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 keepweb
   6
   7 import (
   8         "bytes"
   9         "context"
  10         "crypto/rand"
  11         "crypto/sha256"
  12         "fmt"
  13         "io/ioutil"
  14         "mime"
  15         "net/http"
  16         "net/http/httptest"
  17         "net/url"
  18         "os"
  19         "os/exec"
  20         "strings"
  21         "sync"
  22         "time"
  23
  24         "git.arvados.org/arvados.git/sdk/go/arvados"
  25         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  26         "git.arvados.org/arvados.git/sdk/go/arvadostest"
  27         "git.arvados.org/arvados.git/sdk/go/keepclient"
  28         "github.com/AdRoll/goamz/aws"
  29         "github.com/AdRoll/goamz/s3"
  30         aws_aws "github.com/aws/aws-sdk-go/aws"
  31         aws_credentials "github.com/aws/aws-sdk-go/aws/credentials"
  32         aws_session "github.com/aws/aws-sdk-go/aws/session"
  33         aws_s3 "github.com/aws/aws-sdk-go/service/s3"
  34         check "gopkg.in/check.v1"
  35 )
  36
  37 type s3stage struct {
  38         arv        *arvados.Client
  39         ac         *arvadosclient.ArvadosClient
  40         kc         *keepclient.KeepClient
  41         proj       arvados.Group
  42         projbucket *s3.Bucket
  43         subproj    arvados.Group
  44         coll       arvados.Collection
  45         collbucket *s3.Bucket
  46 }
  47
  48 func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
  49         var proj, subproj arvados.Group
  50         var coll arvados.Collection
  51         arv := arvados.NewClientFromEnv()
  52         arv.AuthToken = arvadostest.ActiveToken
  53         err := arv.RequestAndDecode(&proj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
  54                 "group": map[string]interface{}{
  55                         "group_class": "project",
  56                         "name":        "keep-web s3 test",
  57                         "properties": map[string]interface{}{
  58                                 "project-properties-key": "project properties value",
  59                         },
  60                 },
  61                 "ensure_unique_name": true,
  62         })
  63         c.Assert(err, check.IsNil)
  64         err = arv.RequestAndDecode(&subproj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
  65                 "group": map[string]interface{}{
  66                         "owner_uuid":  proj.UUID,
  67                         "group_class": "project",
  68                         "name":        "keep-web s3 test subproject",
  69                         "properties": map[string]interface{}{
  70                                 "subproject_properties_key": "subproject properties value",
  71                                 "invalid header key":        "this value will not be returned because key contains spaces",
  72                         },
  73                 },
  74         })
  75         c.Assert(err, check.IsNil)
  76         err = arv.RequestAndDecode(&coll, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{
  77                 "owner_uuid":    proj.UUID,
  78                 "name":          "keep-web s3 test collection",
  79                 "manifest_text": ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:emptyfile\n./emptydir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n",
  80                 "properties": map[string]interface{}{
  81                         "string":   "string value",
  82                         "array":    []string{"element1", "element2"},
  83                         "object":   map[string]interface{}{"key": map[string]interface{}{"key2": "value⛵"}},
  84                         "nonascii": "⛵",
  85                         "newline":  "foo\r\nX-Bad: header",
  86                         // This key cannot be expressed as a MIME
  87                         // header key, so it will be silently skipped
  88                         // (see "Inject" in PropertiesAsMetadata test)
  89                         "a: a\r\nInject": "bogus",
  90                 },
  91         }})
  92         c.Assert(err, check.IsNil)
  93         ac, err := arvadosclient.New(arv)
  94         c.Assert(err, check.IsNil)
  95         kc, err := keepclient.MakeKeepClient(ac)
  96         c.Assert(err, check.IsNil)
  97         fs, err := coll.FileSystem(arv, kc)
  98         c.Assert(err, check.IsNil)
  99         f, err := fs.OpenFile("sailboat.txt", os.O_CREATE|os.O_WRONLY, 0644)
 100         c.Assert(err, check.IsNil)
 101         _, err = f.Write([]byte("⛵\n"))
 102         c.Assert(err, check.IsNil)
 103         err = f.Close()
 104         c.Assert(err, check.IsNil)
 105         err = fs.Sync()
 106         c.Assert(err, check.IsNil)
 107         err = arv.RequestAndDecode(&coll, "GET", "arvados/v1/collections/"+coll.UUID, nil, nil)
 108         c.Assert(err, check.IsNil)
 109
 110         auth := aws.NewAuth(arvadostest.ActiveTokenUUID, arvadostest.ActiveToken, "", time.Now().Add(time.Hour))
 111         region := aws.Region{
 112                 Name:       "zzzzz",
 113                 S3Endpoint: s.testServer.URL,
 114         }
 115         client := s3.New(*auth, region)
 116         client.Signature = aws.V4Signature
 117         return s3stage{
 118                 arv:  arv,
 119                 ac:   ac,
 120                 kc:   kc,
 121                 proj: proj,
 122                 projbucket: &s3.Bucket{
 123                         S3:   client,
 124                         Name: proj.UUID,
 125                 },
 126                 subproj: subproj,
 127                 coll:    coll,
 128                 collbucket: &s3.Bucket{
 129                         S3:   client,
 130                         Name: coll.UUID,
 131                 },
 132         }
 133 }
 134
 135 func (stage s3stage) teardown(c *check.C) {
 136         if stage.coll.UUID != "" {
 137                 err := stage.arv.RequestAndDecode(&stage.coll, "DELETE", "arvados/v1/collections/"+stage.coll.UUID, nil, nil)
 138                 c.Check(err, check.IsNil)
 139         }
 140         if stage.proj.UUID != "" {
 141                 err := stage.arv.RequestAndDecode(&stage.proj, "DELETE", "arvados/v1/groups/"+stage.proj.UUID, nil, nil)
 142                 c.Check(err, check.IsNil)
 143         }
 144 }
 145
 146 func (s *IntegrationSuite) TestS3Signatures(c *check.C) {
 147         stage := s.s3setup(c)
 148         defer stage.teardown(c)
 149
 150         bucket := stage.collbucket
 151         for _, trial := range []struct {
 152                 success   bool
 153                 signature int
 154                 accesskey string
 155                 secretkey string
 156         }{
 157                 {true, aws.V2Signature, arvadostest.ActiveToken, "none"},
 158                 {true, aws.V2Signature, url.QueryEscape(arvadostest.ActiveTokenV2), "none"},
 159                 {true, aws.V2Signature, strings.Replace(arvadostest.ActiveTokenV2, "/", "_", -1), "none"},
 160                 {false, aws.V2Signature, "none", "none"},
 161                 {false, aws.V2Signature, "none", arvadostest.ActiveToken},
 162
 163                 {true, aws.V4Signature, arvadostest.ActiveTokenUUID, arvadostest.ActiveToken},
 164                 {true, aws.V4Signature, arvadostest.ActiveToken, arvadostest.ActiveToken},
 165                 {true, aws.V4Signature, url.QueryEscape(arvadostest.ActiveTokenV2), url.QueryEscape(arvadostest.ActiveTokenV2)},
 166                 {true, aws.V4Signature, strings.Replace(arvadostest.ActiveTokenV2, "/", "_", -1), strings.Replace(arvadostest.ActiveTokenV2, "/", "_", -1)},
