16535: Test XML header and content-type.
[arvados.git] / 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         "strings"
15         "sync"
16         "time"
17
18         "git.arvados.org/arvados.git/sdk/go/arvados"
19         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
20         "git.arvados.org/arvados.git/sdk/go/arvadostest"
21         "git.arvados.org/arvados.git/sdk/go/keepclient"
22         "github.com/AdRoll/goamz/aws"
23         "github.com/AdRoll/goamz/s3"
24         check "gopkg.in/check.v1"
25 )
26
27 type s3stage struct {
28         arv        *arvados.Client
29         ac         *arvadosclient.ArvadosClient
30         kc         *keepclient.KeepClient
31         proj       arvados.Group
32         projbucket *s3.Bucket
33         coll       arvados.Collection
34         collbucket *s3.Bucket
35 }
36
37 func (s *IntegrationSuite) s3setup(c *check.C) s3stage {
38         var proj arvados.Group
39         var coll arvados.Collection
40         arv := arvados.NewClientFromEnv()
41         arv.AuthToken = arvadostest.ActiveToken
42         err := arv.RequestAndDecode(&proj, "POST", "arvados/v1/groups", nil, map[string]interface{}{
43                 "group": map[string]interface{}{
44                         "group_class": "project",
45                         "name":        "keep-web s3 test",
46                 },
47                 "ensure_unique_name": true,
48         })
49         c.Assert(err, check.IsNil)
50         err = arv.RequestAndDecode(&coll, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{
51                 "owner_uuid":    proj.UUID,
52                 "name":          "keep-web s3 test collection",
53                 "manifest_text": ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:emptyfile\n./emptydir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n",
54         }})
55         c.Assert(err, check.IsNil)
56         ac, err := arvadosclient.New(arv)
57         c.Assert(err, check.IsNil)
58         kc, err := keepclient.MakeKeepClient(ac)
59         c.Assert(err, check.IsNil)
60         fs, err := coll.FileSystem(arv, kc)
61         c.Assert(err, check.IsNil)
62         f, err := fs.OpenFile("sailboat.txt", os.O_CREATE|os.O_WRONLY, 0644)
63         c.Assert(err, check.IsNil)
64         _, err = f.Write([]byte("⛵\n"))
65         c.Assert(err, check.IsNil)
66         err = f.Close()
67         c.Assert(err, check.IsNil)
68         err = fs.Sync()
69         c.Assert(err, check.IsNil)
70         err = arv.RequestAndDecode(&coll, "GET", "arvados/v1/collections/"+coll.UUID, nil, nil)
71         c.Assert(err, check.IsNil)
72
73         auth := aws.NewAuth(arvadostest.ActiveTokenV2, arvadostest.ActiveTokenV2, "", time.Now().Add(time.Hour))
74         region := aws.Region{
75                 Name:       s.testServer.Addr,
76                 S3Endpoint: "http://" + s.testServer.Addr,
77         }
78         client := s3.New(*auth, region)
79         return s3stage{
80                 arv:  arv,
81                 ac:   ac,
82                 kc:   kc,
83                 proj: proj,
84                 projbucket: &s3.Bucket{
85                         S3:   client,
86                         Name: proj.UUID,
87                 },
88                 coll: coll,
89                 collbucket: &s3.Bucket{
90                         S3:   client,
91                         Name: coll.UUID,
92                 },
93         }
94 }
95
96 func (stage s3stage) teardown(c *check.C) {
97         if stage.coll.UUID != "" {
98                 err := stage.arv.RequestAndDecode(&stage.coll, "DELETE", "arvados/v1/collections/"+stage.coll.UUID, nil, nil)
99                 c.Check(err, check.IsNil)
100         }
101         if stage.proj.UUID != "" {
102                 err := stage.arv.RequestAndDecode(&stage.proj, "DELETE", "arvados/v1/groups/"+stage.proj.UUID, nil, nil)
103                 c.Check(err, check.IsNil)
104         }
105 }
106
107 func (s *IntegrationSuite) TestS3HeadBucket(c *check.C) {
108         stage := s.s3setup(c)
109         defer stage.teardown(c)
110
111         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
112                 c.Logf("bucket %s", bucket.Name)
113                 exists, err := bucket.Exists("")
114                 c.Check(err, check.IsNil)
115                 c.Check(exists, check.Equals, true)
116         }
117 }
118
119 func (s *IntegrationSuite) TestS3CollectionGetObject(c *check.C) {
120         stage := s.s3setup(c)
121         defer stage.teardown(c)
122         s.testS3GetObject(c, stage.collbucket, "")
123 }
124 func (s *IntegrationSuite) TestS3ProjectGetObject(c *check.C) {
125         stage := s.s3setup(c)
126         defer stage.teardown(c)
127         s.testS3GetObject(c, stage.projbucket, stage.coll.Name+"/")
128 }
129 func (s *IntegrationSuite) testS3GetObject(c *check.C, bucket *s3.Bucket, prefix string) {
130         rdr, err := bucket.GetReader(prefix + "emptyfile")
