1
2
3
4
5 package tar
6
7 import (
8 "bytes"
9 "io"
10 "path/filepath"
11 "strconv"
12 "strings"
13 "time"
14 )
15
16
17
18
19 type Reader struct {
20 r io.Reader
21 pad int64
22 curr fileReader
23 blk block
24
25
26
27
28 err error
29 }
30
31 type fileReader interface {
32 io.Reader
33 fileState
34
35 WriteTo(io.Writer) (int64, error)
36 }
37
38
39 func NewReader(r io.Reader) *Reader {
40 return &Reader{r: r, curr: ®FileReader{r, 0}}
41 }
42
43
44
45
46
47
48
49
50
51
52
53
54 func (tr *Reader) Next() (*Header, error) {
55 if tr.err != nil {
56 return nil, tr.err
57 }
58 hdr, err := tr.next()
59 tr.err = err
60 if err == nil && !filepath.IsLocal(hdr.Name) {
61 if tarinsecurepath.Value() == "0" {
62 tarinsecurepath.IncNonDefault()
63 err = ErrInsecurePath
64 }
65 }
66 return hdr, err
67 }
68
69 func (tr *Reader) next() (*Header, error) {
70 var paxHdrs map[string]string
71 var gnuLongName, gnuLongLink string
72
73
74
75
76
77
78 format := FormatUSTAR | FormatPAX | FormatGNU
79 for {
80
81 if err := discard(tr.r, tr.curr.physicalRemaining()); err != nil {
82 return nil, err
83 }
84 if _, err := tryReadFull(tr.r, tr.blk[:tr.pad]); err != nil {
85 return nil, err
86 }
87 tr.pad = 0
88
89 hdr, rawHdr, err := tr.readHeader()
90 if err != nil {
91 return nil, err
92 }
93 if err := tr.handleRegularFile(hdr); err != nil {
94 return nil, err
95 }
96 format.mayOnlyBe(hdr.Format)
97
98
99 switch hdr.Typeflag {
100 case TypeXHeader, TypeXGlobalHeader:
101 format.mayOnlyBe(FormatPAX)
102 paxHdrs, err = parsePAX(tr)
103 if err != nil {
104 return nil, err
105 }
106 if hdr.Typeflag == TypeXGlobalHeader {
107 mergePAX(hdr, paxHdrs)
108 return &Header{
109 Name: hdr.Name,
110 Typeflag: hdr.Typeflag,
111 Xattrs: hdr.Xattrs,
112 PAXRecords: hdr.PAXRecords,
113 Format: format,
114 }, nil
115 }
116 continue
117 case TypeGNULongName, TypeGNULongLink:
118 format.mayOnlyBe(FormatGNU)
119 realname, err := readSpecialFile(tr)
120 if err != nil {
121 return nil, err
122 }
123
124 var p parser
125 switch hdr.Typeflag {
126 case TypeGNULongName:
127 gnuLongName = p.parseString(realname)
128 case TypeGNULongLink:
129 gnuLongLink = p.parseString(realname)
130 }
131 continue
132 default:
133
134
135
136 if err := mergePAX(hdr, paxHdrs); err != nil {
137 return nil, err
138 }
139 if gnuLongName != "" {
140 hdr.Name = gnuLongName
141 }
142 if gnuLongLink != "" {
143 hdr.Linkname = gnuLongLink
144 }
145 if hdr.Typeflag == TypeRegA {
146 if strings.HasSuffix(hdr.Name, "/") {
147 hdr.Typeflag = TypeDir
148 } else {
149 hdr.Typeflag = TypeReg
150 }
151 }
152
153
154
155 if err := tr.handleRegularFile(hdr); err != nil {
156 return nil, err
157 }
158
159
160
161 if err := tr.handleSparseFile(hdr, rawHdr); err != nil {
162 return nil, err
163 }
164
165
166 if format.has(FormatUSTAR) && format.has(FormatPAX) {
167 format.mayOnlyBe(FormatUSTAR)
168 }
169 hdr.Format = format
170 return hdr, nil
171 }
172 }
173 }
174
175
176
177
178 func (tr *Reader) handleRegularFile(hdr *Header) error {
179 nb := hdr.Size
180 if isHeaderOnlyType(hdr.Typeflag) {
181 nb = 0
182 }
183 if nb < 0 {
184 return ErrHeader
