summaryrefslogtreecommitdiff
path: root/vendor/github.com/klauspost/compress/zstd/decoder.go
blob: 62fd373240aa97aebfbd8e75e79aee8112d1bb84 (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
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
// 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"
	"errors"
	"io"
	"sync"
)

// Decoder provides decoding of zstandard streams.
// The decoder has been designed to operate without allocations after a warmup.
// This means that you should store the decoder for best performance.
// To re-use a stream decoder, use the Reset(r io.Reader) error to switch to another stream.
// A decoder can safely be re-used even if the previous stream failed.
// To release the resources, you must call the Close() function on a decoder.
type Decoder struct {
	o decoderOptions

	// Unreferenced decoders, ready for use.
	decoders chan *blockDec

	// Streams ready to be decoded.
	stream chan decodeStream

	// Current read position used for Reader functionality.
	current decoderState

	// Custom dictionaries.
	// Always uses copies.
	dicts map[uint32]dict

	// streamWg is the waitgroup for all streams
	streamWg sync.WaitGroup
}

// decoderState is used for maintaining state when the decoder
// is used for streaming.
type decoderState struct {
	// current block being written to stream.
	decodeOutput

	// output in order to be written to stream.
	output chan decodeOutput

	// cancel remaining output.
	cancel chan struct{}

	flushed bool
}

var (
	// Check the interfaces we want to support.
	_ = io.WriterTo(&Decoder{})
	_ = io.Reader(&Decoder{})
)

// NewReader creates a new decoder.
// A nil Reader can be provided in which case Reset can be used to start a decode.
//
// A Decoder can be used in two modes:
//
// 1) As a stream, or
// 2) For stateless decoding using DecodeAll.
//
// Only a single stream can be decoded concurrently, but the same decoder
// can run multiple concurrent stateless decodes. It is even possible to
// use stateless decodes while a stream is being decoded.
//
// The Reset function can be used to initiate a new stream, which is will considerably
// reduce the allocations normally caused by NewReader.
func NewReader(r io.Reader, opts ...DOption) (*Decoder, error) {
	initPredefined()
	var d Decoder
	d.o.setDefault()
	for _, o := range opts {
		err := o(&d.o)
		if err != nil {
			return nil, err
		}
	}
	d.current.output = make(chan decodeOutput, d.o.concurrent)
	d.current.flushed = true

	if r == nil {
		d.current.err = ErrDecoderNilInput
	}

	// Transfer option dicts.
	d.dicts = make(map[uint32]dict, len(d.o.dicts))
	for _, dc := range d.o.dicts {
		d.dicts[dc.id] = dc
	}
	d.o.dicts = nil

	// Create decoders
	d.decoders = make(chan *blockDec, d.o.concurrent)
	for i := 0; i < d.o.concurrent; i++ {
		dec := newBlockDec(d.o.lowMem)
		dec.localFrame = newFrameDec(d.o)
		d.decoders <- dec
	}

	if r == nil {
		return &d, nil
	}
	return &d, d.Reset(r)
}

// Read bytes from the decompressed stream into p.
// Returns the number of bytes written and any error that occurred.
// When the stream is done, io.EOF will be returned.
func (d *Decoder) Read(p []byte) (int, error) {
	if d.stream == nil {
		return 0, ErrDecoderNilInput
	}
	var n int
	for {
		if len(d.current.b) > 0 {
			filled := copy(p, d.current.b)
			p = p[filled:]
			d.current.b = d.current.b[filled:]
			n += filled
		}
		if len(p) == 0 {
			break
		}
		if len(d.current.b) == 0 {
			// We have an error and no more data
			if d.current.err != nil {
				break
			}
			if !d.nextBlock(n == 0) {
				return n, nil
			}
		}
	}
	if len(d.current.b) > 0 {
		if debug {
			println("returning", n, "still bytes left:", len(d.current.b))
		}
		// Only return error at end of block
		return n, nil
	}
	if d.current.err != nil {
		d.drainOutput()
	}
	if debug {
		println("returning", n, d.current.err, len(d.decoders))
	}
	return n, d.current.err
}

// Reset will reset the decoder the supplied stream after the current has finished processing.
// Note that this functionality cannot be used after Close has been called.
// Reset can be called with a nil reader to release references to the previous reader.
// After being called with a nil reader, no other operations than Reset or DecodeAll or Close
// should be used.
func (d *Decoder) Reset(r io.Reader) error {
	if d.current.err == ErrDecoderClosed {
		return d.current.err
	}

	d.drainOutput()

	if r == nil {
		d.current.err = ErrDecoderNilInput
		d.current.flushed = true
		return nil
	}

	if d.stream == nil {
		d.stream = make(chan decodeStream, 1)
		d.streamWg.Add(1)
		go d.startStreamDecoder(d.stream)
	}

