services/keep-web/s3_test.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         "bytes"
   9         "crypto/rand"
  10         "fmt"
  11         "io/ioutil"
  12         "net/http"
  13         "os"
  14         "os/exec"
  15         "strings"
  16         "sync"
  17         "time"
  18
  19         "git.arvados.org/arvados.git/sdk/go/arvados"
  20         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  21         "git.arvados.org/arvados.git/sdk/go/arvadostest"
  22         "git.arvados.org/arvados.git/sdk/go/keepclient"
  23         "github.com/AdRoll/goamz/aws"
  24         "github.com/AdRoll/goamz/s3"
  25         check "gopkg.in/check.v1"
  26 )
  27
  28 type s3stage struct {
  29         arv        *arvados.Client
  30         ac         *arvadosclient.ArvadosClient
  31         kc         *keepclient.KeepClient
  32         proj       arvados.Group
  33         projbucket *s3.Bucket
  34         coll       arvados.Collection
  35         collbucket *s3.Bucket
  36 }
  37
  38 func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
  39         var proj arvados.Group
  40         var coll arvados.Collection
  41         arv := arvados.NewClientFromEnv()
  42         arv.AuthToken = arvadostest.ActiveToken
  43         err := arv.RequestAndDecode(&proj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
  44                 "group": map[string]interface{}{
  45                         "group_class": "project",
  46                         "name":        "keep-web s3 test",
  47                 },
  48                 "ensure_unique_name": true,
  49         })
  50         c.Assert(err, check.IsNil)
  51         err = arv.RequestAndDecode(&coll, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{
  52                 "owner_uuid":    proj.UUID,
  53                 "name":          "keep-web s3 test collection",
  54                 "manifest_text": ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:emptyfile\n./emptydir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n",
  55         }})
  56         c.Assert(err, check.IsNil)
  57         ac, err := arvadosclient.New(arv)
  58         c.Assert(err, check.IsNil)
  59         kc, err := keepclient.MakeKeepClient(ac)
  60         c.Assert(err, check.IsNil)
  61         fs, err := coll.FileSystem(arv, kc)
  62         c.Assert(err, check.IsNil)
  63         f, err := fs.OpenFile("sailboat.txt", os.O_CREATE|os.O_WRONLY, 0644)
  64         c.Assert(err, check.IsNil)
  65         _, err = f.Write([]byte("⛵\n"))
  66         c.Assert(err, check.IsNil)
  67         err = f.Close()
  68         c.Assert(err, check.IsNil)
  69         err = fs.Sync()
  70         c.Assert(err, check.IsNil)
  71         err = arv.RequestAndDecode(&coll, "GET", "arvados/v1/collections/"+coll.UUID, nil, nil)
  72         c.Assert(err, check.IsNil)
  73
  74         auth := aws.NewAuth(arvadostest.ActiveTokenUUID, arvadostest.ActiveToken, "", time.Now().Add(time.Hour))
  75         region := aws.Region{
  76                 Name:       s.testServer.Addr,
  77                 S3Endpoint: "http://" + s.testServer.Addr,
  78         }
  79         client := s3.New(*auth, region)
  80         client.Signature = aws.V4Signature
  81         return s3stage{
  82                 arv:  arv,
  83                 ac:   ac,
  84                 kc:   kc,
  85                 proj: proj,
  86                 projbucket: &s3.Bucket{
  87                         S3:   client,
  88                         Name: proj.UUID,
  89                 },
  90                 coll: coll,
  91                 collbucket: &s3.Bucket{
  92                         S3:   client,
  93                         Name: coll.UUID,
  94                 },
  95         }
  96 }
  97