 167                 {false, aws.V4Signature, arvadostest.ActiveToken, ""},
 168                 {false, aws.V4Signature, arvadostest.ActiveToken, "none"},
 169                 {false, aws.V4Signature, "none", arvadostest.ActiveToken},
 170                 {false, aws.V4Signature, "none", "none"},
 171         } {
 172                 c.Logf("%#v", trial)
 173                 bucket.S3.Auth = *(aws.NewAuth(trial.accesskey, trial.secretkey, "", time.Now().Add(time.Hour)))
 174                 bucket.S3.Signature = trial.signature
 175                 _, err := bucket.GetReader("emptyfile")
 176                 if trial.success {
 177                         c.Check(err, check.IsNil)
 178                 } else {
 179                         c.Check(err, check.NotNil)
 180                 }
 181         }
 182 }
 183
 184 func (s *IntegrationSuite) TestS3HeadBucket(c *check.C) {
 185         stage := s.s3setup(c)
 186         defer stage.teardown(c)
 187
 188         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 189                 c.Logf("bucket %s", bucket.Name)
 190                 exists, err := bucket.Exists("")
 191                 c.Check(err, check.IsNil)
 192                 c.Check(exists, check.Equals, true)
 193         }
 194 }
 195
 196 func (s *IntegrationSuite) TestS3CollectionGetObject(c *check.C) {
 197         stage := s.s3setup(c)
 198         defer stage.teardown(c)
 199         s.testS3GetObject(c, stage.collbucket, "")
 200 }
 201 func (s *IntegrationSuite) TestS3ProjectGetObject(c *check.C) {
 202         stage := s.s3setup(c)
 203         defer stage.teardown(c)
 204         s.testS3GetObject(c, stage.projbucket, stage.coll.Name+"/")
 205 }
 206 func (s *IntegrationSuite) testS3GetObject(c *check.C, bucket *s3.Bucket, prefix string) {
 207         rdr, err := bucket.GetReader(prefix + "emptyfile")
 208         c.Assert(err, check.IsNil)
 209         buf, err := ioutil.ReadAll(rdr)
 210         c.Check(err, check.IsNil)
 211         c.Check(len(buf), check.Equals, 0)
 212         err = rdr.Close()
 213         c.Check(err, check.IsNil)
 214
 215         // GetObject
 216         rdr, err = bucket.GetReader(prefix + "missingfile")
 217         c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
 218         c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
 219         c.Check(err, check.ErrorMatches, `The specified key does not exist.`)
 220
 221         // HeadObject
 222         exists, err := bucket.Exists(prefix + "missingfile")
 223         c.Check(err, check.IsNil)
 224         c.Check(exists, check.Equals, false)
 225
 226         // GetObject
 227         rdr, err = bucket.GetReader(prefix + "sailboat.txt")
 228         c.Assert(err, check.IsNil)
 229         buf, err = ioutil.ReadAll(rdr)
 230         c.Check(err, check.IsNil)
 231         c.Check(buf, check.DeepEquals, []byte("⛵\n"))
 232         err = rdr.Close()
 233         c.Check(err, check.IsNil)
 234
 235         // HeadObject
 236         resp, err := bucket.Head(prefix+"sailboat.txt", nil)
 237         c.Check(err, check.IsNil)
 238         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 239         c.Check(resp.ContentLength, check.Equals, int64(4))
 240
 241         // HeadObject with superfluous leading slashes
 242         exists, err = bucket.Exists(prefix + "//sailboat.txt")
 243         c.Check(err, check.IsNil)
 244         c.Check(exists, check.Equals, true)
 245 }
 246
 247 func (s *IntegrationSuite) checkMetaEquals(c *check.C, hdr http.Header, expect map[string]string) {
 248         got := map[string]string{}
 249         for hk, hv := range hdr {
 250                 if k := strings.TrimPrefix(hk, "X-Amz-Meta-"); k != hk && len(hv) == 1 {
 251                         got[k] = hv[0]
 252                 }
 253         }
 254         c.Check(got, check.DeepEquals, expect)
 255 }
 256
 257 func (s *IntegrationSuite) TestS3PropertiesAsMetadata(c *check.C) {
 258         stage := s.s3setup(c)
 259         defer stage.teardown(c)
 260
 261         expectCollectionTags := map[string]string{
 262                 "String":   "string value",
 263                 "Array":    `["element1","element2"]`,
 264                 "Object":   mime.BEncoding.Encode("UTF-8", `{"key":{"key2":"value⛵"}}`),
 265                 "Nonascii": "=?UTF-8?b?4pu1?=",
 266                 "Newline":  mime.BEncoding.Encode("UTF-8", "foo\r\nX-Bad: header"),
 267         }
 268         expectSubprojectTags := map[string]string{
 269                 "Subproject_properties_key": "subproject properties value",
 270         }
 271         expectProjectTags := map[string]string{
 272                 "Project-Properties-Key": "project properties value",
 273         }
 274
 275         c.Log("HEAD object with metadata from collection")
 276         resp, err := stage.collbucket.Head("sailboat.txt", nil)
 277         c.Assert(err, check.IsNil)
 278         s.checkMetaEquals(c, resp.Header, expectCollectionTags)
 279
 280         c.Log("GET object with metadata from collection")
 281         rdr, hdr, err := stage.collbucket.GetReaderWithHeaders("sailboat.txt")
 282         c.Assert(err, check.IsNil)
 283         content, err := ioutil.ReadAll(rdr)
 284         c.Check(err, check.IsNil)
 285         rdr.Close()
 286         c.Check(content, check.HasLen, 4)
 287         s.checkMetaEquals(c, hdr, expectCollectionTags)
 288         c.Check(hdr["Inject"], check.IsNil)
 289
 290         c.Log("HEAD bucket with metadata from collection")
 291         resp, err = stage.collbucket.Head("/", nil)
 292         c.Assert(err, check.IsNil)
 293         s.checkMetaEquals(c, resp.Header, expectCollectionTags)
 294
 295         c.Log("HEAD directory placeholder with metadata from collection")
 296         resp, err = stage.projbucket.Head("keep-web s3 test collection/", nil)
 297         c.Assert(err, check.IsNil)
 298         s.checkMetaEquals(c, resp.Header, expectCollectionTags)
 299
 300         c.Log("HEAD file with metadata from collection")
 301         resp, err = stage.projbucket.Head("keep-web s3 test collection/sailboat.txt", nil)
 302         c.Assert(err, check.IsNil)
 303         s.checkMetaEquals(c, resp.Header, expectCollectionTags)
 304
 305         c.Log("HEAD directory placeholder with metadata from subproject")
 306         resp, err = stage.projbucket.Head("keep-web s3 test subproject/", nil)
 307         c.Assert(err, check.IsNil)
 308         s.checkMetaEquals(c, resp.Header, expectSubprojectTags)
 309
 310         c.Log("HEAD bucket with metadata from project")
 311         resp, err = stage.projbucket.Head("/", nil)
 312         c.Assert(err, check.IsNil)
 313         s.checkMetaEquals(c, resp.Header, expectProjectTags)
 314 }
 315
 316 func (s *IntegrationSuite) TestS3CollectionPutObjectSuccess(c *check.C) {
 317         stage := s.s3setup(c)
 318         defer stage.teardown(c)
 319         s.testS3PutObjectSuccess(c, stage.collbucket, "")
 320 }
 321 func (s *IntegrationSuite) TestS3ProjectPutObjectSuccess(c *check.C) {
 322         stage := s.s3setup(c)
 323         defer stage.teardown(c)
 324         s.testS3PutObjectSuccess(c, stage.projbucket, stage.coll.Name+"/")
 325 }
 326 func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket, prefix string) {
 327         for _, trial := range []struct {
 328                 path        string
 329                 size        int
 330                 contentType string
 331         }{
 332                 {
 333                         path:        "newfile",
 334                         size:        128000000,
