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 bytes 6 7 // Simple byte buffer for marshaling data. 8 9 import ( 10 "errors" 11 "io" 12 "unicode/utf8" 13 ) 14 15 // A Buffer is a variable-sized buffer of bytes with Read and Write methods. 16 // The zero value for Buffer is an empty buffer ready to use. 17 type Buffer struct { 18 buf []byte // contents are the bytes buf[off : len(buf)] 19 off int // read at &buf[off], write at &buf[len(buf)] 20 bootstrap [64]byte // memory to hold first slice; helps small buffers avoid allocation. 21 lastRead readOp // last read operation, so that Unread* can work correctly. 22 23 // FIXME: it would be advisable to align Buffer to cachelines to avoid false 24 // sharing. 25 } 26 27 // The readOp constants describe the last action performed on 28 // the buffer, so that UnreadRune and UnreadByte can check for 29 // invalid usage. opReadRuneX constants are chosen such that 30 // converted to int they correspond to the rune size that was read. 31 type readOp int8 32 33 // Don't use iota for these, as the values need to correspond with the 34 // names and comments, which is easier to see when being explicit. 35 const ( 36 opRead readOp = -1 // Any other read operation. 37 opInvalid readOp = 0 // Non-read operation. 38 opReadRune1 readOp = 1 // Read rune of size 1. 39 opReadRune2 readOp = 2 // Read rune of size 2. 40 opReadRune3 readOp = 3 // Read rune of size 3. 41 opReadRune4 readOp = 4 // Read rune of size 4. 42 ) 43 44 // ErrTooLarge is passed to panic if memory cannot be allocated to store data in a buffer. 45 var ErrTooLarge = errors.New("bytes.Buffer: too large") 46 var errNegativeRead = errors.New("bytes.Buffer: reader returned negative count from Read") 47 48 const maxInt = int(^uint(0) >> 1) 49 50 // Bytes returns a slice of length b.Len() holding the unread portion of the buffer. 51 // The slice is valid for use only until the next buffer modification (that is, 52 // only until the next call to a method like Read, Write, Reset, or Truncate). 53 // The slice aliases the buffer content at least until the next buffer modification, 54 // so immediate changes to the slice will affect the result of future reads. 55 func (b *Buffer) Bytes() []byte { return b.buf[b.off:] } 56 57 // String returns the contents of the unread portion of the buffer 58 // as a string. If the Buffer is a nil pointer, it returns "<nil>". 59 // 60 // To build strings more efficiently, see the strings.Builder type. 61 func (b *Buffer) String() string { 62 if b == nil { 63 // Special case, useful in debugging. 64 return "<nil>" 65 } 66 return string(b.buf[b.off:]) 67 } 68 69 // empty returns whether the unread portion of the buffer is empty. 70 func (b *Buffer) empty() bool { return len(b.buf) <= b.off } 71 72 // Len returns the number of bytes of the unread portion of the buffer; 73 // b.Len() == len(b.Bytes()). 74 func (b *Buffer) Len() int { return len(b.buf) - b.off } 75 76 // Cap returns the capacity of the buffer's underlying byte slice, that is, the 77 // total space allocated for the buffer's data. 78 func (b *Buffer) Cap() int { return cap(b.buf) } 79 80 // Truncate discards all but the first n unread bytes from the buffer 81 // but continues to use the same allocated storage. 82 // It panics if n is negative or greater than the length of the buffer. 83 func (b *Buffer) Truncate(n int) { 84 if n == 0 { 85 b.Reset() 86 return 87 } 88 b.lastRead = opInvalid 89 if n < 0 || n > b.Len() { 90 panic("bytes.Buffer: truncation out of range") 91 } 92 b.buf = b.buf[:b.off+n] 93 } 94 95 // Reset resets the buffer to be empty, 96 // but it retains the underlying storage for use by future writes. 97 // Reset is the same as Truncate(0). 98 func (b *Buffer) Reset() { 99 b.buf = b.buf[:0] 100 b.off = 0 101 b.lastRead = opInvalid 102 } 103 104 // tryGrowByReslice is a inlineable version of grow for the fast-case where the 105 // internal buffer only needs to be resliced. 106 // It returns the index where bytes should be written and whether it succeeded. 