services/keep-web/s3.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package main
   6
   7 import (
   8         "encoding/xml"
   9         "errors"
  10         "fmt"
  11         "io"
  12         "net/http"
  13         "os"
  14         "path/filepath"
  15         "sort"
  16         "strconv"
  17         "strings"
  18
  19         "git.arvados.org/arvados.git/sdk/go/arvados"
  20         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  21         "github.com/AdRoll/goamz/s3"
  22 )
  23
  24 const s3MaxKeys = 1000
  25
  26 // serveS3 handles r and returns true if r is a request from an S3
  27 // client, otherwise it returns false.
  28 func (h *handler) serveS3(w http.ResponseWriter, r *http.Request) bool {
  29         var token string
  30         if auth := r.Header.Get("Authorization"); strings.HasPrefix(auth, "AWS ") {
  31                 split := strings.SplitN(auth[4:], ":", 2)
  32                 if len(split) < 2 {
  33                         w.WriteHeader(http.StatusUnauthorized)
  34                         return true
  35                 }
  36                 token = split[0]
  37         } else if strings.HasPrefix(auth, "AWS4-HMAC-SHA256 ") {
  38                 for _, cmpt := range strings.Split(auth[17:], ",") {
  39                         cmpt = strings.TrimSpace(cmpt)
  40                         split := strings.SplitN(cmpt, "=", 2)
  41                         if len(split) == 2 && split[0] == "Credential" {
  42                                 keyandscope := strings.Split(split[1], "/")
  43                                 if len(keyandscope[0]) > 0 {
  44                                         token = keyandscope[0]
  45                                         break
  46                                 }
  47                         }
  48                 }
  49                 if token == "" {
  50                         w.WriteHeader(http.StatusBadRequest)
  51                         fmt.Println(w, "invalid V4 signature")
  52                         return true
  53                 }
  54         } else {
  55                 return false
  56         }
  57
  58         _, kc, client, release, err := h.getClients(r.Header.Get("X-Request-Id"), token)
  59         if err != nil {
  60                 http.Error(w, "Pool failed: "+h.clientPool.Err().Error(), http.StatusInternalServerError)
  61                 return true
  62         }
  63         defer release()
  64
  65         fs := client.SiteFileSystem(kc)
  66         fs.ForwardSlashNameSubstitution(h.Config.cluster.Collections.ForwardSlashNameSubstitution)
  67
  68         objectNameGiven := strings.Count(strings.TrimSuffix(r.URL.Path, "/"), "/") > 1
  69
  70         switch {
  71         case r.Method == "GET" && !objectNameGiven:
  72                 // Path is "/{uuid}" or "/{uuid}/", has no object name
  73                 if _, ok := r.URL.Query()["versioning"]; ok {
  74                         // GetBucketVersioning
  75                         w.Header().Set("Content-Type", "application/xml")
  76                         io.WriteString(w, xml.Header)
  77                         fmt.Fprintln(w, `<VersioningConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/"/>`)
  78                 } else {
  79                         // ListObjects
  80                         h.s3list(w, r, fs)
  81                 }
  82                 return true
  83         case r.Method == "GET" || r.Method == "HEAD":
  84                 fspath := "/by_id" + r.URL.Path
  85                 fi, err := fs.Stat(fspath)
  86                 if r.Method == "HEAD" && !objectNameGiven {
  87                         // HeadBucket
  88                         if err != nil && fi.IsDir() {
  89                                 w.WriteHeader(http.StatusOK)
  90                         } else if os.IsNotExist(err) {
  91                                 w.WriteHeader(http.StatusNotFound)
  92                         } else {
  93                                 http.Error(w, err.Error(), http.StatusBadGateway)
  94                         }
  95                         return true
  96                 }
  97                 if os.IsNotExist(err) ||
  98                         (err != nil && err.Error() == "not a directory") ||
  99                         (fi != nil && fi.IsDir()) {
 100                         http.Error(w, "not found", http.StatusNotFound)
 101                         return true
 102                 }
 103                 // shallow copy r, and change URL path
 104                 r := *r
 105                 r.URL.Path = fspath
 106                 http.FileServer(fs).ServeHTTP(w, &r)
 107                 return true
 108         case r.Method == "PUT":
 109                 if strings.HasSuffix(r.URL.Path, "/") {
 110                         http.Error(w, "invalid object name (trailing '/' char)", http.StatusBadRequest)
 111                         return true
 112                 }
 113                 fspath := "by_id" + r.URL.Path
 114                 _, err = fs.Stat(fspath)
 115                 if err != nil && err.Error() == "not a directory" {
 116                         // requested foo/bar, but foo is a file
 117                         http.Error(w, "object name conflicts with existing object", http.StatusBadRequest)
 118                         return true
