-
Notifications
You must be signed in to change notification settings - Fork 111
/
bucket.go
629 lines (537 loc) · 17 KB
/
bucket.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
package backup
import (
"context"
"crypto/sha256"
"encoding/hex"
"hash"
"io"
"io/ioutil"
"os"
"path"
"path/filepath"
"sort"
"strings"
"go.uber.org/multierr"
"golang.org/x/oauth2/google"
"github.com/aws/aws-sdk-go/aws"
"github.com/aws/aws-sdk-go/aws/session"
"github.com/aws/aws-sdk-go/service/s3"
"gocloud.dev/blob"
"gocloud.dev/blob/gcsblob"
"gocloud.dev/blob/s3blob"
"gocloud.dev/gcerrors"
"gocloud.dev/gcp"
"github.com/chef/automate/lib/io/fileutils"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
)
// BucketObject represents the metadata for a stored object
type BucketObject struct {
// The name is the name of the object in the bucket relative to the bucket
// path. It can be a singular object name or a relative path.
// For example, if we instantiated an s3 bucket with the bucket name
// of foo, and a base path of bar/baz, all access to object would
// be implicitly relative to s3://foo/bar/baz. Calling
// NewReader(ctx, "quuz") would access s3://foo/bar/baz/quux
Name string
}
// SharedPrefix represents a group of object names that exist
type SharedPrefix string
// A BlobWriter is a WriteCloser that can also be failed
type BlobWriter interface {
io.WriteCloser
Fail(error) error
BlobSHA256() string
}
type BlobReader interface {
io.ReadCloser
BlobSHA256() string
}
// A Bucket is a place where blobs can be stored
type Bucket interface {
// NewReader returns a ReadCloser for the given storage key. The reader must be closed
// by the caller. It is provided only when there is not an error.
// If a object with the given name is not found, calling IsNotExist with the error
// must return true
NewReader(ctx context.Context, path string, verifier ObjectVerifier) (io.ReadCloser, error)
// NewWriter returns a BlobWriter that can write to the given storage key. The data may
// not be flushed until Close is called
NewWriter(ctx context.Context, path string) (BlobWriter, error)
// List returns a list of BucketObjects with the given object name prefix.
// If there are no matching objects stored, empty lists will be returned.
//
// If delimited is set to true, SharedPrefix objects will be returned. Doing this is
// equivalent to asking for a 1 level deep directory search. As an example, if pathPrefix
// was set to 'foo/', then the directory 'foo/bar' is a possible return value, but
// not 'foo/bar/baz'. If 'foo/quux' exists and is an object, then it would be returned in the
// BucketObject list.
//
// The pathPrefix provided is assumed to end in '/'
List(ctx context.Context, pathPrefix string, delimited bool) ([]BucketObject, []SharedPrefix, error)
Delete(ctx context.Context, objectPaths []string) error
}
// IsNotExist returns true if the error represents a object access against a objects that
// does not exist
func IsNotExist(err error) bool {
return os.IsNotExist(err) ||
gcerrors.Code(err) == gcerrors.NotFound ||
// blob.IsNotExist returns the wrong answer for minio's NoSuchKey error
strings.Contains(err.Error(), "NoSuchKey")
}
type fsBucket struct {
basePath string
}
// NewFilesystemBucket returns a Bucket backed by a filesystem directory
func NewFilesystemBucket(basePath string) Bucket {
return fsBucket{
basePath: basePath,
}
}
type checksummingAtomicWriter struct {
fileutils.WriteCloserFailer
sha256 hash.Hash
}
func (w *checksummingAtomicWriter) Write(p []byte) (n int, err error) {
w.sha256.Write(p) //nolint: errcheck // docs for hash.Hash say this never errors
return w.WriteCloserFailer.Write(p)
}
func (w *checksummingAtomicWriter) BlobSHA256() string {
return hex.EncodeToString(w.sha256.Sum(nil))
}
func newChecksummingAtomicWriter(writePath string, opts ...fileutils.AtomicWriteOpt) (BlobWriter, error) {
w, err := fileutils.NewAtomicWriter(writePath, opts...)
