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/httpserver"
  25         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  26 )
  27
  28 type handler struct {
  29         Config     *Config
  30         clientPool *arvadosclient.ClientPool
  31         setupOnce  sync.Once
  32 }
  33
  34 // parseCollectionIDFromDNSName returns a UUID or PDH if s begins with
  35 // a UUID or URL-encoded PDH; otherwise "".
  36 func parseCollectionIDFromDNSName(s string) string {
  37         // Strip domain.
  38         if i := strings.IndexRune(s, '.'); i >= 0 {
  39                 s = s[:i]
  40         }
  41         // Names like {uuid}--collections.example.com serve the same
  42         // purpose as {uuid}.collections.example.com but can reduce
  43         // cost/effort of using [additional] wildcard certificates.
  44         if i := strings.Index(s, "--"); i >= 0 {
  45                 s = s[:i]
  46         }
  47         if arvadosclient.UUIDMatch(s) {
  48                 return s
  49         }
  50         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  51                 return pdh
  52         }
  53         return ""
  54 }
  55
  56 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  57
  58 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  59 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  60 // otherwise "".
  61 func parseCollectionIDFromURL(s string) string {
  62         if arvadosclient.UUIDMatch(s) {
  63                 return s
  64         }
  65         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  66                 return pdh
  67         }
  68         return ""
  69 }
  70
  71 func (h *handler) setup() {
  72         h.clientPool = arvadosclient.MakeClientPool()
  73         keepclient.RefreshServiceDiscoveryOnSIGHUP()
  74 }
  75
  76 func (h *handler) serveStatus(w http.ResponseWriter, r *http.Request) {
  77         status := struct {
  78                 cacheStats
  79         }{
  80                 cacheStats: h.Config.Cache.Stats(),
  81         }
  82         json.NewEncoder(w).Encode(status)
  83 }
  84
  85 // ServeHTTP implements http.Handler.
  86 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
  87         h.setupOnce.Do(h.setup)
  88
  89         var statusCode = 0
  90         var statusText string
  91
  92         remoteAddr := r.RemoteAddr
  93         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
  94                 remoteAddr = xff + "," + remoteAddr
  95         }
  96
  97         w := httpserver.WrapResponseWriter(wOrig)
  98         defer func() {
  99                 if statusCode == 0 {
 100                         statusCode = w.WroteStatus()
 101                 } else if w.WroteStatus() == 0 {
 102                         w.WriteHeader(statusCode)
 103                 } else if w.WroteStatus() != statusCode {
 104                         httpserver.Log(r.RemoteAddr, "WARNING",
 105                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
 106                 }
 107                 if statusText == "" {
 108                         statusText = http.StatusText(statusCode)
 109                 }
 110                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
 111         }()
 112
 113         if r.Method == "OPTIONS" {
 114                 method := r.Header.Get("Access-Control-Request-Method")
 115                 if method != "GET" && method != "POST" {
 116                         statusCode = http.StatusMethodNotAllowed
 117                         return
 118                 }
 119                 w.Header().Set("Access-Control-Allow-Headers", "Range")
 120                 w.Header().Set("Access-Control-Allow-Methods", "GET, POST")
 121                 w.Header().Set("Access-Control-Allow-Origin", "*")
 122                 w.Header().Set("Access-Control-Max-Age", "86400")
 123                 statusCode = http.StatusOK
 124                 return
 125         }
 126
 127         if r.Method != "GET" && r.Method != "POST" {
 128                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
 129                 return
 130         }
 131
 132         if r.Header.Get("Origin") != "" {
 133                 // Allow simple cross-origin requests without user
 134                 // credentials ("user credentials" as defined by CORS,
 135                 // i.e., cookies, HTTP authentication, and client-side
 136                 // SSL certificates. See
 137                 // http://www.w3.org/TR/cors/#user-credentials).
 138                 w.Header().Set("Access-Control-Allow-Origin", "*")
 139         }
 140
 141         arv := h.clientPool.Get()
 142         if arv == nil {
 143                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+h.clientPool.Err().Error()
 144                 return
 145         }
 146         defer h.clientPool.Put(arv)
 147
 148         pathParts := strings.Split(r.URL.Path[1:], "/")
 149
 150         var stripParts int
 151         var targetID string
 152         var tokens []string
 153         var reqTokens []string
 154         var pathToken bool
 155         var attachment bool
 156         credentialsOK := h.Config.TrustAllContent
 157
 158         if r.Host != "" && r.Host == h.Config.AttachmentOnlyHost {
 159                 credentialsOK = true
 160                 attachment = true
 161         } else if r.FormValue("disposition") == "attachment" {
 162                 attachment = true
 163         }
 164
 165         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 166                 // http://ID.collections.example/PATH...
