services/keep-web/handler.go

   1 package main
   2
   3 import (
   4         "flag"
   5         "fmt"
   6         "html"
   7         "io"
   8         "mime"
   9         "net/http"
  10         "net/url"
  11         "os"
  12         "regexp"
  13         "strconv"
  14         "strings"
  15
  16         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  17         "git.curoverse.com/arvados.git/sdk/go/auth"
  18         "git.curoverse.com/arvados.git/sdk/go/httpserver"
  19         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  20 )
  21
  22 type handler struct{}
  23
  24 var (
  25         clientPool         = arvadosclient.MakeClientPool()
  26         trustAllContent    = false
  27         attachmentOnlyHost = ""
  28 )
  29
  30 func init() {
  31         flag.StringVar(&attachmentOnlyHost, "attachment-only-host", "",
  32                 "Accept credentials, and add \"Content-Disposition: attachment\" response headers, for requests at this hostname:port. Prohibiting inline display makes it possible to serve untrusted and non-public content from a single origin, i.e., without wildcard DNS or SSL.")
  33         flag.BoolVar(&trustAllContent, "trust-all-content", false,
  34                 "Serve non-public content from a single origin. Dangerous: read docs before using!")
  35 }
  36
  37 // return a UUID or PDH if s begins with a UUID or URL-encoded PDH;
  38 // otherwise return "".
  39 func parseCollectionIDFromDNSName(s string) string {
  40         // Strip domain.
  41         if i := strings.IndexRune(s, '.'); i >= 0 {
  42                 s = s[:i]
  43         }
  44         // Names like {uuid}--collections.example.com serve the same
  45         // purpose as {uuid}.collections.example.com but can reduce
  46         // cost/effort of using [additional] wildcard certificates.
  47         if i := strings.Index(s, "--"); i >= 0 {
  48                 s = s[:i]
  49         }
  50         if arvadosclient.UUIDMatch(s) {
  51                 return s
  52         }
  53         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  54                 return pdh
  55         }
  56         return ""
  57 }
  58
  59 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  60
  61 // return a UUID or PDH if s is a UUID or a PDH (even if it is a PDH
  62 // with "+" replaced by " " or "-"); otherwise return "".
  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) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
  74         var statusCode = 0
  75         var statusText string
  76
  77         remoteAddr := r.RemoteAddr
  78         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
  79                 remoteAddr = xff + "," + remoteAddr
  80         }
  81
  82         w := httpserver.WrapResponseWriter(wOrig)
  83         defer func() {
  84                 if statusCode == 0 {
  85                         statusCode = w.WroteStatus()
  86                 } else if w.WroteStatus() == 0 {
  87                         w.WriteHeader(statusCode)
  88                 } else if w.WroteStatus() != statusCode {
  89                         httpserver.Log(r.RemoteAddr, "WARNING",
  90                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
  91                 }
  92                 if statusText == "" {
  93                         statusText = http.StatusText(statusCode)
  94                 }
  95                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
  96         }()
  97
  98         if r.Method != "GET" && r.Method != "POST" {
  99                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
 100                 return
 101         }
 102
 103         if r.Header.Get("Origin") != "" {
 104                 // Allow simple cross-origin requests without user
 105                 // credentials ("user credentials" as defined by CORS,
 106                 // i.e., cookies, HTTP authentication, and client-side
 107                 // SSL certificates. See
 108                 // http://www.w3.org/TR/cors/#user-credentials).
 109                 w.Header().Set("Access-Control-Allow-Origin", "*")
 110         }
 111
 112         arv := clientPool.Get()
 113         if arv == nil {
 114                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+clientPool.Err().Error()
 115                 return
 116         }
 117         defer clientPool.Put(arv)
 118
 119         pathParts := strings.Split(r.URL.Path[1:], "/")
 120
 121         var targetID string
 122         var targetPath []string
 123         var tokens []string
 124         var reqTokens []string
 125         var pathToken bool
 126         var attachment bool
 127         credentialsOK := trustAllContent
 128
 129         if r.Host != "" && r.Host == attachmentOnlyHost {
 130                 credentialsOK = true
 131                 attachment = true
 132         } else if r.FormValue("disposition") == "attachment" {
 133                 attachment = true
 134         }
 135
 136         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 137                 // http://ID.collections.example/PATH...
 138                 credentialsOK = true
 139                 targetPath = pathParts
 140         } else if len(pathParts) >= 2 && strings.HasPrefix(pathParts[0], "c=") {
 141                 // /c=ID/PATH...
 142                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 143                 targetPath = pathParts[1:]
 144         } else if len(pathParts) >= 3 && pathParts[0] == "collections" {
 145                 if len(pathParts) >= 5 && pathParts[1] == "download" {
 146                         // /collections/download/ID/TOKEN/PATH...
