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         "strings"
  13         "time"
  14
  15         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  16         "git.curoverse.com/arvados.git/sdk/go/auth"
  17         "git.curoverse.com/arvados.git/sdk/go/httpserver"
  18         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  19 )
  20
  21 type handler struct{}
  22
  23 var (
  24         clientPool      = arvadosclient.MakeClientPool()
  25         trustAllContent = false
  26         anonymousTokens []string
  27 )
  28
  29 func init() {
  30         flag.BoolVar(&trustAllContent, "trust-all-content", false,
  31                 "Serve non-public content from a single origin. Dangerous: read docs before using!")
  32 }
  33
  34 // return a UUID or PDH if s begins with a UUID or URL-encoded PDH;
  35 // otherwise return "".
  36 func parseCollectionIdFromDNSName(s string) string {
  37         // Strip domain.
  38         if i := strings.IndexRune(s, '.'); i >= 0 {
  39                 s = s[:i]
  40         }
  41         // Names like {uuid}--dl.example.com serve the same purpose as
  42         // {uuid}.dl.example.com but can reduce cost/effort of using
  43         // [additional] wildcard certificates.
  44         if i := strings.Index(s, "--"); i >= 0 {
  45                 s = s[:i]
  46         }
  47         if arvadosclient.UUIDMatch(s) {
  48                 return s
  49         }
  50         if pdh := strings.Replace(s, "-", "+", 1); arvadosclient.PDHMatch(pdh) {
  51                 return pdh
  52         }
  53         return ""
  54 }
  55
  56 var urlPDHDecoder = strings.NewReplacer(" ", "+", "-", "+")
  57
  58 // return a UUID or PDH if s is a UUID or a PDH (even if it is a PDH
  59 // with "+" replaced by " " or "-"); otherwise return "".
  60 func parseCollectionIdFromURL(s string) string {
  61         if arvadosclient.UUIDMatch(s) {
  62                 return s
  63         }
  64         if pdh := urlPDHDecoder.Replace(s); arvadosclient.PDHMatch(pdh) {
  65                 return pdh
  66         }
  67         return ""
  68 }
  69
  70 func (h *handler) ServeHTTP(wOrig http.ResponseWriter, r *http.Request) {
  71         var statusCode = 0
  72         var statusText string
  73
  74         remoteAddr := r.RemoteAddr
  75         if xff := r.Header.Get("X-Forwarded-For"); xff != "" {
  76                 remoteAddr = xff + "," + remoteAddr
  77         }
  78
  79         w := httpserver.WrapResponseWriter(wOrig)
  80         defer func() {
  81                 if statusCode == 0 {
  82                         statusCode = w.WroteStatus()
  83                 } else if w.WroteStatus() == 0 {
  84                         w.WriteHeader(statusCode)
  85                 } else if w.WroteStatus() != statusCode {
  86                         httpserver.Log(r.RemoteAddr, "WARNING",
  87                                 fmt.Sprintf("Our status changed from %d to %d after we sent headers", w.WroteStatus(), statusCode))
  88                 }
  89                 if statusText == "" {
  90                         statusText = http.StatusText(statusCode)
  91                 }
  92                 httpserver.Log(remoteAddr, statusCode, statusText, w.WroteBodyBytes(), r.Method, r.Host, r.URL.Path, r.URL.RawQuery)
  93         }()
  94
  95         if r.Method != "GET" && r.Method != "POST" {
  96                 statusCode, statusText = http.StatusMethodNotAllowed, r.Method
  97                 return
  98         }
  99
 100         arv := clientPool.Get()
 101         if arv == nil {
 102                 statusCode, statusText = http.StatusInternalServerError, "Pool failed: "+clientPool.Err().Error()
 103                 return
 104         }
 105         defer clientPool.Put(arv)
 106
 107         pathParts := strings.Split(r.URL.Path[1:], "/")
 108
 109         var targetId string
 110         var targetPath []string
 111         var tokens []string
 112         var reqTokens []string
 113         var pathToken bool
 114         credentialsOK := trustAllContent
 115
 116         if targetId = parseCollectionIdFromDNSName(r.Host); targetId != "" {
 117                 // http://ID.dl.example/PATH...