185 }
186
187 tr.pad = blockPadding(nb)
188 tr.curr = ®FileReader{r: tr.r, nb: nb}
189 return nil
190 }
191
192
193
194 func (tr *Reader) handleSparseFile(hdr *Header, rawHdr *block) error {
195 var spd sparseDatas
196 var err error
197 if hdr.Typeflag == TypeGNUSparse {
198 spd, err = tr.readOldGNUSparseMap(hdr, rawHdr)
199 } else {
200 spd, err = tr.readGNUSparsePAXHeaders(hdr)
201 }
202
203
204
205 if err == nil && spd != nil {
206 if isHeaderOnlyType(hdr.Typeflag) || !validateSparseEntries(spd, hdr.Size) {
207 return ErrHeader
208 }
209 sph := invertSparseEntries(spd, hdr.Size)
210 tr.curr = &sparseFileReader{tr.curr, sph, 0}
211 }
212 return err
213 }
214
215
216
217
218
219 func (tr *Reader) readGNUSparsePAXHeaders(hdr *Header) (sparseDatas, error) {
220
221 var is1x0 bool
222 major, minor := hdr.PAXRecords[paxGNUSparseMajor], hdr.PAXRecords[paxGNUSparseMinor]
223 switch {
224 case major == "0" && (minor == "0" || minor == "1"):
225 is1x0 = false
226 case major == "1" && minor == "0":
227 is1x0 = true
228 case major != "" || minor != "":
229 return nil, nil
230 case hdr.PAXRecords[paxGNUSparseMap] != "":
231 is1x0 = false
232 default:
233 return nil, nil
234 }
235 hdr.Format.mayOnlyBe(FormatPAX)
236
237
238 if name := hdr.PAXRecords[paxGNUSparseName]; name != "" {
239 hdr.Name = name
240 }
241 size := hdr.PAXRecords[paxGNUSparseSize]
242 if size == "" {
243 size = hdr.PAXRecords[paxGNUSparseRealSize]
244 }
245 if size != "" {
246 n, err := strconv.ParseInt(size, 10, 64)
247 if err != nil {
248 return nil, ErrHeader
249 }
250 hdr.Size = n
251 }
252
253
254 if is1x0 {
255 return readGNUSparseMap1x0(tr.curr)
256 }
257 return readGNUSparseMap0x1(hdr.PAXRecords)
258 }
259
260
261 func mergePAX(hdr *Header, paxHdrs map[string]string) (err error) {
262 for k, v := range paxHdrs {
263 if v == "" {
264 continue
265 }
266 var id64 int64
267 switch k {
268 case paxPath:
269 hdr.Name = v
270 case paxLinkpath:
271 hdr.Linkname = v
272 case paxUname:
273 hdr.Uname = v
274 case paxGname:
275 hdr.Gname = v
276 case paxUid:
277 id64, err = strconv.ParseInt(v, 10, 64)
278 hdr.Uid = int(id64)
279 case paxGid:
280 id64, err = strconv.ParseInt(v, 10, 64)
281 hdr.Gid = int(id64)
282 case paxAtime:
283 hdr.AccessTime, err = parsePAXTime(v)
284 case paxMtime:
285 hdr.ModTime, err = parsePAXTime(v)
286 case paxCtime:
287 hdr.ChangeTime, err = parsePAXTime(v)
288 case paxSize:
289 hdr.Size, err = strconv.ParseInt(v, 10, 64)
290 default:
291 if strings.HasPrefix(k, paxSchilyXattr) {
292 if hdr.Xattrs == nil {
293 hdr.Xattrs = make(map[string]string)
294 }
295 hdr.Xattrs[k[len(paxSchilyXattr):]] = v
296 }
297 }
298 if err != nil {
299 return ErrHeader
300 }
301 }
302 hdr.PAXRecords = paxHdrs
303 return nil
304 }
305
306
307
308 func parsePAX(r io.Reader) (map[string]string, error) {
309 buf, err := readSpecialFile(r)
310 if err != nil {
311 return nil, err
312 }
313 sbuf := string(buf)
314
315
316
317
318 var sparseMap []string
319
320 paxHdrs := make(map[string]string)
321 for len(sbuf) > 0 {
322 key, value, residual, err := parsePAXRecord(sbuf)
323 if err != nil {
