20726: Fix traversing projects/collections that precede page marker.
[arvados.git] / services / keep-web / s3_test.go
index 261ebb5741388a87a618d7168936e4292052a6c3..9f8650b8ede8a831df79ada8588be10e979a966a 100644 (file)
@@ -11,11 +11,13 @@ import (
        "crypto/sha256"
        "fmt"
        "io/ioutil"
+       "mime"
        "net/http"
        "net/http/httptest"
        "net/url"
        "os"
        "os/exec"
+       "sort"
        "strings"
        "sync"
        "time"
@@ -39,12 +41,13 @@ type s3stage struct {
        kc         *keepclient.KeepClient
        proj       arvados.Group
        projbucket *s3.Bucket
+       subproj    arvados.Group
        coll       arvados.Collection
        collbucket *s3.Bucket
 }
 
 func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
-       var proj arvados.Group
+       var proj, subproj arvados.Group
        var coll arvados.Collection
        arv := arvados.NewClientFromEnv()
        arv.AuthToken = arvadostest.ActiveToken
@@ -52,14 +55,40 @@ func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
                "group": map[string]interface{}{
                        "group_class": "project",
                        "name":        "keep-web s3 test",
+                       "properties": map[string]interface{}{
+                               "project-properties-key": "project properties value",
+                       },
                },
                "ensure_unique_name": true,
        })
        c.Assert(err, check.IsNil)
+       err = arv.RequestAndDecode(&subproj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
+               "group": map[string]interface{}{
+                       "owner_uuid":  proj.UUID,
+                       "group_class": "project",
+                       "name":        "keep-web s3 test subproject",
+                       "properties": map[string]interface{}{
+                               "subproject_properties_key": "subproject properties value",
+                               "invalid header key":        "this value will not be returned because key contains spaces",
+                       },
+               },
+       })
+       c.Assert(err, check.IsNil)
        err = arv.RequestAndDecode(&coll, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{
                "owner_uuid":    proj.UUID,
                "name":          "keep-web s3 test collection",
                "manifest_text": ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:emptyfile\n./emptydir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n",
+               "properties": map[string]interface{}{
+                       "string":   "string value",
+                       "array":    []string{"element1", "element2"},
+                       "object":   map[string]interface{}{"key": map[string]interface{}{"key2": "value⛵"}},
+                       "nonascii": "⛵",
+                       "newline":  "foo\r\nX-Bad: header",
+                       // This key cannot be expressed as a MIME
+                       // header key, so it will be silently skipped
+                       // (see "Inject" in PropertiesAsMetadata test)
+                       "a: a\r\nInject": "bogus",
+               },
        }})
        c.Assert(err, check.IsNil)
        ac, err := arvadosclient.New(arv)
@@ -95,7 +124,8 @@ func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
                        S3:   client,
                        Name: proj.UUID,
                },
-               coll: coll,
+               subproj: subproj,
+               coll:    coll,
                collbucket: &s3.Bucket{
                        S3:   client,
                        Name: coll.UUID,
@@ -215,17 +245,86 @@ func (s *IntegrationSuite) testS3GetObject(c *check.C, bucket *s3.Bucket, prefix
        c.Check(exists, check.Equals, true)
 }
 
