services/keep-web/handler.go

   1 package main
   2
   3 import (
   4         "fmt"
   5         "html"
   6         "io"
   7         "net/http"
   8         "net/url"
   9         "os"
  10         "path"
  11         "strconv"
  12         "strings"
  13         "sync"
  14         "time"
  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         Config     *Config
  24         clientPool *arvadosclient.ClientPool
  25         setupOnce  sync.Once
  26 }
  27
  28 // parseCollectionIDFromDNSName returns a UUID or PDH if s begins with
  29 // a UUID or URL-encoded PDH; otherwise "".
  30 func parseCollectionIDFromDNSName(s string) string {
  31         // Strip domain.
  32         if i := strings.IndexRune(s, '.'); i >= 0 {
  33                 s = s[:i]
  34         }
  35         // Names like {uuid}--collections.example.com serve the same
  36         // purpose as {uuid}.collections.example.com but can reduce
  37         // cost/effort of using [additional] wildcard certificates.
  38         if i := strings.Index(s, "--"); i >= 0 {
  39                 s = s[:i]
  40         }
  41         if arvadosclient.UUIDMatch(s) {
  42                 return s
  43         }
  44         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  45                 return pdh
  46         }
  47         return ""
  48 }
  49
  50 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  51
  52 // parseCollectionIDFromURL returns a UUID or PDH if s is a UUID or a
  53 // PDH (even if it is a PDH with "+" replaced by " " or "-");
  54 // otherwise "".
  55 func parseCollectionIDFromURL(s string) string {
  56         if arvadosclient.UUIDMatch(s) {
  57                 return s
  58         }
  59         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  60                 return pdh
  61         }
  62         return ""
  63 }
  64
  65 func (h *handler) setup() {
  66         h.clientPool = arvadosclient.MakeClientPool()
  67 }
  68
  69 // ServeHTTP implements http.Handler.
  70 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
  71         h.setupOnce.Do(h.setup)
  72
  73         var statusCode = 0
  74         var statusText string
  75
  76         remoteAddr := r.RemoteAddr
  77         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
  78                 remoteAddr = xff + "," + remoteAddr
  79         }
  80
  81         w := httpserver.WrapResponseWriter(wOrig)
  82         defer func() {
  83                 if statusCode == 0 {
  84                         statusCode = w.WroteStatus()
  85                 } else if w.WroteStatus() == 0 {
  86                         w.WriteHeader(statusCode)
  87                 } else if w.WroteStatus() != statusCode {
  88                         httpserver.Log(r.RemoteAddr, "WARNING",
  89                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
  90                 }
  91                 if statusText == "" {
  92                         statusText = http.StatusText(statusCode)
  93                 }
  94                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
  95         }()
  96
  97         if r.Method == "OPTIONS" {
  98                 method := r.Header.Get("Access-Control-Request-Method")
  99                 if method != "GET" && method != "POST" {
 100                         statusCode = http.StatusMethodNotAllowed
 101                         return
 102                 }
 103                 w.Header().Set("Access-Control-Allow-Headers", "Range")
 104                 w.Header().Set("Access-Control-Allow-Methods", "GET, POST")
 105                 w.Header().Set("Access-Control-Allow-Origin", "*")
 106                 w.Header().Set("Access-Control-Max-Age", "86400")
 107                 statusCode = http.StatusOK
 108                 return
 109         }
 110
 111         if r.Method != "GET" && r.Method != "POST" {
 112                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
 113                 return
 114         }
 115
 116         if r.Header.Get("Origin") != "" {
 117                 // Allow simple cross-origin requests without user
 118                 // credentials ("user credentials" as defined by CORS,
 119                 // i.e., cookies, HTTP authentication, and client-side
 120                 // SSL certificates. See
 121                 // http://www.w3.org/TR/cors/#user-credentials).
 122                 w.Header().Set("Access-Control-Allow-Origin", "*")
 123         }
 124
 125         arv := h.clientPool.Get()
 126         if arv == nil {
 127                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+h.clientPool.Err().Error()
 128                 return
 129         }
 130         defer h.clientPool.Put(arv)
 131
 132         pathParts := strings.Split(r.URL.Path[1:], "/")
 133
 134         var targetID string
 135         var targetPath []string
 136         var tokens []string
 137         var reqTokens []string
 138         var pathToken bool
 139         var attachment bool
 140         credentialsOK := h.Config.TrustAllContent
 141
 142         if r.Host != "" && r.Host == h.Config.AttachmentOnlyHost {
 143                 credentialsOK = true
 144                 attachment = true
 145         } else if r.FormValue("disposition") == "attachment" {
 146                 attachment = true
 147         }
 148
 149         if targetID = parseCollectionIDFromDNSName(r.Host); targetID != "" {
 150                 // http://ID.collections.example/PATH...
