summaryrefslogtreecommitdiff
path: root/vendor/github.com/klauspost/compress/zstd/framedec.go
blob: e8cc9a2c223d8a6b7e4fe0bea27067195465ee32 (plain)
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
// Copyright 2019+ Klaus Post. All rights reserved.
// License information can be found in the LICENSE file.
// Based on work by Yann Collet, released under BSD License.

package zstd

import (
	"bytes"
	"encoding/hex"
	"errors"
	"hash"
	"io"
	"sync"

	"github.com/klauspost/compress/zstd/internal/xxhash"
)

type frameDec struct {
	o      decoderOptions
	crc    hash.Hash64
	offset int64

	WindowSize uint64

	// maxWindowSize is the maximum windows size to support.
	// should never be bigger than max-int.
	maxWindowSize uint64

	// In order queue of blocks being decoded.
	decoding chan *blockDec

	// Frame history passed between blocks
	history history

	rawInput byteBuffer

	// Byte buffer that can be reused for small input blocks.
	bBuf byteBuf

	FrameContentSize uint64
	frameDone        sync.WaitGroup

	DictionaryID  *uint32
	HasCheckSum   bool
	SingleSegment bool

	// asyncRunning indicates whether the async routine processes input on 'decoding'.
	asyncRunningMu sync.Mutex
	asyncRunning   bool
}

const (
	// The minimum Window_Size is 1 KB.
	MinWindowSize = 1 << 10
	MaxWindowSize = 1 << 29
)

var (
	frameMagic          = []byte{0x28, 0xb5, 0x2f, 0xfd}
	skippableFrameMagic = []byte{0x2a, 0x4d, 0x18}
)

func newFrameDec(o decoderOptions) *frameDec {
	d := frameDec{
		o:             o,
		maxWindowSize: MaxWindowSize,
	}
	if d.maxWindowSize > o.maxDecodedSize {
		d.maxWindowSize = o.maxDecodedSize
	}
	return &d
}

// reset will read the frame header and prepare for block decoding.
// If nothing can be read from the input, io.EOF will be returned.
// Any other error indicated that the stream contained data, but
// there was a problem.
func (d *frameDec) reset(br byteBuffer) error {
	d.HasCheckSum = false
	d.WindowSize = 0
	var signature [4]byte
	for {
		var err error
		// Check if we can read more...
		b, err := br.readSmall(1)
		switch err {
		case io.EOF, io.ErrUnexpectedEOF:
			return io.EOF
		default:
			return err
		case nil:
			signature[0] = b[0]
		}
		// Read the rest, don't allow io.ErrUnexpectedEOF
		b, err = br.readSmall(3)
		switch err {
		case io.EOF:
			return io.EOF
		default:
			return err
		case nil:
			copy(signature[1:], b)
		}

		if !bytes.Equal(signature[1:4], skippableFrameMagic) || signature[0]&0xf0 != 0x50 {
			if debugDecoder {
				println("Not skippable", hex.EncodeToString(signature[:]), hex.EncodeToString(skippableFrameMagic))
			}
			// Break if not skippable frame.
			break
		}
		// Read size to skip
		b, err = br.readSmall(4)
		if err != nil {
			if debugDecoder {
				println("Reading Frame Size", err)
			}
			return err
		}
		n := uint32(b[0]) | (uint32(b[1]) << 8) | (uint32(b[2]) << 16) | (uint32(b[3]) << 24)
		println("Skipping frame with", n, "bytes.")
		err = br.skipN(int(n))
		if err != nil {
			if debugDecoder {
				println("Reading discarded frame", err)
			}
			return err
		}
	}
	if !bytes.Equal(signature[:], frameMagic) {
		if debugDecoder {
			println("Got magic numbers: ", signature, "want:", frameMagic)
		}
		return ErrMagicMismatch
	}

