Source file src/bytes/buffer.go
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 // smallBufferSize is an initial allocation minimal capacity. 16 const smallBufferSize = 64 17 18 // A Buffer is a variable-sized buffer of bytes with Read and Write methods. 19 // The zero value for Buffer is an empty buffer ready to use. 20 type Buffer struct { 21 buf []byte // contents are the bytes buf[off : len(buf)] 22 off int // read at &buf[off], write at &buf[len(buf)] 23 lastRead readOp // last read operation, so that Unread* can work correctly. 24 } 25 26 // The readOp constants describe the last action performed on 27 // the buffer, so that UnreadRune and UnreadByte can check for 28 // invalid usage. opReadRuneX constants are chosen such that 29 // converted to int they correspond to the rune size that was read. 30 type readOp int8 31 32 // Don't use iota for these, as the values need to correspond with the 33 // names and comments, which is easier to see when being explicit. 34 const ( 35 opRead readOp = -1 // Any other read operation. 36 opInvalid readOp = 0 // Non-read operation. 37 opReadRune1 readOp = 1 // Read rune of size 1. 38 opReadRune2 readOp = 2 // Read rune of size 2. 39 opReadRune3 readOp = 3 // Read rune of size 3. 40 opReadRune4 readOp = 4 // Read rune of size 4. 41 ) 42 43 // ErrTooLarge is passed to panic if memory cannot be allocated to store data in a buffer. 44 var ErrTooLarge = errors.New("bytes.Buffer: too large") 45 var errNegativeRead = errors.New("bytes.Buffer: reader returned negative count from Read") 46 47 const maxInt = int(^uint(0) >> 1) 48 49 // Bytes returns a slice of length b.Len() holding the unread portion of the buffer. 50 // The slice is valid for use only until the next buffer modification (that is, 51 // only until the next call to a method like Read, Write, Reset, or Truncate). 52 // The slice aliases the buffer content at least until the next buffer modification, 53 // so immediate changes to the slice will affect the result of future reads. 54 func (b *Buffer) Bytes() []byte { return b.buf[b.off:] } 55 56 // String returns the contents of the unread portion of the buffer 57 // as a string. If the Buffer is a nil pointer, it returns "<nil>". 58 // 59 // To build strings more efficiently, see the strings.Builder type. 60 func (b *Buffer) String() string { 61 if b == nil { 62 // Special case, useful in debugging. 63 return "<nil>" 64 } 65 return string(b.buf[b.off:]) 66 } 67 68 // empty reports whether the unread portion of the buffer is empty. 69 func (b *Buffer) empty() bool { return len(b.buf) <= b.off } 70 71 // Len returns the number of bytes of the unread portion of the buffer; 72 // b.Len() == len(b.Bytes()). 73 func (b *Buffer) Len() int { return len(b.buf) - b.off } 74 75 // Cap returns the capacity of the buffer's underlying byte slice, that is, the 76 // total space allocated for the buffer's data. 77 func (b *Buffer) Cap() int { return cap(b.buf) } 78 79 // Truncate discards all but the first n unread bytes from the buffer 80 // but continues to use the same allocated storage. 81 // It panics if n is negative or greater than the length of the buffer. 82 func (b *Buffer) Truncate(n int) { 83 if n == 0 { 84 b.Reset() 85 return 86 } 87 b.lastRead = opInvalid 88 if n < 0 || n > b.Len() { 89 panic("bytes.Buffer: truncation out of range") 90 } 91 b.buf = b.buf[:b.off+n] 92 } 93 94 // Reset resets the buffer to be empty, 95 // but it retains the underlying storage for use by future writes. 96 // Reset is the same as Truncate(0). 97 func (b *Buffer) Reset() { 98 b.buf = b.buf[:0] 99 b.off = 0 100 b.lastRead = opInvalid 101 } 102 103 // tryGrowByReslice is a inlineable version of grow for the fast-case where the 104 // internal buffer only needs to be resliced. 105 // It returns the index where bytes should be written and whether it succeeded. 106 func (b *Buffer) tryGrowByReslice(n int) (int, bool) { 107 if l := len(b.buf); n <= cap(b.buf)-l { 108 b.buf = b.buf[:l+n] 109 return l, true 110 } 111 return 0, false 112 } 113 114 // grow grows the buffer to guarantee space for n more bytes. 115 // It returns the index where bytes should be written. 