107 func (b *Buffer) tryGrowByReslice(n int) (int, bool) { 108 if l := len(b.buf); n <= cap(b.buf)-l { 109 b.buf = b.buf[:l+n] 110 return l, true 111 } 112 return 0, false 113 } 114 115 // grow grows the buffer to guarantee space for n more bytes. 116 // It returns the index where bytes should be written. 117 // If the buffer can't grow it will panic with ErrTooLarge. 118 func (b *Buffer) grow(n int) int { 119 m := b.Len() 120 // If buffer is empty, reset to recover space. 121 if m == 0 && b.off != 0 { 122 b.Reset() 123 } 124 // Try to grow by means of a reslice. 125 if i, ok := b.tryGrowByReslice(n); ok { 126 return i 127 } 128 // Check if we can make use of bootstrap array. 129 if b.buf == nil && n <= len(b.bootstrap) { 130 b.buf = b.bootstrap[:n] 131 return 0 132 } 133 c := cap(b.buf) 134 if n <= c/2-m { 135 // We can slide things down instead of allocating a new 136 // slice. We only need m+n <= c to slide, but 137 // we instead let capacity get twice as large so we 138 // don't spend all our time copying. 139 copy(b.buf, b.buf[b.off:]) 140 } else if c > maxInt-c-n { 141 panic(ErrTooLarge) 142 } else { 143 // Not enough space anywhere, we need to allocate. 144 buf := makeSlice(2*c + n) 145 copy(buf, b.buf[b.off:]) 146 b.buf = buf 147 } 148 // Restore b.off and len(b.buf). 149 b.off = 0 150 b.buf = b.buf[:m+n] 151 return m 152 } 153 154 // Grow grows the buffer's capacity, if necessary, to guarantee space for 155 // another n bytes. After Grow(n), at least n bytes can be written to the 156 // buffer without another allocation. 157 // If n is negative, Grow will panic. 158 // If the buffer can't grow it will panic with ErrTooLarge. 159 func (b *Buffer) Grow(n int) { 160 if n < 0 { 161 panic("bytes.Buffer.Grow: negative count") 162 } 163 m := b.grow(n) 164 b.buf = b.buf[:m] 165 } 166 167 // Write appends the contents of p to the buffer, growing the buffer as 168 // needed. The return value n is the length of p; err is always nil. If the 169 // buffer becomes too large, Write will panic with ErrTooLarge. 170 func (b *Buffer) Write(p []byte) (n int, err error) { 171 b.lastRead = opInvalid 172 m, ok := b.tryGrowByReslice(len(p)) 173 if !ok { 174 m = b.grow(len(p)) 175 } 176 return copy(b.buf[m:], p), nil 177 } 178 179 // WriteString appends the contents of s to the buffer, growing the buffer as 180 // needed. The return value n is the length of s; err is always nil. If the 181 // buffer becomes too large, WriteString will panic with ErrTooLarge. 182 func (b *Buffer) WriteString(s string) (n int, err error) { 183 b.lastRead = opInvalid 184 m, ok := b.tryGrowByReslice(len(s)) 185 if !ok { 186 m = b.grow(len(s)) 187 } 188 return copy(b.buf[m:], s), nil 189 } 190 191 // MinRead is the minimum slice size passed to a Read call by 192 // Buffer.ReadFrom. As long as the Buffer has at least MinRead bytes beyond 193 // what is required to hold the contents of r, ReadFrom will not grow the 194 // underlying buffer. 195 const MinRead = 512 196 197 // ReadFrom reads data from r until EOF and appends it to the buffer, growing 198 // the buffer as needed. The return value n is the number of bytes read. Any 199 // error except io.EOF encountered during the read is also returned. If the 200 // buffer becomes too large, ReadFrom will panic with ErrTooLarge. 201 func (b *Buffer) ReadFrom(r io.Reader) (n int64, err error) { 202 b.lastRead = opInvalid 203 for { 204 i := b.grow(MinRead) 205 b.buf = b.buf[:i] 206 m, e := r.Read(b.buf[i:cap(b.buf)]) 207 if m < 0 { 208 panic(errNegativeRead) 209 } 210 211 b.buf = b.buf[:i+m] 212 n += int64(m) 213 if e == io.EOF { 214 return n, nil // e is EOF, so return nil explicitly 215 } 216 if e != nil { 217 return n, e 218 } 219 } 220 } 221 222 // makeSlice allocates a slice of size n. If the allocation fails, it panics 223 // with ErrTooLarge. 224 func makeSlice(n int) []byte { 225 // If the make fails, give a known error. 