 119                 }
 120                 f, err := fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 121                 if os.IsNotExist(err) {
 122                         // create missing intermediate directories, then try again
 123                         for i, c := range fspath {
 124                                 if i > 0 && c == '/' {
 125                                         dir := fspath[:i]
 126                                         if strings.HasSuffix(dir, "/") {
 127                                                 err = errors.New("invalid object name (consecutive '/' chars)")
 128                                                 http.Error(w, err.Error(), http.StatusBadRequest)
 129                                                 return true
 130                                         }
 131                                         err := fs.Mkdir(dir, 0755)
 132                                         if err != nil && err != os.ErrExist {
 133                                                 err = fmt.Errorf("mkdir %q failed: %w", dir, err)
 134                                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 135                                                 return true
 136                                         }
 137                                 }
 138                         }
 139                         f, err = fs.OpenFile(fspath, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
 140                 }
 141                 if err != nil {
 142                         err = fmt.Errorf("open %q failed: %w", r.URL.Path, err)
 143                         http.Error(w, err.Error(), http.StatusBadRequest)
 144                         return true
 145                 }
 146                 defer f.Close()
 147                 _, err = io.Copy(f, r.Body)
 148                 if err != nil {
 149                         err = fmt.Errorf("write to %q failed: %w", r.URL.Path, err)
 150                         http.Error(w, err.Error(), http.StatusBadGateway)
 151                         return true
 152                 }
 153                 err = f.Close()
 154                 if err != nil {
 155                         err = fmt.Errorf("write to %q failed: close: %w", r.URL.Path, err)
 156                         http.Error(w, err.Error(), http.StatusBadGateway)
 157                         return true
 158                 }
 159                 err = fs.Sync()
 160                 if err != nil {
 161                         err = fmt.Errorf("sync failed: %w", err)
 162                         http.Error(w, err.Error(), http.StatusInternalServerError)
 163                         return true
 164                 }
 165                 w.WriteHeader(http.StatusOK)
 166                 return true
 167         default:
 168                 http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
 169                 return true
 170         }
 171 }
 172
 173 func walkFS(fs arvados.CustomFileSystem, path string, ignoreNotFound bool, fn func(path string, fi os.FileInfo) error) error {
 174         f, err := fs.Open(path)
 175         if os.IsNotExist(err) && ignoreNotFound {
 176                 return nil
 177         } else if err != nil {
 178                 return fmt.Errorf("open %q: %w", path, err)
 179         }
 180         defer f.Close()
 181         if path == "/" {
 182                 path = ""
 183         }
 184         fis, err := f.Readdir(-1)
 185         if err != nil {
 186                 return err
 187         }
 188         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 189         for _, fi := range fis {
 190                 err = fn(path+"/"+fi.Name(), fi)
 191                 if err == filepath.SkipDir {
 192                         continue
 193                 } else if err != nil {
 194                         return err
 195                 }
 196                 if fi.IsDir() {
 197                         err = walkFS(fs, path+"/"+fi.Name(), false, fn)
 198                         if err != nil {
 199                                 return err
 200                         }
 201                 }
 202         }
 203         return nil
 204 }
 205
 206 var errDone = errors.New("done")
 207
 208 func (h *handler) s3list(w http.ResponseWriter, r *http.Request, fs arvados.CustomFileSystem) {
 209         var params struct {
 210                 bucket    string
 211                 delimiter string
 212                 marker    string
 213                 maxKeys   int
 214                 prefix    string
 215         }
 216         params.bucket = strings.SplitN(r.URL.Path[1:], "/", 2)[0]
 217         params.delimiter = r.FormValue("delimiter")
 218         params.marker = r.FormValue("marker")
 219         if mk, _ := strconv.ParseInt(r.FormValue("max-keys"), 10, 64); mk > 0 && mk < s3MaxKeys {
 220                 params.maxKeys = int(mk)
 221         } else {
 222                 params.maxKeys = s3MaxKeys
 223         }
 224         params.prefix = r.FormValue("prefix")
 225
 226         bucketdir := "by_id/" + params.bucket
 227         // walkpath is the directory (relative to bucketdir) we need
 228         // to walk: the innermost directory that is guaranteed to
 229         // contain all paths that have the requested prefix. Examples:
 230         // prefix "foo/bar"  => walkpath "foo"
