services/keep-web/handler.go

   1 package main
   2
   3 import (
   4         "fmt"
   5         "html"
   6         "io"
   7         "mime"
   8         "net/http"
   9         "net/url"
  10         "os"
  11         "strings"
  12         "time"
  13
  14         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  15         "git.curoverse.com/arvados.git/sdk/go/auth"
  16         "git.curoverse.com/arvados.git/sdk/go/httpserver"
  17         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  18 )
  19
  20 type handler struct{}
  21
  22 var clientPool = arvadosclient.MakeClientPool()
  23 var anonymousTokens []string
  24
  25 // return a UUID or PDH if s begins with a UUID or URL-encoded PDH;
  26 // otherwise return "".
  27 func parseCollectionIdFromDNSName(s string) string {
  28         // Strip domain.
  29         if i := strings.IndexRune(s, '.'); i >= 0 {
  30                 s = s[:i]
  31         }
  32         // Names like {uuid}--dl.example.com serve the same purpose as
  33         // {uuid}.dl.example.com but can reduce cost/effort of using
  34         // [additional] wildcard certificates.
  35         if i := strings.Index(s, "--"); i >= 0 {
  36                 s = s[:i]
  37         }
  38         if arvadosclient.UUIDMatch(s) {
  39                 return s
  40         }
  41         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  42                 return pdh
  43         }
  44         return ""
  45 }
  46
  47 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  48
  49 // return a UUID or PDH if s is a UUID or a PDH (even if it is a PDH
  50 // with "+" replaced by " " or "-"); otherwise return "".
  51 func parseCollectionIdFromURL(s string) string {
  52         if arvadosclient.UUIDMatch(s) {
  53                 return s
  54         }
  55         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  56                 return pdh
  57         }
  58         return ""
  59 }
  60
  61 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
  62         var statusCode = 0
  63         var statusText string
  64
  65         remoteAddr := r.RemoteAddr
  66         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
  67                 remoteAddr = xff + "," + remoteAddr
  68         }
  69
  70         w := httpserver.WrapResponseWriter(wOrig)
  71         defer func() {
  72                 if statusCode == 0 {
  73                         statusCode = w.WroteStatus()
  74                 } else if w.WroteStatus() == 0 {
  75                         w.WriteHeader(statusCode)
  76                 } else if w.WroteStatus() != statusCode {
  77                         httpserver.Log(r.RemoteAddr, "WARNING",
  78                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
  79                 }
  80                 if statusText == "" {
  81                         statusText = http.StatusText(statusCode)
  82                 }
  83                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
  84         }()
  85
  86         if r.Method != "GET" && r.Method != "POST" {
  87                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
  88                 return
  89         }
  90
  91         arv := clientPool.Get()
  92         if arv == nil {
  93                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+clientPool.Err().Error()
  94                 return
  95         }
  96         defer clientPool.Put(arv)
  97
  98         pathParts := strings.Split(r.URL.Path[1:], "/")
  99
 100         var targetId string
 101         var targetPath []string
 102         var tokens []string
 103         var reqTokens []string
 104         var pathToken bool
 105         var credentialsOK bool
 106
 107         if targetId = parseCollectionIdFromDNSName(r.Host); targetId != "" {
 108                 // http://ID.dl.example/PATH...
 109                 credentialsOK = true
 110                 targetPath = pathParts
 111         } else if len(pathParts) >= 2 && strings.HasPrefix(pathParts[0], "c=") {
 112                 // /c=ID/PATH...
 113                 targetId = parseCollectionIdFromURL(pathParts[0][2:])
 114                 targetPath = pathParts[1:]
 115         } else if len(pathParts) >= 3 && pathParts[0] == "collections" {
 116                 if len(pathParts) >= 5 && pathParts[1] == "download" {
 117                         // /collections/download/ID/TOKEN/PATH...
 118                         targetId = pathParts[2]
 119                         tokens = []string{pathParts[3]}
 120                         targetPath = pathParts[4:]
 121                         pathToken = true
 122                 } else {
 123                         // /collections/ID/PATH...
 124                         targetId = pathParts[1]
 125                         tokens = anonymousTokens
 126                         targetPath = pathParts[2:]
 127                 }
 128         } else {
 129                 statusCode = http.StatusNotFound
 130                 return
 131         }
 132         if t := r.FormValue("api_token"); t != "" {
 133                 // The client provided an explicit token in the query
 134                 // string, or a form in POST body. We must put the
 135                 // token in an HttpOnly cookie, and redirect to the
 136                 // same URL with the query param redacted and method =
 137                 // GET.
 138
 139                 if !credentialsOK {
 140                         // It is not safe to copy the provided token
 141                         // into a cookie unless the current vhost
 142                         // (origin) serves only a single collection.
 143                         statusCode = http.StatusBadRequest
 144                         return
 145                 }
 146
 147                 // The HttpOnly flag is necessary to prevent
 148                 // JavaScript code (included in, or loaded by, a page
 149                 // in the collection being served) from employing the
 150                 // user's token beyond reading other files in the same
 151                 // domain, i.e., same collection.