131         c.Assert(err, check.IsNil)
132         buf, err := ioutil.ReadAll(rdr)
133         c.Check(err, check.IsNil)
134         c.Check(len(buf), check.Equals, 0)
135         err = rdr.Close()
136         c.Check(err, check.IsNil)
137
138         // GetObject
139         rdr, err = bucket.GetReader(prefix + "missingfile")
140         c.Check(err, check.ErrorMatches, `404 Not Found`)
141
142         // HeadObject
143         exists, err := bucket.Exists(prefix + "missingfile")
144         c.Check(err, check.IsNil)
145         c.Check(exists, check.Equals, false)
146
147         // GetObject
148         rdr, err = bucket.GetReader(prefix + "sailboat.txt")
149         c.Assert(err, check.IsNil)
150         buf, err = ioutil.ReadAll(rdr)
151         c.Check(err, check.IsNil)
152         c.Check(buf, check.DeepEquals, []byte("⛵\n"))
153         err = rdr.Close()
154         c.Check(err, check.IsNil)
155
156         // HeadObject
157         exists, err = bucket.Exists(prefix + "sailboat.txt")
158         c.Check(err, check.IsNil)
159         c.Check(exists, check.Equals, true)
160 }
161
162 func (s *IntegrationSuite) TestS3CollectionPutObjectSuccess(c *check.C) {
163         stage := s.s3setup(c)
164         defer stage.teardown(c)
165         s.testS3PutObjectSuccess(c, stage.collbucket, "")
166 }
167 func (s *IntegrationSuite) TestS3ProjectPutObjectSuccess(c *check.C) {
168         stage := s.s3setup(c)
169         defer stage.teardown(c)
170         s.testS3PutObjectSuccess(c, stage.projbucket, stage.coll.Name+"/")
171 }
172 func (s *IntegrationSuite) testS3PutObjectSuccess(c *check.C, bucket *s3.Bucket, prefix string) {
173         for _, trial := range []struct {
174                 path string
175                 size int
176         }{
177                 {
178                         path: "newfile",
179                         size: 128000000,
180                 }, {
181                         path: "newdir/newfile",
182                         size: 1 << 26,
183                 }, {
184                         path: "newdir1/newdir2/newfile",
185                         size: 0,
186                 },
187         } {
188                 c.Logf("=== %v", trial)
189
190                 objname := prefix + trial.path
191
192                 _, err := bucket.GetReader(objname)
193                 c.Assert(err, check.ErrorMatches, `404 Not Found`)
194
195                 buf := make([]byte, trial.size)
196                 rand.Read(buf)
197
198                 err = bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), "application/octet-stream", s3.Private, s3.Options{})
199                 c.Check(err, check.IsNil)
200
201                 rdr, err := bucket.GetReader(objname)
202                 if !c.Check(err, check.IsNil) {
203                         continue
204                 }
205                 buf2, err := ioutil.ReadAll(rdr)
206                 c.Check(err, check.IsNil)
207                 c.Check(buf2, check.HasLen, len(buf))
208                 c.Check(bytes.Equal(buf, buf2), check.Equals, true)
209         }
210 }
211
212 func (s *IntegrationSuite) TestS3CollectionPutObjectFailure(c *check.C) {
213         stage := s.s3setup(c)
214         defer stage.teardown(c)
215         s.testS3PutObjectFailure(c, stage.collbucket, "")
216 }
217 func (s *IntegrationSuite) TestS3ProjectPutObjectFailure(c *check.C) {
218         stage := s.s3setup(c)
219         defer stage.teardown(c)
220         s.testS3PutObjectFailure(c, stage.projbucket, stage.coll.Name+"/")
221 }
222 func (s *IntegrationSuite) testS3PutObjectFailure(c *check.C, bucket *s3.Bucket, prefix string) {
223         var wg sync.WaitGroup
224         for _, trial := range []struct {
225                 path string
226         }{
227                 {
228                         path: "emptyfile/newname", // emptyfile exists, see s3setup()
229                 }, {
230                         path: "emptyfile/", // emptyfile exists, see s3setup()
231                 }, {
232                         path: "emptydir", // dir already exists, see s3setup()
233                 }, {
234                         path: "emptydir/",
235                 }, {
236                         path: "emptydir//",
237                 }, {
238                         path: "newdir/",
239                 }, {
240                         path: "newdir//",
241                 }, {
242                         path: "/",
243                 }, {
244                         path: "//",
245                 }, {
246                         path: "foo//bar",
247                 }, {
248                         path: "",
249                 },
250         } {
251                 trial := trial
252                 wg.Add(1)
253                 go func() {
254                         defer wg.Done()
255                         c.Logf("=== %v", trial)
256
257                         objname := prefix + trial.path
258