226 defer func() { 227 if recover() != nil { 228 panic(ErrTooLarge) 229 } 230 }() 231 return make([]byte, n) 232 } 233 234 // WriteTo writes data to w until the buffer is drained or an error occurs. 235 // The return value n is the number of bytes written; it always fits into an 236 // int, but it is int64 to match the io.WriterTo interface. Any error 237 // encountered during the write is also returned. 238 func (b *Buffer) WriteTo(w io.Writer) (n int64, err error) { 239 b.lastRead = opInvalid 240 if nBytes := b.Len(); nBytes > 0 { 241 m, e := w.Write(b.buf[b.off:]) 242 if m > nBytes { 243 panic("bytes.Buffer.WriteTo: invalid Write count") 244 } 245 b.off += m 246 n = int64(m) 247 if e != nil { 248 return n, e 249 } 250 // all bytes should have been written, by definition of 251 // Write method in io.Writer 252 if m != nBytes { 253 return n, io.ErrShortWrite 254 } 255 } 256 // Buffer is now empty; reset. 257 b.Reset() 258 return n, nil 259 } 260 261 // WriteByte appends the byte c to the buffer, growing the buffer as needed. 262 // The returned error is always nil, but is included to match bufio.Writer's 263 // WriteByte. If the buffer becomes too large, WriteByte will panic with 264 // ErrTooLarge. 265 func (b *Buffer) WriteByte(c byte) error { 266 b.lastRead = opInvalid 267 m, ok := b.tryGrowByReslice(1) 268 if !ok { 269 m = b.grow(1) 270 } 271 b.buf[m] = c 272 return nil 273 } 274 275 // WriteRune appends the UTF-8 encoding of Unicode code point r to the 276 // buffer, returning its length and an error, which is always nil but is 277 // included to match bufio.Writer's WriteRune. The buffer is grown as needed; 278 // if it becomes too large, WriteRune will panic with ErrTooLarge. 279 func (b *Buffer) WriteRune(r rune) (n int, err error) { 280 if r < utf8.RuneSelf { 281 b.WriteByte(byte(r)) 282 return 1, nil 283 } 284 b.lastRead = opInvalid 285 m, ok := b.tryGrowByReslice(utf8.UTFMax) 286 if !ok { 287 m = b.grow(utf8.UTFMax) 288 } 289 n = utf8.EncodeRune(b.buf[m:m+utf8.UTFMax], r) 290 b.buf = b.buf[:m+n] 291 return n, nil 292 } 293 294 // Read reads the next len(p) bytes from the buffer or until the buffer 295 // is drained. The return value n is the number of bytes read. If the 296 // buffer has no data to return, err is io.EOF (unless len(p) is zero); 297 // otherwise it is nil. 298 func (b *Buffer) Read(p []byte) (n int, err error) { 299 b.lastRead = opInvalid 300 if b.empty() { 301 // Buffer is empty, reset to recover space. 302 b.Reset() 303 if len(p) == 0 { 304 return 0, nil 305 } 306 return 0, io.EOF 307 } 308 n = copy(p, b.buf[b.off:]) 309 b.off += n 310 if n > 0 { 311 b.lastRead = opRead 312 } 313 return n, nil 314 } 315 316 // Next returns a slice containing the next n bytes from the buffer, 317 // advancing the buffer as if the bytes had been returned by Read. 318 // If there are fewer than n bytes in the buffer, Next returns the entire buffer. 319 // The slice is only valid until the next call to a read or write method. 320 func (b *Buffer) Next(n int) []byte { 321 b.lastRead = opInvalid 322 m := b.Len() 323 if n > m { 324 n = m 325 } 326 data := b.buf[b.off : b.off+n] 327 b.off += n 328 if n > 0 { 329 b.lastRead = opRead 330 } 331 return data 332 } 333 334 // ReadByte reads and returns the next byte from the buffer. 335 // If no byte is available, it returns error io.EOF. 336 func (b *Buffer) ReadByte() (byte, error) { 337 if b.empty() { 338 // Buffer is empty, reset to recover space. 339 b.Reset() 340 return 0, io.EOF 341 } 342 c := b.buf[b.off] 343 b.off++ 344 b.lastRead = opRead 345 return c, nil 346 } 347 348 // ReadRune reads and returns the next UTF-8-encoded 349 // Unicode code point from the buffer. 350 // If no bytes are available, the error returned is io.EOF. 351 // If the bytes are an erroneous UTF-8 encoding, it 352 // consumes one byte and returns U+FFFD, 1. 353 func (b *Buffer) ReadRune() (r rune, size int, err error) { 354 if b.empty() { 355 // Buffer is empty, reset to recover space. 