324 return nil, ErrHeader
325 }
326 sbuf = residual
327
328 switch key {
329 case paxGNUSparseOffset, paxGNUSparseNumBytes:
330
331 if (len(sparseMap)%2 == 0 && key != paxGNUSparseOffset) ||
332 (len(sparseMap)%2 == 1 && key != paxGNUSparseNumBytes) ||
333 strings.Contains(value, ",") {
334 return nil, ErrHeader
335 }
336 sparseMap = append(sparseMap, value)
337 default:
338 paxHdrs[key] = value
339 }
340 }
341 if len(sparseMap) > 0 {
342 paxHdrs[paxGNUSparseMap] = strings.Join(sparseMap, ",")
343 }
344 return paxHdrs, nil
345 }
346
347
348
349
350
351
352
353
354
355 func (tr *Reader) readHeader() (*Header, *block, error) {
356
357 if _, err := io.ReadFull(tr.r, tr.blk[:]); err != nil {
358 return nil, nil, err
359 }
360 if bytes.Equal(tr.blk[:], zeroBlock[:]) {
361 if _, err := io.ReadFull(tr.r, tr.blk[:]); err != nil {
362 return nil, nil, err
363 }
364 if bytes.Equal(tr.blk[:], zeroBlock[:]) {
365 return nil, nil, io.EOF
366 }
367 return nil, nil, ErrHeader
368 }
369
370
371 format := tr.blk.getFormat()
372 if format == FormatUnknown {
373 return nil, nil, ErrHeader
374 }
375
376 var p parser
377 hdr := new(Header)
378
379
380 v7 := tr.blk.toV7()
381 hdr.Typeflag = v7.typeFlag()[0]
382 hdr.Name = p.parseString(v7.name())
383 hdr.Linkname = p.parseString(v7.linkName())
384 hdr.Size = p.parseNumeric(v7.size())
385 hdr.Mode = p.parseNumeric(v7.mode())
386 hdr.Uid = int(p.parseNumeric(v7.uid()))
387 hdr.Gid = int(p.parseNumeric(v7.gid()))
388 hdr.ModTime = time.Unix(p.parseNumeric(v7.modTime()), 0)
389
390
391 if format > formatV7 {
392 ustar := tr.blk.toUSTAR()
393 hdr.Uname = p.parseString(ustar.userName())
394 hdr.Gname = p.parseString(ustar.groupName())
395 hdr.Devmajor = p.parseNumeric(ustar.devMajor())
396 hdr.Devminor = p.parseNumeric(ustar.devMinor())
397
398 var prefix string
399 switch {
400 case format.has(FormatUSTAR | FormatPAX):
401 hdr.Format = format
402 ustar := tr.blk.toUSTAR()
403 prefix = p.parseString(ustar.prefix())
404
405
406
407 notASCII := func(r rune) bool { return r >= 0x80 }
408 if bytes.IndexFunc(tr.blk[:], notASCII) >= 0 {
409 hdr.Format = FormatUnknown
410 }
411 nul := func(b []byte) bool { return int(b[len(b)-1]) == 0 }
412 if !(nul(v7.size()) && nul(v7.mode()) && nul(v7.uid()) && nul(v7.gid()) &&
413 nul(v7.modTime()) && nul(ustar.devMajor()) && nul(ustar.devMinor())) {
414 hdr.Format = FormatUnknown
415 }
416 case format.has(formatSTAR):
417 star := tr.blk.toSTAR()
418 prefix = p.parseString(star.prefix())
419 hdr.AccessTime = time.Unix(p.parseNumeric(star.accessTime()), 0)
420 hdr.ChangeTime = time.Unix(p.parseNumeric(star.changeTime()), 0)
421 case format.has(FormatGNU):
422 hdr.Format = format
423 var p2 parser
424 gnu := tr.blk.toGNU()
425 if b := gnu.accessTime(); b[0] != 0 {
426 hdr.AccessTime = time.Unix(p2.parseNumeric(b), 0)
427 }
428 if b := gnu.changeTime(); b[0] != 0 {
429 hdr.ChangeTime = time.Unix(p2.parseNumeric(b), 0)
430 }
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453 if p2.err != nil {
454 hdr.AccessTime, hdr.ChangeTime = time.Time{}, time.Time{}
455 ustar := tr.blk.toUSTAR()
