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