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