 147                         targetID = pathParts[2]
 148                         tokens = []string{pathParts[3]}
 149                         targetPath = pathParts[4:]
 150                         pathToken = true
 151                 } else {
 152                         // /collections/ID/PATH...
 153                         targetID = pathParts[1]
 154                         tokens = anonymousTokens
 155                         targetPath = pathParts[2:]
 156                 }
 157         } else {
 158                 statusCode = http.StatusNotFound
 159                 return
 160         }
 161
 162         formToken := r.FormValue("api_token")
 163         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 164                 // The client provided an explicit token in the POST
 165                 // body. The Origin header indicates this *might* be
 166                 // an AJAX request, in which case redirect-with-cookie
 167                 // won't work: we should just serve the content in the
 168                 // POST response. This is safe because:
 169                 //
 170                 // * We're supplying an attachment, not inline
 171                 //   content, so we don't need to convert the POST to
 172                 //   a GET and avoid the "really resubmit form?"
 173                 //   problem.
 174                 //
 175                 // * The token isn't embedded in the URL, so we don't
 176                 //   need to worry about bookmarks and copy/paste.
 177                 tokens = append(tokens, formToken)
 178         } else if formToken != "" {
 179                 // The client provided an explicit token in the query
 180                 // string, or a form in POST body. We must put the
 181                 // token in an HttpOnly cookie, and redirect to the
 182                 // same URL with the query param redacted and method =
 183                 // GET.
 184
 185                 if !credentialsOK {
 186                         // It is not safe to copy the provided token
 187                         // into a cookie unless the current vhost
 188                         // (origin) serves only a single collection or
 189                         // we are in trustAllContent mode.
 190                         statusCode = http.StatusBadRequest
 191                         return
 192                 }
 193
 194                 // The HttpOnly flag is necessary to prevent
 195                 // JavaScript code (included in, or loaded by, a page
 196                 // in the collection being served) from employing the
 197                 // user's token beyond reading other files in the same
 198                 // domain, i.e., same collection.
 199                 //
 200                 // The 303 redirect is necessary in the case of a GET
 201                 // request to avoid exposing the token in the Location
 202                 // bar, and in the case of a POST request to avoid
 203                 // raising warnings when the user refreshes the
 204                 // resulting page.
 205
 206                 http.SetCookie(w, &http.Cookie{
 207                         Name:     "arvados_api_token",
 208                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 209                         Path:     "/",
 210                         HttpOnly: true,
 211                 })
 212
 213                 // Propagate query parameters (except api_token) from
 214                 // the original request.
 215                 redirQuery := r.URL.Query()
 216                 redirQuery.Del("api_token")
 217
 218                 redir := (&url.URL{
 219                         Host:     r.Host,
 220                         Path:     r.URL.Path,
 221                         RawQuery: redirQuery.Encode(),
 222                 }).String()
 223
 224                 w.Header().Add("Location", redir)
 225                 statusCode, statusText = http.StatusSeeOther, redir
 226                 w.WriteHeader(statusCode)
 227                 io.WriteString(w, `<A href="`)
 228                 io.WriteString(w, html.EscapeString(redir))
 229                 io.WriteString(w, `">Continue</A>`)
 230                 return
 231         }
 232
 233         if tokens == nil && strings.HasPrefix(targetPath[0], "t=") {
 234                 // http://ID.example/t=TOKEN/PATH...
 235                 // /c=ID/t=TOKEN/PATH...
 236                 //
 237                 // This form must only be used to pass scoped tokens
 238                 // that give permission for a single collection. See
 239                 // FormValue case above.
 240                 tokens = []string{targetPath[0][2:]}
 241                 pathToken = true
 242                 targetPath = targetPath[1:]
 243         }
 244
 245         if tokens == nil {
 246                 if credentialsOK {
 247                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 248                 }
 249                 tokens = append(reqTokens, anonymousTokens...)
 250         }
 251
 252         if len(targetPath) > 0 && targetPath[0] == "_" {
 253                 // If a collection has a directory called "t=foo" or
 254                 // "_", it can be served at
 255                 // //collections.example/_/t=foo/ or
 256                 // //collections.example/_/_/ respectively:
 257                 // //collections.example/t=foo/ won't work because
 258                 // t=foo will be interpreted as a token "foo".
 259                 targetPath = targetPath[1:]
 260         }
 261
 262         tokenResult := make(map[string]int)
 263         collection := make(map[string]interface{})
 264         found := false
 265         for _, arv.ApiToken = range tokens {
 266                 err := arv.Get("collections", targetID, nil, &collection)
 267                 if err == nil {
 268                         // Success
 269                         found = true
 270                         break
 271                 }
 272                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 273                         switch srvErr.HttpStatusCode {
 274                         case 404, 401:
 275                                 // Token broken or insufficient to
 276                                 // retrieve collection
 277                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 278                                 continue
 279                         }
 280                 }
 281                 // Something more serious is wrong
 282                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 283                 return
 284         }
 285         if !found {
 286                 if pathToken || !credentialsOK {
 287                         // Either the URL is a "secret sharing link"
 288                         // that didn't work out (and asking the client
 289                         // for additional credentials would just be
 290                         // confusing), or we don't even accept
 291                         // credentials at this path.
