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         arv := clientPool.Get()
 104         if arv == nil {
 105                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+clientPool.Err().Error()
 106                 return
 107         }
 108         defer clientPool.Put(arv)
 109
 110         pathParts := strings.Split(r.URL.Path[1:], "/")
 111
 112         var targetID string
 113         var targetPath []string
 114         var tokens []string
 115         var reqTokens []string
 116         var pathToken bool
 117         var attachment bool
 118         credentialsOK := trustAllContent
 119
 120         if r.Host != "" && r.Host == attachmentOnlyHost {
 121                 credentialsOK = true
 122                 attachment = true
 123         } else if r.FormValue("disposition") == "attachment" {
 124                 attachment = true
 125         }
 126
 127         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 128                 // http://ID.collections.example/PATH...
 129                 credentialsOK = true
 130                 targetPath = pathParts
 131         } else if len(pathParts) >= 2 && strings.HasPrefix(pathParts[0], "c=") {
 132                 // /c=ID/PATH...
 133                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 134                 targetPath = pathParts[1:]
 135         } else if len(pathParts) >= 3 && pathParts[0] == "collections" {
 136                 if len(pathParts) >= 5 && pathParts[1] == "download" {
 137                         // /collections/download/ID/TOKEN/PATH...
 138                         targetID = pathParts[2]
 139                         tokens = []string{pathParts[3]}
 140                         targetPath = pathParts[4:]
 141                         pathToken = true
 142                 } else {
 143                         // /collections/ID/PATH...
 144                         targetID = pathParts[1]
 145                         tokens = anonymousTokens
 146                         targetPath = pathParts[2:]
 147                 }
 148         } else {
 149                 statusCode = http.StatusNotFound
 150                 return
 151         }
 152         if t := r.FormValue("api_token"); t != "" {
 153                 // The client provided an explicit token in the query
 154                 // string, or a form in POST body. We must put the
 155                 // token in an HttpOnly cookie, and redirect to the
 156                 // same URL with the query param redacted and method =
 157                 // GET.
 158
 159                 if !credentialsOK {
 160                         // It is not safe to copy the provided token
 161                         // into a cookie unless the current vhost
 162                         // (origin) serves only a single collection or
 163                         // we are in trustAllContent mode.
 164                         statusCode = http.StatusBadRequest
 165                         return
 166                 }
 167
 168                 // The HttpOnly flag is necessary to prevent
 169                 // JavaScript code (included in, or loaded by, a page
 170                 // in the collection being served) from employing the
 171                 // user's token beyond reading other files in the same
 172                 // domain, i.e., same collection.
 173                 //
 174                 // The 303 redirect is necessary in the case of a GET
 175                 // request to avoid exposing the token in the Location
 176                 // bar, and in the case of a POST request to avoid
 177                 // raising warnings when the user refreshes the
 178                 // resulting page.
 179
 180                 http.SetCookie(w, &http.Cookie{
 181                         Name:     "arvados_api_token",
 182                         Value:    auth.EncodeTokenCookie([]byte(t)),
 183                         Path:     "/",
 184                         HttpOnly: true,
 185                 })
 186
 187                 // Propagate query parameters (except api_token) from
 188                 // the original request.
 189                 redirQuery := r.URL.Query()
 190                 redirQuery.Del("api_token")
 191
 192                 redir := (&url.URL{
 193                         Host:     r.Host,
 194                         Path:     r.URL.Path,
 195                         RawQuery: redirQuery.Encode(),
 196                 }).String()
 197
 198                 w.Header().Add("Location", redir)
 199                 statusCode, statusText = http.StatusSeeOther, redir
 200                 w.WriteHeader(statusCode)
 201                 io.WriteString(w, `<A href="`)
 202                 io.WriteString(w, html.EscapeString(redir))
 203                 io.WriteString(w, `">Continue</A>`)
 204                 return
 205         }
 206
 207         if tokens == nil && strings.HasPrefix(targetPath[0], "t=") {
 208                 // http://ID.example/t=TOKEN/PATH...
 209                 // /c=ID/t=TOKEN/PATH...
 210                 //
 211                 // This form must only be used to pass scoped tokens
 212                 // that give permission for a single collection. See
 213                 // FormValue case above.
 214                 tokens = []string{targetPath[0][2:]}
 215                 pathToken = true
 216                 targetPath = targetPath[1:]
 217         }
 218
 219         if tokens == nil {
 220                 if credentialsOK {
 221                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 222                 }
 223                 tokens = append(reqTokens, anonymousTokens...)
 224         }
 225
 226         if len(targetPath) > 0 && targetPath[0] == "_" {
 227                 // If a collection has a directory called "t=foo" or
 228                 // "_", it can be served at
 229                 // //collections.example/_/t=foo/ or
 230                 // //collections.example/_/_/ respectively:
 231                 // //collections.example/t=foo/ won't work because
 232                 // t=foo will be interpreted as a token "foo".