116 // If the buffer can't grow it will panic with ErrTooLarge. 117 func (b *Buffer) grow(n int) int { 118 m := b.Len() 119 // If buffer is empty, reset to recover space. 120 if m == 0 && b.off != 0 { 121 b.Reset() 122 } 123 // Try to grow by means of a reslice. 124 if i, ok := b.tryGrowByReslice(n); ok { 125 return i 126 } 127 if b.buf == nil && n <= smallBufferSize { 128 b.buf = make([]byte, n, smallBufferSize) 129 return 0 130 } 131 c := cap(b.buf) 132 if n <= c/2-m { 133 // We can slide things down instead of allocating a new 134 // slice. We only need m+n <= c to slide, but 135 // we instead let capacity get twice as large so we 136 // don't spend all our time copying. 137 copy(b.buf, b.buf[b.off:]) 138 } else if c > maxInt-c-n { 139 panic(ErrTooLarge) 140 } else { 141 // Add b.off to account for b.buf[:b.off] being sliced off the front. 142 b.buf = growSlice(b.buf[b.off:], b.off+n) 143 } 144 // Restore b.off and len(b.buf). 145 b.off = 0 146 b.buf = b.buf[:m+n] 147 return m 148 } 149 150 // Grow grows the buffer's capacity, if necessary, to guarantee space for 151 // another n bytes. After Grow(n), at least n bytes can be written to the 152 // buffer without another allocation. 153 // If n is negative, Grow will panic. 154 // If the buffer can't grow it will panic with ErrTooLarge. 155 func (b *Buffer) Grow(n int) { 156 if n < 0 { 157 panic("bytes.Buffer.Grow: negative count") 158 } 159 m := b.grow(n) 160 b.buf = b.buf[:m] 161 } 162 163 // Write appends the contents of p to the buffer, growing the buffer as 164 // needed. The return value n is the length of p; err is always nil. If the 165 // buffer becomes too large, Write will panic with ErrTooLarge. 166 func (b *Buffer) Write(p []byte) (n int, err error) { 167 b.lastRead = opInvalid 168 m, ok := b.tryGrowByReslice(len(p)) 169 if !ok { 170 m = b.grow(len(p)) 171 } 172 return copy(b.buf[m:], p), nil 173 } 174 175 // WriteString appends the contents of s to the buffer, growing the buffer as 176 // needed. The return value n is the length of s; err is always nil. If the 177 // buffer becomes too large, WriteString will panic with ErrTooLarge. 178 func (b *Buffer) WriteString(s string) (n int, err error) { 179 b.lastRead = opInvalid 180 m, ok := b.tryGrowByReslice(len(s)) 181 if !ok { 182 m = b.grow(len(s)) 183 } 184 return copy(b.buf[m:], s), nil 185 } 186 187 // MinRead is the minimum slice size passed to a Read call by 188 // Buffer.ReadFrom. As long as the Buffer has at least MinRead bytes beyond 189 // what is required to hold the contents of r, ReadFrom will not grow the 190 // underlying buffer. 191 const MinRead = 512 192 193 // ReadFrom reads data from r until EOF and appends it to the buffer, growing 194 // the buffer as needed. The return value n is the number of bytes read. Any 195 // error except io.EOF encountered during the read is also returned. If the 196 // buffer becomes too large, ReadFrom will panic with ErrTooLarge. 197 func (b *Buffer) ReadFrom(r io.Reader) (n int64, err error) { 198 b.lastRead = opInvalid 199 for { 200 i := b.grow(MinRead) 201 b.buf = b.buf[:i] 202 m, e := r.Read(b.buf[i:cap(b.buf)]) 203 if m < 0 { 204 panic(errNegativeRead) 205 } 206 207 b.buf = b.buf[:i+m] 208 n += int64(m) 209 if e == io.EOF { 210 return n, nil // e is EOF, so return nil explicitly 211 } 212 if e != nil { 213 return n, e 214 } 215 } 216 } 217 218 // growSlice grows b by n, preserving the original content of b. 219 // If the allocation fails, it panics with ErrTooLarge. 220 func growSlice(b []byte, n int) []byte { 221 defer func() { 222 if recover() != nil { 223 panic(ErrTooLarge) 224 } 225 }() 226 // TODO(http://golang.org/issue/51462): We should rely on the append-make 227 // pattern so that the compiler can call runtime.growslice. For example: 228 // return append(b, make([]byte, n)...) 229 // This avoids unnecessary zero-ing of the first len(b) bytes of the 230 // allocated slice, but this pattern causes b to escape onto the heap. 231 // 232 // Instead use the append-make pattern with a nil slice to ensure that 233 // we allocate buffers rounded up to the closest size class. 