 152                 //
 153                 // The 303 redirect is necessary in the case of a GET
 154                 // request to avoid exposing the token in the Location
 155                 // bar, and in the case of a POST request to avoid
 156                 // raising warnings when the user refreshes the
 157                 // resulting page.
 158
 159                 http.SetCookie(w, &http.Cookie{
 160                         Name:     "api_token",
 161                         Value:    auth.EncodeTokenCookie([]byte(t)),
 162                         Path:     "/",
 163                         Expires:  time.Now().AddDate(10,0,0),
 164                         HttpOnly: true,
 165                 })
 166                 redir := (&url.URL{Host: r.Host, Path: r.URL.Path}).String()
 167
 168                 w.Header().Add("Location", redir)
 169                 statusCode, statusText = http.StatusSeeOther, redir
 170                 w.WriteHeader(statusCode)
 171                 io.WriteString(w, `<A href="`)
 172                 io.WriteString(w, html.EscapeString(redir))
 173                 io.WriteString(w, `">Continue</A>`)
 174                 return
 175         }
 176
 177         if tokens == nil && strings.HasPrefix(targetPath[0], "t=") {
 178                 // http://ID.example/t=TOKEN/PATH...
 179                 // /c=ID/t=TOKEN/PATH...
 180                 //
 181                 // This form must only be used to pass scoped tokens
 182                 // that give permission for a single collection. See
 183                 // FormValue case above.
 184                 tokens = []string{targetPath[0][2:]}
 185                 pathToken = true
 186                 targetPath = targetPath[1:]
 187         }
 188
 189         if tokens == nil {
 190                 if credentialsOK {
 191                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 192                 }
 193                 tokens = append(reqTokens, anonymousTokens...)
 194         }
 195
 196         if len(targetPath) > 0 && targetPath[0] == "_" {
 197                 // If a collection has a directory called "t=foo" or
 198                 // "_", it can be served at //dl.example/_/t=foo/ or
 199                 // //dl.example/_/_/ respectively: //dl.example/t=foo/
 200                 // won't work because t=foo will be interpreted as a
 201                 // token "foo".
 202                 targetPath = targetPath[1:]
 203         }
 204
 205         tokenResult := make(map[string]int)
 206         collection := make(map[string]interface{})
 207         found := false
 208         for _, arv.ApiToken = range tokens {
 209                 err := arv.Get("collections", targetId, nil, &collection)
 210                 if err == nil {
 211                         // Success
 212                         found = true
 213                         break
 214                 }
 215                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 216                         switch srvErr.HttpStatusCode {
 217                         case 404, 401:
 218                                 // Token broken or insufficient to
 219                                 // retrieve collection
 220                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 221                                 continue
 222                         }
 223                 }
 224                 // Something more serious is wrong
 225                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 226                 return
 227         }
 228         if !found {
 229                 if pathToken || !credentialsOK {
 230                         // Either the URL is a "secret sharing link"
 231                         // that didn't work out (and asking the client
 232                         // for additional credentials would just be
 233                         // confusing), or we don't even accept
 234                         // credentials at this path.
 235                         statusCode = http.StatusNotFound
 236                         return
 237                 }
 238                 for _, t := range reqTokens {
 239                         if tokenResult[t] == 404 {
 240                                 // The client provided valid token(s), but the
 241                                 // collection was not found.
 242                                 statusCode = http.StatusNotFound
 243                                 return
 244                         }
 245                 }
 246                 // The client's token was invalid (e.g., expired), or
 247                 // the client didn't even provide one.  Propagate the
 248                 // 401 to encourage the client to use a [different]
 249                 // token.
 250                 //
 251                 // TODO(TC): This response would be confusing to
 252                 // someone trying (anonymously) to download public
 253                 // data that has been deleted.  Allow a referrer to
 254                 // provide this context somehow?
 255                 w.Header().Add("WWW-Authenticate", "Basic realm=\"dl\"")
 256                 statusCode = http.StatusUnauthorized
 257                 return
 258         }
 259
 260         filename := strings.Join(targetPath, "/")
 261         kc, err := keepclient.MakeKeepClient(arv)
 262         if err != nil {
 263                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 264                 return
 265         }
 266         rdr, err := kc.CollectionFileReader(collection, filename)
 267         if os.IsNotExist(err) {
 268                 statusCode = http.StatusNotFound
 269                 return
 270         } else if err != nil {
 271                 statusCode, statusText = http.StatusBadGateway, err.Error()
 272                 return
 273         }
 274         defer rdr.Close()
 275
 276         // One or both of these can be -1 if not found:
 277         basenamePos := strings.LastIndex(filename, "/")
 278         extPos := strings.LastIndex(filename, ".")
 279         if extPos > basenamePos {
 280                 // Now extPos is safely >= 0.
 281                 if t := mime.TypeByExtension(filename[extPos:]); t != "" {
 282                         w.Header().Set("Content-Type", t)
 283                 }
 284         }
 285         w.Header().Set("Content-Length", fmt.Sprintf("%d", rdr.Len()))
 286
 287         w.WriteHeader(http.StatusOK)
 288         _, err = io.Copy(w, rdr)
 289         if err != nil {
 290                 statusCode, statusText = http.StatusBadGateway, err.Error()
 291         }
 292 }