	// If bytes buffer and < 1MB, do sync decoding anyway.
	if bb, ok := r.(*bytes.Buffer); ok && bb.Len() < 1<<20 {
		if debug {
			println("*bytes.Buffer detected, doing sync decode, len:", bb.Len())
		}
		b := bb.Bytes()
		var dst []byte
		if cap(d.current.b) > 0 {
			dst = d.current.b
		}

		dst, err := d.DecodeAll(b, dst[:0])
		if err == nil {
			err = io.EOF
		}
		d.current.b = dst
		d.current.err = err
		d.current.flushed = true
		if debug {
			println("sync decode to", len(dst), "bytes, err:", err)
		}
		return nil
	}

	// Remove current block.
	d.current.decodeOutput = decodeOutput{}
	d.current.err = nil
	d.current.cancel = make(chan struct{})
	d.current.flushed = false
	d.current.d = nil

	d.stream <- decodeStream{
		r:      r,
		output: d.current.output,
		cancel: d.current.cancel,
	}
	return nil
}

// drainOutput will drain the output until errEndOfStream is sent.
func (d *Decoder) drainOutput() {
	if d.current.cancel != nil {
		println("cancelling current")
		close(d.current.cancel)
		d.current.cancel = nil
	}
	if d.current.d != nil {
		if debug {
			printf("re-adding current decoder %p, decoders: %d", d.current.d, len(d.decoders))
		}
		d.decoders <- d.current.d
		d.current.d = nil
		d.current.b = nil
	}
	if d.current.output == nil || d.current.flushed {
		println("current already flushed")
		return
	}
	for {
		select {
		case v := <-d.current.output:
			if v.d != nil {
				if debug {
					printf("re-adding decoder %p", v.d)
				}
				d.decoders <- v.d
			}
			if v.err == errEndOfStream {
				println("current flushed")
				d.current.flushed = true
				return
			}
		}
	}
}

// WriteTo writes data to w until there's no more data to write or when an error occurs.
// The return value n is the number of bytes written.
// Any error encountered during the write is also returned.
func (d *Decoder) WriteTo(w io.Writer) (int64, error) {
	if d.stream == nil {
		return 0, ErrDecoderNilInput
	}
	var n int64
	for {
		if len(d.current.b) > 0 {
			n2, err2 := w.Write(d.current.b)
			n += int64(n2)
			if err2 != nil && d.current.err == nil {
				d.current.err = err2
				break
			}
		}
		if d.current.err != nil {
			break
		}
		d.nextBlock(true)
	}
	err := d.current.err
	if err != nil {
		d.drainOutput()
	}
	if err == io.EOF {
		err = nil
	}
	return n, err
}

// DecodeAll allows stateless decoding of a blob of bytes.
// Output will be appended to dst, so if the destination size is known
// you can pre-allocate the destination slice to avoid allocations.
// DecodeAll can be used concurrently.
// The Decoder concurrency limits will be respected.
func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) {
	if d.current.err == ErrDecoderClosed {
		return dst, ErrDecoderClosed
	}

	// Grab a block decoder and frame decoder.
	block := <-d.decoders
	frame := block.localFrame
	defer func() {
		if debug {
			printf("re-adding decoder: %p", block)
		}
		frame.rawInput = nil
		frame.bBuf = nil
		d.decoders <- block
	}()
	frame.bBuf = input

	for {
		frame.history.reset()
		err := frame.reset(&frame.bBuf)
		if err == io.EOF {
			if debug {
				println("frame reset return EOF")
			}
			return dst, nil
		}
		if frame.DictionaryID != nil {
			dict, ok := d.dicts[*frame.DictionaryID]
			if !ok {
				return nil, ErrUnknownDictionary
			}
			frame.history.setDict(&dict)
		}
		if err != nil {
			return dst, err
		}
		if frame.FrameContentSize > d.o.maxDecodedSize-uint64(len(dst)) {
			return dst, ErrDecoderSizeExceeded
		}
		if frame.FrameContentSize > 0 && frame.FrameContentSize < 1<<30 {
			// Never preallocate moe than 1 GB up front.
			if cap(dst)-len(dst) < int(frame.FrameContentSize) {
				dst2 := make([]byte, len(dst), len(dst)+int(frame.FrameContentSize))
				copy(dst2, dst)
				dst = dst2
			}
		}
		if cap(dst) == 0 {
			// Allocate len(input) * 2 by default if nothing is provided
			// and we didn't get frame content size.
			size := len(input) * 2
			// Cap to 1 MB.
			if size > 1<<20 {
				size = 1 << 20
			}
			if uint64(size) > d.o.maxDecodedSize {
				size = int(d.o.maxDecodedSize)
			}
			dst = make([]byte, 0, size)
		}

		dst, err = frame.runDecoder(dst, block)
		if err != nil {
			return dst, err
		}
		if len(frame.bBuf) == 0 {
			if debug {
				println("frame dbuf empty")
			}
			break
		}
	}
	return dst, nil
}