  98 func (stage s3stage) teardown(c *check.C) {
  99         if stage.coll.UUID != "" {
 100                 err := stage.arv.RequestAndDecode(&stage.coll, "DELETE", "arvados/v1/collections/"+stage.coll.UUID, nil, nil)
 101                 c.Check(err, check.IsNil)
 102         }
 103         if stage.proj.UUID != "" {
 104                 err := stage.arv.RequestAndDecode(&stage.proj, "DELETE", "arvados/v1/groups/"+stage.proj.UUID, nil, nil)
 105                 c.Check(err, check.IsNil)
 106         }
 107 }
 108
 109 func (s *IntegrationSuite) TestS3Signatures(c *check.C) {
 110         stage := s.s3setup(c)
 111         defer stage.teardown(c)
 112
 113         bucket := stage.collbucket
 114         for _, trial := range []struct {
 115                 success   bool
 116                 signature int
 117                 accesskey string
 118                 secretkey string
 119         }{
 120                 {true, aws.V2Signature, arvadostest.ActiveToken, "none"},
 121                 {false, aws.V2Signature, "none", "none"},
 122                 {false, aws.V2Signature, "none", arvadostest.ActiveToken},
 123
 124                 {true, aws.V4Signature, arvadostest.ActiveTokenUUID, arvadostest.ActiveToken},
 125                 {true, aws.V4Signature, arvadostest.ActiveToken, arvadostest.ActiveToken},
 126                 {false, aws.V4Signature, arvadostest.ActiveToken, ""},
 127                 {false, aws.V4Signature, arvadostest.ActiveToken, "none"},
 128                 {false, aws.V4Signature, "none", arvadostest.ActiveToken},
 129                 {false, aws.V4Signature, "none", "none"},
 130         } {
 131                 c.Logf("%#v", trial)
 132                 bucket.S3.Auth = *(aws.NewAuth(trial.accesskey, trial.secretkey, "", time.Now().Add(time.Hour)))
 133                 bucket.S3.Signature = trial.signature
 134                 _, err := bucket.GetReader("emptyfile")
 135                 if trial.success {
 136                         c.Check(err, check.IsNil)
 137                 } else {
 138                         c.Check(err, check.NotNil)
 139                 }
 140         }
 141 }
 142
 143 func (s *IntegrationSuite) TestS3HeadBucket(c *check.C) {
 144         stage := s.s3setup(c)
 145         defer stage.teardown(c)
 146
 147         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 148                 c.Logf("bucket %s", bucket.Name)
 149                 exists, err := bucket.Exists("")
 150                 c.Check(err, check.IsNil)
 151                 c.Check(exists, check.Equals, true)
 152         }
 153 }
 154
 155 func (s *IntegrationSuite) TestS3CollectionGetObject(c *check.C) {
 156         stage := s.s3setup(c)
 157         defer stage.teardown(c)
 158         s.testS3GetObject(c, stage.collbucket, "")
 159 }
 160 func (s *IntegrationSuite) TestS3ProjectGetObject(c *check.C) {
 161         stage := s.s3setup(c)
 162         defer stage.teardown(c)
 163         s.testS3GetObject(c, stage.projbucket, stage.coll.Name+"/")
 164 }
 165 func (s *IntegrationSuite) testS3GetObject(c *check.C, bucket *s3.Bucket, prefix string) {
 166         rdr, err := bucket.GetReader(prefix + "emptyfile")
 167         c.Assert(err, check.IsNil)
 168         buf, err := ioutil.ReadAll(rdr)
 169         c.Check(err, check.IsNil)
 170         c.Check(len(buf), check.Equals, 0)
 171         err = rdr.Close()
 172         c.Check(err, check.IsNil)
 173
 174         // GetObject
 175         rdr, err = bucket.GetReader(prefix + "missingfile")
 176         c.Check(err, check.ErrorMatches, `404 Not Found`)
 177
 178         // HeadObject
 179         exists, err := bucket.Exists(prefix + "missingfile")
 180         c.Check(err, check.IsNil)
 181         c.Check(exists, check.Equals, false)
 182