234 c := len(b) + n // ensure enough space for n elements 235 if c < 2*cap(b) { 236 // The growth rate has historically always been 2x. In the future, 237 // we could rely purely on append to determine the growth rate. 238 c = 2 * cap(b) 239 } 240 b2 := append([]byte(nil), make([]byte, c)...) 241 copy(b2, b) 242 return b2[:len(b)] 243 } 244 245 // WriteTo writes data to w until the buffer is drained or an error occurs. 246 // The return value n is the number of bytes written; it always fits into an 247 // int, but it is int64 to match the io.WriterTo interface. Any error 248 // encountered during the write is also returned. 249 func (b *Buffer) WriteTo(w io.Writer) (n int64, err error) { 250 b.lastRead = opInvalid 251 if nBytes := b.Len(); nBytes > 0 { 252 m, e := w.Write(b.buf[b.off:]) 253 if m > nBytes { 254 panic("bytes.Buffer.WriteTo: invalid Write count") 255 } 256 b.off += m 257 n = int64(m) 258 if e != nil { 259 return n, e 260 } 261 // all bytes should have been written, by definition of 262 // Write method in io.Writer 263 if m != nBytes { 264 return n, io.ErrShortWrite 265 } 266 } 267 // Buffer is now empty; reset. 268 b.Reset() 269 return n, nil 270 } 271 272 // WriteByte appends the byte c to the buffer, growing the buffer as needed. 273 // The returned error is always nil, but is included to match bufio.Writer's 274 // WriteByte. If the buffer becomes too large, WriteByte will panic with 275 // ErrTooLarge. 276 func (b *Buffer) WriteByte(c byte) error { 277 b.lastRead = opInvalid 278 m, ok := b.tryGrowByReslice(1) 279 if !ok { 280 m = b.grow(1) 281 } 282 b.buf[m] = c 283 return nil 284 } 285 286 // WriteRune appends the UTF-8 encoding of Unicode code point r to the 287 // buffer, returning its length and an error, which is always nil but is 288 // included to match bufio.Writer's WriteRune. The buffer is grown as needed; 289 // if it becomes too large, WriteRune will panic with ErrTooLarge. 290 func (b *Buffer) WriteRune(r rune) (n int, err error) { 291 // Compare as uint32 to correctly handle negative runes. 292 if uint32(r) < utf8.RuneSelf { 293 b.WriteByte(byte(r)) 294 return 1, nil 295 } 296 b.lastRead = opInvalid 297 m, ok := b.tryGrowByReslice(utf8.UTFMax) 298 if !ok { 299 m = b.grow(utf8.UTFMax) 300 } 301 b.buf = utf8.AppendRune(b.buf[:m], r) 302 return len(b.buf) - m, nil 303 } 304 305 // Read reads the next len(p) bytes from the buffer or until the buffer 306 // is drained. The return value n is the number of bytes read. If the 307 // buffer has no data to return, err is io.EOF (unless len(p) is zero); 308 // otherwise it is nil. 309 func (b *Buffer) Read(p []byte) (n int, err error) { 310 b.lastRead = opInvalid 311 if b.empty() { 312 // Buffer is empty, reset to recover space. 313 b.Reset() 314 if len(p) == 0 { 315 return 0, nil 316 } 317 return 0, io.EOF 318 } 319 n = copy(p, b.buf[b.off:]) 320 b.off += n 321 if n > 0 { 322 b.lastRead = opRead 323 } 324 return n, nil 325 } 326 327 // Next returns a slice containing the next n bytes from the buffer, 328 // advancing the buffer as if the bytes had been returned by Read. 329 // If there are fewer than n bytes in the buffer, Next returns the entire buffer. 330 // The slice is only valid until the next call to a read or write method. 331 func (b *Buffer) Next(n int) []byte { 332 b.lastRead = opInvalid 333 m := b.Len() 334 if n > m { 335 n = m 336 } 337 data := b.buf[b.off : b.off+n] 338 b.off += n 339 if n > 0 { 340 b.lastRead = opRead 341 } 342 return data 343 } 344 345 // ReadByte reads and returns the next byte from the buffer. 346 // If no byte is available, it returns error io.EOF. 347 func (b *Buffer) ReadByte() (byte, error) { 348 if b.empty() { 349 // Buffer is empty, reset to recover space. 350 b.Reset() 351 return 0, io.EOF 352 } 353 c := b.buf[b.off] 354 b.off++ 355 b.lastRead = opRead 356 return c, nil 357 } 358 359 // ReadRune reads and returns the next UTF-8-encoded 360 // Unicode code point from the buffer. 361 // If no bytes are available, the error returned is io.EOF. 362 // If the bytes are an erroneous UTF-8 encoding, it 363 // consumes one byte and returns U+FFFD, 1. 364 func (b *Buffer) ReadRune() (r rune, size int, err error) { 365 if b.empty() { 366 // Buffer is empty, reset to recover space. 