+func (s *IntegrationSuite) checkMetaEquals(c *check.C, hdr http.Header, expect map[string]string) {
+       got := map[string]string{}
+       for hk, hv := range hdr {
+               if k := strings.TrimPrefix(hk, "X-Amz-Meta-"); k != hk && len(hv) == 1 {
+                       got[k] = hv[0]
+               }
+       }
+       c.Check(got, check.DeepEquals, expect)
+}
+
+func (s *IntegrationSuite) TestS3PropertiesAsMetadata(c *check.C) {
+       stage := s.s3setup(c)
+       defer stage.teardown(c)
+
+       expectCollectionTags := map[string]string{
+               "String":   "string value",
+               "Array":    `["element1","element2"]`,
+               "Object":   mime.BEncoding.Encode("UTF-8", `{"key":{"key2":"value⛵"}}`),
+               "Nonascii": "=?UTF-8?b?4pu1?=",
+               "Newline":  mime.BEncoding.Encode("UTF-8", "foo\r\nX-Bad: header"),
+       }
+       expectSubprojectTags := map[string]string{
+               "Subproject_properties_key": "subproject properties value",
+       }
+       expectProjectTags := map[string]string{
+               "Project-Properties-Key": "project properties value",
+       }
+
+       c.Log("HEAD object with metadata from collection")
+       resp, err := stage.collbucket.Head("sailboat.txt", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectCollectionTags)
+
+       c.Log("GET object with metadata from collection")
+       rdr, hdr, err := stage.collbucket.GetReaderWithHeaders("sailboat.txt")
+       c.Assert(err, check.IsNil)
+       content, err := ioutil.ReadAll(rdr)
+       c.Check(err, check.IsNil)
+       rdr.Close()
+       c.Check(content, check.HasLen, 4)
+       s.checkMetaEquals(c, hdr, expectCollectionTags)
+       c.Check(hdr["Inject"], check.IsNil)
+
+       c.Log("HEAD bucket with metadata from collection")
+       resp, err = stage.collbucket.Head("/", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectCollectionTags)
+
+       c.Log("HEAD directory placeholder with metadata from collection")
+       resp, err = stage.projbucket.Head("keep-web s3 test collection/", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectCollectionTags)
+
+       c.Log("HEAD file with metadata from collection")
+       resp, err = stage.projbucket.Head("keep-web s3 test collection/sailboat.txt", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectCollectionTags)
+
+       c.Log("HEAD directory placeholder with metadata from subproject")
+       resp, err = stage.projbucket.Head("keep-web s3 test subproject/", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectSubprojectTags)
+
+       c.Log("HEAD bucket with metadata from project")
+       resp, err = stage.projbucket.Head("/", nil)
+       c.Assert(err, check.IsNil)
+       s.checkMetaEquals(c, resp.Header, expectProjectTags)
+}
+
 func (s *IntegrationSuite) TestS3CollectionPutObjectSuccess(c *check.C) {
        stage := s.s3setup(c)
        defer stage.teardown(c)
-       s.testS3PutObjectSuccess(c, stage.collbucket, "")
+       s.testS3PutObjectSuccess(c, stage.collbucket, "", stage.coll.UUID)
 }
 func (s *IntegrationSuite) TestS3ProjectPutObjectSuccess(c *check.C) {
        stage := s.s3setup(c)
        defer stage.teardown(c)
-       s.testS3PutObjectSuccess(c, stage.projbucket, stage.coll.Name+"/")
+       s.testS3PutObjectSuccess(c, stage.projbucket, stage.coll.Name+"/", stage.coll.UUID)
 }
-func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket, prefix string) {
+func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket, prefix string, collUUID string) {
        for _, trial := range []struct {
                path        string
                size        int
@@ -269,7 +368,7 @@ func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket,
                if !c.Check(err, check.NotNil) {
                        continue
                }
-               c.Check(err.(*s3.Error).StatusCode, check.Equals, 404)
+               c.Check(err.(*s3.Error).StatusCode, check.Equals, http.StatusNotFound)
                c.Check(err.(*s3.Error).Code, check.Equals, `NoSuchKey`)
                if !c.Check(err, check.ErrorMatches, `The specified key does not exist.`) {
                        continue
@@ -292,6 +391,14 @@ func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket,
                c.Check(err, check.IsNil)
                c.Check(buf2, check.HasLen, len(buf))
                c.Check(bytes.Equal(buf, buf2), check.Equals, true)
+
+               // Check that the change is immediately visible via
+               // (non-S3) webdav request.
+               _, resp := s.do("GET", "http://"+collUUID+".keep-web.example/"+trial.path, arvadostest.ActiveTokenV2, nil)
+               c.Check(resp.Code, check.Equals, http.StatusOK)
+               if !strings.HasSuffix(trial.path, "/") {
+                       c.Check(resp.Body.Len(), check.Equals, trial.size)
+               }
        }
 }
 