 118                 credentialsOK = true
 119                 targetPath = pathParts
 120         } else if len(pathParts) >= 2 && strings.HasPrefix(pathParts[0], "c=") {
 121                 // /c=ID/PATH...
 122                 targetId = parseCollectionIdFromURL(pathParts[0][2:])
 123                 targetPath = pathParts[1:]
 124         } else if len(pathParts) >= 3 && pathParts[0] == "collections" {
 125                 if len(pathParts) >= 5 && pathParts[1] == "download" {
 126                         // /collections/download/ID/TOKEN/PATH...
 127                         targetId = pathParts[2]
 128                         tokens = []string{pathParts[3]}
 129                         targetPath = pathParts[4:]
 130                         pathToken = true
 131                 } else {
 132                         // /collections/ID/PATH...
 133                         targetId = pathParts[1]
 134                         tokens = anonymousTokens
 135                         targetPath = pathParts[2:]
 136                 }
 137         } else {
 138                 statusCode = http.StatusNotFound
 139                 return
 140         }
 141         if t := r.FormValue("api_token"); t != "" {
 142                 // The client provided an explicit token in the query
 143                 // string, or a form in POST body. We must put the
 144                 // token in an HttpOnly cookie, and redirect to the
 145                 // same URL with the query param redacted and method =
 146                 // GET.
 147
 148                 if !credentialsOK {
 149                         // It is not safe to copy the provided token
 150                         // into a cookie unless the current vhost
 151                         // (origin) serves only a single collection or
 152                         // we are in trustAllContent mode.
 153                         statusCode = http.StatusBadRequest
 154                         return
 155                 }
 156
 157                 // The HttpOnly flag is necessary to prevent
 158                 // JavaScript code (included in, or loaded by, a page
 159                 // in the collection being served) from employing the
 160                 // user's token beyond reading other files in the same
 161                 // domain, i.e., same collection.
 162                 //
 163                 // The 303 redirect is necessary in the case of a GET
 164                 // request to avoid exposing the token in the Location
 165                 // bar, and in the case of a POST request to avoid
 166                 // raising warnings when the user refreshes the
 167                 // resulting page.
 168
 169                 http.SetCookie(w, &http.Cookie{
 170                         Name:     "api_token",
 171                         Value:    auth.EncodeTokenCookie([]byte(t)),
 172                         Path:     "/",
 173                         Expires:  time.Now().AddDate(10, 0, 0),
 174                         HttpOnly: true,
 175                 })
 176                 redir := (&url.URL{Host: r.Host, Path: r.URL.Path}).String()
 177
 178                 w.Header().Add("Location", redir)
 179                 statusCode, statusText = http.StatusSeeOther, redir
 180                 w.WriteHeader(statusCode)
 181                 io.WriteString(w, `<A href="`)
 182                 io.WriteString(w, html.EscapeString(redir))
 183                 io.WriteString(w, `">Continue</A>`)
 184                 return
 185         }
 186
 187         if tokens == nil && strings.HasPrefix(targetPath[0], "t=") {
 188                 // http://ID.example/t=TOKEN/PATH...
 189                 // /c=ID/t=TOKEN/PATH...
 190                 //
 191                 // This form must only be used to pass scoped tokens
 192                 // that give permission for a single collection. See
 193                 // FormValue case above.
 194                 tokens = []string{targetPath[0][2:]}
 195                 pathToken = true
 196                 targetPath = targetPath[1:]
 197         }
 198
 199         if tokens == nil {
 200                 if credentialsOK {
 201                         reqTokens = auth.NewCredentialsFromHTTPRequest(r).Tokens
 202                 }
 203                 tokens = append(reqTokens, anonymousTokens...)
