514         funcs := template.FuncMap{
 515                 "nbsp": func(s string) template.HTML {
 516                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 517                 },
 518         }
 519         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 520         if err != nil {
 521                 http.Error(w, err.Error(), http.StatusInternalServerError)
 522                 return
 523         }
 524         sort.Slice(files, func(i, j int) bool {
 525                 return files[i].Name < files[j].Name
 526         })
 527         w.WriteHeader(http.StatusOK)
 528         tmpl.Execute(w, map[string]interface{}{
 529                 "CollectionName": collectionName,
 530                 "Files":          files,
 531                 "Request":        r,
 532                 "StripParts":     stripParts,
 533         })
 534 }
 535
 536 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 537         disposition := "inline"
 538         if isAttachment {
 539                 disposition = "attachment"
 540         }
 541         if strings.ContainsRune(r.RequestURI, '?') {
 542                 // Help the UA realize that the filename is just
 543                 // "filename.txt", not
 544                 // "filename.txt?disposition=attachment".
 545                 //
 546                 // TODO(TC): Follow advice at RFC 6266 appendix D
 547                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 548         }
 549         if disposition != "inline" {
 550                 w.Header().Set("Content-Disposition", disposition)
 551         }
 552 }
 553
 554 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 555         if formToken := r.FormValue("api_token"); formToken != "" {
 556                 if !credentialsOK {
 557                         // It is not safe to copy the provided token
 558                         // into a cookie unless the current vhost
 559                         // (origin) serves only a single collection or
 560                         // we are in TrustAllContent mode.
 561                         w.WriteHeader(http.StatusBadRequest)
 562                         return
 563                 }
 564
 565                 // The HttpOnly flag is necessary to prevent
 566                 // JavaScript code (included in, or loaded by, a page
 567                 // in the collection being served) from employing the
 568                 // user's token beyond reading other files in the same
 569                 // domain, i.e., same collection.
 570                 //
 571                 // The 303 redirect is necessary in the case of a GET
 572                 // request to avoid exposing the token in the Location
 573                 // bar, and in the case of a POST request to avoid
 574                 // raising warnings when the user refreshes the
 575                 // resulting page.
 576                 http.SetCookie(w, &http.Cookie{
 577                         Name:     "arvados_api_token",
 578                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 579                         Path:     "/",
 580                         HttpOnly: true,
 581                 })
 582         }
 583
 584         // Propagate query parameters (except api_token) from
 585         // the original request.
 586         redirQuery := r.URL.Query()
 587         redirQuery.Del("api_token")
 588
 589         u := r.URL
 590         if location != "" {
 591                 newu, err := u.Parse(location)
 592                 if err != nil {
 593                         w.WriteHeader(http.StatusInternalServerError)
 594                         return
 595                 }
 596                 u = newu
 597         }
 598         redir := (&url.URL{
 599                 Host:     r.Host,
 600                 Path:     u.Path,
 601                 RawQuery: redirQuery.Encode(),
 602         }).String()
 603
 604         w.Header().Add("Location", redir)
 605         w.WriteHeader(http.StatusSeeOther)
 606         io.WriteString(w, `<A href="`)
 607         io.WriteString(w, html.EscapeString(redir))
 608         io.WriteString(w, `">Continue</A>`)
 609 }