@@ -711,8 +818,8 @@ func (s *IntegrationSuite) TestS3CollectionList(c *check.C) {
 
        var markers int
        for markers, s.handler.Cluster.Collections.S3FolderObjects = range []bool{false, true} {
-               dirs := 2
-               filesPerDir := 1001
+               dirs := 2000
+               filesPerDir := 2
                stage.writeBigDirs(c, dirs, filesPerDir)
                // Total # objects is:
                //                 2 file entries from s3setup (emptyfile and sailboat.txt)
@@ -721,6 +828,7 @@ func (s *IntegrationSuite) TestS3CollectionList(c *check.C) {
                // +filesPerDir*dirs file entries from writeBigDirs (dir0/file0.txt, etc.)
                s.testS3List(c, stage.collbucket, "", 4000, markers+2+(filesPerDir+markers)*dirs)
                s.testS3List(c, stage.collbucket, "", 131, markers+2+(filesPerDir+markers)*dirs)
+               s.testS3List(c, stage.collbucket, "", 51, markers+2+(filesPerDir+markers)*dirs)
                s.testS3List(c, stage.collbucket, "dir0/", 71, filesPerDir+markers)
        }
 }
@@ -743,6 +851,9 @@ func (s *IntegrationSuite) testS3List(c *check.C, bucket *s3.Bucket, prefix stri
                        break
                }
                for _, key := range resp.Contents {
+                       if _, dup := gotKeys[key.Key]; dup {
+                               c.Errorf("got duplicate key %q on page %d", key.Key, pages)
+                       }
                        gotKeys[key.Key] = key
                        if strings.Contains(key.Key, "sailboat.txt") {
                                c.Check(key.Size, check.Equals, int64(4))
@@ -757,7 +868,16 @@ func (s *IntegrationSuite) testS3List(c *check.C, bucket *s3.Bucket, prefix stri
                }
                nextMarker = resp.NextMarker
        }
-       c.Check(len(gotKeys), check.Equals, expectFiles)
+       if !c.Check(len(gotKeys), check.Equals, expectFiles) {
+               var sorted []string
+               for k := range gotKeys {
+                       sorted = append(sorted, k)
+               }
+               sort.Strings(sorted)
+               for _, k := range sorted {
+                       c.Logf("got %s", k)
+               }
+       }
 }
 
 func (s *IntegrationSuite) TestS3CollectionListRollup(c *check.C) {
@@ -823,7 +943,8 @@ func (s *IntegrationSuite) testS3CollectionListRollup(c *check.C) {
                {"dir0", "", ""},
                {"dir0/", "", ""},
                {"dir0/f", "", ""},
-               {"dir0", "/", "dir0/file14.txt"},       // no commonprefixes
+               {"dir0", "/", "dir0/file14.txt"},       // one commonprefix, "dir0/"
+               {"dir0", "/", "dir0/zzzzfile.txt"},     // no commonprefixes
                {"", "", "dir0/file14.txt"},            // middle page, skip walking dir1
                {"", "", "dir1/file14.txt"},            // middle page, skip walking dir0
                {"", "", "dir1/file498.txt"},           // last page of results
@@ -854,28 +975,31 @@ func (s *IntegrationSuite) testS3CollectionListRollup(c *check.C) {
                var expectTruncated bool
                for _, key := range allfiles {
                        full := len(expectKeys)+len(expectPrefixes) >= maxKeys
-                       if !strings.HasPrefix(key, trial.prefix) || key < trial.marker {
+                       if !strings.HasPrefix(key, trial.prefix) || key <= trial.marker {
                                continue
                        } else if idx := strings.Index(key[len(trial.prefix):], trial.delimiter); trial.delimiter != "" && idx >= 0 {
                                prefix := key[:len(trial.prefix)+idx+1]
                                if len(expectPrefixes) > 0 && expectPrefixes[len(expectPrefixes)-1] == prefix {
                                        // same prefix as previous key
                                } else if full {
-                                       expectNextMarker = key
                                        expectTruncated = true
                                } else {
                                        expectPrefixes = append(expectPrefixes, prefix)
+                                       expectNextMarker = prefix
                                }
                        } else if full {
-                               if trial.delimiter != "" {
-                                       expectNextMarker = key
-                               }
                                expectTruncated = true
                                break
                        } else {
                                expectKeys = append(expectKeys, key)
+                               if trial.delimiter != "" {
+                                       expectNextMarker = key
+                               }
                        }
                }
+               if !expectTruncated {
+                       expectNextMarker = ""
+               }
 