	// Read Frame_Header_Descriptor
	fhd, err := br.readByte()
	if err != nil {
		if debugDecoder {
			println("Reading Frame_Header_Descriptor", err)
		}
		return err
	}
	d.SingleSegment = fhd&(1<<5) != 0

	if fhd&(1<<3) != 0 {
		return errors.New("reserved bit set on frame header")
	}

	// Read Window_Descriptor
	// https://github.com/facebook/zstd/blob/dev/doc/zstd_compression_format.md#window_descriptor
	d.WindowSize = 0
	if !d.SingleSegment {
		wd, err := br.readByte()
		if err != nil {
			if debugDecoder {
				println("Reading Window_Descriptor", err)
			}
			return err
		}
		printf("raw: %x, mantissa: %d, exponent: %d\n", wd, wd&7, wd>>3)
		windowLog := 10 + (wd >> 3)
		windowBase := uint64(1) << windowLog
		windowAdd := (windowBase / 8) * uint64(wd&0x7)
		d.WindowSize = windowBase + windowAdd
	}

	// Read Dictionary_ID
	// https://github.com/facebook/zstd/blob/dev/doc/zstd_compression_format.md#dictionary_id
	d.DictionaryID = nil
	if size := fhd & 3; size != 0 {
		if size == 3 {
			size = 4
		}

		b, err := br.readSmall(int(size))
		if err != nil {
			println("Reading Dictionary_ID", err)
			return err
		}
		var id uint32
		switch size {
		case 1:
			id = uint32(b[0])
		case 2:
			id = uint32(b[0]) | (uint32(b[1]) << 8)
		case 4:
			id = uint32(b[0]) | (uint32(b[1]) << 8) | (uint32(b[2]) << 16) | (uint32(b[3]) << 24)
		}
		if debugDecoder {
			println("Dict size", size, "ID:", id)
		}
		if id > 0 {
			// ID 0 means "sorry, no dictionary anyway".
			// https://github.com/facebook/zstd/blob/dev/doc/zstd_compression_format.md#dictionary-format
			d.DictionaryID = &id
		}
	}

	// Read Frame_Content_Size
	// https://github.com/facebook/zstd/blob/dev/doc/zstd_compression_format.md#frame_content_size
	var fcsSize int
	v := fhd >> 6
	switch v {
	case 0:
		if d.SingleSegment {
			fcsSize = 1
		}
	default:
		fcsSize = 1 << v
	}
	d.FrameContentSize = 0
	if fcsSize > 0 {
		b, err := br.readSmall(fcsSize)
		if err != nil {
			println("Reading Frame content", err)
			return err
		}
		switch fcsSize {
		case 1:
			d.FrameContentSize = uint64(b[0])
		case 2:
			// When FCS_Field_Size is 2, the offset of 256 is added.
			d.FrameContentSize = uint64(b[0]) | (uint64(b[1]) << 8) + 256
		case 4:
			d.FrameContentSize = uint64(b[0]) | (uint64(b[1]) << 8) | (uint64(b[2]) << 16) | (uint64(b[3]) << 24)
		case 8:
			d1 := uint32(b[0]) | (uint32(b[1]) << 8) | (uint32(b[2]) << 16) | (uint32(b[3]) << 24)
			d2 := uint32(b[4]) | (uint32(b[5]) << 8) | (uint32(b[6]) << 16) | (uint32(b[7]) << 24)
			d.FrameContentSize = uint64(d1) | (uint64(d2) << 32)
		}
		if debugDecoder {
			println("field size bits:", v, "fcsSize:", fcsSize, "FrameContentSize:", d.FrameContentSize, hex.EncodeToString(b[:fcsSize]), "singleseg:", d.SingleSegment, "window:", d.WindowSize)
		}
	}
	// Move this to shared.
	d.HasCheckSum = fhd&(1<<2) != 0
	if d.HasCheckSum {
		if d.crc == nil {
			d.crc = xxhash.New()
		}
		d.crc.Reset()
	}

