utf8.go

Documentation: unicode/utf8

     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 utf8 implements functions and constants to support text encoded in
     6  // UTF-8. It includes functions to translate between runes and UTF-8 byte sequences.
     7  package utf8
     8  
     9  // The conditions RuneError==unicode.ReplacementChar and
    10  // MaxRune==unicode.MaxRune are verified in the tests.
    11  // Defining them locally avoids this package depending on package unicode.
    12  
    13  // Numbers fundamental to the encoding.
    14  const (
    15  	RuneError = '\uFFFD'     // the "error" Rune or "Unicode replacement character"
    16  	RuneSelf  = 0x80         // characters below Runeself are represented as themselves in a single byte.
    17  	MaxRune   = '\U0010FFFF' // Maximum valid Unicode code point.
    18  	UTFMax    = 4            // maximum number of bytes of a UTF-8 encoded Unicode character.
    19  )
    20  
    21  // Code points in the surrogate range are not valid for UTF-8.
    22  const (
    23  	surrogateMin = 0xD800
    24  	surrogateMax = 0xDFFF
    25  )
    26  
    27  const (
    28  	t1 = 0x00 // 0000 0000
    29  	tx = 0x80 // 1000 0000
    30  	t2 = 0xC0 // 1100 0000
    31  	t3 = 0xE0 // 1110 0000
    32  	t4 = 0xF0 // 1111 0000
    33  	t5 = 0xF8 // 1111 1000
    34  
    35  	maskx = 0x3F // 0011 1111
    36  	mask2 = 0x1F // 0001 1111
    37  	mask3 = 0x0F // 0000 1111
    38  	mask4 = 0x07 // 0000 0111
    39  
    40  	rune1Max = 1<<7 - 1
    41  	rune2Max = 1<<11 - 1
    42  	rune3Max = 1<<16 - 1
    43  
    44  	// The default lowest and highest continuation byte.
    45  	locb = 0x80 // 1000 0000
    46  	hicb = 0xBF // 1011 1111
    47  
    48  	// These names of these constants are chosen to give nice alignment in the
    49  	// table below. The first nibble is an index into acceptRanges or F for
    50  	// special one-byte cases. The second nibble is the Rune length or the
    51  	// Status for the special one-byte case.
    52  	xx = 0xF1 // invalid: size 1
    53  	as = 0xF0 // ASCII: size 1
    54  	s1 = 0x02 // accept 0, size 2
    55  	s2 = 0x13 // accept 1, size 3
    56  	s3 = 0x03 // accept 0, size 3
    57  	s4 = 0x23 // accept 2, size 3
    58  	s5 = 0x34 // accept 3, size 4
    59  	s6 = 0x04 // accept 0, size 4
    60  	s7 = 0x44 // accept 4, size 4
    61  )
    62  
    63  // first is information about the first byte in a UTF-8 sequence.
    64  var first = [256]uint8{
    65  	//   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
    66  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x00-0x0F
    67  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x10-0x1F
    68  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x20-0x2F
    69  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x30-0x3F
    70  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x40-0x4F
    71  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x50-0x5F
    72  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x60-0x6F
    73  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x70-0x7F
    74  	//   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
    75  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x80-0x8F
    76  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x90-0x9F
    77  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xA0-0xAF
    78  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xB0-0xBF
    79  	xx, xx, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xC0-0xCF
    80  	s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xD0-0xDF
    81  	s2, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s4, s3, s3, // 0xE0-0xEF
    82  	s5, s6, s6, s6, s7, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xF0-0xFF
    83  }
    84  
    85  // acceptRange gives the range of valid values for the second byte in a UTF-8
    86  // sequence.
    87  type acceptRange struct {
    88  	lo uint8 // lowest value for second byte.
    89  	hi uint8 // highest value for second byte.
    90  }
    91  
    92  var acceptRanges = [...]acceptRange{
    93  	0: {locb, hicb},
    94  	1: {0xA0, hicb},
    95  	2: {locb, 0x9F},
    96  	3: {0x90, hicb},
    97  	4: {locb, 0x8F},
    98  }
    99  
   100  // FullRune reports whether the bytes in p begin with a full UTF-8 encoding of a rune.
   101  // An invalid encoding is considered a full Rune since it will convert as a width-1 error rune.
   102  func FullRune(p []byte) bool {
   103  	n := len(p)
   104  	if n == 0 {
   105  		return false
   106  	}
   107  	x := first[p[0]]
   108  	if n >= int(x&7) {
   109  		return true // ASCII, invalid or valid.
   110  	}
   111  	// Must be short or invalid.
   112  	accept := acceptRanges[x>>4]
   113  	if n > 1 && (p[1] < accept.lo || accept.hi < p[1]) {
   114  		return true
   115  	} else if n > 2 && (p[2] < locb || hicb < p[2]) {
   116  		return true
   117  	}
   118  	return false
   119  }
   120  
   121  // FullRuneInString is like FullRune but its input is a string.