if err != nil {
return nil, err
}
return &checksummingAtomicWriter{
WriteCloserFailer: w,
sha256: sha256.New(),
}, nil
}
type checksummingReader struct {
io.ReadCloser
sha256 hash.Hash
tee io.Reader
}
func newChecksummingReader(r io.ReadCloser) BlobReader {
s := sha256.New()
return &checksummingReader{
ReadCloser: r,
sha256: s,
tee: io.TeeReader(r, s),
}
}
func (r *checksummingReader) Read(p []byte) (int, error) {
return r.tee.Read(p)
}
func (r *checksummingReader) BlobSHA256() string {
return hex.EncodeToString(r.sha256.Sum(nil))
}
func (bkt fsBucket) NewReader(ctx context.Context, relPath string, verifier ObjectVerifier) (io.ReadCloser, error) {
validatePath(relPath)
r, err := os.Open(path.Join(bkt.basePath, relPath))
logrus.Debugf("fsBucket NewReader %s ", relPath)
if err != nil {
return nil, err
}
cksumReader := newChecksummingReader(r)
defer cksumReader.Close()
return newVerifiedReader(relPath, cksumReader, verifier)
}
func (bkt fsBucket) NewWriter(ctx context.Context, relPath string) (BlobWriter, error) {
validatePath(relPath)
writePath := path.Join(bkt.basePath, relPath)
writeDir := path.Dir(writePath)
if err := os.MkdirAll(writeDir, defaultDirPerms); err != nil {
return nil, errors.Wrapf(err, "failed to create directory %s", writeDir)
}
return newChecksummingAtomicWriter(writePath, fileutils.WithAtomicWriteFileMode(0600), fileutils.WithAtomicWriteNoSync(true))
}
func (bkt fsBucket) List(ctx context.Context, pathPrefix string, delimited bool) ([]BucketObject, []SharedPrefix, error) {
validatePath(pathPrefix)
dir := path.Join(bkt.basePath, pathPrefix)
objs := []BucketObject{}
sharedPrefixes := []SharedPrefix{}
// Check if the directory where the objects would exist exists
if _, err := os.Stat(dir); err != nil {
if os.IsNotExist(err) {
// If it doesn't exist, there is no error. We return empty lists
return objs, sharedPrefixes, nil
}
return objs, sharedPrefixes, err
}
// When we're in the delimited mode, we don't need to talk the directory recursively. We
// can list the files in the given directory (pathPrefix) and break those into the
// BlobObject and SharedPrefix lists
if delimited {
dirs, err := ioutil.ReadDir(dir)
if err != nil {
return objs, sharedPrefixes, err
}
for _, d := range dirs {
relPath := path.Join(pathPrefix, d.Name())
if d.IsDir() {
// Directories are not a real thing in the object world, but they represent
// that a group of objects that exist under path name.
//
// NOTE: This is slightly a lie. It's possible that under d, there are only
// directories, or d itself is empty. We should really prune these out
// since no objects exist under it. In S3, this case cannot exist since
// it's not actually a filesystem.
// By not pruning, we expose a slight race condition. It is possible
// that the backups have started writing the status file. The directory
// will get created. If the backups are listed before the status file
// is written, the listing will be unable to distinguish between an
// old completed backup that did not have a status file and a new
// incomplete one.
if relPath[len(relPath)-1] != '/' {
relPath = relPath + "/"
}
sharedPrefixes = append(sharedPrefixes, SharedPrefix(relPath))
} else {
// There is an actual file there. return it as a BlobObject
objs = append(objs, BucketObject{
Name: relPath,
})
}
}
return objs, sharedPrefixes, nil
}
// If delimited mode was not asked for, we need to recursively find all the objects
err := filepath.Walk(dir, func(p string, info os.FileInfo, err error) error {
if err != nil {
return err
}
// Ignore the .tmp files because they're not complete
if strings.HasPrefix(path.Base(p), ".tmp.") {
return nil
}
// Directories aren't actual things, so skip them
if !info.IsDir() {
relPath, err := filepath.Rel(bkt.basePath, p)
if err != nil {
return err
}
objs = append(objs, BucketObject{
Name: relPath,
})
}
return nil
})
return objs, sharedPrefixes, err
}
func (bkt fsBucket) Delete(_ context.Context, objectPaths []string) error {
dirsToCheck := []string{}
for _, objectPath := range objectPaths {
