Scott Baker | ed4efab | 2020-01-13 19:12:25 -0800 | [diff] [blame] | 1 | // Package zstd provides decompression of zstandard files. |
| 2 | // |
| 3 | // For advanced usage and examples, go to the README: https://github.com/klauspost/compress/tree/master/zstd#zstd |
| 4 | package zstd |
| 5 | |
| 6 | import ( |
| 7 | "errors" |
| 8 | "log" |
| 9 | "math/bits" |
| 10 | ) |
| 11 | |
| 12 | const debug = false |
| 13 | const debugSequences = false |
| 14 | const debugMatches = false |
| 15 | |
| 16 | // force encoder to use predefined tables. |
| 17 | const forcePreDef = false |
| 18 | |
| 19 | // zstdMinMatch is the minimum zstd match length. |
| 20 | const zstdMinMatch = 3 |
| 21 | |
| 22 | var ( |
| 23 | // ErrReservedBlockType is returned when a reserved block type is found. |
| 24 | // Typically this indicates wrong or corrupted input. |
| 25 | ErrReservedBlockType = errors.New("invalid input: reserved block type encountered") |
| 26 | |
| 27 | // ErrCompressedSizeTooBig is returned when a block is bigger than allowed. |
| 28 | // Typically this indicates wrong or corrupted input. |
| 29 | ErrCompressedSizeTooBig = errors.New("invalid input: compressed size too big") |
| 30 | |
| 31 | // ErrBlockTooSmall is returned when a block is too small to be decoded. |
| 32 | // Typically returned on invalid input. |
| 33 | ErrBlockTooSmall = errors.New("block too small") |
| 34 | |
| 35 | // ErrMagicMismatch is returned when a "magic" number isn't what is expected. |
| 36 | // Typically this indicates wrong or corrupted input. |
| 37 | ErrMagicMismatch = errors.New("invalid input: magic number mismatch") |
| 38 | |
| 39 | // ErrWindowSizeExceeded is returned when a reference exceeds the valid window size. |
| 40 | // Typically this indicates wrong or corrupted input. |
| 41 | ErrWindowSizeExceeded = errors.New("window size exceeded") |
| 42 | |
| 43 | // ErrWindowSizeTooSmall is returned when no window size is specified. |
| 44 | // Typically this indicates wrong or corrupted input. |
| 45 | ErrWindowSizeTooSmall = errors.New("invalid input: window size was too small") |
| 46 | |
| 47 | // ErrDecoderSizeExceeded is returned if decompressed size exceeds the configured limit. |
| 48 | ErrDecoderSizeExceeded = errors.New("decompressed size exceeds configured limit") |
| 49 | |
| 50 | // ErrUnknownDictionary is returned if the dictionary ID is unknown. |
| 51 | // For the time being dictionaries are not supported. |
| 52 | ErrUnknownDictionary = errors.New("unknown dictionary") |
| 53 | |
| 54 | // ErrFrameSizeExceeded is returned if the stated frame size is exceeded. |
| 55 | // This is only returned if SingleSegment is specified on the frame. |
| 56 | ErrFrameSizeExceeded = errors.New("frame size exceeded") |
| 57 | |
| 58 | // ErrCRCMismatch is returned if CRC mismatches. |
| 59 | ErrCRCMismatch = errors.New("CRC check failed") |
| 60 | |
| 61 | // ErrDecoderClosed will be returned if the Decoder was used after |
| 62 | // Close has been called. |
| 63 | ErrDecoderClosed = errors.New("decoder used after Close") |
| 64 | ) |
| 65 | |
| 66 | func println(a ...interface{}) { |
| 67 | if debug { |
| 68 | log.Println(a...) |
| 69 | } |
| 70 | } |
| 71 | |
| 72 | func printf(format string, a ...interface{}) { |
| 73 | if debug { |
| 74 | log.Printf(format, a...) |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | // matchLen returns the maximum length. |
| 79 | // a must be the shortest of the two. |
| 80 | // The function also returns whether all bytes matched. |
| 81 | func matchLen(a, b []byte) int { |
| 82 | b = b[:len(a)] |
| 83 | for i := 0; i < len(a)-7; i += 8 { |
| 84 | if diff := load64(a, i) ^ load64(b, i); diff != 0 { |
| 85 | return i + (bits.TrailingZeros64(diff) >> 3) |
| 86 | } |
| 87 | } |
| 88 | checked := (len(a) >> 3) << 3 |
| 89 | a = a[checked:] |
| 90 | b = b[checked:] |
| 91 | // TODO: We could do a 4 check. |
| 92 | for i := range a { |
| 93 | if a[i] != b[i] { |
| 94 | return int(i) + checked |
| 95 | } |
| 96 | } |
| 97 | return len(a) + checked |
| 98 | } |
| 99 | |
| 100 | // matchLen returns a match length in src between index s and t |
| 101 | func matchLenIn(src []byte, s, t int32) int32 { |
| 102 | s1 := len(src) |
| 103 | b := src[t:] |
| 104 | a := src[s:s1] |
| 105 | b = b[:len(a)] |
| 106 | // Extend the match to be as long as possible. |
| 107 | for i := range a { |
| 108 | if a[i] != b[i] { |
| 109 | return int32(i) |
| 110 | } |
| 111 | } |
| 112 | return int32(len(a)) |
| 113 | } |
| 114 | |
| 115 | func load3232(b []byte, i int32) uint32 { |
| 116 | // Help the compiler eliminate bounds checks on the read so it can be done in a single read. |
| 117 | b = b[i:] |
| 118 | b = b[:4] |
| 119 | return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24 |
| 120 | } |
| 121 | |
| 122 | func load6432(b []byte, i int32) uint64 { |
| 123 | // Help the compiler eliminate bounds checks on the read so it can be done in a single read. |
| 124 | b = b[i:] |
| 125 | b = b[:8] |
| 126 | return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 | |
| 127 | uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56 |
| 128 | } |
| 129 | |
| 130 | func load64(b []byte, i int) uint64 { |
| 131 | // Help the compiler eliminate bounds checks on the read so it can be done in a single read. |
| 132 | b = b[i:] |
| 133 | b = b[:8] |
| 134 | return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 | |
| 135 | uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56 |
| 136 | } |