 151                 credentialsOK = true
 152                 targetPath = pathParts
 153         } else if len(pathParts) >= 2 && strings.HasPrefix(pathParts[0], "c=") {
 154                 // /c=ID/PATH...
 155                 targetID = parseCollectionIDFromURL(pathParts[0][2:])
 156                 targetPath = pathParts[1:]
 157         } else if len(pathParts) >= 3 && pathParts[0] == "collections" {
 158                 if len(pathParts) >= 5 && pathParts[1] == "download" {
 159                         // /collections/download/ID/TOKEN/PATH...
 160                         targetID = parseCollectionIDFromURL(pathParts[2])
 161                         tokens = []string{pathParts[3]}
 162                         targetPath = pathParts[4:]
 163                         pathToken = true
 164                 } else {
 165                         // /collections/ID/PATH...
 166                         targetID = parseCollectionIDFromURL(pathParts[1])
 167                         tokens = h.Config.AnonymousTokens
 168                         targetPath = pathParts[2:]
 169                 }
 170         }
 171
 172         if targetID == "" {
 173                 statusCode = http.StatusNotFound
 174                 return
 175         }
 176
 177         formToken := r.FormValue("api_token")
 178         if formToken != "" && r.Header.Get("Origin") != "" && attachment && r.URL.Query().Get("api_token") == "" {
 179                 // The client provided an explicit token in the POST
 180                 // body. The Origin header indicates this *might* be
 181                 // an AJAX request, in which case redirect-with-cookie
 182                 // won't work: we should just serve the content in the
 183                 // POST response. This is safe because:
 184                 //
 185                 // * We're supplying an attachment, not inline
 186                 //   content, so we don't need to convert the POST to
 187                 //   a GET and avoid the "really resubmit form?"
 188                 //   problem.
 189                 //
 190                 // * The token isn't embedded in the URL, so we don't
 191                 //   need to worry about bookmarks and copy/paste.
 192                 tokens = append(tokens, formToken)
 193         } else if formToken != "" {
 194                 // The client provided an explicit token in the query
 195                 // string, or a form in POST body. We must put the
 196                 // token in an HttpOnly cookie, and redirect to the
 197                 // same URL with the query param redacted and method =
 198                 // GET.
 199
 200                 if !credentialsOK {
 201                         // It is not safe to copy the provided token
 202                         // into a cookie unless the current vhost
 203                         // (origin) serves only a single collection or
 204                         // we are in TrustAllContent mode.
 205                         statusCode = http.StatusBadRequest
 206                         return
 207                 }
 208
 209                 // The HttpOnly flag is necessary to prevent
 210                 // JavaScript code (included in, or loaded by, a page
 211                 // in the collection being served) from employing the
 212                 // user's token beyond reading other files in the same
 213                 // domain, i.e., same collection.
 214                 //
 215                 // The 303 redirect is necessary in the case of a GET
 216                 // request to avoid exposing the token in the Location
 217                 // bar, and in the case of a POST request to avoid
 218                 // raising warnings when the user refreshes the
 219                 // resulting page.
 220
 221                 http.SetCookie(w, &http.Cookie{
 222                         Name:     "arvados_api_token",
 223                         Value:    auth.EncodeTokenCookie([]byte(formToken)),
 224                         Path:     "/",
 225                         HttpOnly: true,
 226                 })
 227
 228                 // Propagate query parameters (except api_token) from
 229                 // the original request.
 230                 redirQuery := r.URL.Query()
 231                 redirQuery.Del("api_token")
 232
 233                 redir := (&url.URL{
 234                         Host:     r.Host,
 235                         Path:     r.URL.Path,
 236                         RawQuery: redirQuery.Encode(),
 237                 }).String()
 238
 239                 w.Header().Add("Location", redir)
 240                 statusCode, statusText = http.StatusSeeOther, redir
 241                 w.WriteHeader(statusCode)
 242                 io.WriteString(w, `<A href="`)
 243                 io.WriteString(w, html.EscapeString(redir))
 244                 io.WriteString(w, `">Continue</A>`)
 245                 return
 246         }
 247
 248         if tokens == nil && strings.HasPrefix(targetPath[0], "t=") {
 249                 // http://ID.example/t=TOKEN/PATH...
 250                 // /c=ID/t=TOKEN/PATH...
 251                 //
 252                 // This form must only be used to pass scoped tokens
 253                 // that give permission for a single collection. See
 254                 // FormValue case above.
