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 )
  28
  29 type handler struct {
  30         Config     *Config
  31         clientPool *arvadosclient.ClientPool
  32         setupOnce  sync.Once
  33         hmux       *http.ServeMux
  34 }
  35
  36 // parseCollectionIDFromDNSName returns a UUID or PDH if s begins with
  37 // a UUID or URL-encoded PDH; otherwise "".
  38 func parseCollectionIDFromDNSName(s string) string {
  39         // Strip domain.
  40         if i := strings.IndexRune(s, '.'); i >= 0 {
  41                 s = s[:i]
  42         }
  43         // Names like {uuid}--collections.example.com serve the same
  44         // purpose as {uuid}.collections.example.com but can reduce
  45         // cost/effort of using [additional] wildcard certificates.
  46         if i := strings.Index(s, "--"); i >= 0 {
  47                 s = s[:i]
  48         }
  49         if arvadosclient.UUIDMatch(s) {
  50                 return s
  51         }
  52         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  53                 return pdh
  54         }
  55         return ""
  56 }
  57
  58 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  59
  60 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  61 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  62 // otherwise "".
  63 func parseCollectionIDFromURL(s string) string {
  64         if arvadosclient.UUIDMatch(s) {
  65                 return s
  66         }
  67         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  68                 return pdh
  69         }
  70         return ""
  71 }
  72
  73 func (h *handler) setup() {
  74         h.clientPool = arvadosclient.MakeClientPool()
  75
  76         keepclient.RefreshServiceDiscoveryOnSIGHUP()
  77
  78         h.hmux = http.NewServeMux()
  79         h.hmux.Handle("/_health/", &health.Handler{
  80                 Token:  h.Config.ManagementToken,
  81                 Prefix: "/_health/",
  82         })
  83 }
  84
  85 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  86         status := struct {
  87                 cacheStats
  88         }{
  89                 cacheStats: h.Config.Cache.Stats(),
  90         }
  91         json.NewEncoder(w).Encode(status)
  92 }
  93
  94 func (h *handler) healthCheck(w http.ResponseWriter, r *http.Request) {
  95         h.hmux.ServeHTTP(w, r)
  96 }
  97
  98 // ServeHTTP implements http.Handler.
  99 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
 100         h.setupOnce.Do(h.setup)
 101
 102         var statusCode = 0
 103         var statusText string
 104
 105         remoteAddr := r.RemoteAddr
 106         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
 107                 remoteAddr = xff + "," + remoteAddr
 108         }
 109
 110         w := httpserver.WrapResponseWriter(wOrig)
 111         defer func() {
 112                 if statusCode == 0 {
 113                         statusCode = w.WroteStatus()
 114                 } else if w.WroteStatus() == 0 {
 115                         w.WriteHeader(statusCode)
 116                 } else if w.WroteStatus() != statusCode {
 117                         httpserver.Log(r.RemoteAddr, "WARNING",
 118                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
 119                 }
 120                 if statusText == "" {
 121                         statusText = http.StatusText(statusCode)
 122                 }
 123                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
 124         }()
 125
 126         if strings.HasPrefix(r.URL.Path, "/_health/") && r.Method == "GET" {
 127                 h.healthCheck(w, r)
 128                 return
 129         }
 130
 131         if r.Method == "OPTIONS" {
 132                 method := r.Header.Get("Access-Control-Request-Method")
 133                 if method != "GET" && method != "POST" {
 134                         statusCode = http.StatusMethodNotAllowed
 135                         return
 136                 }
 137                 w.Header().Set("Access-Control-Allow-Headers", "Range")
 138                 w.Header().Set("Access-Control-Allow-Methods", "GET, POST")
 139                 w.Header().Set("Access-Control-Allow-Origin", "*")
 140                 w.Header().Set("Access-Control-Max-Age", "86400")
 141                 statusCode = http.StatusOK
 142                 return
 143         }
 144
 145         if r.Method != "GET" && r.Method != "POST" {
 146                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
 147                 return
 148         }
 149
 150         if r.Header.Get("Origin") != "" {
 151                 // Allow simple cross-origin requests without user
 152                 // credentials ("user credentials" as defined by CORS,
 153                 // i.e., cookies, HTTP authentication, and client-side
 154                 // SSL certificates. See
 155                 // http://www.w3.org/TR/cors/#user-credentials).