 183         // GetObject
 184         rdr, err = bucket.GetReader(prefix + "sailboat.txt")
 185         c.Assert(err, check.IsNil)
 186         buf, err = ioutil.ReadAll(rdr)
 187         c.Check(err, check.IsNil)
 188         c.Check(buf, check.DeepEquals, []byte("⛵\n"))
 189         err = rdr.Close()
 190         c.Check(err, check.IsNil)
 191
 192         // HeadObject
 193         resp, err := bucket.Head(prefix+"sailboat.txt", nil)
 194         c.Check(err, check.IsNil)
 195         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 196         c.Check(resp.ContentLength, check.Equals, int64(4))
 197 }
 198
 199 func (s *IntegrationSuite) TestS3CollectionPutObjectSuccess(c *check.C) {
 200         stage := s.s3setup(c)
 201         defer stage.teardown(c)
 202         s.testS3PutObjectSuccess(c, stage.collbucket, "")
 203 }
 204 func (s *IntegrationSuite) TestS3ProjectPutObjectSuccess(c *check.C) {
 205         stage := s.s3setup(c)
 206         defer stage.teardown(c)
 207         s.testS3PutObjectSuccess(c, stage.projbucket, stage.coll.Name+"/")
 208 }
 209 func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket, prefix string) {
 210         for _, trial := range []struct {
 211                 path        string
 212                 size        int
 213                 contentType string
 214         }{
 215                 {
 216                         path:        "newfile",
 217                         size:        128000000,
 218                         contentType: "application/octet-stream",
 219                 }, {
 220                         path:        "newdir/newfile",
 221                         size:        1 << 26,
 222                         contentType: "application/octet-stream",
 223                 }, {
 224                         path:        "newdir1/newdir2/newfile",
 225                         size:        0,
 226                         contentType: "application/octet-stream",
 227                 }, {
 228                         path:        "newdir1/newdir2/newdir3/",
 229                         size:        0,
 230                         contentType: "application/x-directory",
 231                 },
 232         } {
 233                 c.Logf("=== %v", trial)
 234
 235                 objname := prefix + trial.path
 236
 237                 _, err := bucket.GetReader(objname)
 238                 c.Assert(err, check.ErrorMatches, `404 Not Found`)
 239
 240                 buf := make([]byte, trial.size)
 241                 rand.Read(buf)
 242
 243                 err = bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), trial.contentType, s3.Private, s3.Options{})
 244                 c.Check(err, check.IsNil)
 245
 246                 rdr, err := bucket.GetReader(objname)
 247                 if strings.HasSuffix(trial.path, "/") && !s.testServer.Config.cluster.Collections.S3FolderObjects {
 248                         c.Check(err, check.NotNil)
 249                         continue
 250                 } else if !c.Check(err, check.IsNil) {
 251                         continue
 252                 }
 253                 buf2, err := ioutil.ReadAll(rdr)
 254                 c.Check(err, check.IsNil)
 255                 c.Check(buf2, check.HasLen, len(buf))
 256                 c.Check(bytes.Equal(buf, buf2), check.Equals, true)
 257         }
 258 }
 259
 260 func (s *IntegrationSuite) TestS3ProjectPutObjectNotSupported(c *check.C) {
 261         stage := s.s3setup(c)
 262         defer stage.teardown(c)
 263         bucket := stage.projbucket
 264
 265         for _, trial := range []struct {
 266                 path        string
 267                 size        int
 268                 contentType string
 269         }{
 270                 {