	if d.WindowSize == 0 && d.SingleSegment {
		// We may not need window in this case.
		d.WindowSize = d.FrameContentSize
		if d.WindowSize < MinWindowSize {
			d.WindowSize = MinWindowSize
		}
	}

	if d.WindowSize > d.maxWindowSize {
		printf("window size %d > max %d\n", d.WindowSize, d.maxWindowSize)
		return ErrWindowSizeExceeded
	}
	// The minimum Window_Size is 1 KB.
	if d.WindowSize < MinWindowSize {
		println("got window size: ", d.WindowSize)
		return ErrWindowSizeTooSmall
	}
	d.history.windowSize = int(d.WindowSize)
	if d.o.lowMem && d.history.windowSize < maxBlockSize {
		d.history.maxSize = d.history.windowSize * 2
	} else {
		d.history.maxSize = d.history.windowSize + maxBlockSize
	}
	// history contains input - maybe we do something
	d.rawInput = br
	return nil
}

// next will start decoding the next block from stream.
func (d *frameDec) next(block *blockDec) error {
	if debugDecoder {
		printf("decoding new block %p:%p", block, block.data)
	}
	err := block.reset(d.rawInput, d.WindowSize)
	if err != nil {
		println("block error:", err)
		// Signal the frame decoder we have a problem.
		d.sendErr(block, err)
		return err
	}
	block.input <- struct{}{}
	if debugDecoder {
		println("next block:", block)
	}
	d.asyncRunningMu.Lock()
	defer d.asyncRunningMu.Unlock()
	if !d.asyncRunning {
		return nil
	}
	if block.Last {
		// We indicate the frame is done by sending io.EOF
		d.decoding <- block
		return io.EOF
	}
	d.decoding <- block
	return nil
}

// sendEOF will queue an error block on the frame.
// This will cause the frame decoder to return when it encounters the block.
// Returns true if the decoder was added.
func (d *frameDec) sendErr(block *blockDec, err error) bool {
	d.asyncRunningMu.Lock()
	defer d.asyncRunningMu.Unlock()
	if !d.asyncRunning {
		return false
	}

	println("sending error", err.Error())
	block.sendErr(err)
	d.decoding <- block
	return true
}

// checkCRC will check the checksum if the frame has one.
// Will return ErrCRCMismatch if crc check failed, otherwise nil.
func (d *frameDec) checkCRC() error {
	if !d.HasCheckSum {
		return nil
	}
	var tmp [4]byte
	got := d.crc.Sum64()
	// Flip to match file order.
	tmp[0] = byte(got >> 0)
	tmp[1] = byte(got >> 8)
	tmp[2] = byte(got >> 16)
	tmp[3] = byte(got >> 24)

	// We can overwrite upper tmp now
	want, err := d.rawInput.readSmall(4)
	if err != nil {
		println("CRC missing?", err)
		return err
	}

	if !bytes.Equal(tmp[:], want) {
		if debugDecoder {
			println("CRC Check Failed:", tmp[:], "!=", want)
		}
		return ErrCRCMismatch
	}
	if debugDecoder {
		println("CRC ok", tmp[:])
	}
	return nil
}

func (d *frameDec) initAsync() {
	if !d.o.lowMem && !d.SingleSegment {
		// set max extra size history to 10MB.
		d.history.maxSize = d.history.windowSize + maxBlockSize*5
	}
	// re-alloc if more than one extra block size.
	if d.o.lowMem && cap(d.history.b) > d.history.maxSize+maxBlockSize {
		d.history.b = make([]byte, 0, d.history.maxSize)
	}
	if cap(d.history.b) < d.history.maxSize {
		d.history.b = make([]byte, 0, d.history.maxSize)
	}
	if cap(d.decoding) < d.o.concurrent {
		d.decoding = make(chan *blockDec, d.o.concurrent)
	}
	if debugDecoder {
		h := d.history
		printf("history init. len: %d, cap: %d", len(h.b), cap(h.b))
	}
	d.asyncRunningMu.Lock()
	d.asyncRunning = true
	d.asyncRunningMu.Unlock()
}