                var gotKeys []string
                for _, key := range resp.Contents {
@@ -894,6 +1018,61 @@ func (s *IntegrationSuite) testS3CollectionListRollup(c *check.C) {
        }
 }
 
+func (s *IntegrationSuite) TestS3ListObjectsV2ManySubprojects(c *check.C) {
+       stage := s.s3setup(c)
+       defer stage.teardown(c)
+       projects := 50
+       collectionsPerProject := 2
+       for i := 0; i < projects; i++ {
+               var subproj arvados.Group
+               err := stage.arv.RequestAndDecode(&subproj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
+                       "group": map[string]interface{}{
+                               "owner_uuid":  stage.subproj.UUID,
+                               "group_class": "project",
+                               "name":        fmt.Sprintf("keep-web s3 test subproject %d", i),
+                       },
+               })
+               c.Assert(err, check.IsNil)
+               for j := 0; j < collectionsPerProject; j++ {
+                       err = stage.arv.RequestAndDecode(nil, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{
+                               "owner_uuid":    subproj.UUID,
+                               "name":          fmt.Sprintf("keep-web s3 test collection %d", j),
+                               "manifest_text": ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:emptyfile\n./emptydir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n",
+                       }})
+                       c.Assert(err, check.IsNil)
+               }
+       }
+       c.Logf("setup complete")
+
+       sess := aws_session.Must(aws_session.NewSession(&aws_aws.Config{
+               Region:           aws_aws.String("auto"),
+               Endpoint:         aws_aws.String(s.testServer.URL),
+               Credentials:      aws_credentials.NewStaticCredentials(url.QueryEscape(arvadostest.ActiveTokenV2), url.QueryEscape(arvadostest.ActiveTokenV2), ""),
+               S3ForcePathStyle: aws_aws.Bool(true),
+       }))
+       client := aws_s3.New(sess)
+       ctx := context.Background()
+       params := aws_s3.ListObjectsV2Input{
+               Bucket:    aws_aws.String(stage.proj.UUID),
+               Delimiter: aws_aws.String("/"),
+               Prefix:    aws_aws.String("keep-web s3 test subproject/"),
+               MaxKeys:   aws_aws.Int64(int64(projects / 2)),
+       }
+       for page := 1; ; page++ {
+               t0 := time.Now()
+               result, err := client.ListObjectsV2WithContext(ctx, &params)
+               if !c.Check(err, check.IsNil) {
+                       break
+               }
+               c.Logf("got page %d in %v with len(Contents) == %d, len(CommonPrefixes) == %d", page, time.Since(t0), len(result.Contents), len(result.CommonPrefixes))
+               if !*result.IsTruncated {
+                       break
+               }
+               params.ContinuationToken = result.NextContinuationToken
+               *params.MaxKeys = *params.MaxKeys/2 + 1
+       }
+}
+
 func (s *IntegrationSuite) TestS3ListObjectsV2(c *check.C) {
        stage := s.s3setup(c)
        defer stage.teardown(c)