 167                 credentialsOK = true
 168         } else if r.URL.Path == "/status.json" {
 169                 h.serveStatus(w, r)
 170                 return
 171         } else if len(pathParts) >= 1 && strings.HasPrefix(pathParts[0], "c=") {
 172                 // /c=ID[/PATH...]
 173                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 174                 stripParts = 1
 175         } else if len(pathParts) >= 2 && pathParts[0] == "collections" {
 176                 if len(pathParts) >= 4 && pathParts[1] == "download" {
 177                         // /collections/download/ID/TOKEN/PATH...
 178                         targetID = parseCollectionIDFromURL(pathParts[2])
 179                         tokens = []string{pathParts[3]}
 180                         stripParts = 4
 181                         pathToken = true
 182                 } else {
 183                         // /collections/ID/PATH...
 184                         targetID = parseCollectionIDFromURL(pathParts[1])
 185                         tokens = h.Config.AnonymousTokens
 186                         stripParts = 2
 187                 }
 188         }
 189
 190         if targetID == "" {
 191                 statusCode = http.StatusNotFound
 192                 return
 193         }
 194
 195         formToken := r.FormValue("api_token")
 196         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 197                 // The client provided an explicit token in the POST
 198                 // body. The Origin header indicates this *might* be
 199                 // an AJAX request, in which case redirect-with-cookie
 200                 // won't work: we should just serve the content in the
 201                 // POST response. This is safe because:
 202                 //
 203                 // * We're supplying an attachment, not inline
 204                 //   content, so we don't need to convert the POST to
 205                 //   a GET and avoid the "really resubmit form?"
 206                 //   problem.
 207                 //
 208                 // * The token isn't embedded in the URL, so we don't
 209                 //   need to worry about bookmarks and copy/paste.
 210                 tokens = append(tokens, formToken)
 211         } else if formToken != "" {
 212                 // The client provided an explicit token in the query
 213                 // string, or a form in POST body. We must put the
 214                 // token in an HttpOnly cookie, and redirect to the
 215                 // same URL with the query param redacted and method =
 216                 // GET.
 217                 h.seeOtherWithCookie(w, r, "", credentialsOK)
 218                 return
 219         }
 220
 221         targetPath := pathParts[stripParts:]
 222         if tokens == nil && len(targetPath) > 0 && strings.HasPrefix(targetPath[0], "t=") {
 223                 // http://ID.example/t=TOKEN/PATH...
 224                 // /c=ID/t=TOKEN/PATH...
 225                 //
 226                 // This form must only be used to pass scoped tokens
 227                 // that give permission for a single collection. See
 228                 // FormValue case above.
 229                 tokens = []string{targetPath[0][2:]}
 230                 pathToken = true
 231                 targetPath = targetPath[1:]
 232                 stripParts++
 233         }
 234
 235         if tokens == nil {
 236                 if credentialsOK {
 237                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 238                 }
 239                 tokens = append(reqTokens, h.Config.AnonymousTokens...)
 240         }
 241
 242         if len(targetPath) > 0 && targetPath[0] == "_" {
 243                 // If a collection has a directory called "t=foo" or
 244                 // "_", it can be served at
 245                 // //collections.example/_/t=foo/ or
 246                 // //collections.example/_/_/ respectively:
 247                 // //collections.example/t=foo/ won't work because
 248                 // t=foo will be interpreted as a token "foo".
 249                 targetPath = targetPath[1:]
 250                 stripParts++
 251         }
 252
 253         forceReload := false
 254         if cc := r.Header.Get("Cache-Control"); strings.Contains(cc, "no-cache") || strings.Contains(cc, "must-revalidate") {
 255                 forceReload = true
 256         }
 257
 258         var collection *arvados.Collection
 259         tokenResult := make(map[string]int)
 260         for _, arv.ApiToken = range tokens {
 261                 var err error
 262                 collection, err = h.Config.Cache.Get(arv, targetID, forceReload)
 263                 if err == nil {
 264                         // Success
 265                         break
 266                 }
 267                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 268                         switch srvErr.HttpStatusCode {
 269                         case 404, 401:
 270                                 // Token broken or insufficient to
 271                                 // retrieve collection
 272                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 273                                 continue
 274                         }
 275                 }
 276                 // Something more serious is wrong
 277                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 278                 return
 279         }
 280         if collection == nil {
 281                 if pathToken || !credentialsOK {
 282                         // Either the URL is a "secret sharing link"
 283                         // that didn't work out (and asking the client
 284                         // for additional credentials would just be
 285                         // confusing), or we don't even accept
 286                         // credentials at this path.
 287                         statusCode = http.StatusNotFound
 288                         return
 289                 }
 290                 for _, t := range reqTokens {
 291                         if tokenResult[t] == 404 {
 292                                 // The client provided valid token(s), but the
 293                                 // collection was not found.