// nextBlock returns the next block.
// If an error occurs d.err will be set.
// Optionally the function can block for new output.
// If non-blocking mode is used the returned boolean will be false
// if no data was available without blocking.
func (d *Decoder) nextBlock(blocking bool) (ok bool) {
	if d.current.d != nil {
		if debug {
			printf("re-adding current decoder %p", d.current.d)
		}
		d.decoders <- d.current.d
		d.current.d = nil
	}
	if d.current.err != nil {
		// Keep error state.
		return blocking
	}

	if blocking {
		d.current.decodeOutput = <-d.current.output
	} else {
		select {
		case d.current.decodeOutput = <-d.current.output:
		default:
			return false
		}
	}
	if debug {
		println("got", len(d.current.b), "bytes, error:", d.current.err)
	}
	return true
}

// Close will release all resources.
// It is NOT possible to reuse the decoder after this.
func (d *Decoder) Close() {
	if d.current.err == ErrDecoderClosed {
		return
	}
	d.drainOutput()
	if d.stream != nil {
		close(d.stream)
		d.streamWg.Wait()
		d.stream = nil
	}
	if d.decoders != nil {
		close(d.decoders)
		for dec := range d.decoders {
			dec.Close()
		}
		d.decoders = nil
	}
	if d.current.d != nil {
		d.current.d.Close()
		d.current.d = nil
	}
	d.current.err = ErrDecoderClosed
}

// IOReadCloser returns the decoder as an io.ReadCloser for convenience.
// Any changes to the decoder will be reflected, so the returned ReadCloser
// can be reused along with the decoder.
// io.WriterTo is also supported by the returned ReadCloser.
func (d *Decoder) IOReadCloser() io.ReadCloser {
	return closeWrapper{d: d}
}

// closeWrapper wraps a function call as a closer.
type closeWrapper struct {
	d *Decoder
}

// WriteTo forwards WriteTo calls to the decoder.
func (c closeWrapper) WriteTo(w io.Writer) (n int64, err error) {
	return c.d.WriteTo(w)
}

// Read forwards read calls to the decoder.
func (c closeWrapper) Read(p []byte) (n int, err error) {
	return c.d.Read(p)
}

// Close closes the decoder.
func (c closeWrapper) Close() error {
	c.d.Close()
	return nil
}

type decodeOutput struct {
	d   *blockDec
	b   []byte
	err error
}

type decodeStream struct {
	r io.Reader

	// Blocks ready to be written to output.
	output chan decodeOutput

	// cancel reading from the input
	cancel chan struct{}
}

// errEndOfStream indicates that everything from the stream was read.
var errEndOfStream = errors.New("end-of-stream")

// Create Decoder:
// Spawn n block decoders. These accept tasks to decode a block.
// Create goroutine that handles stream processing, this will send history to decoders as they are available.
// Decoders update the history as they decode.
// When a block is returned:
// 		a) history is sent to the next decoder,
// 		b) content written to CRC.
// 		c) return data to WRITER.
// 		d) wait for next block to return data.
// Once WRITTEN, the decoders reused by the writer frame decoder for re-use.
func (d *Decoder) startStreamDecoder(inStream chan decodeStream) {
	defer d.streamWg.Done()
	frame := newFrameDec(d.o)
	for stream := range inStream {
		if debug {
			println("got new stream")
		}
		br := readerWrapper{r: stream.r}
	decodeStream:
		for {
			frame.history.reset()
			err := frame.reset(&br)
			if debug && err != nil {
				println("Frame decoder returned", err)
			}
			if err == nil && frame.DictionaryID != nil {
				dict, ok := d.dicts[*frame.DictionaryID]
				if !ok {
					err = ErrUnknownDictionary
				} else {
					frame.history.setDict(&dict)
				}
			}
			if err != nil {
				stream.output <- decodeOutput{
					err: err,
				}
				break
			}
			if debug {
				println("starting frame decoder")
			}

			// This goroutine will forward history between frames.
			frame.frameDone.Add(1)
			frame.initAsync()

			go frame.startDecoder(stream.output)
		decodeFrame:
			// Go through all blocks of the frame.
			for {
				dec := <-d.decoders
				select {
				case <-stream.cancel:
					if !frame.sendErr(dec, io.EOF) {
						// To not let the decoder dangle, send it back.
						stream.output <- decodeOutput{d: dec}
					}
					break decodeStream
				default:
				}
				err := frame.next(dec)
				switch err {
				case io.EOF:
					// End of current frame, no error
					println("EOF on next block")
					break decodeFrame
				case nil:
					continue
				default:
					println("block decoder returned", err)
					break decodeStream
				}
			}
			// All blocks have started decoding, check if there are more frames.
			println("waiting for done")
			frame.frameDone.Wait()
			println("done waiting...")
		}
		frame.frameDone.Wait()
		println("Sending EOS")
		stream.output <- decodeOutput{err: errEndOfStream}
	}
}