 271                         path:        "newfile",
 272                         size:        1234,
 273                         contentType: "application/octet-stream",
 274                 }, {
 275                         path:        "newdir/newfile",
 276                         size:        1234,
 277                         contentType: "application/octet-stream",
 278                 }, {
 279                         path:        "newdir2/",
 280                         size:        0,
 281                         contentType: "application/x-directory",
 282                 },
 283         } {
 284                 c.Logf("=== %v", trial)
 285
 286                 _, err := bucket.GetReader(trial.path)
 287                 c.Assert(err, check.ErrorMatches, `404 Not Found`)
 288
 289                 buf := make([]byte, trial.size)
 290                 rand.Read(buf)
 291
 292                 err = bucket.PutReader(trial.path, bytes.NewReader(buf), int64(len(buf)), trial.contentType, s3.Private, s3.Options{})
 293                 c.Check(err, check.ErrorMatches, `400 Bad Request`)
 294
 295                 _, err = bucket.GetReader(trial.path)
 296                 c.Assert(err, check.ErrorMatches, `404 Not Found`)
 297         }
 298 }
 299
 300 func (s *IntegrationSuite) TestS3CollectionDeleteObject(c *check.C) {
 301         stage := s.s3setup(c)
 302         defer stage.teardown(c)
 303         s.testS3DeleteObject(c, stage.collbucket, "")
 304 }
 305 func (s *IntegrationSuite) TestS3ProjectDeleteObject(c *check.C) {
 306         stage := s.s3setup(c)
 307         defer stage.teardown(c)
 308         s.testS3DeleteObject(c, stage.projbucket, stage.coll.Name+"/")
 309 }
 310 func (s *IntegrationSuite) testS3DeleteObject(c *check.C, bucket *s3.Bucket, prefix string) {
 311         s.testServer.Config.cluster.Collections.S3FolderObjects = true
 312         for _, trial := range []struct {
 313                 path string
 314         }{
 315                 {"/"},
 316                 {"nonexistentfile"},
 317                 {"emptyfile"},
 318                 {"sailboat.txt"},
 319                 {"sailboat.txt/"},
 320                 {"emptydir"},
 321                 {"emptydir/"},
 322         } {
 323                 objname := prefix + trial.path
 324                 comment := check.Commentf("objname %q", objname)
 325
 326                 err := bucket.Del(objname)
 327                 if trial.path == "/" {
 328                         c.Check(err, check.NotNil)
 329                         continue
 330                 }
 331                 c.Check(err, check.IsNil, comment)
 332                 _, err = bucket.GetReader(objname)
 333                 c.Check(err, check.NotNil, comment)
 334         }
 335 }
 336
 337 func (s *IntegrationSuite) TestS3CollectionPutObjectFailure(c *check.C) {
 338         stage := s.s3setup(c)
 339         defer stage.teardown(c)
 340         s.testS3PutObjectFailure(c, stage.collbucket, "")
 341 }
 342 func (s *IntegrationSuite) TestS3ProjectPutObjectFailure(c *check.C) {
 343         stage := s.s3setup(c)
 344         defer stage.teardown(c)
 345         s.testS3PutObjectFailure(c, stage.projbucket, stage.coll.Name+"/")
 346 }
 347 func (s *IntegrationSuite) testS3PutObjectFailure(c *check.C, bucket *s3.Bucket, prefix string) {
 348         s.testServer.Config.cluster.Collections.S3FolderObjects = false
 349
 350         // Can't use V4 signature for these tests, because
 351         // double-slash is incorrectly cleaned by the aws.V4Signature,
 352         // resulting in a "bad signature" error. (Cleaning the path is
 353         // appropriate for other services, but not in S3 where object
 354         // names "foo//bar" and "foo/bar" are semantically different.)
