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