 294                                 statusCode = http.StatusNotFound
 295                                 return
 296                         }
 297                 }
 298                 // The client's token was invalid (e.g., expired), or
 299                 // the client didn't even provide one.  Propagate the
 300                 // 401 to encourage the client to use a [different]
 301                 // token.
 302                 //
 303                 // TODO(TC): This response would be confusing to
 304                 // someone trying (anonymously) to download public
 305                 // data that has been deleted.  Allow a referrer to
 306                 // provide this context somehow?
 307                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 308                 statusCode = http.StatusUnauthorized
 309                 return
 310         }
 311
 312         kc, err := keepclient.MakeKeepClient(arv)
 313         if err != nil {
 314                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 315                 return
 316         }
 317
 318         basename := targetPath[len(targetPath)-1]
 319         applyContentDispositionHdr(w, r, basename, attachment)
 320
 321         fs := collection.FileSystem(&arvados.Client{
 322                 APIHost:   arv.ApiServer,
 323                 AuthToken: arv.ApiToken,
 324                 Insecure:  arv.ApiInsecure,
 325         }, kc)
 326         openPath := "/" + strings.Join(targetPath, "/")
 327         if f, err := fs.Open(openPath); os.IsNotExist(err) {
 328                 // Requested non-existent path
 329                 statusCode = http.StatusNotFound
 330         } else if err != nil {
 331                 // Some other (unexpected) error
 332                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 333         } else if stat, err := f.Stat(); err != nil {
 334                 // Can't get Size/IsDir (shouldn't happen with a collectionFS!)
 335                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 336         } else if stat.IsDir() && !strings.HasSuffix(r.URL.Path, "/") {
 337                 // If client requests ".../dirname", redirect to
 338                 // ".../dirname/". This way, relative links in the
 339                 // listing for "dirname" can always be "fnm", never
 340                 // "dirname/fnm".
 341                 h.seeOtherWithCookie(w, r, basename+"/", credentialsOK)
 342         } else if stat.IsDir() {
 343                 h.serveDirectory(w, r, collection.Name, fs, openPath, stripParts)
 344         } else {
 345                 http.ServeContent(w, r, basename, stat.ModTime(), f)
 346                 if r.Header.Get("Range") == "" && int64(w.WroteBodyBytes()) != stat.Size() {
 347                         // If we wrote fewer bytes than expected, it's
 348                         // too late to change the real response code
 349                         // or send an error message to the client, but
 350                         // at least we can try to put some useful
 351                         // debugging info in the logs.
 352                         n, err := f.Read(make([]byte, 1024))
 353                         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)
 354
 355                 }
 356         }
 357 }
 358
 359 var dirListingTemplate = `<!DOCTYPE HTML>
 360 <HTML><HEAD>
 361   <META name="robots" content="NOINDEX">
 362   <TITLE>{{ .Collection.Name }}</TITLE>
 363   <STYLE type="text/css">
 364     body {
 365       margin: 1.5em;
 366     }
 367     pre {
 368       background-color: #D9EDF7;
 369       border-radius: .25em;
 370       padding: .75em;
 371       overflow: auto;
 372     }
 373     .footer p {
 374       font-size: 82%;
 375     }
 376     ul {
 377       padding: 0;
 378     }
 379     ul li {
 380       font-family: monospace;
 381       list-style: none;
 382     }
 383   </STYLE>
 384 </HEAD>
 385 <BODY>
 386 <H1>{{ .CollectionName }}</H1>
 387
 388 <P>This collection of data files is being shared with you through
 389 Arvados.  You can download individual files listed below.  To download
 390 the entire collection with wget, try:</P>
 391
 392 <PRE>$ wget --mirror --no-parent --no-host --cut-dirs={{ .StripParts }} https://{{ .Request.Host }}{{ .Request.URL }}</PRE>
 393
 394 <H2>File Listing</H2>
 395
 396 {{if .Files}}
 397 <UL>
 398 {{range .Files}}  <LI>{{.Size | printf "%15d  " | nbsp}}<A href="{{.Name}}">{{.Name}}</A></LI>{{end}}