367 b.Reset() 368 return 0, 0, io.EOF 369 } 370 c := b.buf[b.off] 371 if c < utf8.RuneSelf { 372 b.off++ 373 b.lastRead = opReadRune1 374 return rune(c), 1, nil 375 } 376 r, n := utf8.DecodeRune(b.buf[b.off:]) 377 b.off += n 378 b.lastRead = readOp(n) 379 return r, n, nil 380 } 381 382 // UnreadRune unreads the last rune returned by ReadRune. 383 // If the most recent read or write operation on the buffer was 384 // not a successful ReadRune, UnreadRune returns an error. (In this regard 385 // it is stricter than UnreadByte, which will unread the last byte 386 // from any read operation.) 387 func (b *Buffer) UnreadRune() error { 388 if b.lastRead <= opInvalid { 389 return errors.New("bytes.Buffer: UnreadRune: previous operation was not a successful ReadRune") 390 } 391 if b.off >= int(b.lastRead) { 392 b.off -= int(b.lastRead) 393 } 394 b.lastRead = opInvalid 395 return nil 396 } 397 398 var errUnreadByte = errors.New("bytes.Buffer: UnreadByte: previous operation was not a successful read") 399 400 // UnreadByte unreads the last byte returned by the most recent successful 401 // read operation that read at least one byte. If a write has happened since 402 // the last read, if the last read returned an error, or if the read read zero 403 // bytes, UnreadByte returns an error. 404 func (b *Buffer) UnreadByte() error { 405 if b.lastRead == opInvalid { 406 return errUnreadByte 407 } 408 b.lastRead = opInvalid 409 if b.off > 0 { 410 b.off-- 411 } 412 return nil 413 } 414 415 // ReadBytes reads until the first occurrence of delim in the input, 416 // returning a slice containing the data up to and including the delimiter. 417 // If ReadBytes encounters an error before finding a delimiter, 418 // it returns the data read before the error and the error itself (often io.EOF). 419 // ReadBytes returns err != nil if and only if the returned data does not end in 420 // delim. 421 func (b *Buffer) ReadBytes(delim byte) (line []byte, err error) { 422 slice, err := b.readSlice(delim) 423 // return a copy of slice. The buffer's backing array may 424 // be overwritten by later calls. 425 line = append(line, slice...) 426 return line, err 427 } 428 429 // readSlice is like ReadBytes but returns a reference to internal buffer data. 430 func (b *Buffer) readSlice(delim byte) (line []byte, err error) { 431 i := IndexByte(b.buf[b.off:], delim) 432 end := b.off + i + 1 433 if i < 0 { 434 end = len(b.buf) 435 err = io.EOF 436 } 437 line = b.buf[b.off:end] 438 b.off = end 439 b.lastRead = opRead 440 return line, err 441 } 442 443 // ReadString reads until the first occurrence of delim in the input, 444 // returning a string containing the data up to and including the delimiter. 445 // If ReadString encounters an error before finding a delimiter, 446 // it returns the data read before the error and the error itself (often io.EOF). 447 // ReadString returns err != nil if and only if the returned data does not end 448 // in delim. 449 func (b *Buffer) ReadString(delim byte) (line string, err error) { 450 slice, err := b.readSlice(delim) 451 return string(slice), err 452 } 453 454 // NewBuffer creates and initializes a new Buffer using buf as its 455 // initial contents. The new Buffer takes ownership of buf, and the 456 // caller should not use buf after this call. NewBuffer is intended to 457 // prepare a Buffer to read existing data. It can also be used to set 458 // the initial size of the internal buffer for writing. To do that, 459 // buf should have the desired capacity but a length of zero. 460 // 461 // In most cases, new(Buffer) (or just declaring a Buffer variable) is 462 // sufficient to initialize a Buffer. 463 func NewBuffer(buf []byte) *Buffer { return &Buffer{buf: buf} } 464 465 // NewBufferString creates and initializes a new Buffer using string s as its 466 // initial contents. It is intended to prepare a buffer to read an existing 467 // string. 468 // 469 // In most cases, new(Buffer) (or just declaring a Buffer variable) is 470 // sufficient to initialize a Buffer. 471 func NewBufferString(s string) *Buffer { 472 return &Buffer{buf: []byte(s)} 473 } 474