 355         bucket.S3.Auth = *(aws.NewAuth(arvadostest.ActiveToken, "none", "", time.Now().Add(time.Hour)))
 356         bucket.S3.Signature = aws.V2Signature
 357
 358         var wg sync.WaitGroup
 359         for _, trial := range []struct {
 360                 path string
 361         }{
 362                 {
 363                         path: "emptyfile/newname", // emptyfile exists, see s3setup()
 364                 }, {
 365                         path: "emptyfile/", // emptyfile exists, see s3setup()
 366                 }, {
 367                         path: "emptydir", // dir already exists, see s3setup()
 368                 }, {
 369                         path: "emptydir/",
 370                 }, {
 371                         path: "emptydir//",
 372                 }, {
 373                         path: "newdir/",
 374                 }, {
 375                         path: "newdir//",
 376                 }, {
 377                         path: "/",
 378                 }, {
 379                         path: "//",
 380                 }, {
 381                         path: "foo//bar",
 382                 }, {
 383                         path: "",
 384                 },
 385         } {
 386                 trial := trial
 387                 wg.Add(1)
 388                 go func() {
 389                         defer wg.Done()
 390                         c.Logf("=== %v", trial)
 391
 392                         objname := prefix + trial.path
 393
 394                         buf := make([]byte, 1234)
 395                         rand.Read(buf)
 396
 397                         err := bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), "application/octet-stream", s3.Private, s3.Options{})
 398                         if !c.Check(err, check.ErrorMatches, `400 Bad.*`, check.Commentf("PUT %q should fail", objname)) {
 399                                 return
 400                         }
 401
 402                         if objname != "" && objname != "/" {
 403                                 _, err = bucket.GetReader(objname)
 404                                 c.Check(err, check.ErrorMatches, `404 Not Found`, check.Commentf("GET %q should return 404", objname))
 405                         }
 406                 }()
 407         }
 408         wg.Wait()
 409 }
 410
 411 func (stage *s3stage) writeBigDirs(c *check.C, dirs int, filesPerDir int) {
 412         fs, err := stage.coll.FileSystem(stage.arv, stage.kc)
 413         c.Assert(err, check.IsNil)
 414         for d := 0; d < dirs; d++ {
 415                 dir := fmt.Sprintf("dir%d", d)
 416                 c.Assert(fs.Mkdir(dir, 0755), check.IsNil)
 417                 for i := 0; i < filesPerDir; i++ {
 418                         f, err := fs.OpenFile(fmt.Sprintf("%s/file%d.txt", dir, i), os.O_CREATE|os.O_WRONLY, 0644)
 419                         c.Assert(err, check.IsNil)
 420                         c.Assert(f.Close(), check.IsNil)
 421                 }
 422         }
 423         c.Assert(fs.Sync(), check.IsNil)
 424 }
 425
 426 func (s *IntegrationSuite) TestS3GetBucketVersioning(c *check.C) {
 427         stage := s.s3setup(c)
 428         defer stage.teardown(c)
 429         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 430                 req, err := http.NewRequest("GET", bucket.URL("/"), nil)
 431                 c.Check(err, check.IsNil)
 432                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 433                 req.URL.RawQuery = "versioning"
 434                 resp, err := http.DefaultClient.Do(req)
 435                 c.Assert(err, check.IsNil)
 436                 c.Check(resp.Header.Get("Content-Type"), check.Equals, "application/xml")
 437                 buf, err := ioutil.ReadAll(resp.Body)
 438                 c.Assert(err, check.IsNil)
 439                 c.Check(string(buf), check.Equals, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<VersioningConfiguration xmlns=\"http://s3.amazonaws.com/doc/2006-03-01/\"/>\n")
 440         }
 441 }
 442
 443 // If there are no CommonPrefixes entries, the CommonPrefixes XML tag
 444 // should not appear at all.
 445 func (s *IntegrationSuite) TestS3ListNoCommonPrefixes(c *check.C) {
 446         stage := s.s3setup(c)
 447         defer stage.teardown(c)
 448
 449         req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 450         c.Assert(err, check.IsNil)
 451         req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 452         req.URL.RawQuery = "prefix=asdfasdfasdf&delimiter=/"
 453         resp, err := http.DefaultClient.Do(req)
 454         c.Assert(err, check.IsNil)
 455         buf, err := ioutil.ReadAll(resp.Body)
 456         c.Assert(err, check.IsNil)
 457         c.Check(string(buf), check.Not(check.Matches), `(?ms).*CommonPrefixes.*`)
 458 }
 459
 460 // If there is no delimiter in the request, or the results are not
 461 // truncated, the NextMarker XML tag should not appear in the response
 462 // body.
 463 func (s *IntegrationSuite) TestS3ListNoNextMarker(c *check.C) {
 464         stage := s.s3setup(c)
 465         defer stage.teardown(c)
 466
 467         for _, query := range []string{"prefix=e&delimiter=/", ""} {
 468                 req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 469                 c.Assert(err, check.IsNil)
 470                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 471                 req.URL.RawQuery = query
 472                 resp, err := http.DefaultClient.Do(req)
 473                 c.Assert(err, check.IsNil)
 474                 buf, err := ioutil.ReadAll(resp.Body)
 475                 c.Assert(err, check.IsNil)
 476                 c.Check(string(buf), check.Not(check.Matches), `(?ms).*NextMarker.*`)
 477         }
 478 }
 479
 480 // List response should include KeyCount field.
