...
Run Format

Source file src/bufio/scan.go

Documentation: bufio

     1  // Copyright 2013 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 bufio
     6  
     7  import (
     8  	"bytes"
     9  	"errors"
    10  	"io"
    11  	"unicode/utf8"
    12  )
    13  
    14  // Scanner provides a convenient interface for reading data such as
    15  // a file of newline-delimited lines of text. Successive calls to
    16  // the Scan method will step through the 'tokens' of a file, skipping
    17  // the bytes between the tokens. The specification of a token is
    18  // defined by a split function of type SplitFunc; the default split
    19  // function breaks the input into lines with line termination stripped. Split
    20  // functions are defined in this package for scanning a file into
    21  // lines, bytes, UTF-8-encoded runes, and space-delimited words. The
    22  // client may instead provide a custom split function.
    23  //
    24  // Scanning stops unrecoverably at EOF, the first I/O error, or a token too
    25  // large to fit in the buffer. When a scan stops, the reader may have
    26  // advanced arbitrarily far past the last token. Programs that need more
    27  // control over error handling or large tokens, or must run sequential scans
    28  // on a reader, should use bufio.Reader instead.
    29  //
    30  type Scanner struct {
    31  	r            io.Reader // The reader provided by the client.
    32  	split        SplitFunc // The function to split the tokens.
    33  	maxTokenSize int       // Maximum size of a token; modified by tests.
    34  	token        []byte    // Last token returned by split.
    35  	buf          []byte    // Buffer used as argument to split.
    36  	start        int       // First non-processed byte in buf.
    37  	end          int       // End of data in buf.
    38  	err          error     // Sticky error.
    39  	empties      int       // Count of successive empty tokens.
    40  	scanCalled   bool      // Scan has been called; buffer is in use.
    41  	done         bool      // Scan has finished.
    42  }
    43  
    44  // SplitFunc is the signature of the split function used to tokenize the
    45  // input. The arguments are an initial substring of the remaining unprocessed
    46  // data and a flag, atEOF, that reports whether the Reader has no more data
    47  // to give. The return values are the number of bytes to advance the input
    48  // and the next token to return to the user, if any, plus an error, if any.
    49  //
    50  // Scanning stops if the function returns an error, in which case some of
    51  // the input may be discarded.
    52  //
    53  // Otherwise, the Scanner advances the input. If the token is not nil,
    54  // the Scanner returns it to the user. If the token is nil, the
    55  // Scanner reads more data and continues scanning; if there is no more
    56  // data--if atEOF was true--the Scanner returns. If the data does not
    57  // yet hold a complete token, for instance if it has no newline while
    58  // scanning lines, a SplitFunc can return (0, nil, nil) to signal the
    59  // Scanner to read more data into the slice and try again with a
    60  // longer slice starting at the same point in the input.
    61  //
    62  // The function is never called with an empty data slice unless atEOF
    63  // is true. If atEOF is true, however, data may be non-empty and,
    64  // as always, holds unprocessed text.
    65  type SplitFunc func(data []byte, atEOF bool) (advance int, token []byte, err error)
    66  
    67  // Errors returned by Scanner.
    68  var (
    69  	ErrTooLong         = errors.New("bufio.Scanner: token too long")
    70  	ErrNegativeAdvance = errors.New("bufio.Scanner: SplitFunc returns negative advance count")
    71  	ErrAdvanceTooFar   = errors.New("bufio.Scanner: SplitFunc returns advance count beyond input")
    72  )
    73  
    74  const (
    75  	// MaxScanTokenSize is the maximum size used to buffer a token
    76  	// unless the user provides an explicit buffer with Scan.Buffer.
    77  	// The actual maximum token size may be smaller as the buffer
    78  	// may need to include, for instance, a newline.
    79  	MaxScanTokenSize = 64 * 1024
    80  
    81  	startBufSize = 4096 // Size of initial allocation for buffer.
    82  )
    83  
    84  // NewScanner returns a new Scanner to read from r.
    85  // The split function defaults to ScanLines.
    86  func NewScanner(r io.Reader) *Scanner {
    87  	return &Scanner{
    88  		r:            r,
    89  		split:        ScanLines,
    90  		maxTokenSize: MaxScanTokenSize,
    91  	}
    92  }
    93  
    94  // Err returns the first non-EOF error that was encountered by the Scanner.