 204         }
 205
 206         if len(targetPath) > 0 && targetPath[0] == "_" {
 207                 // If a collection has a directory called "t=foo" or
 208                 // "_", it can be served at //dl.example/_/t=foo/ or
 209                 // //dl.example/_/_/ respectively: //dl.example/t=foo/
 210                 // won't work because t=foo will be interpreted as a
 211                 // token "foo".
 212                 targetPath = targetPath[1:]
 213         }
 214
 215         tokenResult := make(map[string]int)
 216         collection := make(map[string]interface{})
 217         found := false
 218         for _, arv.ApiToken = range tokens {
 219                 err := arv.Get("collections", targetId, nil, &collection)
 220                 if err == nil {
 221                         // Success
 222                         found = true
 223                         break
 224                 }
 225                 if srvErr, ok := err.(arvadosclient.APIServerError); ok {
 226                         switch srvErr.HttpStatusCode {
 227                         case 404, 401:
 228                                 // Token broken or insufficient to
 229                                 // retrieve collection
 230                                 tokenResult[arv.ApiToken] = srvErr.HttpStatusCode
 231                                 continue
 232                         }
 233                 }
 234                 // Something more serious is wrong
 235                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 236                 return
 237         }
 238         if !found {
 239                 if pathToken || !credentialsOK {
 240                         // Either the URL is a "secret sharing link"
 241                         // that didn't work out (and asking the client
 242                         // for additional credentials would just be
 243                         // confusing), or we don't even accept
 244                         // credentials at this path.
 245                         statusCode = http.StatusNotFound
 246                         return
 247                 }
 248                 for _, t := range reqTokens {
 249                         if tokenResult[t] == 404 {
 250                                 // The client provided valid token(s), but the
 251                                 // collection was not found.
 252                                 statusCode = http.StatusNotFound
 253                                 return
 254                         }
 255                 }
 256                 // The client's token was invalid (e.g., expired), or
 257                 // the client didn't even provide one.  Propagate the
 258                 // 401 to encourage the client to use a [different]
 259                 // token.
 260                 //
 261                 // TODO(TC): This response would be confusing to
 262                 // someone trying (anonymously) to download public
 263                 // data that has been deleted.  Allow a referrer to
 264                 // provide this context somehow?
 265                 w.Header().Add("WWW-Authenticate", "Basic realm=\"dl\"")
 266                 statusCode = http.StatusUnauthorized
 267                 return
 268         }
 269
 270         filename := strings.Join(targetPath, "/")
 271         kc, err := keepclient.MakeKeepClient(arv)
 272         if err != nil {
 273                 statusCode, statusText = http.StatusInternalServerError, err.Error()
 274                 return
 275         }
 276         rdr, err := kc.CollectionFileReader(collection, filename)
 277         if os.IsNotExist(err) {
 278                 statusCode = http.StatusNotFound
 279                 return
 280         } else if err != nil {
 281                 statusCode, statusText = http.StatusBadGateway, err.Error()
 282                 return
 283         }
 284         defer rdr.Close()
 285
 286         // One or both of these can be -1 if not found:
 287         basenamePos := strings.LastIndex(filename, "/")
 288         extPos := strings.LastIndex(filename, ".")
 289         if extPos > basenamePos {
 290                 // Now extPos is safely >= 0.
 291                 if t := mime.TypeByExtension(filename[extPos:]); t != "" {
 292                         w.Header().Set("Content-Type", t)
 293                 }
 294         }
 295         w.Header().Set("Content-Length", fmt.Sprintf("%d", rdr.Len()))
 296
 297         w.WriteHeader(http.StatusOK)
 298         _, err = io.Copy(w, rdr)
 299         if err != nil {
 300                 statusCode, statusText = http.StatusBadGateway, err.Error()
 301         }
 302 }