 233                 targetPath = targetPath[1:]
 234         }
 235
 236         tokenResult := make(map[string]int)
 237         collection := make(map[string]interface{})
 238         found := false
 239         for _, arv.ApiToken = range tokens {
 240                 err := arv.Get("collections", targetID, nil, &collection)
 241                 if err == nil {
 242                         // Success
 243                         found = true
 244                         break
 245                 }
 246                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 247                         switch srvErr.HttpStatusCode {
 248                         case 404, 401:
 249                                 // Token broken or insufficient to
 250                                 // retrieve collection
 251                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 252                                 continue
 253                         }
 254                 }
 255                 // Something more serious is wrong
 256                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 257                 return
 258         }
 259         if !found {
 260                 if pathToken || !credentialsOK {
 261                         // Either the URL is a "secret sharing link"
 262                         // that didn't work out (and asking the client
 263                         // for additional credentials would just be
 264                         // confusing), or we don't even accept
 265                         // credentials at this path.
 266                         statusCode = http.StatusNotFound
 267                         return
 268                 }
 269                 for _, t := range reqTokens {
 270                         if tokenResult[t] == 404 {
 271                                 // The client provided valid token(s), but the
 272                                 // collection was not found.
 273                                 statusCode = http.StatusNotFound
 274                                 return
 275                         }
 276                 }
 277                 // The client's token was invalid (e.g., expired), or
 278                 // the client didn't even provide one.  Propagate the
 279                 // 401 to encourage the client to use a [different]
 280                 // token.
 281                 //
 282                 // TODO(TC): This response would be confusing to
 283                 // someone trying (anonymously) to download public
 284                 // data that has been deleted.  Allow a referrer to
 285                 // provide this context somehow?
 286                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 287                 statusCode = http.StatusUnauthorized
 288                 return
 289         }
 290
 291         filename := strings.Join(targetPath, "/")
 292         kc, err := keepclient.MakeKeepClient(arv)
 293         if err != nil {
 294                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 295                 return
 296         }
 297         rdr, err := kc.CollectionFileReader(collection, filename)
 298         if os.IsNotExist(err) {
 299                 statusCode = http.StatusNotFound
 300                 return
 301         } else if err != nil {
 302                 statusCode, statusText = http.StatusBadGateway, err.Error()
 303                 return
 304         }
 305         defer rdr.Close()
 306
 307         basenamePos := strings.LastIndex(filename, "/")
 308         if basenamePos < 0 {
 309                 basenamePos = 0
 310         }
 311         extPos := strings.LastIndex(filename, ".")
 312         if extPos > basenamePos {
 313                 // Now extPos is safely >= 0.
 314                 if t := mime.TypeByExtension(filename[extPos:]); t != "" {
 315                         w.Header().Set("Content-Type", t)
 316                 }
 317         }
 318         if rdr, ok := rdr.(keepclient.ReadCloserWithLen); ok {
 319                 w.Header().Set("Content-Length", fmt.Sprintf("%d", rdr.Len()))
 320         }
 321
 322         applyContentDispositionHdr(w, r, filename[basenamePos:], attachment)
 323         rangeRdr, statusCode := applyRangeHdr(w, r, rdr)
 324
 325         w.WriteHeader(statusCode)
 326         _, err = io.Copy(w, rangeRdr)
 327         if err != nil {
 328                 statusCode, statusText = http.StatusBadGateway, err.Error()
 329         }
 330 }
 331
 332 var rangeRe = regexp.MustCompile(`^bytes=0-([0-9]*)$`)
 333
 334 func applyRangeHdr(w http.ResponseWriter, r *http.Request, rdr keepclient.ReadCloserWithLen) (io.Reader, int) {
 335         w.Header().Set("Accept-Ranges", "bytes")
 336         hdr := r.Header.Get("Range")
 337         fields := rangeRe.FindStringSubmatch(hdr)
 338         if fields == nil {
 339                 return rdr, http.StatusOK
 340         }
 341         rangeEnd, err := strconv.ParseInt(fields[1], 10, 64)
 342         if err != nil {
 343                 // Empty or too big for int64 == send entire content
 344                 return rdr, http.StatusOK
 345         }
 346         if uint64(rangeEnd) >= rdr.Len() {
 347                 return rdr, http.StatusOK
 348         }
 349         w.Header().Set("Content-Length", fmt.Sprintf("%d", rangeEnd+1))
 350         w.Header().Set("Content-Range", fmt.Sprintf("bytes %d-%d/%d", 0, rangeEnd, rdr.Len()))
 351         return &io.LimitedReader{R: rdr, N: rangeEnd + 1}, http.StatusPartialContent
 352 }
 353
 354 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 355         disposition := "inline"
 356         if isAttachment {
 357                 disposition = "attachment"
 358         }
 359         if strings.ContainsRune(r.RequestURI, '?') {
 360                 // Help the UA realize that the filename is just
 361                 // "filename.txt", not
 362                 // "filename.txt?disposition=attachment".
 363                 //
 364                 // TODO(TC): Follow advice at RFC 6266 appendix D
 365                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 366         }
 367         if disposition != "inline" {
 368                 w.Header().Set("Content-Disposition", disposition)
 369         }
 370 }