 481 func (s *IntegrationSuite) TestS3ListKeyCount(c *check.C) {
 482         stage := s.s3setup(c)
 483         defer stage.teardown(c)
 484
 485         req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 486         c.Assert(err, check.IsNil)
 487         req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 488         req.URL.RawQuery = "prefix=&delimiter=/"
 489         resp, err := http.DefaultClient.Do(req)
 490         c.Assert(err, check.IsNil)
 491         buf, err := ioutil.ReadAll(resp.Body)
 492         c.Assert(err, check.IsNil)
 493         c.Check(string(buf), check.Matches, `(?ms).*<KeyCount>2</KeyCount>.*`)
 494 }
 495
 496 func (s *IntegrationSuite) TestS3CollectionList(c *check.C) {
 497         stage := s.s3setup(c)
 498         defer stage.teardown(c)
 499
 500         var markers int
 501         for markers, s.testServer.Config.cluster.Collections.S3FolderObjects = range []bool{false, true} {
 502                 dirs := 2
 503                 filesPerDir := 1001
 504                 stage.writeBigDirs(c, dirs, filesPerDir)
 505                 // Total # objects is:
 506                 //                 2 file entries from s3setup (emptyfile and sailboat.txt)
 507                 //                +1 fake "directory" marker from s3setup (emptydir) (if enabled)
 508                 //             +dirs fake "directory" marker from writeBigDirs (dir0/, dir1/) (if enabled)
 509                 // +filesPerDir*dirs file entries from writeBigDirs (dir0/file0.txt, etc.)
 510                 s.testS3List(c, stage.collbucket, "", 4000, markers+2+(filesPerDir+markers)*dirs)
 511                 s.testS3List(c, stage.collbucket, "", 131, markers+2+(filesPerDir+markers)*dirs)
 512                 s.testS3List(c, stage.collbucket, "dir0/", 71, filesPerDir+markers)
 513         }
 514 }
 515 func (s *IntegrationSuite) testS3List(c *check.C, bucket *s3.Bucket, prefix string, pageSize, expectFiles int) {
 516         c.Logf("testS3List: prefix=%q pageSize=%d S3FolderObjects=%v", prefix, pageSize, s.testServer.Config.cluster.Collections.S3FolderObjects)
 517         expectPageSize := pageSize
 518         if expectPageSize > 1000 {
 519                 expectPageSize = 1000
 520         }
 521         gotKeys := map[string]s3.Key{}
 522         nextMarker := ""
 523         pages := 0
 524         for {
 525                 resp, err := bucket.List(prefix, "", nextMarker, pageSize)
 526                 if !c.Check(err, check.IsNil) {
 527                         break
 528                 }
 529                 c.Check(len(resp.Contents) <= expectPageSize, check.Equals, true)
 530                 if pages++; !c.Check(pages <= (expectFiles/expectPageSize)+1, check.Equals, true) {
 531                         break
 532                 }
 533                 for _, key := range resp.Contents {
 534                         gotKeys[key.Key] = key
 535                         if strings.Contains(key.Key, "sailboat.txt") {
 536                                 c.Check(key.Size, check.Equals, int64(4))
 537                         }
 538                 }
 539                 if !resp.IsTruncated {
 540                         c.Check(resp.NextMarker, check.Equals, "")
 541                         break
 542                 }
 543                 if !c.Check(resp.NextMarker, check.Not(check.Equals), "") {
 544                         break
 545                 }
 546                 nextMarker = resp.NextMarker
 547         }
 548         c.Check(len(gotKeys), check.Equals, expectFiles)
 549 }
 550
 551 func (s *IntegrationSuite) TestS3CollectionListRollup(c *check.C) {
 552         for _, s.testServer.Config.cluster.Collections.S3FolderObjects = range []bool{false, true} {
 553                 s.testS3CollectionListRollup(c)
 554         }
 555 }
 556
 557 func (s *IntegrationSuite) testS3CollectionListRollup(c *check.C) {
 558         stage := s.s3setup(c)
 559         defer stage.teardown(c)
 560
 561         dirs := 2
