1 // Copyright 2009 The Go Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style 3 // license that can be found in the LICENSE file. 4 5 // Package gzip implements reading and writing of gzip format compressed files, 6 // as specified in RFC 1952. 7 package gzip 8 9 import ( 10 "bufio" 11 "compress/flate" 12 "errors" 13 "hash" 14 "hash/crc32" 15 "io" 16 "time" 17 ) 18 19 const ( 20 gzipID1 = 0x1f 21 gzipID2 = 0x8b 22 gzipDeflate = 8 23 flagText = 1 << 0 24 flagHdrCrc = 1 << 1 25 flagExtra = 1 << 2 26 flagName = 1 << 3 27 flagComment = 1 << 4 28 ) 29 30 func makeReader(r io.Reader) flate.Reader { 31 if rr, ok := r.(flate.Reader); ok { 32 return rr 33 } 34 return bufio.NewReader(r) 35 } 36 37 var ( 38 // ErrChecksum is returned when reading GZIP data that has an invalid checksum. 39 ErrChecksum = errors.New("gzip: invalid checksum") 40 // ErrHeader is returned when reading GZIP data that has an invalid header. 41 ErrHeader = errors.New("gzip: invalid header") 42 ) 43 44 // The gzip file stores a header giving metadata about the compressed file. 45 // That header is exposed as the fields of the Writer and Reader structs. 46 type Header struct { 47 Comment string // comment 48 Extra []byte // "extra data" 49 ModTime time.Time // modification time 50 Name string // file name 51 OS byte // operating system type 52 } 53 54 // A Reader is an io.Reader that can be read to retrieve 55 // uncompressed data from a gzip-format compressed file. 56 // 57 // In general, a gzip file can be a concatenation of gzip files, 58 // each with its own header. Reads from the Reader 59 // return the concatenation of the uncompressed data of each. 60 // Only the first header is recorded in the Reader fields. 61 // 62 // Gzip files store a length and checksum of the uncompressed data. 63 // The Reader will return a ErrChecksum when Read 64 // reaches the end of the uncompressed data if it does not 65 // have the expected length or checksum. Clients should treat data 66 // returned by Read as tentative until they receive the io.EOF 67 // marking the end of the data. 68 type Reader struct { 69 Header 70 r flate.Reader 71 decompressor io.ReadCloser 72 digest hash.Hash32 73 size uint32 74 flg byte 75 buf [512]byte 76 err error 77 multistream bool 78 } 79 80 // NewReader creates a new Reader reading the given reader. 81 // If r does not also implement io.ByteReader, 82 // the decompressor may read more data than necessary from r. 83 // It is the caller's responsibility to call Close on the Reader when done. 84 func NewReader(r io.Reader) (*Reader, error) { 85 z := new(Reader) 86 z.r = makeReader(r) 87 z.multistream = true 88 z.digest = crc32.NewIEEE() 89 if err := z.readHeader(true); err != nil { 90 return nil, err 91 } 92 return z, nil 93 } 94 95 // Reset discards the Reader z's state and makes it equivalent to the 96 // result of its original state from NewReader, but reading from r instead. 97 // This permits reusing a Reader rather than allocating a new one. 98 func (z *Reader) Reset(r io.Reader) error { 99 z.r = makeReader(r) 100 if z.digest == nil { 101 z.digest = crc32.NewIEEE() 102 } else { 103 z.digest.Reset() 104 } 105 z.size = 0 106 z.err = nil 107 z.multistream = true 108 return z.readHeader(true) 109 } 110 111 // Multistream controls whether the reader supports multistream files. 112 // 113 // If enabled (the default), the Reader expects the input to be a sequence 114 // of individually gzipped data streams, each with its own header and 115 // trailer, ending at EOF. The effect is that the concatenation of a sequence 116 // of gzipped files is treated as equivalent to the gzip of the concatenation 117 // of the sequence. This is standard behavior for gzip readers. 118 // 119 // Calling Multistream(false) disables this behavior; disabling the behavior 120 // can be useful when reading file formats that distinguish individual gzip 121 // data streams or mix gzip data streams with other data streams. 