 399 </UL>
 400 {{else}}
 401 <P>(No files; this collection is empty.)</P>
 402 {{end}}
 403
 404 <HR noshade>
 405 <DIV class="footer">
 406   <P>
 407     About Arvados:
 408     Arvados is a free and open source software bioinformatics platform.
 409     To learn more, visit arvados.org.
 410     Arvados is not responsible for the files listed on this page.
 411   </P>
 412 </DIV>
 413
 414 </BODY>
 415 `
 416
 417 type fileListEnt struct {
 418         Name string
 419         Size int64
 420 }
 421
 422 func (h *handler) serveDirectory(w http.ResponseWriter, r *http.Request, collectionName string, fs http.FileSystem, base string, stripParts int) {
 423         var files []fileListEnt
 424         var walk func(string) error
 425         if !strings.HasSuffix(base, "/") {
 426                 base = base + "/"
 427         }
 428         walk = func(path string) error {
 429                 dirname := base + path
 430                 if dirname != "/" {
 431                         dirname = strings.TrimSuffix(dirname, "/")
 432                 }
 433                 d, err := fs.Open(dirname)
 434                 if err != nil {
 435                         return err
 436                 }
 437                 ents, err := d.Readdir(-1)
 438                 if err != nil {
 439                         return err
 440                 }
 441                 for _, ent := range ents {
 442                         if ent.IsDir() {
 443                                 err = walk(path + ent.Name() + "/")
 444                                 if err != nil {
 445                                         return err
 446                                 }
 447                         } else {
 448                                 files = append(files, fileListEnt{
 449                                         Name: path + ent.Name(),
 450                                         Size: ent.Size(),
 451                                 })
 452                         }
 453                 }
 454                 return nil
 455         }
 456         if err := walk(""); err != nil {
 457                 http.Error(w, err.Error(), http.StatusInternalServerError)
 458                 return
 459         }
 460
 461         funcs := template.FuncMap{
 462                 "nbsp": func(s string) template.HTML {
 463                         return template.HTML(strings.Replace(s, " ", "&nbsp;", -1))
 464                 },
 465         }
 466         tmpl, err := template.New("dir").Funcs(funcs).Parse(dirListingTemplate)
 467         if err != nil {
 468                 http.Error(w, err.Error(), http.StatusInternalServerError)
 469                 return
 470         }
 471         sort.Slice(files, func(i, j int) bool {
 472                 return files[i].Name < files[j].Name
 473         })
 474         w.WriteHeader(http.StatusOK)
 475         tmpl.Execute(w, map[string]interface{}{
 476                 "CollectionName": collectionName,
 477                 "Files":          files,
 478                 "Request":        r,
 479                 "StripParts":     stripParts,
 480         })
 481 }
 482
 483 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 484         disposition := "inline"
 485         if isAttachment {
 486                 disposition = "attachment"
 487         }
 488         if strings.ContainsRune(r.RequestURI, '?') {
 489                 // Help the UA realize that the filename is just
 490                 // "filename.txt", not
 491                 // "filename.txt?disposition=attachment".
 492                 //
 493                 // TODO(TC): Follow advice at RFC 6266 appendix D
 494                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 495         }
 496         if disposition != "inline" {
 497                 w.Header().Set("Content-Disposition", disposition)
 498         }
 499 }
 500
 501 func (h *handler) seeOtherWithCookie(w http.ResponseWriter, r *http.Request, location string, credentialsOK bool) {
 502         if !credentialsOK {
 503                 // It is not safe to copy the provided token
 504                 // into a cookie unless the current vhost
 505                 // (origin) serves only a single collection or
 506                 // we are in TrustAllContent mode.
 507                 w.WriteHeader(http.StatusBadRequest)
 508                 return
 509         }
 510
 511         if formToken := r.FormValue("api_token"); formToken != "" {
 512                 // The HttpOnly flag is necessary to prevent
 513                 // JavaScript code (included in, or loaded by, a page
 514                 // in the collection being served) from employing the
 515                 // user's token beyond reading other files in the same
 516                 // domain, i.e., same collection.
 517                 //
 518                 // The 303 redirect is necessary in the case of a GET
 519                 // request to avoid exposing the token in the Location
 520                 // bar, and in the case of a POST request to avoid
 521                 // raising warnings when the user refreshes the
 522                 // resulting page.
 523
 524                 http.SetCookie(w, &http.Cookie{
 525                         Name:     "arvados_api_token",
 526                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 527                         Path:     "/",
 528                         HttpOnly: true,
 529                 })
 530         }
 531
 532         // Propagate query parameters (except api_token) from
 533         // the original request.
 534         redirQuery := r.URL.Query()
 535         redirQuery.Del("api_token")
 536
 537         u := r.URL
 538         if location != "" {
 539                 newu, err := u.Parse(location)
 540                 if err != nil {
 541                         w.WriteHeader(http.StatusInternalServerError)
 542                         return
 543                 }
 544                 u = newu
 545         }
 546         redir := (&url.URL{
 547                 Host:     r.Host,
 548                 Path:     u.Path,
 549                 RawQuery: redirQuery.Encode(),
 550         }).String()
 551
 552         w.Header().Add("Location", redir)
 553         w.WriteHeader(http.StatusSeeOther)
 554         io.WriteString(w, `<A href="`)
 555         io.WriteString(w, html.EscapeString(redir))
 556         io.WriteString(w, `">Continue</A>`)
 557 }