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