259                         buf := make([]byte, 1234)
260                         rand.Read(buf)
261
262                         err := bucket.PutReader(objname, bytes.NewReader(buf), int64(len(buf)), "application/octet-stream", s3.Private, s3.Options{})
263                         if !c.Check(err, check.ErrorMatches, `400 Bad.*`, check.Commentf("PUT %q should fail", objname)) {
264                                 return
265                         }
266
267                         if objname != "" && objname != "/" {
268                                 _, err = bucket.GetReader(objname)
269                                 c.Check(err, check.ErrorMatches, `404 Not Found`, check.Commentf("GET %q should return 404", objname))
270                         }
271                 }()
272         }
273         wg.Wait()
274 }
275
276 func (stage *s3stage) writeBigDirs(c *check.C, dirs int, filesPerDir int) {
277         fs, err := stage.coll.FileSystem(stage.arv, stage.kc)
278         c.Assert(err, check.IsNil)
279         for d := 0; d < dirs; d++ {
280                 dir := fmt.Sprintf("dir%d", d)
281                 c.Assert(fs.Mkdir(dir, 0755), check.IsNil)
282                 for i := 0; i < filesPerDir; i++ {
283                         f, err := fs.OpenFile(fmt.Sprintf("%s/file%d.txt", dir, i), os.O_CREATE|os.O_WRONLY, 0644)
284                         c.Assert(err, check.IsNil)
285                         c.Assert(f.Close(), check.IsNil)
286                 }
287         }
288         c.Assert(fs.Sync(), check.IsNil)
289 }
290
291 func (s *IntegrationSuite) TestS3GetBucketVersioning(c *check.C) {
292         stage := s.s3setup(c)
293         defer stage.teardown(c)
294         for _, bucket := range []*s3.Bucket{stage.collbucket, stage.projbucket} {
295                 req, err := http.NewRequest("GET", bucket.URL("/"), nil)
296                 req.Header.Set("Authorization", "AWS "+arvadostest.ActiveTokenV2+":none")
297                 req.URL.RawQuery = "versioning"
298                 resp, err := http.DefaultClient.Do(req)
299                 c.Assert(err, check.IsNil)
300                 c.Check(resp.Header.Get("Content-Type"), check.Equals, "application/xml")
301                 buf, err := ioutil.ReadAll(resp.Body)
302                 c.Assert(err, check.IsNil)
303                 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")
304         }
305 }
306
307 func (s *IntegrationSuite) TestS3CollectionList(c *check.C) {
308         stage := s.s3setup(c)
309         defer stage.teardown(c)
310
311         filesPerDir := 1001
312         stage.writeBigDirs(c, 2, filesPerDir)
313         s.testS3List(c, stage.collbucket, "", 4000, 2+filesPerDir*2)
314         s.testS3List(c, stage.collbucket, "", 131, 2+filesPerDir*2)
315         s.testS3List(c, stage.collbucket, "dir0/", 71, filesPerDir)
316 }
317 func (s *IntegrationSuite) testS3List(c *check.C, bucket *s3.Bucket, prefix string, pageSize, expectFiles int) {
318         expectPageSize := pageSize
319         if expectPageSize > 1000 {
320                 expectPageSize = 1000
321         }
322         gotKeys := map[string]s3.Key{}
323         nextMarker := ""
324         pages := 0
325         for {
326                 resp, err := bucket.List(prefix, "", nextMarker, pageSize)
327                 if !c.Check(err, check.IsNil) {
328                         break
329                 }
330                 c.Check(len(resp.Contents) <= expectPageSize, check.Equals, true)
331                 if pages++; !c.Check(pages <= (expectFiles/expectPageSize)+1, check.Equals, true) {
332                         break
333                 }
334                 for _, key := range resp.Contents {
335                         gotKeys[key.Key] = key
336                         if strings.Contains(key.Key, "sailboat.txt") {
337                                 c.Check(key.Size, check.Equals, int64(4))
338                         }
339                 }
340                 if !resp.IsTruncated {
341                         c.Check(resp.NextMarker, check.Equals, "")
342                         break
343                 }
344                 if !c.Check(resp.NextMarker, check.Not(check.Equals), "") {
345                         break
346                 }
347                 nextMarker = resp.NextMarker
348         }
349         c.Check(len(gotKeys), check.Equals, expectFiles)
350 }
351
352 func (s *IntegrationSuite) TestS3CollectionListRollup(c *check.C) {
353         stage := s.s3setup(c)
354         defer stage.teardown(c)
355
356         dirs := 2
357         filesPerDir := 500
358         stage.writeBigDirs(c, dirs, filesPerDir)
359         err := stage.collbucket.PutReader("dingbats", &bytes.Buffer{}, 0, "application/octet-stream", s3.Private, s3.Options{})
360         c.Assert(err, check.IsNil)
361         var allfiles []string
362         for marker := ""; ; {
363                 resp, err := stage.collbucket.List("", "", marker, 20000)