 156                 w.Header().Set("Access-Control-Allow-Origin", "*")
 157         }
 158
 159         arv := h.clientPool.Get()
 160         if arv == nil {
 161                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+h.clientPool.Err().Error()
 162                 return
 163         }
 164         defer h.clientPool.Put(arv)
 165
 166         pathParts := strings.Split(r.URL.Path[1:], "/")
 167
 168         var stripParts int
 169         var targetID string
 170         var tokens []string
 171         var reqTokens []string
 172         var pathToken bool
 173         var attachment bool
 174         credentialsOK := h.Config.TrustAllContent
 175
 176         if r.Host != "" && r.Host == h.Config.AttachmentOnlyHost {
 177                 credentialsOK = true
 178                 attachment = true
 179         } else if r.FormValue("disposition") == "attachment" {
 180                 attachment = true
 181         }
 182
 183         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 184                 // http://ID.collections.example/PATH...
 185                 credentialsOK = true
 186         } else if r.URL.Path == "/status.json" {
 187                 h.serveStatus(w, r)
 188                 return
 189         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 190                 // /c=ID[/PATH...]
 191                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 192                 stripParts = 1
 193         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 194                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 195                         // /collections/download/ID/TOKEN/PATH...
 196                         targetID = parseCollectionIDFromURL(pathParts[2])
 197                         tokens = []string{pathParts[3]}
 198                         stripParts = 4
 199                         pathToken = true
 200                 } else {
 201                         // /collections/ID/PATH...
 202                         targetID = parseCollectionIDFromURL(pathParts[1])
 203                         tokens = h.Config.AnonymousTokens
 204                         stripParts = 2
 205                 }
 206         }
 207
 208         if targetID == "" {
 209                 statusCode = http.StatusNotFound
 210                 return
 211         }
 212
 213         formToken := r.FormValue("api_token")
 214         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 215                 // The client provided an explicit token in the POST
 216                 // body. The Origin header indicates this *might* be
 217                 // an AJAX request, in which case redirect-with-cookie
 218                 // won't work: we should just serve the content in the
 219                 // POST response. This is safe because:
 220                 //
 221                 // * We're supplying an attachment, not inline
 222                 //   content, so we don't need to convert the POST to
 223                 //   a GET and avoid the "really resubmit form?"
 224                 //   problem.
 225                 //
 226                 // * The token isn't embedded in the URL, so we don't
 227                 //   need to worry about bookmarks and copy/paste.
 228                 tokens = append(tokens, formToken)
 229         } else if formToken != "" {
 230                 // The client provided an explicit token in the query
 231                 // string, or a form in POST body. We must put the
 232                 // token in an HttpOnly cookie, and redirect to the
 233                 // same URL with the query param redacted and method =
 234                 // GET.
 235                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 236                 return
 237         }
 238
 239         targetPath := pathParts[stripParts:]
 240         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 241                 // http://ID.example/t=TOKEN/PATH...
 242                 // /c=ID/t=TOKEN/PATH...
 243                 //
 244                 // This form must only be used to pass scoped tokens
 245                 // that give permission for a single collection. See
 246                 // FormValue case above.
 247                 tokens = []string{targetPath[0][2:]}
 248                 pathToken = true
 249                 targetPath = targetPath[1:]
 250                 stripParts++
 251         }
 252
 253         if tokens == nil {
 254                 if credentialsOK {
 255                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 256                 }
 257                 tokens = append(reqTokens, h.Config.AnonymousTokens...)
 258         }
 259
 260         if len(targetPath) > 0 && targetPath[0] == "_" {
 261                 // If a collection has a directory called "t=foo" or
 262                 // "_", it can be served at
 263                 // //collections.example/_/t=foo/ or
 264                 // //collections.example/_/_/ respectively:
 265                 // //collections.example/t=foo/ won't work because
 266                 // t=foo will be interpreted as a token "foo".
 267                 targetPath = targetPath[1:]
 268                 stripParts++
 269         }
 270
 271         forceReload := false
 272         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 273                 forceReload = true
 274         }
 275
 276         var collection *arvados.Collection
 277         tokenResult := make(map[string]int)
 278         for _, arv.ApiToken = range tokens {
 279                 var err error
 280                 collection, err = h.Config.Cache.Get(arv, targetID, forceReload)
 281                 if err == nil {
 282                         // Success
 283                         break
 284                 }
 285                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 286                         switch srvErr.HttpStatusCode {
 287                         case 404, 401:
 288                                 // Token broken or insufficient to
 289                                 // retrieve collection
 290                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 291                                 continue
 292                         }
 293                 }
 294                 // Something more serious is wrong
 295                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 296                 return
 297         }
 298         if collection == nil {
 299                 if pathToken || !credentialsOK {
 300                         // Either the URL is a "secret sharing link"
 301                         // that didn't work out (and asking the client
 302                         // for additional credentials would just be
 303                         // confusing), or we don't even accept
 304                         // credentials at this path.