    95  func (s *Scanner) Err() error {
    96  	if s.err == io.EOF {
    97  		return nil
    98  	}
    99  	return s.err
   100  }
   101  
   102  // Bytes returns the most recent token generated by a call to Scan.
   103  // The underlying array may point to data that will be overwritten
   104  // by a subsequent call to Scan. It does no allocation.
   105  func (s *Scanner) Bytes() []byte {
   106  	return s.token
   107  }
   108  
   109  // Text returns the most recent token generated by a call to Scan
   110  // as a newly allocated string holding its bytes.
   111  func (s *Scanner) Text() string {
   112  	return string(s.token)
   113  }
   114  
   115  // ErrFinalToken is a special sentinel error value. It is intended to be
   116  // returned by a Split function to indicate that the token being delivered
   117  // with the error is the last token and scanning should stop after this one.
   118  // After ErrFinalToken is received by Scan, scanning stops with no error.
   119  // The value is useful to stop processing early or when it is necessary to
   120  // deliver a final empty token. One could achieve the same behavior
   121  // with a custom error value but providing one here is tidier.
   122  // See the emptyFinalToken example for a use of this value.
   123  var ErrFinalToken = errors.New("final token")
   124  
   125  // Scan advances the Scanner to the next token, which will then be
   126  // available through the Bytes or Text method. It returns false when the
   127  // scan stops, either by reaching the end of the input or an error.
   128  // After Scan returns false, the Err method will return any error that
   129  // occurred during scanning, except that if it was io.EOF, Err
   130  // will return nil.
   131  // Scan panics if the split function returns too many empty
   132  // tokens without advancing the input. This is a common error mode for
   133  // scanners.
   134  func (s *Scanner) Scan() bool {
   135  	if s.done {
   136  		return false
   137  	}
   138  	s.scanCalled = true
   139  	// Loop until we have a token.
   140  	for {
   141  		// See if we can get a token with what we already have.
   142  		// If we've run out of data but have an error, give the split function
   143  		// a chance to recover any remaining, possibly empty token.
   144  		if s.end > s.start || s.err != nil {
   145  			advance, token, err := s.split(s.buf[s.start:s.end], s.err != nil)
   146  			if err != nil {
   147  				if err == ErrFinalToken {
   148  					s.token = token
   149  					s.done = true
   150  					return true
   151  				}
   152  				s.setErr(err)
   153  				return false
   154  			}
   155  			if !s.advance(advance) {
   156  				return false
   157  			}
   158  			s.token = token
   159  			if token != nil {
   160  				if s.err == nil || advance > 0 {
   161  					s.empties = 0
   162  				} else {
   163  					// Returning tokens not advancing input at EOF.
   164  					s.empties++
   165  					if s.empties > maxConsecutiveEmptyReads {
   166  						panic("bufio.Scan: too many empty tokens without progressing")
   167  					}
   168  				}
   169  				return true
   170  			}
   171  		}
   172  		// We cannot generate a token with what we are holding.
   173  		// If we've already hit EOF or an I/O error, we are done.
   174  		if s.err != nil {
   175  			// Shut it down.
   176  			s.start = 0
   177  			s.end = 0
   178  			return false
   179  		}
   180  		// Must read more data.
   181  		// First, shift data to beginning of buffer if there's lots of empty space
   182  		// or space is needed.
   183  		if s.start > 0 && (s.end == len(s.buf) || s.start > len(s.buf)/2) {
   184  			copy(s.buf, s.buf[s.start:s.end])
   185  			s.end -= s.start
   186  			s.start = 0
   187  		}
   188  		// Is the buffer full? If so, resize.
   189  		if s.end == len(s.buf) {
   190  			// Guarantee no overflow in the multiplication below.