 335                         contentType: "application/octet-stream",
 336                 }, {
 337                         path:        "newdir/newfile",
 338                         size:        1 << 26,
 339                         contentType: "application/octet-stream",
 340                 }, {
 341                         path:        "/aaa",
 342                         size:        2,
 343                         contentType: "application/octet-stream",
 344                 }, {
 345                         path:        "//bbb",
 346                         size:        2,
 347                         contentType: "application/octet-stream",
 348                 }, {
 349                         path:        "ccc//",
 350                         size:        0,
 351                         contentType: "application/x-directory",
 352                 }, {
 353                         path:        "newdir1/newdir2/newfile",
 354                         size:        0,
 355                         contentType: "application/octet-stream",
 356                 }, {
 357                         path:        "newdir1/newdir2/newdir3/",
 358                         size:        0,
 359                         contentType: "application/x-directory",
 360                 },
 361         } {
 362                 c.Logf("=== %v", trial)
 363
 364                 objname := prefix + trial.path
 365
 366                 _, err := bucket.GetReader(objname)
 367                 if !c.Check(err, check.NotNil) {
 368                         continue
 369                 }
 370                 c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
 371                 c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
 372                 if !c.Check(err, check.ErrorMatches, `The specified key does not exist.`) {
 373                         continue
 374                 }
 375
 376                 buf := make([]byte, trial.size)
 377                 rand.Read(buf)
 378
 379                 err = bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), trial.contentType, s3.Private, s3.Options{})
 380                 c.Check(err, check.IsNil)
 381
 382                 rdr, err := bucket.GetReader(objname)
 383                 if strings.HasSuffix(trial.path, "/") && !s.handler.Cluster.Collections.S3FolderObjects {
 384                         c.Check(err, check.NotNil)
 385                         continue
 386                 } else if !c.Check(err, check.IsNil) {
 387                         continue
 388                 }
 389                 buf2, err := ioutil.ReadAll(rdr)
 390                 c.Check(err, check.IsNil)
 391                 c.Check(buf2, check.HasLen, len(buf))
 392                 c.Check(bytes.Equal(buf, buf2), check.Equals, true)
 393         }
 394 }
 395
 396 func (s *IntegrationSuite) TestS3ProjectPutObjectNotSupported(c *check.C) {
 397         stage := s.s3setup(c)
 398         defer stage.teardown(c)
 399         bucket := stage.projbucket
 400
 401         for _, trial := range []struct {
 402                 path        string
 403                 size        int
 404                 contentType string
 405         }{
 406                 {
 407                         path:        "newfile",
 408                         size:        1234,
 409                         contentType: "application/octet-stream",
 410                 }, {
 411                         path:        "newdir/newfile",
 412                         size:        1234,
 413                         contentType: "application/octet-stream",
 414                 }, {
 415                         path:        "newdir2/",
 416                         size:        0,
 417                         contentType: "application/x-directory",
 418                 },
 419         } {
 420                 c.Logf("=== %v", trial)
 421
 422                 _, err := bucket.GetReader(trial.path)
 423                 c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
 424                 c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
 425                 c.Assert(err, check.ErrorMatches, `The specified key does not exist.`)
 426
 427                 buf := make([]byte, trial.size)
 428                 rand.Read(buf)
 429
 430                 err = bucket.PutReader(trial.path, bytes.NewReader(buf), int64(len(buf)), trial.contentType, s3.Private, s3.Options{})
 431                 c.Check(err.(*s3.Error).StatusCode, check.Equals, 400)
 432                 c.Check(err.(*s3.Error).Code, check.Equals, `InvalidArgument`)
 433                 c.Check(err, check.ErrorMatches, `(mkdir "/by_id/zzzzz-j7d0g-[a-z0-9]{15}/newdir2?"|open "/zzzzz-j7d0g-[a-z0-9]{15}/newfile") failed: invalid (argument|operation)`)
 434
 435                 _, err = bucket.GetReader(trial.path)
 436                 c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
 437                 c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
 438                 c.Assert(err, check.ErrorMatches, `The specified key does not exist.`)
 439         }
 440 }
 441
 442 func (s *IntegrationSuite) TestS3CollectionDeleteObject(c *check.C) {
 443         stage := s.s3setup(c)
 444         defer stage.teardown(c)
 445         s.testS3DeleteObject(c, stage.collbucket, "")
 446 }
 447 func (s *IntegrationSuite) TestS3ProjectDeleteObject(c *check.C) {
 448         stage := s.s3setup(c)
 449         defer stage.teardown(c)
 450         s.testS3DeleteObject(c, stage.projbucket, stage.coll.Name+"/")
 451 }
 452 func (s *IntegrationSuite) testS3DeleteObject(c *check.C, bucket *s3.Bucket, prefix string) {
 453         s.handler.Cluster.Collections.S3FolderObjects = true
 454         for _, trial := range []struct {
 455                 path string
 456         }{
 457                 {"/"},
 458                 {"nonexistentfile"},
 459                 {"emptyfile"},
 460                 {"sailboat.txt"},
 461                 {"sailboat.txt/"},
 462                 {"emptydir"},
 463                 {"emptydir/"},
 464         } {
 465                 objname := prefix + trial.path
 466                 comment := check.Commentf("objname %q", objname)
 467
 468                 err := bucket.Del(objname)
 469                 if trial.path == "/" {
 470                         c.Check(err, check.NotNil)
 471                         continue
 472                 }
 473                 c.Check(err, check.IsNil, comment)
 474                 _, err = bucket.GetReader(objname)
 475                 c.Check(err, check.NotNil, comment)
 476         }
 477 }
 478
 479 func (s *IntegrationSuite) TestS3CollectionPutObjectFailure(c *check.C) {
 480         stage := s.s3setup(c)
 481         defer stage.teardown(c)
 482         s.testS3PutObjectFailure(c, stage.collbucket, "")
 483 }
 484 func (s *IntegrationSuite) TestS3ProjectPutObjectFailure(c *check.C) {
 485         stage := s.s3setup(c)
 486         defer stage.teardown(c)
 487         s.testS3PutObjectFailure(c, stage.projbucket, stage.coll.Name+"/")
 488 }
 489 func (s *IntegrationSuite) testS3PutObjectFailure(c *check.C, bucket *s3.Bucket, prefix string) {
 490         s.handler.Cluster.Collections.S3FolderObjects = false
 491
 492         var wg sync.WaitGroup
 493         for _, trial := range []struct {
 494                 path string
 495         }{
 496                 {
 497                         path: "emptyfile/newname", // emptyfile exists, see s3setup()
 498                 }, {
 499                         path: "emptyfile/", // emptyfile exists, see s3setup()
 500                 }, {