 562         filesPerDir := 500
 563         stage.writeBigDirs(c, dirs, filesPerDir)
 564         err := stage.collbucket.PutReader("dingbats", &bytes.Buffer{}, 0, "application/octet-stream", s3.Private, s3.Options{})
 565         c.Assert(err, check.IsNil)
 566         var allfiles []string
 567         for marker := ""; ; {
 568                 resp, err := stage.collbucket.List("", "", marker, 20000)
 569                 c.Check(err, check.IsNil)
 570                 for _, key := range resp.Contents {
 571                         if len(allfiles) == 0 || allfiles[len(allfiles)-1] != key.Key {
 572                                 allfiles = append(allfiles, key.Key)
 573                         }
 574                 }
 575                 marker = resp.NextMarker
 576                 if marker == "" {
 577                         break
 578                 }
 579         }
 580         markers := 0
 581         if s.testServer.Config.cluster.Collections.S3FolderObjects {
 582                 markers = 1
 583         }
 584         c.Check(allfiles, check.HasLen, dirs*(filesPerDir+markers)+3+markers)
 585
 586         gotDirMarker := map[string]bool{}
 587         for _, name := range allfiles {
 588                 isDirMarker := strings.HasSuffix(name, "/")
 589                 if markers == 0 {
 590                         c.Check(isDirMarker, check.Equals, false, check.Commentf("name %q", name))
 591                 } else if isDirMarker {
 592                         gotDirMarker[name] = true
 593                 } else if i := strings.LastIndex(name, "/"); i >= 0 {
 594                         c.Check(gotDirMarker[name[:i+1]], check.Equals, true, check.Commentf("name %q", name))
 595                         gotDirMarker[name[:i+1]] = true // skip redundant complaints about this dir marker
 596                 }
 597         }
 598
 599         for _, trial := range []struct {
 600                 prefix    string
 601                 delimiter string
 602                 marker    string
 603         }{
 604                 {"", "", ""},
 605                 {"di", "/", ""},
 606                 {"di", "r", ""},
 607                 {"di", "n", ""},
 608                 {"dir0", "/", ""},
 609                 {"dir0/", "/", ""},
 610                 {"dir0/f", "/", ""},
 611                 {"dir0", "", ""},
 612                 {"dir0/", "", ""},
 613                 {"dir0/f", "", ""},
 614                 {"dir0", "/", "dir0/file14.txt"},       // no commonprefixes
 615                 {"", "", "dir0/file14.txt"},            // middle page, skip walking dir1
 616                 {"", "", "dir1/file14.txt"},            // middle page, skip walking dir0
 617                 {"", "", "dir1/file498.txt"},           // last page of results
 618                 {"dir1/file", "", "dir1/file498.txt"},  // last page of results, with prefix
 619                 {"dir1/file", "/", "dir1/file498.txt"}, // last page of results, with prefix + delimiter
 620                 {"dir1", "Z", "dir1/file498.txt"},      // delimiter "Z" never appears
 621                 {"dir2", "/", ""},                      // prefix "dir2" does not exist
 622                 {"", "/", ""},
 623         } {
 624                 c.Logf("\n\n=== trial %+v markers=%d", trial, markers)
 625
 626                 maxKeys := 20
 627                 resp, err := stage.collbucket.List(trial.prefix, trial.delimiter, trial.marker, maxKeys)
 628                 c.Check(err, check.IsNil)
 629                 if resp.IsTruncated && trial.delimiter == "" {
 630                         // goamz List method fills in the missing
 631                         // NextMarker field if resp.IsTruncated, so
 632                         // now we can't really tell whether it was
 633                         // sent by the server or by goamz. In cases
 634                         // where it should be empty but isn't, assume
 635                         // it's goamz's fault.