364                 c.Check(err, check.IsNil)
365                 for _, key := range resp.Contents {
366                         if len(allfiles) == 0 || allfiles[len(allfiles)-1] != key.Key {
367                                 allfiles = append(allfiles, key.Key)
368                         }
369                 }
370                 marker = resp.NextMarker
371                 if marker == "" {
372                         break
373                 }
374         }
375         c.Check(allfiles, check.HasLen, dirs*filesPerDir+3)
376
377         for _, trial := range []struct {
378                 prefix    string
379                 delimiter string
380                 marker    string
381         }{
382                 {"di", "/", ""},
383                 {"di", "r", ""},
384                 {"di", "n", ""},
385                 {"dir0", "/", ""},
386                 {"dir0", "/", "dir0/file14.txt"},       // no commonprefixes
387                 {"", "", "dir0/file14.txt"},            // middle page, skip walking dir1
388                 {"", "", "dir1/file14.txt"},            // middle page, skip walking dir0
389                 {"", "", "dir1/file498.txt"},           // last page of results
390                 {"dir1/file", "", "dir1/file498.txt"},  // last page of results, with prefix
391                 {"dir1/file", "/", "dir1/file498.txt"}, // last page of results, with prefix + delimiter
392                 {"dir1", "Z", "dir1/file498.txt"},      // delimiter "Z" never appears
393                 {"dir2", "/", ""},                      // prefix "dir2" does not exist
394                 {"", "/", ""},
395         } {
396                 c.Logf("\n\n=== trial %+v", trial)
397
398                 maxKeys := 20
399                 resp, err := stage.collbucket.List(trial.prefix, trial.delimiter, trial.marker, maxKeys)
400                 c.Check(err, check.IsNil)
401                 if resp.IsTruncated && trial.delimiter == "" {
402                         // goamz List method fills in the missing
403                         // NextMarker field if resp.IsTruncated, so
404                         // now we can't really tell whether it was
405                         // sent by the server or by goamz. In cases
406                         // where it should be empty but isn't, assume
407                         // it's goamz's fault.
408                         resp.NextMarker = ""
409                 }
410
411                 var expectKeys []string
412                 var expectPrefixes []string
413                 var expectNextMarker string
414                 var expectTruncated bool
415                 for _, key := range allfiles {
416                         full := len(expectKeys)+len(expectPrefixes) >= maxKeys
417                         if !strings.HasPrefix(key, trial.prefix) || key < trial.marker {
418                                 continue
419                         } else if idx := strings.Index(key[len(trial.prefix):], trial.delimiter); trial.delimiter != "" && idx >= 0 {
420                                 prefix := key[:len(trial.prefix)+idx+1]
421                                 if len(expectPrefixes) > 0 && expectPrefixes[len(expectPrefixes)-1] == prefix {
422                                         // same prefix as previous key
423                                 } else if full {
424                                         expectNextMarker = key
425                                         expectTruncated = true
426                                 } else {
427                                         expectPrefixes = append(expectPrefixes, prefix)
428                                 }
429                         } else if full {
430                                 if trial.delimiter != "" {
431                                         expectNextMarker = key
432                                 }
433                                 expectTruncated = true
434                                 break
435                         } else {
436                                 expectKeys = append(expectKeys, key)
437                         }
438                 }
439
440                 var gotKeys []string
441                 for _, key := range resp.Contents {
442                         gotKeys = append(gotKeys, key.Key)
443                 }
444                 var gotPrefixes []string
445                 for _, prefix := range resp.CommonPrefixes {
446                         gotPrefixes = append(gotPrefixes, prefix)
447                 }
448                 c.Check(gotKeys, check.DeepEquals, expectKeys)
449                 c.Check(gotPrefixes, check.DeepEquals, expectPrefixes)
450                 c.Check(resp.NextMarker, check.Equals, expectNextMarker)
451                 c.Check(resp.IsTruncated, check.Equals, expectTruncated)
452                 c.Logf("=== trial %+v keys %q prefixes %q nextMarker %q", trial, gotKeys, gotPrefixes, resp.NextMarker)
453         }
454 }