   191  			const maxInt = int(^uint(0) >> 1)
   192  			if len(s.buf) >= s.maxTokenSize || len(s.buf) > maxInt/2 {
   193  				s.setErr(ErrTooLong)
   194  				return false
   195  			}
   196  			newSize := len(s.buf) * 2
   197  			if newSize == 0 {
   198  				newSize = startBufSize
   199  			}
   200  			if newSize > s.maxTokenSize {
   201  				newSize = s.maxTokenSize
   202  			}
   203  			newBuf := make([]byte, newSize)
   204  			copy(newBuf, s.buf[s.start:s.end])
   205  			s.buf = newBuf
   206  			s.end -= s.start
   207  			s.start = 0
   208  		}
   209  		// Finally we can read some input. Make sure we don't get stuck with
   210  		// a misbehaving Reader. Officially we don't need to do this, but let's
   211  		// be extra careful: Scanner is for safe, simple jobs.
   212  		for loop := 0; ; {
   213  			n, err := s.r.Read(s.buf[s.end:len(s.buf)])
   214  			s.end += n
   215  			if err != nil {
   216  				s.setErr(err)
   217  				break
   218  			}
   219  			if n > 0 {
   220  				s.empties = 0
   221  				break
   222  			}
   223  			loop++
   224  			if loop > maxConsecutiveEmptyReads {
   225  				s.setErr(io.ErrNoProgress)
   226  				break
   227  			}
   228  		}
   229  	}
   230  }
   231  
   232  // advance consumes n bytes of the buffer. It reports whether the advance was legal.
   233  func (s *Scanner) advance(n int) bool {
   234  	if n < 0 {
   235  		s.setErr(ErrNegativeAdvance)
   236  		return false
   237  	}
   238  	if n > s.end-s.start {
   239  		s.setErr(ErrAdvanceTooFar)
   240  		return false
   241  	}
   242  	s.start += n
   243  	return true
   244  }
   245  
   246  // setErr records the first error encountered.
   247  func (s *Scanner) setErr(err error) {
   248  	if s.err == nil || s.err == io.EOF {
   249  		s.err = err
   250  	}
   251  }
   252  
   253  // Buffer sets the initial buffer to use when scanning and the maximum
   254  // size of buffer that may be allocated during scanning. The maximum
   255  // token size is the larger of max and cap(buf). If max <= cap(buf),
   256  // Scan will use this buffer only and do no allocation.
   257  //
   258  // By default, Scan uses an internal buffer and sets the
   259  // maximum token size to MaxScanTokenSize.
   260  //
   261  // Buffer panics if it is called after scanning has started.
   262  func (s *Scanner) Buffer(buf []byte, max int) {
   263  	if s.scanCalled {
   264  		panic("Buffer called after Scan")
   265  	}
   266  	s.buf = buf[0:cap(buf)]
   267  	s.maxTokenSize = max
   268  }
   269  
   270  // Split sets the split function for the Scanner.
   271  // The default split function is ScanLines.
   272  //
   273  // Split panics if it is called after scanning has started.
   274  func (s *Scanner) Split(split SplitFunc) {
   275  	if s.scanCalled {
   276  		panic("Split called after Scan")
   277  	}
   278  	s.split = split
   279  }
   280  
   281  // Split functions
   282  
   283  // ScanBytes is a split function for a Scanner that returns each byte as a token.
   284  func ScanBytes(data []byte, atEOF bool) (advance int, token []byte, err error) {
   285  	if atEOF && len(data) == 0 {
   286  		return 0, nil, nil
   287  	}
   288  	return 1, data[0:1], nil
   289  }
   290  
   291  var errorRune = []byte(string(utf8.RuneError))
   292  
   293  // ScanRunes is a split function for a Scanner that returns each
   294  // UTF-8-encoded rune as a token. The sequence of runes returned is
   295  // equivalent to that from a range loop over the input as a string, which
   296  // means that erroneous UTF-8 encodings translate to U+FFFD = "\xef\xbf\xbd".
   297  // Because of the Scan interface, this makes it impossible for the client to
   298  // distinguish correctly encoded replacement runes from encoding errors.
   299  func ScanRunes(data []byte, atEOF bool) (advance int, token []byte, err error) {
   300  	if atEOF && len(data) == 0 {
   301  		return 0, nil, nil
   302  	}
   303  
   304  	// Fast path 1: ASCII.
   305  	if data[0] < utf8.RuneSelf {
   306  		return 1, data[0:1], nil
   307  	}
   308  
   309  	// Fast path 2: Correct UTF-8 decode without error.