456 if s := p.parseString(ustar.prefix()); isASCII(s) {
457 prefix = s
458 }
459 hdr.Format = FormatUnknown
460 }
461 }
462 if len(prefix) > 0 {
463 hdr.Name = prefix + "/" + hdr.Name
464 }
465 }
466 return hdr, &tr.blk, p.err
467 }
468
469
470
471
472
473
474
475
476
477 func (tr *Reader) readOldGNUSparseMap(hdr *Header, blk *block) (sparseDatas, error) {
478
479
480
481 if blk.getFormat() != FormatGNU {
482 return nil, ErrHeader
483 }
484 hdr.Format.mayOnlyBe(FormatGNU)
485
486 var p parser
487 hdr.Size = p.parseNumeric(blk.toGNU().realSize())
488 if p.err != nil {
489 return nil, p.err
490 }
491 s := blk.toGNU().sparse()
492 spd := make(sparseDatas, 0, s.maxEntries())
493 totalSize := len(s)
494 for totalSize < maxSpecialFileSize {
495 for i := 0; i < s.maxEntries(); i++ {
496
497 if s.entry(i).offset()[0] == 0x00 {
498 break
499 }
500 offset := p.parseNumeric(s.entry(i).offset())
501 length := p.parseNumeric(s.entry(i).length())
502 if p.err != nil {
503 return nil, p.err
504 }
505 var err error
506 spd, err = appendSparseEntry(spd, sparseEntry{Offset: offset, Length: length})
507 if err != nil {
508 return nil, err
509 }
510 }
511
512 if s.isExtended()[0] > 0 {
513
514 if _, err := mustReadFull(tr.r, blk[:]); err != nil {
515 return nil, err
516 }
517 s = blk.toSparse()
518 totalSize += len(s)
519 continue
520 }
521 return spd, nil
522 }
523 return nil, errSparseTooLong
524 }
525
526
527
528
529
530
531
532
533
534
535
536 func readGNUSparseMap1x0(r io.Reader) (sparseDatas, error) {
537 var (
538 cntNewline int64
539 buf bytes.Buffer
540 blk block
541 totalSize int
542 )
543
544
545
546 feedTokens := func(n int64) error {
547 for cntNewline < n {
548 totalSize += len(blk)
549 if totalSize > maxSpecialFileSize {
550 return errSparseTooLong
551 }
552 if _, err := mustReadFull(r, blk[:]); err != nil {
553 return err
554 }
555 buf.Write(blk[:])
556 for _, c := range blk {
557 if c == '\n' {
558 cntNewline++
559 }
560 }
561 }
562 return nil
563 }
564
565
566
567 nextToken := func() string {
568 cntNewline--
569 tok, _ := buf.ReadString('\n')
570 return strings.TrimRight(tok, "\n")
571 }
572
573
574
575 if err := feedTokens(1); err != nil {
576 return nil, err
577 }
578 numEntries, err := strconv.ParseInt(nextToken(), 10, 0)
579 if err != nil || numEntries < 0 || int(2*numEntries) < int(numEntries) {
580 return nil, ErrHeader
581 }
582
583
584
585
586 if err := feedTokens(2 * numEntries); err != nil {
587 return nil, err
588 }
589 spd := make(sparseDatas, 0, numEntries)
590 for i := int64(0); i < numEntries; i++ {
591 offset, err1 := strconv.ParseInt(nextToken(), 10, 64)
592 length, err2 := strconv.ParseInt(nextToken(), 10, 64)
593 if err1 != nil || err2 != nil {
594 return nil, ErrHeader
595 }
596 spd, err = appendSparseEntry(spd, sparseEntry{Offset: offset, Length: length})
597 if err != nil {
598 return nil, err
599 }
600 }
601 return spd, nil
602 }
603
604
605
606 func readGNUSparseMap0x1(paxHdrs map[string]string) (sparseDatas, error) {
607
608
609 numEntriesStr := paxHdrs[paxGNUSparseNumBlocks]
610 numEntries, err := strconv.ParseInt(numEntriesStr, 10, 0)
