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