validatePath(objectPath)
filePath := path.Join(bkt.basePath, objectPath)
if err := os.Remove(filePath); err != nil {
if !os.IsNotExist(err) {
return errors.Wrapf(err, "failed to delete %s", filePath)
}
} else {
dirPath := path.Dir(filePath)
dirsToCheck = append(dirsToCheck, dirPath)
}
}
sort.Slice(dirsToCheck, func(i, j int) bool { return len(dirsToCheck[i]) > len(dirsToCheck[j]) })
for _, dir := range dirsToCheck {
errContainsNonDir := errors.New("contains non dir")
err := filepath.Walk(dir, func(path string, info os.FileInfo, err error) error {
if err != nil {
return err
}
if !info.IsDir() {
return errContainsNonDir
}
return nil
})
if err != nil {
if os.IsNotExist(err) || err == errContainsNonDir {
continue
}
return errors.Wrap(err, "failed to cleanup filesystem bucket")
}
logrus.Debugf("Deleting empty directory %s. No files found under it", dir)
if err := os.RemoveAll(dir); err != nil {
return errors.Wrapf(err, "failed to delete %s", dir)
}
}
return nil
}
func validatePath(path string) {
if path == "" {
return
}
// In a object store, `foo//bar` is not the same as `foo/bar`
// There are other similar cases that I may or may not enumerate,
// hopefully this will help catch some bugs
if path[0] == '/' || strings.Contains(path, "//") || strings.Contains(path, "./") || strings.Contains(path, "..") {
logrus.WithField("path", path).Fatal("Invalid storage. Contains // or starts with /")
}
}
type writeCloserFailerWrapper struct {
io.WriteCloser
sha256 hash.Hash
}
func newWriteCloserFailerWrapper(w io.WriteCloser) *writeCloserFailerWrapper {
return &writeCloserFailerWrapper{
WriteCloser: w,
sha256: sha256.New(),
}
}
func (w *writeCloserFailerWrapper) Fail(err error) error {
return multierr.Combine(
err,
w.Close(),
)
}
func (w *writeCloserFailerWrapper) Write(p []byte) (n int, err error) {
w.sha256.Write(p) //nolint: errcheck // docs for hash.Hash say this never errors
return w.WriteCloser.Write(p)
}
func (w *writeCloserFailerWrapper) BlobSHA256() string {
return hex.EncodeToString(w.sha256.Sum(nil))
}
type s3Bucket struct {
bucket *blob.Bucket
s3Svc *s3.S3
name string
basePath string
}
func NewS3Bucket(name string, basePath string, c *aws.Config) (Bucket, error) {
logrus.Debugf("NewS3Bucket : basePath = %s : name = %s", basePath, name)
s, err := session.NewSession(c)
if err != nil {
return nil, err
}
bucket, err := s3blob.OpenBucket(context.Background(), s, name, nil)
if err != nil {
return nil, err
}
s3Svc := s3.New(s)
return &s3Bucket{
bucket: bucket,
s3Svc: s3Svc,
name: name,
basePath: basePath,
}, nil
}
func (bkt *s3Bucket) NewReader(ctx context.Context, name string, verifier ObjectVerifier) (io.ReadCloser, error) {
relPath := path.Join(bkt.basePath, name)
validatePath(relPath)
logrus.Debugf("s3Bucket : NewReader : relPath = %s : name = %s", bkt.basePath, name)
r, err := bkt.bucket.NewReader(ctx, relPath, nil)
if err != nil {
return nil, err
}
cksumReader := newChecksummingReader(r)
defer cksumReader.Close()
return newVerifiedReader(name, cksumReader, verifier)
}
func (bkt *s3Bucket) NewWriter(ctx context.Context, name string) (BlobWriter, error) {
relPath := path.Join(bkt.basePath, name)
validatePath(relPath)
writer, err := bkt.bucket.NewWriter(ctx, relPath, nil)
if err != nil {
return nil, err
}
return newWriteCloserFailerWrapper(writer), nil
}
func (bkt *s3Bucket) Delete(ctx context.Context, objectPaths []string) error {
if len(objectPaths) == 0 {
return nil
}
objectIdents := make([]*s3.ObjectIdentifier, len(objectPaths))
for i, objectPath := range objectPaths {
fullObjectPath := path.Join(bkt.basePath, objectPath)
validatePath(fullObjectPath)
objectIdents[i] = &s3.ObjectIdentifier{
Key: aws.String(fullObjectPath),
}
logrus.Debugf("marking %s for delete", fullObjectPath)
}
req := &s3.DeleteObjectsInput{
Bucket: aws.String(bkt.name),
Delete: &s3.Delete{
Objects: objectIdents,
Quiet: aws.Bool(true),
},
}
resp, err := bkt.s3Svc.DeleteObjectsWithContext(ctx, req)
if err != nil {
logrus.
WithField("objectPaths", objectPaths).
WithError(err).