122 // In this mode, when the Reader reaches the end of the data stream, 123 // Read returns io.EOF. If the underlying reader implements io.ByteReader, 124 // it will be left positioned just after the gzip stream. 125 // To start the next stream, call z.Reset(r) followed by z.Multistream(false). 126 // If there is no next stream, z.Reset(r) will return io.EOF. 127 func (z *Reader) Multistream(ok bool) { 128 z.multistream = ok 129 } 130 131 // GZIP (RFC 1952) is little-endian, unlike ZLIB (RFC 1950). 132 func get4(p []byte) uint32 { 133 return uint32(p[0]) | uint32(p[1])<<8 | uint32(p[2])<<16 | uint32(p[3])<<24 134 } 135 136 func (z *Reader) readString() (string, error) { 137 var err error 138 needconv := false 139 for i := 0; ; i++ { 140 if i >= len(z.buf) { 141 return "", ErrHeader 142 } 143 z.buf[i], err = z.r.ReadByte() 144 if err != nil { 145 return "", err 146 } 147 if z.buf[i] > 0x7f { 148 needconv = true 149 } 150 if z.buf[i] == 0 { 151 // GZIP (RFC 1952) specifies that strings are NUL-terminated ISO 8859-1 (Latin-1). 152 if needconv { 153 s := make([]rune, 0, i) 154 for _, v := range z.buf[0:i] { 155 s = append(s, rune(v)) 156 } 157 return string(s), nil 158 } 159 return string(z.buf[0:i]), nil 160 } 161 } 162 } 163 164 func (z *Reader) read2() (uint32, error) { 165 _, err := io.ReadFull(z.r, z.buf[0:2]) 166 if err != nil { 167 return 0, err 168 } 169 return uint32(z.buf[0]) | uint32(z.buf[1])<<8, nil 170 } 171 172 func (z *Reader) readHeader(save bool) error { 173 _, err := io.ReadFull(z.r, z.buf[0:10]) 174 if err != nil { 175 return err 176 } 177 if z.buf[0] != gzipID1 || z.buf[1] != gzipID2 || z.buf[2] != gzipDeflate { 178 return ErrHeader 179 } 180 z.flg = z.buf[3] 181 if save { 182 z.ModTime = time.Unix(int64(get4(z.buf[4:8])), 0) 183 // z.buf[8] is xfl, ignored 184 z.OS = z.buf[9] 185 } 186 z.digest.Reset() 187 z.digest.Write(z.buf[0:10]) 188 189 if z.flg&flagExtra != 0 { 190 n, err := z.read2() 191 if err != nil { 192 return err 193 } 194 data := make([]byte, n) 195 if _, err = io.ReadFull(z.r, data); err != nil { 196 return err 197 } 198 if save { 199 z.Extra = data 200 } 201 } 202 203 var s string 204 if z.flg&flagName != 0 { 205 if s, err = z.readString(); err != nil { 206 return err 207 } 208 if save { 209 z.Name = s 210 } 211 } 212 213 if z.flg&flagComment != 0 { 214 if s, err = z.readString(); err != nil { 215 return err 216 } 217 if save { 218 z.Comment = s 219 } 220 } 221 222 if z.flg&flagHdrCrc != 0 { 223 n, err := z.read2() 224 if err != nil { 225 return err 226 } 227 sum := z.digest.Sum32() & 0xFFFF 228 if n != sum { 229 return ErrHeader 230 } 231 } 232 233 z.digest.Reset() 234 if z.decompressor == nil { 235 z.decompressor = flate.NewReader(z.r) 236 } else { 237 z.decompressor.(flate.Resetter).Reset(z.r, nil) 238 } 239 return nil 240 } 241 242 func (z *Reader) Read(p []byte) (n int, err error) { 243 if z.err != nil { 244 return 0, z.err 245 } 246 if len(p) == 0 { 247 return 0, nil 248 } 249 250 n, err = z.decompressor.Read(p) 251 z.digest.Write(p[0:n]) 252 z.size += uint32(n) 253 if n != 0 || err != io.EOF { 254 z.err = err 255 return 256 } 257 258 // Finished file; check checksum + size. 259 if _, err := io.ReadFull(z.r, z.buf[0:8]); err != nil { 260 z.err = err 261 return 0, err 262 } 263 crc32, isize := get4(z.buf[0:4]), get4(z.buf[4:8]) 264 sum := z.digest.Sum32() 265 if sum != crc32 || isize != z.size { 266 z.err = ErrChecksum 267 return 0, z.err 268 } 269 270 // File is ok; is there another? 271 if !z.multistream { 272 return 0, io.EOF 273 } 274 275 if err = z.readHeader(false); err != nil { 276 z.err = err 277 return 278 } 279 280 // Yes. Reset and read from it. 281 z.digest.Reset() 282 z.size = 0 283 return z.Read(p) 284 } 285 286 // Close closes the Reader. It does not close the underlying io.Reader. 287 func (z *Reader) Close() error { return z.decompressor.Close() } 288