 501                         path: "emptydir", // dir already exists, see s3setup()
 502                 }, {
 503                         path: "emptydir/",
 504                 }, {
 505                         path: "emptydir//",
 506                 }, {
 507                         path: "newdir/",
 508                 }, {
 509                         path: "newdir//",
 510                 }, {
 511                         path: "/",
 512                 }, {
 513                         path: "//",
 514                 }, {
 515                         path: "",
 516                 },
 517         } {
 518                 trial := trial
 519                 wg.Add(1)
 520                 go func() {
 521                         defer wg.Done()
 522                         c.Logf("=== %v", trial)
 523
 524                         objname := prefix + trial.path
 525
 526                         buf := make([]byte, 1234)
 527                         rand.Read(buf)
 528
 529                         err := bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), "application/octet-stream", s3.Private, s3.Options{})
 530                         if !c.Check(err, check.ErrorMatches, `(invalid object name.*|open ".*" failed.*|object name conflicts with existing object|Missing object name in PUT request.)`, check.Commentf("PUT %q should fail", objname)) {
 531                                 return
 532                         }
 533
 534                         if objname != "" && objname != "/" {
 535                                 _, err = bucket.GetReader(objname)
 536                                 c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
 537                                 c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
 538                                 c.Check(err, check.ErrorMatches, `The specified key does not exist.`, check.Commentf("GET %q should return 404", objname))
 539                         }
 540                 }()
 541         }
 542         wg.Wait()
 543 }
 544
 545 func (stage *s3stage) writeBigDirs(c *check.C, dirs int, filesPerDir int) {
 546         fs, err := stage.coll.FileSystem(stage.arv, stage.kc)
 547         c.Assert(err, check.IsNil)
 548         for d := 0; d < dirs; d++ {
 549                 dir := fmt.Sprintf("dir%d", d)
 550                 c.Assert(fs.Mkdir(dir, 0755), check.IsNil)
 551                 for i := 0; i < filesPerDir; i++ {
 552                         f, err := fs.OpenFile(fmt.Sprintf("%s/file%d.txt", dir, i), os.O_CREATE|os.O_WRONLY, 0644)
 553                         c.Assert(err, check.IsNil)
 554                         c.Assert(f.Close(), check.IsNil)
 555                 }
 556         }
 557         c.Assert(fs.Sync(), check.IsNil)
 558 }
 559
 560 func (s *IntegrationSuite) sign(c *check.C, req *http.Request, key, secret string) {
 561         scope := "20200202/zzzzz/service/aws4_request"
 562         signedHeaders := "date"
 563         req.Header.Set("Date", time.Now().UTC().Format(time.RFC1123))
 564         stringToSign, err := s3stringToSign(s3SignAlgorithm, scope, signedHeaders, req)
 565         c.Assert(err, check.IsNil)
 566         sig, err := s3signature(secret, scope, signedHeaders, stringToSign)
 567         c.Assert(err, check.IsNil)
 568         req.Header.Set("Authorization", s3SignAlgorithm+" Credential="+key+"/"+scope+", SignedHeaders="+signedHeaders+", Signature="+sig)
 569 }
 570
 571 func (s *IntegrationSuite) TestS3VirtualHostStyleRequests(c *check.C) {
 572         stage := s.s3setup(c)
 573         defer stage.teardown(c)
 574         for _, trial := range []struct {
 575                 url            string
 576                 method         string
 577                 body           string
 578                 responseCode   int
 579                 responseRegexp []string
 580         }{
 581                 {
 582                         url:            "https://" + stage.collbucket.Name + ".example.com/",
 583                         method:         "GET",
 584                         responseCode:   http.StatusOK,
 585                         responseRegexp: []string{`(?ms).*sailboat\.txt.*`},
 586                 },
 587                 {
 588                         url:            "https://" + strings.Replace(stage.coll.PortableDataHash, "+", "-", -1) + ".example.com/",
 589                         method:         "GET",
 590                         responseCode:   http.StatusOK,
 591                         responseRegexp: []string{`(?ms).*sailboat\.txt.*`},
 592                 },
 593                 {
 594                         url:            "https://" + stage.projbucket.Name + ".example.com/?prefix=" + stage.coll.Name + "/&delimiter=/",
 595                         method:         "GET",
 596                         responseCode:   http.StatusOK,
 597                         responseRegexp: []string{`(?ms).*sailboat\.txt.*`},
 598                 },
 599                 {
 600                         url:            "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "/sailboat.txt",
 601                         method:         "GET",
 602                         responseCode:   http.StatusOK,
 603                         responseRegexp: []string{`⛵\n`},
 604                 },
 605                 {
 606                         url:          "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "/beep",
 607                         method:       "PUT",
 608                         body:         "boop",
 609                         responseCode: http.StatusOK,
 610                 },
 611                 {
 612                         url:            "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "/beep",
 613                         method:         "GET",
 614                         responseCode:   http.StatusOK,
 615                         responseRegexp: []string{`boop`},
 616                 },
 617                 {
 618                         url:          "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "//boop",
 619                         method:       "GET",
 620                         responseCode: http.StatusNotFound,
 621                 },
 622                 {
 623                         url:          "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "//boop",
 624                         method:       "PUT",
 625                         body:         "boop",
 626                         responseCode: http.StatusOK,
 627                 },
 628                 {
 629                         url:            "https://" + stage.projbucket.Name + ".example.com/" + stage.coll.Name + "//boop",
 630                         method:         "GET",
 631                         responseCode:   http.StatusOK,
 632                         responseRegexp: []string{`boop`},
 633                 },
 634         } {
 635                 url, err := url.Parse(trial.url)
 636                 c.Assert(err, check.IsNil)
 637                 req, err := http.NewRequest(trial.method, url.String(), bytes.NewReader([]byte(trial.body)))
 638                 c.Assert(err, check.IsNil)
 639                 s.sign(c, req, arvadostest.ActiveTokenUUID, arvadostest.ActiveToken)
 640                 rr := httptest.NewRecorder()
 641                 s.handler.ServeHTTP(rr, req)
 642                 resp := rr.Result()
 643                 c.Check(resp.StatusCode, check.Equals, trial.responseCode)
 644                 body, err := ioutil.ReadAll(resp.Body)
 645                 c.Assert(err, check.IsNil)
 646                 for _, re := range trial.responseRegexp {