 305                         statusCode = http.StatusNotFound
 306                         return
 307                 }
 308                 for _, t := range reqTokens {
 309                         if tokenResult[t] == 404 {
 310                                 // The client provided valid token(s), but the
 311                                 // collection was not found.
 312                                 statusCode = http.StatusNotFound
 313                                 return
 314                         }
 315                 }
 316                 // The client's token was invalid (e.g., expired), or
 317                 // the client didn't even provide one.  Propagate the
 318                 // 401 to encourage the client to use a [different]
 319                 // token.
 320                 //
 321                 // TODO(TC): This response would be confusing to
 322                 // someone trying (anonymously) to download public
 323                 // data that has been deleted.  Allow a referrer to
 324                 // provide this context somehow?
 325                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 326                 statusCode = http.StatusUnauthorized
 327                 return
 328         }
 329
 330         kc, err := keepclient.MakeKeepClient(arv)
 331         if err != nil {
 332                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 333                 return
 334         }
 335
 336         basename := targetPath[len(targetPath)-1]
 337         applyContentDispositionHdr(w, r, basename, attachment)
 338
 339         fs := collection.FileSystem(&arvados.Client{
 340                 APIHost:   arv.ApiServer,
 341                 AuthToken: arv.ApiToken,
 342                 Insecure:  arv.ApiInsecure,
 343         }, kc)
 344         openPath := "/" + strings.Join(targetPath, "/")
 345         if f, err := fs.Open(openPath); os.IsNotExist(err) {
 346                 // Requested non-existent path
 347                 statusCode = http.StatusNotFound
 348         } else if err != nil {
 349                 // Some other (unexpected) error
 350                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 351         } else if stat, err := f.Stat(); err != nil {
 352                 // Can't get Size/IsDir (shouldn't happen with a collectionFS!)
 353                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 354         } else if stat.IsDir() && !strings.HasSuffix(r.URL.Path, "/") {
 355                 // If client requests ".../dirname", redirect to
 356                 // ".../dirname/". This way, relative links in the
 357                 // listing for "dirname" can always be "fnm", never
 358                 // "dirname/fnm".
 359                 h.seeOtherWithCookie(w, r, basename+"/", credentialsOK)
 360         } else if stat.IsDir() {
 361                 h.serveDirectory(w, r, collection.Name, fs, openPath, stripParts)
 362         } else {
 363                 http.ServeContent(w, r, basename, stat.ModTime(), f)
 364                 if r.Header.Get("Range") == "" && int64(w.WroteBodyBytes()) != stat.Size() {
 365                         // If we wrote fewer bytes than expected, it's
 366                         // too late to change the real response code
 367                         // or send an error message to the client, but
 368                         // at least we can try to put some useful
 369                         // debugging info in the logs.
 370                         n, err := f.Read(make([]byte, 1024))
 371                         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)
 372
 373                 }
 374         }
 375 }
 376
 377 var dirListingTemplate = `<!DOCTYPE HTML>
 378 <HTML><HEAD>
 379   <META name="robots" content="NOINDEX">
 380   <TITLE>{{ .Collection.Name }}</TITLE>
 381   <STYLE type="text/css">
 382     body {
 383       margin: 1.5em;
 384     }
 385     pre {
 386       background-color: #D9EDF7;
 387       border-radius: .25em;
 388       padding: .75em;
 389       overflow: auto;
 390     }
 391     .footer p {
 392       font-size: 82%;
 393     }
 394     ul {
 395       padding: 0;
 396     }
 397     ul li {
 398       font-family: monospace;
 399       list-style: none;
 400     }
 401   </STYLE>
 402 </HEAD>
 403 <BODY>
 404 <H1>{{ .CollectionName }}</H1>
 405
 406 <P>This collection of data files is being shared with you through
 407 Arvados.  You can download individual files listed below.  To download
 408 the entire collection with wget, try:</P>
 409
 410 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL }}</PRE>
 411
 412 <H2>File Listing</H2>
 413
 414 {{if .Files}}
 415 <UL>
 416 {{range .Files}}  <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{.Name}}">{{.Name}}</A></LI>{{end}}