// startDecoder will start decoding blocks and write them to the writer.
// The decoder will stop as soon as an error occurs or at end of frame.
// When the frame has finished decoding the *bufio.Reader
// containing the remaining input will be sent on frameDec.frameDone.
func (d *frameDec) startDecoder(output chan decodeOutput) {
	written := int64(0)

	defer func() {
		d.asyncRunningMu.Lock()
		d.asyncRunning = false
		d.asyncRunningMu.Unlock()

		// Drain the currently decoding.
		d.history.error = true
	flushdone:
		for {
			select {
			case b := <-d.decoding:
				b.history <- &d.history
				output <- <-b.result
			default:
				break flushdone
			}
		}
		println("frame decoder done, signalling done")
		d.frameDone.Done()
	}()
	// Get decoder for first block.
	block := <-d.decoding
	block.history <- &d.history
	for {
		var next *blockDec
		// Get result
		r := <-block.result
		if r.err != nil {
			println("Result contained error", r.err)
			output <- r
			return
		}
		if debugDecoder {
			println("got result, from ", d.offset, "to", d.offset+int64(len(r.b)))
			d.offset += int64(len(r.b))
		}
		if !block.Last {
			// Send history to next block
			select {
			case next = <-d.decoding:
				if debugDecoder {
					println("Sending ", len(d.history.b), "bytes as history")
				}
				next.history <- &d.history
			default:
				// Wait until we have sent the block, so
				// other decoders can potentially get the decoder.
				next = nil
			}
		}

		// Add checksum, async to decoding.
		if d.HasCheckSum {
			n, err := d.crc.Write(r.b)
			if err != nil {
				r.err = err
				if n != len(r.b) {
					r.err = io.ErrShortWrite
				}
				output <- r
				return
			}
		}
		written += int64(len(r.b))
		if d.SingleSegment && uint64(written) > d.FrameContentSize {
			println("runDecoder: single segment and", uint64(written), ">", d.FrameContentSize)
			r.err = ErrFrameSizeExceeded
			output <- r
			return
		}
		if block.Last {
			r.err = d.checkCRC()
			output <- r
			return
		}
		output <- r
		if next == nil {
			// There was no decoder available, we wait for one now that we have sent to the writer.
			if debugDecoder {
				println("Sending ", len(d.history.b), " bytes as history")
			}
			next = <-d.decoding
			next.history <- &d.history
		}
		block = next
	}
}

// runDecoder will create a sync decoder that will decode a block of data.
func (d *frameDec) runDecoder(dst []byte, dec *blockDec) ([]byte, error) {
	saved := d.history.b

	// We use the history for output to avoid copying it.
	d.history.b = dst
	// Store input length, so we only check new data.
	crcStart := len(dst)
	var err error
	for {
		err = dec.reset(d.rawInput, d.WindowSize)
		if err != nil {
			break
		}
		if debugDecoder {
			println("next block:", dec)
		}
		err = dec.decodeBuf(&d.history)
		if err != nil || dec.Last {
			break
		}
		if uint64(len(d.history.b)) > d.o.maxDecodedSize {
			err = ErrDecoderSizeExceeded
			break
		}
		if d.SingleSegment && uint64(len(d.history.b)) > d.o.maxDecodedSize {
			println("runDecoder: single segment and", uint64(len(d.history.b)), ">", d.o.maxDecodedSize)
			err = ErrFrameSizeExceeded
			break
		}
	}
	dst = d.history.b
	if err == nil {
		if d.HasCheckSum {
			var n int
			n, err = d.crc.Write(dst[crcStart:])
			if err == nil {
				if n != len(dst)-crcStart {
					err = io.ErrShortWrite
				} else {
					err = d.checkCRC()
				}
			}
		}
	}
	d.history.b = saved
	return dst, err
}