 647                         c.Check(string(body), check.Matches, re)
 648                 }
 649         }
 650 }
 651
 652 func (s *IntegrationSuite) TestS3NormalizeURIForSignature(c *check.C) {
 653         stage := s.s3setup(c)
 654         defer stage.teardown(c)
 655         for _, trial := range []struct {
 656                 rawPath        string
 657                 normalizedPath string
 658         }{
 659                 {"/foo", "/foo"},                           // boring case
 660                 {"/foo%5fbar", "/foo_bar"},                 // _ must not be escaped
 661                 {"/foo%2fbar", "/foo/bar"},                 // / must not be escaped
 662                 {"/(foo)/[];,", "/%28foo%29/%5B%5D%3B%2C"}, // ()[];, must be escaped
 663                 {"/foo%5bbar", "/foo%5Bbar"},               // %XX must be uppercase
 664                 {"//foo///.bar", "/foo/.bar"},              // "//" and "///" must be squashed to "/"
 665         } {
 666                 c.Logf("trial %q", trial)
 667
 668                 date := time.Now().UTC().Format("20060102T150405Z")
 669                 scope := "20200202/zzzzz/S3/aws4_request"
 670                 canonicalRequest := fmt.Sprintf("%s\n%s\n%s\n%s\n%s\n%s", "GET", trial.normalizedPath, "", "host:host.example.com\n", "host", "")
 671                 c.Logf("canonicalRequest %q", canonicalRequest)
 672                 expect := fmt.Sprintf("%s\n%s\n%s\n%s", s3SignAlgorithm, date, scope, hashdigest(sha256.New(), canonicalRequest))
 673                 c.Logf("expected stringToSign %q", expect)
 674
 675                 req, err := http.NewRequest("GET", "https://host.example.com"+trial.rawPath, nil)
 676                 req.Header.Set("X-Amz-Date", date)
 677                 req.Host = "host.example.com"
 678                 c.Assert(err, check.IsNil)
 679
 680                 obtained, err := s3stringToSign(s3SignAlgorithm, scope, "host", req)
 681                 if !c.Check(err, check.IsNil) {
 682                         continue
 683                 }
 684                 c.Check(obtained, check.Equals, expect)
 685         }
 686 }
 687
 688 func (s *IntegrationSuite) TestS3GetBucketLocation(c *check.C) {
 689         stage := s.s3setup(c)
 690         defer stage.teardown(c)
 691         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 692                 req, err := http.NewRequest("GET", bucket.URL("/"), nil)
 693                 c.Check(err, check.IsNil)
 694                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 695                 req.URL.RawQuery = "location"
 696                 resp, err := http.DefaultClient.Do(req)
 697                 c.Assert(err, check.IsNil)
 698                 c.Check(resp.Header.Get("Content-Type"), check.Equals, "application/xml")
 699                 buf, err := ioutil.ReadAll(resp.Body)
 700                 c.Assert(err, check.IsNil)
 701                 c.Check(string(buf), check.Equals, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<LocationConstraint><LocationConstraint xmlns=\"http://s3.amazonaws.com/doc/2006-03-01/\">zzzzz</LocationConstraint></LocationConstraint>\n")
 702         }
 703 }
 704
 705 func (s *IntegrationSuite) TestS3GetBucketVersioning(c *check.C) {
 706         stage := s.s3setup(c)
 707         defer stage.teardown(c)
 708         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 709                 req, err := http.NewRequest("GET", bucket.URL("/"), nil)
 710                 c.Check(err, check.IsNil)
 711                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 712                 req.URL.RawQuery = "versioning"
 713                 resp, err := http.DefaultClient.Do(req)
 714                 c.Assert(err, check.IsNil)
 715                 c.Check(resp.Header.Get("Content-Type"), check.Equals, "application/xml")
 716                 buf, err := ioutil.ReadAll(resp.Body)
 717                 c.Assert(err, check.IsNil)
 718                 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")
 719         }
 720 }
 721
 722 func (s *IntegrationSuite) TestS3UnsupportedAPIs(c *check.C) {
 723         stage := s.s3setup(c)
 724         defer stage.teardown(c)
 725         for _, trial := range []struct {
 726                 method   string
 727                 path     string
 728                 rawquery string
 729         }{
 730                 {"GET", "/", "acl&versionId=1234"},    // GetBucketAcl
 731                 {"GET", "/foo", "acl&versionId=1234"}, // GetObjectAcl
 732                 {"PUT", "/", "acl"},                   // PutBucketAcl
 733                 {"PUT", "/foo", "acl"},                // PutObjectAcl
 734                 {"DELETE", "/", "tagging"},            // DeleteBucketTagging
 735                 {"DELETE", "/foo", "tagging"},         // DeleteObjectTagging
 736         } {
 737                 for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
 738                         c.Logf("trial %v bucket %v", trial, bucket)
 739                         req, err := http.NewRequest(trial.method, bucket.URL(trial.path), nil)
 740                         c.Check(err, check.IsNil)
 741                         req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 742                         req.URL.RawQuery = trial.rawquery
 743                         resp, err := http.DefaultClient.Do(req)
 744                         c.Assert(err, check.IsNil)
 745                         c.Check(resp.Header.Get("Content-Type"), check.Equals, "application/xml")
 746                         buf, err := ioutil.ReadAll(resp.Body)
 747                         c.Assert(err, check.IsNil)
 748                         c.Check(string(buf), check.Matches, "(?ms).*InvalidRequest.*API not supported.*")
 749                 }
 750         }
 751 }
 752
 753 // If there are no CommonPrefixes entries, the CommonPrefixes XML tag
 754 // should not appear at all.
 755 func (s *IntegrationSuite) TestS3ListNoCommonPrefixes(c *check.C) {
 756         stage := s.s3setup(c)
 757         defer stage.teardown(c)
 758
 759         req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 760         c.Assert(err, check.IsNil)
 761         req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 762         req.URL.RawQuery = "prefix=asdfasdfasdf&delimiter=/"
 763         resp, err := http.DefaultClient.Do(req)
 764         c.Assert(err, check.IsNil)
 765         buf, err := ioutil.ReadAll(resp.Body)
 766         c.Assert(err, check.IsNil)
 767         c.Check(string(buf), check.Not(check.Matches), `(?ms).*CommonPrefixes.*`)
 768 }
 769
 770 // If there is no delimiter in the request, or the results are not
 771 // truncated, the NextMarker XML tag should not appear in the response
 772 // body.
 773 func (s *IntegrationSuite) TestS3ListNoNextMarker(c *check.C) {
 774         stage := s.s3setup(c)
 775         defer stage.teardown(c)
 776
 777         for _, query := range []string{"prefix=e&delimiter=/", ""} {
 778                 req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 779                 c.Assert(err, check.IsNil)
 780                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 781                 req.URL.RawQuery = query
 782                 resp, err := http.DefaultClient.Do(req)
 783                 c.Assert(err, check.IsNil)
 784                 buf, err := ioutil.ReadAll(resp.Body)
 785                 c.Assert(err, check.IsNil)
 786                 c.Check(string(buf), check.Not(check.Matches), `(?ms).*NextMarker.*`)
 787         }
 788 }
 789
 790 // List response should include KeyCount field.