356 b.Reset() 357 return 0, 0, io.EOF 358 } 359 c := b.buf[b.off] 360 if c < utf8.RuneSelf { 361 b.off++ 362 b.lastRead = opReadRune1 363 return rune(c), 1, nil 364 } 365 r, n := utf8.DecodeRune(b.buf[b.off:]) 366 b.off += n 367 b.lastRead = readOp(n) 368 return r, n, nil 369 } 370 371 // UnreadRune unreads the last rune returned by ReadRune. 372 // If the most recent read or write operation on the buffer was 373 // not a successful ReadRune, UnreadRune returns an error. (In this regard 374 // it is stricter than UnreadByte, which will unread the last byte 375 // from any read operation.) 376 func (b *Buffer) UnreadRune() error { 377 if b.lastRead <= opInvalid { 378 return errors.New("bytes.Buffer: UnreadRune: previous operation was not a successful ReadRune") 379 } 380 if b.off >= int(b.lastRead) { 381 b.off -= int(b.lastRead) 382 } 383 b.lastRead = opInvalid 384 return nil 385 } 386 387 // UnreadByte unreads the last byte returned by the most recent successful 388 // read operation that read at least one byte. If a write has happened since 389 // the last read, if the last read returned an error, or if the read read zero 390 // bytes, UnreadByte returns an error. 391 func (b *Buffer) UnreadByte() error { 392 if b.lastRead == opInvalid { 393 return errors.New("bytes.Buffer: UnreadByte: previous operation was not a successful read") 394 } 395 b.lastRead = opInvalid 396 if b.off > 0 { 397 b.off-- 398 } 399 return nil 400 } 401 402 // ReadBytes reads until the first occurrence of delim in the input, 403 // returning a slice containing the data up to and including the delimiter. 404 // If ReadBytes encounters an error before finding a delimiter, 405 // it returns the data read before the error and the error itself (often io.EOF). 406 // ReadBytes returns err != nil if and only if the returned data does not end in 407 // delim. 408 func (b *Buffer) ReadBytes(delim byte) (line []byte, err error) { 409 slice, err := b.readSlice(delim) 410 // return a copy of slice. The buffer's backing array may 411 // be overwritten by later calls. 412 line = append(line, slice...) 413 return line, err 414 } 415 416 // readSlice is like ReadBytes but returns a reference to internal buffer data. 417 func (b *Buffer) readSlice(delim byte) (line []byte, err error) { 418 i := IndexByte(b.buf[b.off:], delim) 419 end := b.off + i + 1 420 if i < 0 { 421 end = len(b.buf) 422 err = io.EOF 423 } 424 line = b.buf[b.off:end] 425 b.off = end 426 b.lastRead = opRead 427 return line, err 428 } 429 430 // ReadString reads until the first occurrence of delim in the input, 431 // returning a string containing the data up to and including the delimiter. 432 // If ReadString encounters an error before finding a delimiter, 433 // it returns the data read before the error and the error itself (often io.EOF). 434 // ReadString returns err != nil if and only if the returned data does not end 435 // in delim. 436 func (b *Buffer) ReadString(delim byte) (line string, err error) { 437 slice, err := b.readSlice(delim) 438 return string(slice), err 439 } 440 441 // NewBuffer creates and initializes a new Buffer using buf as its 442 // initial contents. The new Buffer takes ownership of buf, and the 443 // caller should not use buf after this call. NewBuffer is intended to 444 // prepare a Buffer to read existing data. It can also be used to size 445 // the internal buffer for writing. To do that, buf should have the 446 // desired capacity but a length of zero. 447 // 448 // In most cases, new(Buffer) (or just declaring a Buffer variable) is 449 // sufficient to initialize a Buffer. 450 func NewBuffer(buf []byte) *Buffer { return &Buffer{buf: buf} } 451 452 // NewBufferString creates and initializes a new Buffer using string s as its 453 // initial contents. It is intended to prepare a buffer to read an existing 454 // string. 455 // 456 // In most cases, new(Buffer) (or just declaring a Buffer variable) is 457 // sufficient to initialize a Buffer. 458 func NewBufferString(s string) *Buffer { 459 return &Buffer{buf: []byte(s)} 460 } 461