 231         // prefix "foo/bar/" => walkpath "foo/bar"
 232         // prefix "foo"      => walkpath ""
 233         // prefix ""         => walkpath ""
 234         walkpath := params.prefix
 235         if cut := strings.LastIndex(walkpath, "/"); cut >= 0 {
 236                 walkpath = walkpath[:cut]
 237         } else {
 238                 walkpath = ""
 239         }
 240
 241         resp := s3.ListResp{
 242                 Name:      strings.SplitN(r.URL.Path[1:], "/", 2)[0],
 243                 Prefix:    params.prefix,
 244                 Delimiter: params.delimiter,
 245                 Marker:    params.marker,
 246                 MaxKeys:   params.maxKeys,
 247         }
 248         commonPrefixes := map[string]bool{}
 249         err := walkFS(fs, strings.TrimSuffix(bucketdir+"/"+walkpath, "/"), true, func(path string, fi os.FileInfo) error {
 250                 path = path[len(bucketdir)+1:]
 251                 if len(path) <= len(params.prefix) {
 252                         if path > params.prefix[:len(path)] {
 253                                 // with prefix "foobar", walking "fooz" means we're done
 254                                 return errDone
 255                         }
 256                         if path < params.prefix[:len(path)] {
 257                                 // with prefix "foobar", walking "foobag" is pointless
 258                                 return filepath.SkipDir
 259                         }
 260                         if fi.IsDir() && !strings.HasPrefix(params.prefix+"/", path+"/") {
 261                                 // with prefix "foo/bar", walking "fo"
 262                                 // is pointless (but walking "foo" or
 263                                 // "foo/bar" is necessary)
 264                                 return filepath.SkipDir
 265                         }
 266                         if len(path) < len(params.prefix) {
 267                                 // can't skip anything, and this entry
 268                                 // isn't in the results, so just
 269                                 // continue descent
 270                                 return nil
 271                         }
 272                 } else {
 273                         if path[:len(params.prefix)] > params.prefix {
 274                                 // with prefix "foobar", nothing we
 275                                 // see after "foozzz" is relevant
 276                                 return errDone
 277                         }
 278                 }
 279                 if path < params.marker || path < params.prefix {
 280                         return nil
 281                 }
 282                 if fi.IsDir() {
 283                         return nil
 284                 }
 285                 if params.delimiter != "" {
 286                         idx := strings.Index(path[len(params.prefix):], params.delimiter)
 287                         if idx >= 0 {
 288                                 // with prefix "foobar" and delimiter
 289                                 // "z", when we hit "foobar/baz", we
 290                                 // add "/baz" to commonPrefixes and
 291                                 // stop descending (note that even if
 292                                 // delimiter is "/" we don't add
 293                                 // anything to commonPrefixes when
 294                                 // seeing a dir: we wait until we see
 295                                 // a file, so we don't incorrectly
 296                                 // return results for empty dirs)
 297                                 commonPrefixes[path[:len(params.prefix)+idx+1]] = true
 298                                 return filepath.SkipDir
 299                         }
 300                 }
 301                 if len(resp.Contents)+len(commonPrefixes) >= params.maxKeys {
 302                         resp.IsTruncated = true
 303                         if params.delimiter != "" {
 304                                 resp.NextMarker = path
 305                         }
 306                         return errDone
 307                 }
 308                 resp.Contents = append(resp.Contents, s3.Key{
 309                         Key: path,
 310                 })
 311                 return nil
 312         })
 313         if err != nil && err != errDone {
 314                 http.Error(w, err.Error(), http.StatusInternalServerError)
 315                 return
 316         }
 317         if params.delimiter != "" {
 318                 for prefix := range commonPrefixes {
 319                         resp.CommonPrefixes = append(resp.CommonPrefixes, prefix)
 320                         sort.Strings(resp.CommonPrefixes)
 321                 }
 322         }
 323         wrappedResp := struct {
 324                 XMLName string `xml:"http://s3.amazonaws.com/doc/2006-03-01/ ListBucketResult"`
 325                 s3.ListResp
 326         }{"", resp}
 327         w.Header().Set("Content-Type", "application/xml")
 328         io.WriteString(w, xml.Header)
 329         if err := xml.NewEncoder(w).Encode(wrappedResp); err != nil {
 330                 ctxlog.FromContext(r.Context()).WithError(err).Error("error writing xml response")
 331         }
 332 }