 636                         resp.NextMarker = ""
 637                 }
 638
 639                 var expectKeys []string
 640                 var expectPrefixes []string
 641                 var expectNextMarker string
 642                 var expectTruncated bool
 643                 for _, key := range allfiles {
 644                         full := len(expectKeys)+len(expectPrefixes) >= maxKeys
 645                         if !strings.HasPrefix(key, trial.prefix) || key < trial.marker {
 646                                 continue
 647                         } else if idx := strings.Index(key[len(trial.prefix):], trial.delimiter); trial.delimiter != "" && idx >= 0 {
 648                                 prefix := key[:len(trial.prefix)+idx+1]
 649                                 if len(expectPrefixes) > 0 && expectPrefixes[len(expectPrefixes)-1] == prefix {
 650                                         // same prefix as previous key
 651                                 } else if full {
 652                                         expectNextMarker = key
 653                                         expectTruncated = true
 654                                 } else {
 655                                         expectPrefixes = append(expectPrefixes, prefix)
 656                                 }
 657                         } else if full {
 658                                 if trial.delimiter != "" {
 659                                         expectNextMarker = key
 660                                 }
 661                                 expectTruncated = true
 662                                 break
 663                         } else {
 664                                 expectKeys = append(expectKeys, key)
 665                         }
 666                 }
 667
 668                 var gotKeys []string
 669                 for _, key := range resp.Contents {
 670                         gotKeys = append(gotKeys, key.Key)
 671                 }
 672                 var gotPrefixes []string
 673                 for _, prefix := range resp.CommonPrefixes {
 674                         gotPrefixes = append(gotPrefixes, prefix)
 675                 }
 676                 commentf := check.Commentf("trial %+v markers=%d", trial, markers)
 677                 c.Check(gotKeys, check.DeepEquals, expectKeys, commentf)
 678                 c.Check(gotPrefixes, check.DeepEquals, expectPrefixes, commentf)
 679                 c.Check(resp.NextMarker, check.Equals, expectNextMarker, commentf)
 680                 c.Check(resp.IsTruncated, check.Equals, expectTruncated, commentf)
 681                 c.Logf("=== trial %+v keys %q prefixes %q nextMarker %q", trial, gotKeys, gotPrefixes, resp.NextMarker)
 682         }
 683 }
 684
 685 // TestS3cmd checks compatibility with the s3cmd command line tool, if
 686 // it's installed. As of Debian buster, s3cmd is only in backports, so
 687 // `arvados-server install` don't install it, and this test skips if
 688 // it's not installed.
 689 func (s *IntegrationSuite) TestS3cmd(c *check.C) {
 690         if _, err := exec.LookPath("s3cmd"); err != nil {
 691                 c.Skip("s3cmd not found")
 692                 return
 693         }
 694
 695         stage := s.s3setup(c)
 696         defer stage.teardown(c)
 697
 698         cmd := exec.Command("s3cmd", "--no-ssl", "--host="+s.testServer.Addr, "--host-bucket="+s.testServer.Addr, "--access_key="+arvadostest.ActiveTokenUUID, "--secret_key="+arvadostest.ActiveToken, "ls", "s3://"+arvadostest.FooCollection)
 699         buf, err := cmd.CombinedOutput()
 700         c.Check(err, check.IsNil)
 701         c.Check(string(buf), check.Matches, `.* 3 +s3://`+arvadostest.FooCollection+`/foo\n`)
 702 }
 703
 704 func (s *IntegrationSuite) TestS3BucketInHost(c *check.C) {
 705         stage := s.s3setup(c)
 706         defer stage.teardown(c)
 707
 708         hdr, body, _ := s.runCurl(c, "AWS "+arvadostest.ActiveTokenV2+":none", stage.coll.UUID+".collections.example.com", "/sailboat.txt")
 709         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 710         c.Check(body, check.Equals, "⛵\n")
 711 }