 791 func (s *IntegrationSuite) TestS3ListKeyCount(c *check.C) {
 792         stage := s.s3setup(c)
 793         defer stage.teardown(c)
 794
 795         req, err := http.NewRequest("GET", stage.collbucket.URL("/"), nil)
 796         c.Assert(err, check.IsNil)
 797         req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
 798         req.URL.RawQuery = "prefix=&delimiter=/"
 799         resp, err := http.DefaultClient.Do(req)
 800         c.Assert(err, check.IsNil)
 801         buf, err := ioutil.ReadAll(resp.Body)
 802         c.Assert(err, check.IsNil)
 803         c.Check(string(buf), check.Matches, `(?ms).*<KeyCount>2</KeyCount>.*`)
 804 }
 805
 806 func (s *IntegrationSuite) TestS3CollectionList(c *check.C) {
 807         stage := s.s3setup(c)
 808         defer stage.teardown(c)
 809
 810         var markers int
 811         for markers, s.handler.Cluster.Collections.S3FolderObjects = range []bool{false, true} {
 812                 dirs := 2
 813                 filesPerDir := 1001
 814                 stage.writeBigDirs(c, dirs, filesPerDir)
 815                 // Total # objects is:
 816                 //                 2 file entries from s3setup (emptyfile and sailboat.txt)
 817                 //                +1 fake "directory" marker from s3setup (emptydir) (if enabled)
 818                 //             +dirs fake "directory" marker from writeBigDirs (dir0/, dir1/) (if enabled)
 819                 // +filesPerDir*dirs file entries from writeBigDirs (dir0/file0.txt, etc.)
 820                 s.testS3List(c, stage.collbucket, "", 4000, markers+2+(filesPerDir+markers)*dirs)
 821                 s.testS3List(c, stage.collbucket, "", 131, markers+2+(filesPerDir+markers)*dirs)
 822                 s.testS3List(c, stage.collbucket, "dir0/", 71, filesPerDir+markers)
 823         }
 824 }
 825 func (s *IntegrationSuite) testS3List(c *check.C, bucket *s3.Bucket, prefix string, pageSize, expectFiles int) {
 826         c.Logf("testS3List: prefix=%q pageSize=%d S3FolderObjects=%v", prefix, pageSize, s.handler.Cluster.Collections.S3FolderObjects)
 827         expectPageSize := pageSize
 828         if expectPageSize > 1000 {
 829                 expectPageSize = 1000
 830         }
 831         gotKeys := map[string]s3.Key{}
 832         nextMarker := ""
 833         pages := 0
 834         for {
 835                 resp, err := bucket.List(prefix, "", nextMarker, pageSize)
 836                 if !c.Check(err, check.IsNil) {
 837                         break
 838                 }
 839                 c.Check(len(resp.Contents) <= expectPageSize, check.Equals, true)
 840                 if pages++; !c.Check(pages <= (expectFiles/expectPageSize)+1, check.Equals, true) {
 841                         break
 842                 }
 843                 for _, key := range resp.Contents {
 844                         gotKeys[key.Key] = key
 845                         if strings.Contains(key.Key, "sailboat.txt") {
 846                                 c.Check(key.Size, check.Equals, int64(4))
 847                         }
 848                 }
 849                 if !resp.IsTruncated {
 850                         c.Check(resp.NextMarker, check.Equals, "")
 851                         break
 852                 }
 853                 if !c.Check(resp.NextMarker, check.Not(check.Equals), "") {
 854                         break
 855                 }
 856                 nextMarker = resp.NextMarker
 857         }
 858         c.Check(len(gotKeys), check.Equals, expectFiles)
 859 }
 860
 861 func (s *IntegrationSuite) TestS3CollectionListRollup(c *check.C) {
 862         for _, s.handler.Cluster.Collections.S3FolderObjects = range []bool{false, true} {
 863                 s.testS3CollectionListRollup(c)
 864         }
 865 }
 866
 867 func (s *IntegrationSuite) testS3CollectionListRollup(c *check.C) {
 868         stage := s.s3setup(c)
 869         defer stage.teardown(c)
 870
 871         dirs := 2
 872         filesPerDir := 500
 873         stage.writeBigDirs(c, dirs, filesPerDir)
 874         err := stage.collbucket.PutReader("dingbats", &bytes.Buffer{}, 0, "application/octet-stream", s3.Private, s3.Options{})
 875         c.Assert(err, check.IsNil)
 876         var allfiles []string
 877         for marker := ""; ; {
 878                 resp, err := stage.collbucket.List("", "", marker, 20000)
 879                 c.Check(err, check.IsNil)
 880                 for _, key := range resp.Contents {
 881                         if len(allfiles) == 0 || allfiles[len(allfiles)-1] != key.Key {
 882                                 allfiles = append(allfiles, key.Key)
 883                         }
 884                 }
 885                 marker = resp.NextMarker
 886                 if marker == "" {
 887                         break
 888                 }
 889         }
 890         markers := 0
 891         if s.handler.Cluster.Collections.S3FolderObjects {
 892                 markers = 1
 893         }
 894         c.Check(allfiles, check.HasLen, dirs*(filesPerDir+markers)+3+markers)
 895
 896         gotDirMarker := map[string]bool{}
 897         for _, name := range allfiles {
 898                 isDirMarker := strings.HasSuffix(name, "/")
 899                 if markers == 0 {
 900                         c.Check(isDirMarker, check.Equals, false, check.Commentf("name %q", name))
 901                 } else if isDirMarker {
 902                         gotDirMarker[name] = true
 903                 } else if i := strings.LastIndex(name, "/"); i >= 0 {
 904                         c.Check(gotDirMarker[name[:i+1]], check.Equals, true, check.Commentf("name %q", name))
 905                         gotDirMarker[name[:i+1]] = true // skip redundant complaints about this dir marker
 906                 }
 907         }
 908
 909         for _, trial := range []struct {
 910                 prefix    string
 911                 delimiter string
 912                 marker    string
 913         }{
 914                 {"", "", ""},
 915                 {"di", "/", ""},
 916                 {"di", "r", ""},
 917                 {"di", "n", ""},
 918                 {"dir0", "/", ""},
 919                 {"dir0/", "/", ""},
 920                 {"dir0/f", "/", ""},
 921                 {"dir0", "", ""},
 922                 {"dir0/", "", ""},
 923                 {"dir0/f", "", ""},
 924                 {"dir0", "/", "dir0/file14.txt"},       // no commonprefixes
 925                 {"", "", "dir0/file14.txt"},            // middle page, skip walking dir1
 926                 {"", "", "dir1/file14.txt"},            // middle page, skip walking dir0
 927                 {"", "", "dir1/file498.txt"},           // last page of results
 928                 {"dir1/file", "", "dir1/file498.txt"},  // last page of results, with prefix
 929                 {"dir1/file", "/", "dir1/file498.txt"}, // last page of results, with prefix + delimiter
 930                 {"dir1", "Z", "dir1/file498.txt"},      // delimiter "Z" never appears
 931                 {"dir2", "/", ""},                      // prefix "dir2" does not exist
 932                 {"", "/", ""},
 933         } {
 934                 c.Logf("\n\n=== trial %+v markers=%d", trial, markers)
 935
 936                 maxKeys := 20
 937                 resp, err := stage.collbucket.List(trial.prefix, trial.delimiter, trial.marker, maxKeys)
 938                 c.Check(err, check.IsNil)
 939                 if resp.IsTruncated && trial.delimiter == "" {
 940                         // goamz List method fills in the missing
 941                         // NextMarker field if resp.IsTruncated, so
 942                         // now we can't really tell whether it was
 943                         // sent by the server or by goamz. In cases
 944                         // where it should be empty but isn't, assume
 945                         // it's goamz's fault.