611 if err != nil || numEntries < 0 || int(2*numEntries) < int(numEntries) {
612 return nil, ErrHeader
613 }
614
615
616 sparseMap := strings.Split(paxHdrs[paxGNUSparseMap], ",")
617 if len(sparseMap) == 1 && sparseMap[0] == "" {
618 sparseMap = sparseMap[:0]
619 }
620 if int64(len(sparseMap)) != 2*numEntries {
621 return nil, ErrHeader
622 }
623
624
625
626 spd := make(sparseDatas, 0, numEntries)
627 for len(sparseMap) >= 2 {
628 offset, err1 := strconv.ParseInt(sparseMap[0], 10, 64)
629 length, err2 := strconv.ParseInt(sparseMap[1], 10, 64)
630 if err1 != nil || err2 != nil {
631 return nil, ErrHeader
632 }
633 spd, err = appendSparseEntry(spd, sparseEntry{Offset: offset, Length: length})
634 if err != nil {
635 return nil, err
636 }
637 sparseMap = sparseMap[2:]
638 }
639 return spd, nil
640 }
641
642 func appendSparseEntry(spd sparseDatas, ent sparseEntry) (sparseDatas, error) {
643 if len(spd) >= maxSparseFileEntries {
644 return nil, errSparseTooLong
645 }
646 return append(spd, ent), nil
647 }
648
649
650
651
652
653
654
655
656
657
658
659 func (tr *Reader) Read(b []byte) (int, error) {
660 if tr.err != nil {
661 return 0, tr.err
662 }
663 n, err := tr.curr.Read(b)
664 if err != nil && err != io.EOF {
665 tr.err = err
666 }
667 return n, err
668 }
669
670
671
672
673
674
675
676
677
678
679
680 func (tr *Reader) writeTo(w io.Writer) (int64, error) {
681 if tr.err != nil {
682 return 0, tr.err
683 }
684 n, err := tr.curr.WriteTo(w)
685 if err != nil {
686 tr.err = err
687 }
688 return n, err
689 }
690
691
692 type regFileReader struct {
693 r io.Reader
694 nb int64
695 }
696
697 func (fr *regFileReader) Read(b []byte) (n int, err error) {
698 if int64(len(b)) > fr.nb {
699 b = b[:fr.nb]
700 }
701 if len(b) > 0 {
702 n, err = fr.r.Read(b)
703 fr.nb -= int64(n)
704 }
705 switch {
706 case err == io.EOF && fr.nb > 0:
707 return n, io.ErrUnexpectedEOF
708 case err == nil && fr.nb == 0:
709 return n, io.EOF
710 default:
711 return n, err
712 }
713 }
714
715 func (fr *regFileReader) WriteTo(w io.Writer) (int64, error) {
716 return io.Copy(w, struct{ io.Reader }{fr})
717 }
718
719
720 func (fr regFileReader) logicalRemaining() int64 {
721 return fr.nb
722 }
723
724
725 func (fr regFileReader) physicalRemaining() int64 {
726 return fr.nb
727 }
728
729
730 type sparseFileReader struct {
731 fr fileReader
732 sp sparseHoles
733 pos int64
734 }
735
736 func (sr *sparseFileReader) Read(b []byte) (n int, err error) {
737 finished := int64(len(b)) >= sr.logicalRemaining()
738 if finished {
739 b = b[:sr.logicalRemaining()]
740 }
741
742 b0 := b
743 endPos := sr.pos + int64(len(b))
744 for endPos > sr.pos && err == nil {
745 var nf int
746 holeStart, holeEnd := sr.sp[0].Offset, sr.sp[0].endOffset()
747 if sr.pos < holeStart {
748 bf := b[:min(int64(len(b)), holeStart-sr.pos)]
749 nf, err = tryReadFull(sr.fr, bf)
750 } else {
751 bf := b[:min(int64(len(b)), holeEnd-sr.pos)]
752 nf, err = tryReadFull(zeroReader{}, bf)
753 }
754 b = b[nf:]
755 sr.pos += int64(nf)
756 if sr.pos >= holeEnd && len(sr.sp) > 1 {
757 sr.sp = sr.sp[1:]
758 }
759 }
760