 255                 tokens = []string{targetPath[0][2:]}
 256                 pathToken = true
 257                 targetPath = targetPath[1:]
 258         }
 259
 260         if tokens == nil {
 261                 if credentialsOK {
 262                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 263                 }
 264                 tokens = append(reqTokens, h.Config.AnonymousTokens...)
 265         }
 266
 267         if len(targetPath) > 0 && targetPath[0] == "_" {
 268                 // If a collection has a directory called "t=foo" or
 269                 // "_", it can be served at
 270                 // //collections.example/_/t=foo/ or
 271                 // //collections.example/_/_/ respectively:
 272                 // //collections.example/t=foo/ won't work because
 273                 // t=foo will be interpreted as a token "foo".
 274                 targetPath = targetPath[1:]
 275         }
 276
 277         tokenResult := make(map[string]int)
 278         collection := make(map[string]interface{})
 279         found := false
 280         for _, arv.ApiToken = range tokens {
 281                 err := arv.Get("collections", targetID, nil, &collection)
 282                 if err == nil {
 283                         // Success
 284                         found = true
 285                         break
 286                 }
 287                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 288                         switch srvErr.HttpStatusCode {
 289                         case 404, 401:
 290                                 // Token broken or insufficient to
 291                                 // retrieve collection
 292                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 293                                 continue
 294                         }
 295                 }
 296                 // Something more serious is wrong
 297                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 298                 return
 299         }
 300         if !found {
 301                 if pathToken || !credentialsOK {
 302                         // Either the URL is a "secret sharing link"
 303                         // that didn't work out (and asking the client
 304                         // for additional credentials would just be
 305                         // confusing), or we don't even accept
 306                         // credentials at this path.
 307                         statusCode = http.StatusNotFound
 308                         return
 309                 }
 310                 for _, t := range reqTokens {
 311                         if tokenResult[t] == 404 {
 312                                 // The client provided valid token(s), but the
 313                                 // collection was not found.
 314                                 statusCode = http.StatusNotFound
 315                                 return
 316                         }
 317                 }
 318                 // The client's token was invalid (e.g., expired), or
 319                 // the client didn't even provide one.  Propagate the
 320                 // 401 to encourage the client to use a [different]
 321                 // token.
 322                 //
 323                 // TODO(TC): This response would be confusing to
 324                 // someone trying (anonymously) to download public
 325                 // data that has been deleted.  Allow a referrer to
 326                 // provide this context somehow?
 327                 w.Header().Add("WWW-Authenticate", "Basic realm=\"collections\"")
 328                 statusCode = http.StatusUnauthorized
 329                 return
 330         }
 331
 332         filename := strings.Join(targetPath, "/")
 333         kc, err := keepclient.MakeKeepClient(arv)
 334         if err != nil {
 335                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 336                 return
 337         }
 338         if client, ok := kc.Client.(*http.Client); ok && client.Transport != nil {
 339                 // Workaround for https://dev.arvados.org/issues/9005
 340                 if t, ok := client.Transport.(*http.Transport); ok {
 341                         defer t.CloseIdleConnections()
 342                 }
 343         }
 344         rdr, err := kc.CollectionFileReader(collection, filename)
 345         if os.IsNotExist(err) {
 346                 statusCode = http.StatusNotFound
 347                 return
 348         } else if err != nil {
 349                 statusCode, statusText = http.StatusBadGateway, err.Error()
 350                 return
 351         }
 352         defer rdr.Close()
 353
 354         basename := path.Base(filename)
 355         applyContentDispositionHdr(w, r, basename, attachment)
 356
 357         modstr, _ := collection["modified_at"].(string)
 358         modtime, err := time.Parse(time.RFC3339Nano, modstr)
 359         if err != nil {
 360                 modtime = time.Now()
 361         }
 362         http.ServeContent(w, r, basename, modtime, rdr)
 363 }
 364
 365 func applyContentDispositionHdr(w http.ResponseWriter, r *http.Request, filename string, isAttachment bool) {
 366         disposition := "inline"
 367         if isAttachment {
 368                 disposition = "attachment"
 369         }
 370         if strings.ContainsRune(r.RequestURI, '?') {
 371                 // Help the UA realize that the filename is just
 372                 // "filename.txt", not
 373                 // "filename.txt?disposition=attachment".
 374                 //
 375                 // TODO(TC): Follow advice at RFC 6266 appendix D
 376                 disposition += "; filename=" + strconv.QuoteToASCII(filename)
 377         }
 378         if disposition != "inline" {
 379                 w.Header().Set("Content-Disposition", disposition)
 380         }
 381 }