 417 </UL>
 418 {{else}}
 419 <P>(No files; this collection is empty.)</P>
 420 {{end}}
 421
 422 <HR noshade>
 423 <DIV class="footer">
 424   <P>
 425     About Arvados:
 426     Arvados is a free and open source software bioinformatics platform.
 427     To learn more, visit arvados.org.
 428     Arvados is not responsible for the files listed on this page.
 429   </P>
 430 </DIV>
 431
 432 </BODY>
 433 `
 434
 435 type fileListEnt struct {
 436         Name string
 437         Size int64
 438 }
 439
 440 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, stripParts int) {
 441         var files []fileListEnt
 442         var walk func(string) error
 443         if !strings.HasSuffix(base, "/") {
 444                 base = base + "/"
 445         }
 446         walk = func(path string) error {
 447                 dirname := base + path
 448                 if dirname != "/" {
 449                         dirname = strings.TrimSuffix(dirname, "/")
 450                 }
 451                 d, err := fs.Open(dirname)
 452                 if err != nil {
 453                         return err
 454                 }
 455                 ents, err := d.Readdir(-1)
 456                 if err != nil {
 457                         return err
 458                 }
 459                 for _, ent := range ents {
 460                         if ent.IsDir() {
 461                                 err = walk(path + ent.Name() + "/")
 462                                 if err != nil {
 463                                         return err
 464                                 }
 465                         } else {
 466                                 files = append(files, fileListEnt{
 467                                         Name: path + ent.Name(),
 468                                         Size: ent.Size(),
 469                                 })
 470                         }
 471                 }
 472                 return nil
 473         }
 474         if err := walk(""); err != nil {
 475                 http.Error(w, err.Error(), http.StatusInternalServerError)
 476                 return
 477         }
 478
 479         funcs := template.FuncMap{
 480                 "nbsp": func(s string) template.HTML {
 481                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 482                 },
 483         }
 484         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 485         if err != nil {
 486                 http.Error(w, err.Error(), http.StatusInternalServerError)
 487                 return
 488         }
 489         sort.Slice(files, func(i, j int) bool {
 490                 return files[i].Name < files[j].Name
 491         })
 492         w.WriteHeader(http.StatusOK)
 493         tmpl.Execute(w, map[string]interface{}{
 494                 "CollectionName": collectionName,
 495                 "Files":          files,
 496                 "Request":        r,
 497                 "StripParts":     stripParts,
 498         })
 499 }
 500
 501 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 502         disposition := "inline"
 503         if isAttachment {
 504                 disposition = "attachment"
 505         }
 506         if strings.ContainsRune(r.RequestURI, '?') {
 507                 // Help the UA realize that the filename is just
 508                 // "filename.txt", not
 509                 // "filename.txt?disposition=attachment".
 510                 //
 511                 // TODO(TC): Follow advice at RFC 6266 appendix D
 512                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 513         }
 514         if disposition != "inline" {
 515                 w.Header().Set("Content-Disposition", disposition)
 516         }
 517 }
 518
 519 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 520         if !credentialsOK {
 521                 // It is not safe to copy the provided token
 522                 // into a cookie unless the current vhost
 523                 // (origin) serves only a single collection or
 524                 // we are in TrustAllContent mode.
 525                 w.WriteHeader(http.StatusBadRequest)
 526                 return
 527         }
 528
 529         if formToken := r.FormValue("api_token"); formToken != "" {
 530                 // The HttpOnly flag is necessary to prevent
 531                 // JavaScript code (included in, or loaded by, a page
 532                 // in the collection being served) from employing the
 533                 // user's token beyond reading other files in the same
 534                 // domain, i.e., same collection.
 535                 //
 536                 // The 303 redirect is necessary in the case of a GET
 537                 // request to avoid exposing the token in the Location
 538                 // bar, and in the case of a POST request to avoid
 539                 // raising warnings when the user refreshes the
 540                 // resulting page.
 541
 542                 http.SetCookie(w, &http.Cookie{
 543                         Name:     "arvados_api_token",
 544                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 545                         Path:     "/",
 546                         HttpOnly: true,
 547                 })
 548         }
 549
 550         // Propagate query parameters (except api_token) from
 551         // the original request.
 552         redirQuery := r.URL.Query()
 553         redirQuery.Del("api_token")
 554
 555         u := r.URL
 556         if location != "" {
 557                 newu, err := u.Parse(location)
 558                 if err != nil {
 559                         w.WriteHeader(http.StatusInternalServerError)
 560                         return
 561                 }
 562                 u = newu
 563         }
 564         redir := (&url.URL{
 565                 Host:     r.Host,
 566                 Path:     u.Path,
 567                 RawQuery: redirQuery.Encode(),
 568         }).String()
 569
 570         w.Header().Add("Location", redir)
 571         w.WriteHeader(http.StatusSeeOther)
 572         io.WriteString(w, `<A href="`)
 573         io.WriteString(w, html.EscapeString(redir))
 574         io.WriteString(w, `">Continue</A>`)
 575 }