Error("Failed to delete objects")
return errors.Wrap(err, "failed to delete objects")
}
for _, e := range resp.Errors {
path := ""
reason := ""
if e.Key != nil {
path = *e.Key
}
if e.Message != nil {
reason = *e.Message
}
logrus.WithFields(logrus.Fields{
"path": path,
"reason": reason,
}).Warn("Failed to delete object")
}
return nil
}
func (bkt *s3Bucket) List(ctx context.Context, pathPrefix string, delimited bool) ([]BucketObject, []SharedPrefix, error) {
pathPrefix = path.Join(bkt.basePath, pathPrefix)
validatePath(pathPrefix)
prefix := pathPrefix
if prefix != "" && prefix[len(prefix)-1] != '/' {
prefix = prefix + "/"
}
objs := []BucketObject{}
prefixes := []SharedPrefix{}
req := &s3.ListObjectsV2Input{
Bucket: aws.String(bkt.name),
Prefix: aws.String(prefix),
}
if delimited {
req.Delimiter = aws.String("/")
}
err := bkt.s3Svc.ListObjectsV2Pages(req, func(page *s3.ListObjectsV2Output, lastPage bool) bool {
if page.Contents != nil {
for _, s3obj := range page.Contents {
relKey := strings.TrimPrefix(strings.TrimPrefix(*s3obj.Key, bkt.basePath), "/")
objs = append(objs, BucketObject{
Name: relKey,
})
}
}
if page.CommonPrefixes != nil {
for _, prefix := range page.CommonPrefixes {
relKey := strings.TrimPrefix(strings.TrimPrefix(*prefix.Prefix, bkt.basePath), "/")
prefixes = append(prefixes, SharedPrefix(relKey))
}
}
return true
})
return objs, prefixes, err
}
type errBucket struct {
err error
}
func (bkt errBucket) NewReader(ctx context.Context, path string, verifier ObjectVerifier) (io.ReadCloser, error) {
logrus.Debugf("errBucket : NewReader : path = %s", path)
return nil, bkt.err
}
func (bkt errBucket) NewWriter(ctx context.Context, path string) (BlobWriter, error) {
return nil, bkt.err
}
func (bkt errBucket) List(ctx context.Context, pathPrefix string, delimited bool) ([]BucketObject, []SharedPrefix, error) {
return nil, nil, bkt.err
}
func (bkt errBucket) Delete(ctx context.Context, objectPaths []string) error {
return bkt.err
}
func ToObjectPaths(deleteObjs []BucketObject) []string {
objPaths := make([]string, len(deleteObjs))
for i, o := range deleteObjs {
objPaths[i] = o.Name
}
return objPaths
}
type gcsBucket struct {
bucket *blob.Bucket
name string
basePath string
}
type GCSConfig struct {
GoogleApplicationCredentials string
}
func NewGCSBucket(name string, basePath string, c *GCSConfig) (Bucket, error) {
ctx := context.Background()
logrus.Debugf("NewGCSBucket : NewReader : basePath = %s : name = %s", basePath, name)
var creds *google.Credentials
var err error
if c.GoogleApplicationCredentials != "" {
creds, err = google.CredentialsFromJSON(ctx, []byte(c.GoogleApplicationCredentials),
"https://www.googleapis.com/auth/cloud-platform")
} else {
creds, err = gcp.DefaultCredentials(ctx)
if err != nil {
return nil, err
}
}
if err != nil {
return nil, errors.Wrap(err, "failed to get GCS credentials")
}
client, err := gcp.NewHTTPClient(
gcp.DefaultTransport(),
gcp.CredentialsTokenSource(creds))
if err != nil {
return nil, err
}
bucket, err := gcsblob.OpenBucket(ctx, client, name, nil)
if err != nil {
return nil, err
}
return &gcsBucket{
bucket: bucket,
name: name,
basePath: basePath,
}, nil
}
func (bkt *gcsBucket) NewReader(ctx context.Context, name string, verifier ObjectVerifier) (io.ReadCloser, error) {
relPath := path.Join(bkt.basePath, name)
validatePath(relPath)
logrus.Debugf("gcsBucket : NewReader : relPath = %s : name = %s", relPath, name)
r, err := bkt.bucket.NewReader(ctx, relPath, nil)
if err != nil {
return nil, err
}
cksumReader := newChecksummingReader(r)
defer cksumReader.Close()
return newVerifiedReader(name, cksumReader, verifier)
}
func (bkt *gcsBucket) NewWriter(ctx context.Context, name string) (BlobWriter, error) {
return nil, errors.New("Unimplemented")
}
func (bkt *gcsBucket) Delete(ctx context.Context, objectPaths []string) error {
return errors.New("Unimplemented")
}
func (bkt *gcsBucket) List(ctx context.Context, pathPrefix string, delimited bool) ([]BucketObject, []SharedPrefix, error) {
pathPrefix = path.Join(bkt.basePath, pathPrefix)
validatePath(pathPrefix)
prefix := pathPrefix
if prefix != "" && prefix[len(prefix)-1] != '/' {
prefix = prefix + "/"
}
objs := []BucketObject{}
prefixes := []SharedPrefix{}
delimiter := ""
if delimited {
delimiter = "/"
}
iter := bkt.bucket.List(&blob.ListOptions{
Prefix: prefix,
Delimiter: delimiter,
})
for {
obj, err := iter.Next(ctx)
if err == io.EOF {
break
}
if err != nil {
return nil, nil, err
}
relKey := strings.TrimPrefix(strings.TrimPrefix(obj.Key, bkt.basePath), "/")
if obj.IsDir {
prefixes = append(prefixes, SharedPrefix(relKey))
} else {
objs = append(objs, BucketObject{
Name: relKey,
})
}
}
return objs, prefixes, nil
}