761 n = len(b0) - len(b)
762 switch {
763 case err == io.EOF:
764 return n, errMissData
765 case err != nil:
766 return n, err
767 case sr.logicalRemaining() == 0 && sr.physicalRemaining() > 0:
768 return n, errUnrefData
769 case finished:
770 return n, io.EOF
771 default:
772 return n, nil
773 }
774 }
775
776 func (sr *sparseFileReader) WriteTo(w io.Writer) (n int64, err error) {
777 ws, ok := w.(io.WriteSeeker)
778 if ok {
779 if _, err := ws.Seek(0, io.SeekCurrent); err != nil {
780 ok = false
781 }
782 }
783 if !ok {
784 return io.Copy(w, struct{ io.Reader }{sr})
785 }
786
787 var writeLastByte bool
788 pos0 := sr.pos
789 for sr.logicalRemaining() > 0 && !writeLastByte && err == nil {
790 var nf int64
791 holeStart, holeEnd := sr.sp[0].Offset, sr.sp[0].endOffset()
792 if sr.pos < holeStart {
793 nf = holeStart - sr.pos
794 nf, err = io.CopyN(ws, sr.fr, nf)
795 } else {
796 nf = holeEnd - sr.pos
797 if sr.physicalRemaining() == 0 {
798 writeLastByte = true
799 nf--
800 }
801 _, err = ws.Seek(nf, io.SeekCurrent)
802 }
803 sr.pos += nf
804 if sr.pos >= holeEnd && len(sr.sp) > 1 {
805 sr.sp = sr.sp[1:]
806 }
807 }
808
809
810
811 if writeLastByte && err == nil {
812 _, err = ws.Write([]byte{0})
813 sr.pos++
814 }
815
816 n = sr.pos - pos0
817 switch {
818 case err == io.EOF:
819 return n, errMissData
820 case err != nil:
821 return n, err
822 case sr.logicalRemaining() == 0 && sr.physicalRemaining() > 0:
823 return n, errUnrefData
824 default:
825 return n, nil
826 }
827 }
828
829 func (sr sparseFileReader) logicalRemaining() int64 {
830 return sr.sp[len(sr.sp)-1].endOffset() - sr.pos
831 }
832 func (sr sparseFileReader) physicalRemaining() int64 {
833 return sr.fr.physicalRemaining()
834 }
835
836 type zeroReader struct{}
837
838 func (zeroReader) Read(b []byte) (int, error) {
839 clear(b)
840 return len(b), nil
841 }
842
843
844
845 func mustReadFull(r io.Reader, b []byte) (int, error) {
846 n, err := tryReadFull(r, b)
847 if err == io.EOF {
848 err = io.ErrUnexpectedEOF
849 }
850 return n, err
851 }
852
853
854
855 func tryReadFull(r io.Reader, b []byte) (n int, err error) {
856 for len(b) > n && err == nil {
857 var nn int
858 nn, err = r.Read(b[n:])
859 n += nn
860 }
861 if len(b) == n && err == io.EOF {
862 err = nil
863 }
864 return n, err
865 }
866
867
868
869 func readSpecialFile(r io.Reader) ([]byte, error) {
870 buf, err := io.ReadAll(io.LimitReader(r, maxSpecialFileSize+1))
871 if len(buf) > maxSpecialFileSize {
872 return nil, ErrFieldTooLong
873 }
874 return buf, err
875 }
876
877
878 func discard(r io.Reader, n int64) error {
879
880
881
882
883 var seekSkipped int64
884 if sr, ok := r.(io.Seeker); ok && n > 1 {
885
886
887
888
889 pos1, err := sr.Seek(0, io.SeekCurrent)
890 if pos1 >= 0 && err == nil {
891
892 pos2, err := sr.Seek(n-1, io.SeekCurrent)
893 if pos2 < 0 || err != nil {
894 return err
895 }
896 seekSkipped = pos2 - pos1
897 }
898 }
899
900 copySkipped, err := io.CopyN(io.Discard, r, n-seekSkipped)
901 if err == io.EOF && seekSkipped+copySkipped < n {
902 err = io.ErrUnexpectedEOF
903 }
904 return err
905 }
906
View as plain text