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