 292                         statusCode = http.StatusNotFound
 293                         return
 294                 }
 295                 for _, t := range reqTokens {
 296                         if tokenResult[t] == 404 {
 297                                 // The client provided valid token(s), but the
 298                                 // collection was not found.
 299                                 statusCode = http.StatusNotFound
 300                                 return
 301                         }
 302                 }
 303                 // The client's token was invalid (e.g., expired), or
 304                 // the client didn't even provide one.  Propagate the
 305                 // 401 to encourage the client to use a [different]
 306                 // token.
 307                 //
 308                 // TODO(TC): This response would be confusing to
 309                 // someone trying (anonymously) to download public
 310                 // data that has been deleted.  Allow a referrer to
 311                 // provide this context somehow?
 312                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 313                 statusCode = http.StatusUnauthorized
 314                 return
 315         }
 316
 317         filename := strings.Join(targetPath, "/")
 318         kc, err := keepclient.MakeKeepClient(arv)
 319         if err != nil {
 320                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 321                 return
 322         }
 323         if kc.Client != nil && kc.Client.Transport != nil {
 324                 // Workaround for https://dev.arvados.org/issues/9005
 325                 if t, ok := kc.Client.Transport.(*http.Transport); ok {
 326                         defer t.CloseIdleConnections()
 327                 }
 328         }
 329         rdr, err := kc.CollectionFileReader(collection, filename)
 330         if os.IsNotExist(err) {
 331                 statusCode = http.StatusNotFound
 332                 return
 333         } else if err != nil {
 334                 statusCode, statusText = http.StatusBadGateway, err.Error()
 335                 return
 336         }
 337         defer rdr.Close()
 338
 339         basenamePos := strings.LastIndex(filename, "/")
 340         if basenamePos < 0 {
 341                 basenamePos = 0
 342         }
 343         extPos := strings.LastIndex(filename, ".")
 344         if extPos > basenamePos {
 345                 // Now extPos is safely >= 0.
 346                 if t := mime.TypeByExtension(filename[extPos:]); t != "" {
 347                         w.Header().Set("Content-Type", t)
 348                 }
 349         }
 350         if rdr, ok := rdr.(keepclient.ReadCloserWithLen); ok {
 351                 w.Header().Set("Content-Length", fmt.Sprintf("%d", rdr.Len()))
 352         }
 353
 354         applyContentDispositionHdr(w, r, filename[basenamePos:], attachment)
 355         rangeRdr, statusCode := applyRangeHdr(w, r, rdr)
 356
 357         w.WriteHeader(statusCode)
 358         _, err = io.Copy(w, rangeRdr)
 359         if err != nil {
 360                 statusCode, statusText = http.StatusBadGateway, err.Error()
 361         }
 362 }
 363
 364 var rangeRe = regexp.MustCompile(`^bytes=0-([0-9]*)$`)
 365
 366 func applyRangeHdr(w http.ResponseWriter, r *http.Request, rdr keepclient.ReadCloserWithLen) (io.Reader, int) {
 367         w.Header().Set("Accept-Ranges", "bytes")
 368         hdr := r.Header.Get("Range")
 369         fields := rangeRe.FindStringSubmatch(hdr)
 370         if fields == nil {
 371                 return rdr, http.StatusOK
 372         }
 373         rangeEnd, err := strconv.ParseInt(fields[1], 10, 64)
 374         if err != nil {
 375                 // Empty or too big for int64 == send entire content
 376                 return rdr, http.StatusOK
 377         }
 378         if uint64(rangeEnd) >= rdr.Len() {
 379                 return rdr, http.StatusOK
 380         }
 381         w.Header().Set("Content-Length", fmt.Sprintf("%d", rangeEnd+1))
 382         w.Header().Set("Content-Range", fmt.Sprintf("bytes %d-%d/%d", 0, rangeEnd, rdr.Len()))
 383         return &io.LimitedReader{R: rdr, N: rangeEnd + 1}, http.StatusPartialContent
 384 }
 385
 386 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 387         disposition := "inline"
 388         if isAttachment {
 389                 disposition = "attachment"
 390         }
 391         if strings.ContainsRune(r.RequestURI, '?') {
 392                 // Help the UA realize that the filename is just
 393                 // "filename.txt", not
 394                 // "filename.txt?disposition=attachment".
 395                 //
 396                 // TODO(TC): Follow advice at RFC 6266 appendix D
 397                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 398         }
 399         if disposition != "inline" {
 400                 w.Header().Set("Content-Disposition", disposition)
 401         }
 402 }