 946                         resp.NextMarker = ""
 947                 }
 948
 949                 var expectKeys []string
 950                 var expectPrefixes []string
 951                 var expectNextMarker string
 952                 var expectTruncated bool
 953                 for _, key := range allfiles {
 954                         full := len(expectKeys)+len(expectPrefixes) >= maxKeys
 955                         if !strings.HasPrefix(key, trial.prefix) || key < trial.marker {
 956                                 continue
 957                         } else if idx := strings.Index(key[len(trial.prefix):], trial.delimiter); trial.delimiter != "" && idx >= 0 {
 958                                 prefix := key[:len(trial.prefix)+idx+1]
 959                                 if len(expectPrefixes) > 0 && expectPrefixes[len(expectPrefixes)-1] == prefix {
 960                                         // same prefix as previous key
 961                                 } else if full {
 962                                         expectNextMarker = key
 963                                         expectTruncated = true
 964                                 } else {
 965                                         expectPrefixes = append(expectPrefixes, prefix)
 966                                 }
 967                         } else if full {
 968                                 if trial.delimiter != "" {
 969                                         expectNextMarker = key
 970                                 }
 971                                 expectTruncated = true
 972                                 break
 973                         } else {
 974                                 expectKeys = append(expectKeys, key)
 975                         }
 976                 }
 977
 978                 var gotKeys []string
 979                 for _, key := range resp.Contents {
 980                         gotKeys = append(gotKeys, key.Key)
 981                 }
 982                 var gotPrefixes []string
 983                 for _, prefix := range resp.CommonPrefixes {
 984                         gotPrefixes = append(gotPrefixes, prefix)
 985                 }
 986                 commentf := check.Commentf("trial %+v markers=%d", trial, markers)
 987                 c.Check(gotKeys, check.DeepEquals, expectKeys, commentf)
 988                 c.Check(gotPrefixes, check.DeepEquals, expectPrefixes, commentf)
 989                 c.Check(resp.NextMarker, check.Equals, expectNextMarker, commentf)
 990                 c.Check(resp.IsTruncated, check.Equals, expectTruncated, commentf)
 991                 c.Logf("=== trial %+v keys %q prefixes %q nextMarker %q", trial, gotKeys, gotPrefixes, resp.NextMarker)
 992         }
 993 }
 994
 995 func (s *IntegrationSuite) TestS3ListObjectsV2(c *check.C) {
 996         stage := s.s3setup(c)
 997         defer stage.teardown(c)
 998         dirs := 2
 999         filesPerDir := 40
1000         stage.writeBigDirs(c, dirs, filesPerDir)
1001
1002         sess := aws_session.Must(aws_session.NewSession(&aws_aws.Config{
1003                 Region:           aws_aws.String("auto"),
1004                 Endpoint:         aws_aws.String(s.testServer.URL),
1005                 Credentials:      aws_credentials.NewStaticCredentials(url.QueryEscape(arvadostest.ActiveTokenV2), url.QueryEscape(arvadostest.ActiveTokenV2), ""),
1006                 S3ForcePathStyle: aws_aws.Bool(true),
1007         }))
1008
1009         stringOrNil := func(s string) *string {
1010                 if s == "" {
1011                         return nil
1012                 } else {
1013                         return &s
1014                 }
1015         }
1016
1017         client := aws_s3.New(sess)
1018         ctx := context.Background()
1019
1020         for _, trial := range []struct {
1021                 prefix               string
1022                 delimiter            string
1023                 startAfter           string
1024                 maxKeys              int
1025                 expectKeys           int
1026                 expectCommonPrefixes map[string]bool
1027         }{
1028                 {
1029                         // Expect {filesPerDir plus the dir itself}
1030                         // for each dir, plus emptydir, emptyfile, and
1031                         // sailboat.txt.
1032                         expectKeys: (filesPerDir+1)*dirs + 3,
1033                 },
1034                 {
1035                         maxKeys:    15,
1036                         expectKeys: (filesPerDir+1)*dirs + 3,
1037                 },
1038                 {
1039                         startAfter: "dir0/z",
1040                         maxKeys:    15,
1041                         // Expect {filesPerDir plus the dir itself}
1042                         // for each dir except dir0, plus emptydir,
1043                         // emptyfile, and sailboat.txt.
1044                         expectKeys: (filesPerDir+1)*(dirs-1) + 3,
1045                 },
1046                 {
1047                         maxKeys:              1,
1048                         delimiter:            "/",
1049                         expectKeys:           2, // emptyfile, sailboat.txt
1050                         expectCommonPrefixes: map[string]bool{"dir0/": true, "dir1/": true, "emptydir/": true},
1051                 },
1052                 {
1053                         startAfter:           "dir0/z",
1054                         maxKeys:              15,
1055                         delimiter:            "/",
1056                         expectKeys:           2, // emptyfile, sailboat.txt
1057                         expectCommonPrefixes: map[string]bool{"dir1/": true, "emptydir/": true},
1058                 },
1059                 {
1060                         startAfter:           "dir0/file10.txt",
1061                         maxKeys:              15,
1062                         delimiter:            "/",
1063                         expectKeys:           2,
1064                         expectCommonPrefixes: map[string]bool{"dir0/": true, "dir1/": true, "emptydir/": true},
1065                 },
1066                 {
1067                         startAfter:           "dir0/file10.txt",
1068                         maxKeys:              15,
1069                         prefix:               "d",
1070                         delimiter:            "/",
1071                         expectKeys:           0,
1072                         expectCommonPrefixes: map[string]bool{"dir0/": true, "dir1/": true},
1073                 },
1074         } {
1075                 c.Logf("[trial %+v]", trial)
1076                 params := aws_s3.ListObjectsV2Input{
1077                         Bucket:     aws_aws.String(stage.collbucket.Name),
1078                         Prefix:     stringOrNil(trial.prefix),
1079                         Delimiter:  stringOrNil(trial.delimiter),
1080                         StartAfter: stringOrNil(trial.startAfter),
1081                         MaxKeys:    aws_aws.Int64(int64(trial.maxKeys)),
1082                 }
1083                 keySeen := map[string]bool{}
1084                 prefixSeen := map[string]bool{}
1085                 for {
1086                         result, err := client.ListObjectsV2WithContext(ctx, &params)
1087                         if !c.Check(err, check.IsNil) {
1088                                 break
1089                         }
1090                         c.Check(result.Name, check.DeepEquals, aws_aws.String(stage.collbucket.Name))
1091                         c.Check(result.Prefix, check.DeepEquals, aws_aws.String(trial.prefix))
1092                         c.Check(result.Delimiter, check.DeepEquals, aws_aws.String(trial.delimiter))
1093                         // The following two fields are expected to be
1094                         // nil (i.e., no tag in XML response) rather
1095                         // than "" when the corresponding request
1096                         // field was empty or nil.