   310  	_, width := utf8.DecodeRune(data)
   311  	if width > 1 {
   312  		// It's a valid encoding. Width cannot be one for a correctly encoded
   313  		// non-ASCII rune.
   314  		return width, data[0:width], nil
   315  	}
   316  
   317  	// We know it's an error: we have width==1 and implicitly r==utf8.RuneError.
   318  	// Is the error because there wasn't a full rune to be decoded?
   319  	// FullRune distinguishes correctly between erroneous and incomplete encodings.
   320  	if !atEOF && !utf8.FullRune(data) {
   321  		// Incomplete; get more bytes.
   322  		return 0, nil, nil
   323  	}
   324  
   325  	// We have a real UTF-8 encoding error. Return a properly encoded error rune
   326  	// but advance only one byte. This matches the behavior of a range loop over
   327  	// an incorrectly encoded string.
   328  	return 1, errorRune, nil
   329  }
   330  
   331  // dropCR drops a terminal \r from the data.
   332  func dropCR(data []byte) []byte {
   333  	if len(data) > 0 && data[len(data)-1] == '\r' {
   334  		return data[0 : len(data)-1]
   335  	}
   336  	return data
   337  }
   338  
   339  // ScanLines is a split function for a Scanner that returns each line of
   340  // text, stripped of any trailing end-of-line marker. The returned line may
   341  // be empty. The end-of-line marker is one optional carriage return followed
   342  // by one mandatory newline. In regular expression notation, it is `\r?\n`.
   343  // The last non-empty line of input will be returned even if it has no
   344  // newline.
   345  func ScanLines(data []byte, atEOF bool) (advance int, token []byte, err error) {
   346  	if atEOF && len(data) == 0 {
   347  		return 0, nil, nil
   348  	}
   349  	if i := bytes.IndexByte(data, '\n'); i >= 0 {
   350  		// We have a full newline-terminated line.
   351  		return i + 1, dropCR(data[0:i]), nil
   352  	}
   353  	// If we're at EOF, we have a final, non-terminated line. Return it.
   354  	if atEOF {
   355  		return len(data), dropCR(data), nil
   356  	}
   357  	// Request more data.
   358  	return 0, nil, nil
   359  }
   360  
   361  // isSpace reports whether the character is a Unicode white space character.
   362  // We avoid dependency on the unicode package, but check validity of the implementation
   363  // in the tests.
   364  func isSpace(r rune) bool {
   365  	if r <= '\u00FF' {
   366  		// Obvious ASCII ones: \t through \r plus space. Plus two Latin-1 oddballs.
   367  		switch r {
   368  		case ' ', '\t', '\n', '\v', '\f', '\r':
   369  			return true
   370  		case '\u0085', '\u00A0':
   371  			return true
   372  		}
   373  		return false
   374  	}
   375  	// High-valued ones.
   376  	if '\u2000' <= r && r <= '\u200a' {
   377  		return true
   378  	}
   379  	switch r {
   380  	case '\u1680', '\u2028', '\u2029', '\u202f', '\u205f', '\u3000':
   381  		return true
   382  	}
   383  	return false
   384  }
   385  
   386  // ScanWords is a split function for a Scanner that returns each
   387  // space-separated word of text, with surrounding spaces deleted. It will
   388  // never return an empty string. The definition of space is set by
   389  // unicode.IsSpace.
   390  func ScanWords(data []byte, atEOF bool) (advance int, token []byte, err error) {
   391  	// Skip leading spaces.
   392  	start := 0
   393  	for width := 0; start < len(data); start += width {
   394  		var r rune
   395  		r, width = utf8.DecodeRune(data[start:])
   396  		if !isSpace(r) {
   397  			break
   398  		}
   399  	}
   400  	// Scan until space, marking end of word.
   401  	for width, i := 0, start; i < len(data); i += width {
   402  		var r rune
   403  		r, width = utf8.DecodeRune(data[i:])
   404  		if isSpace(r) {
   405  			return i + width, data[start:i], nil
   406  		}
   407  	}
   408  	// If we're at EOF, we have a final, non-empty, non-terminated word. Return it.
   409  	if atEOF && len(data) > start {
   410  		return len(data), data[start:], nil
   411  	}
   412  	// Request more data.
   413  	return start, nil, nil
   414  }
   415  

View as plain text