1097                         c.Check(result.StartAfter, check.DeepEquals, stringOrNil(trial.startAfter))
1098                         c.Check(result.ContinuationToken, check.DeepEquals, params.ContinuationToken)
1099
1100                         if trial.maxKeys > 0 {
1101                                 c.Check(result.MaxKeys, check.DeepEquals, aws_aws.Int64(int64(trial.maxKeys)))
1102                                 c.Check(len(result.Contents)+len(result.CommonPrefixes) <= trial.maxKeys, check.Equals, true)
1103                         } else {
1104                                 c.Check(result.MaxKeys, check.DeepEquals, aws_aws.Int64(int64(s3MaxKeys)))
1105                         }
1106
1107                         for _, ent := range result.Contents {
1108                                 c.Assert(ent.Key, check.NotNil)
1109                                 c.Check(*ent.Key > trial.startAfter, check.Equals, true)
1110                                 c.Check(keySeen[*ent.Key], check.Equals, false, check.Commentf("dup key %q", *ent.Key))
1111                                 keySeen[*ent.Key] = true
1112                         }
1113                         for _, ent := range result.CommonPrefixes {
1114                                 c.Assert(ent.Prefix, check.NotNil)
1115                                 c.Check(strings.HasSuffix(*ent.Prefix, trial.delimiter), check.Equals, true, check.Commentf("bad CommonPrefix %q", *ent.Prefix))
1116                                 if strings.HasPrefix(trial.startAfter, *ent.Prefix) {
1117                                         // If we asked for
1118                                         // startAfter=dir0/file10.txt,
1119                                         // we expect dir0/ to be
1120                                         // returned as a common prefix
1121                                 } else {
1122                                         c.Check(*ent.Prefix > trial.startAfter, check.Equals, true)
1123                                 }
1124                                 c.Check(prefixSeen[*ent.Prefix], check.Equals, false, check.Commentf("dup common prefix %q", *ent.Prefix))
1125                                 prefixSeen[*ent.Prefix] = true
1126                         }
1127                         if *result.IsTruncated && c.Check(result.NextContinuationToken, check.Not(check.Equals), "") {
1128                                 params.ContinuationToken = aws_aws.String(*result.NextContinuationToken)
1129                         } else {
1130                                 break
1131                         }
1132                 }
1133                 c.Check(keySeen, check.HasLen, trial.expectKeys)
1134                 c.Check(prefixSeen, check.HasLen, len(trial.expectCommonPrefixes))
1135                 if len(trial.expectCommonPrefixes) > 0 {
1136                         c.Check(prefixSeen, check.DeepEquals, trial.expectCommonPrefixes)
1137                 }
1138         }
1139 }
1140
1141 func (s *IntegrationSuite) TestS3ListObjectsV2EncodingTypeURL(c *check.C) {
1142         stage := s.s3setup(c)
1143         defer stage.teardown(c)
1144         dirs := 2
1145         filesPerDir := 40
1146         stage.writeBigDirs(c, dirs, filesPerDir)
1147
1148         sess := aws_session.Must(aws_session.NewSession(&aws_aws.Config{
1149                 Region:           aws_aws.String("auto"),
1150                 Endpoint:         aws_aws.String(s.testServer.URL),
1151                 Credentials:      aws_credentials.NewStaticCredentials(url.QueryEscape(arvadostest.ActiveTokenV2), url.QueryEscape(arvadostest.ActiveTokenV2), ""),
1152                 S3ForcePathStyle: aws_aws.Bool(true),
1153         }))
1154
1155         client := aws_s3.New(sess)
1156         ctx := context.Background()
1157
1158         result, err := client.ListObjectsV2WithContext(ctx, &aws_s3.ListObjectsV2Input{
1159                 Bucket:       aws_aws.String(stage.collbucket.Name),
1160                 Prefix:       aws_aws.String("dir0/"),
1161                 Delimiter:    aws_aws.String("/"),
1162                 StartAfter:   aws_aws.String("dir0/"),
1163                 EncodingType: aws_aws.String("url"),
1164         })
1165         c.Assert(err, check.IsNil)
1166         c.Check(*result.Prefix, check.Equals, "dir0%2F")
1167         c.Check(*result.Delimiter, check.Equals, "%2F")
1168         c.Check(*result.StartAfter, check.Equals, "dir0%2F")
1169         for _, ent := range result.Contents {
1170                 c.Check(*ent.Key, check.Matches, "dir0%2F.*")
1171         }
1172         result, err = client.ListObjectsV2WithContext(ctx, &aws_s3.ListObjectsV2Input{
1173                 Bucket:       aws_aws.String(stage.collbucket.Name),
1174                 Delimiter:    aws_aws.String("/"),
1175                 EncodingType: aws_aws.String("url"),
1176         })
1177         c.Assert(err, check.IsNil)
1178         c.Check(*result.Delimiter, check.Equals, "%2F")
1179         c.Check(result.CommonPrefixes, check.HasLen, dirs+1)
1180         for _, ent := range result.CommonPrefixes {
1181                 c.Check(*ent.Prefix, check.Matches, ".*%2F")
1182         }
1183 }
1184
1185 // TestS3cmd checks compatibility with the s3cmd command line tool, if
1186 // it's installed. As of Debian buster, s3cmd is only in backports, so
1187 // `arvados-server install` don't install it, and this test skips if
1188 // it's not installed.
1189 func (s *IntegrationSuite) TestS3cmd(c *check.C) {
1190         if _, err := exec.LookPath("s3cmd"); err != nil {
1191                 c.Skip("s3cmd not found")
1192                 return
1193         }
1194
1195         stage := s.s3setup(c)
1196         defer stage.teardown(c)
1197
1198         cmd := exec.Command("s3cmd", "--no-ssl", "--host="+s.testServer.URL[7:], "--host-bucket="+s.testServer.URL[7:], "--access_key="+arvadostest.ActiveTokenUUID, "--secret_key="+arvadostest.ActiveToken, "ls", "s3://"+arvadostest.FooCollection)
1199         buf, err := cmd.CombinedOutput()
1200         c.Check(err, check.IsNil)
1201         c.Check(string(buf), check.Matches, `.* 3 +s3://`+arvadostest.FooCollection+`/foo\n`)
1202
1203         // This tests whether s3cmd's path normalization agrees with
1204         // keep-web's signature verification wrt chars like "|"
1205         // (neither reserved nor unreserved) and "," (not normally
1206         // percent-encoded in a path).
1207         tmpfile := c.MkDir() + "/dstfile"
1208         cmd = exec.Command("s3cmd", "--no-ssl", "--host="+s.testServer.URL[7:], "--host-bucket="+s.testServer.URL[7:], "--access_key="+arvadostest.ActiveTokenUUID, "--secret_key="+arvadostest.ActiveToken, "get", "s3://"+arvadostest.FooCollection+"/foo,;$[|]bar", tmpfile)
1209         buf, err = cmd.CombinedOutput()
1210         c.Check(err, check.NotNil)
1211         c.Check(string(buf), check.Matches, `(?ms).*NoSuchKey.*\n`)
1212 }
1213
1214 func (s *IntegrationSuite) TestS3BucketInHost(c *check.C) {
1215         stage := s.s3setup(c)
1216         defer stage.teardown(c)
1217
1218         hdr, body, _ := s.runCurl(c, "AWS "+arvadostest.ActiveTokenV2+":none", stage.coll.UUID+".collections.example.com", "/sailboat.txt")
1219         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
1220         c.Check(body, check.Equals, "⛵\n")
1221 }