   122  func FullRuneInString(s string) bool {
   123  	n := len(s)
   124  	if n == 0 {
   125  		return false
   126  	}
   127  	x := first[s[0]]
   128  	if n >= int(x&7) {
   129  		return true // ASCII, invalid, or valid.
   130  	}
   131  	// Must be short or invalid.
   132  	accept := acceptRanges[x>>4]
   133  	if n > 1 && (s[1] < accept.lo || accept.hi < s[1]) {
   134  		return true
   135  	} else if n > 2 && (s[2] < locb || hicb < s[2]) {
   136  		return true
   137  	}
   138  	return false
   139  }
   140  
   141  // DecodeRune unpacks the first UTF-8 encoding in p and returns the rune and
   142  // its width in bytes. If p is empty it returns (RuneError, 0). Otherwise, if
   143  // the encoding is invalid, it returns (RuneError, 1). Both are impossible
   144  // results for correct, non-empty UTF-8.
   145  //
   146  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
   147  // out of range, or is not the shortest possible UTF-8 encoding for the
   148  // value. No other validation is performed.
   149  func DecodeRune(p []byte) (r rune, size int) {
   150  	n := len(p)
   151  	if n < 1 {
   152  		return RuneError, 0
   153  	}
   154  	p0 := p[0]
   155  	x := first[p0]
   156  	if x >= as {
   157  		// The following code simulates an additional check for x == xx and
   158  		// handling the ASCII and invalid cases accordingly. This mask-and-or
   159  		// approach prevents an additional branch.
   160  		mask := rune(x) << 31 >> 31 // Create 0x0000 or 0xFFFF.
   161  		return rune(p[0])&^mask | RuneError&mask, 1
   162  	}
   163  	sz := x & 7
   164  	accept := acceptRanges[x>>4]
   165  	if n < int(sz) {
   166  		return RuneError, 1
   167  	}
   168  	b1 := p[1]
   169  	if b1 < accept.lo || accept.hi < b1 {
   170  		return RuneError, 1
   171  	}
   172  	if sz == 2 {
   173  		return rune(p0&mask2)<<6 | rune(b1&maskx), 2
   174  	}
   175  	b2 := p[2]
   176  	if b2 < locb || hicb < b2 {
   177  		return RuneError, 1
   178  	}
   179  	if sz == 3 {
   180  		return rune(p0&mask3)<<12 | rune(b1&maskx)<<6 | rune(b2&maskx), 3
   181  	}
   182  	b3 := p[3]
   183  	if b3 < locb || hicb < b3 {
   184  		return RuneError, 1
   185  	}
   186  	return rune(p0&mask4)<<18 | rune(b1&maskx)<<12 | rune(b2&maskx)<<6 | rune(b3&maskx), 4
   187  }
   188  
   189  // DecodeRuneInString is like DecodeRune but its input is a string. If s is
   190  // empty it returns (RuneError, 0). Otherwise, if the encoding is invalid, it
   191  // returns (RuneError, 1). Both are impossible results for correct, non-empty
   192  // UTF-8.
   193  //
   194  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
   195  // out of range, or is not the shortest possible UTF-8 encoding for the
   196  // value. No other validation is performed.
   197  func DecodeRuneInString(s string) (r rune, size int) {
   198  	n := len(s)
   199  	if n < 1 {
   200  		return RuneError, 0
   201  	}
   202  	s0 := s[0]
   203  	x := first[s0]
   204  	if x >= as {
   205  		// The following code simulates an additional check for x == xx and
   206  		// handling the ASCII and invalid cases accordingly. This mask-and-or
   207  		// approach prevents an additional branch.
   208  		mask := rune(x) << 31 >> 31 // Create 0x0000 or 0xFFFF.
   209  		return rune(s[0])&^mask | RuneError&mask, 1
   210  	}
   211  	sz := x & 7
   212  	accept := acceptRanges[x>>4]
   213  	if n < int(sz) {
   214  		return RuneError, 1
   215  	}
   216  	s1 := s[1]
   217  	if s1 < accept.lo || accept.hi < s1 {
   218  		return RuneError, 1
   219  	}
   220  	if sz == 2 {
   221  		return rune(s0&mask2)<<6 | rune(s1&maskx), 2
   222  	}
   223  	s2 := s[2]
   224  	if s2 < locb || hicb < s2 {
   225  		return RuneError, 1
   226  	}
   227  	if sz == 3 {
   228  		return rune(s0&mask3)<<12 | rune(s1&maskx)<<6 | rune(s2&maskx), 3
   229  	}
   230  	s3 := s[3]
   231  	if s3 < locb || hicb < s3 {
   232  		return RuneError, 1
   233  	}
   234  	return rune(s0&mask4)<<18 | rune(s1&maskx)<<12 | rune(s2&maskx)<<6 | rune(s3&maskx), 4
   235  }
   236  
   237  // DecodeLastRune unpacks the last UTF-8 encoding in p and returns the rune and
   238  // its width in bytes. If p is empty it returns (RuneError, 0). Otherwise, if
   239  // the encoding is invalid, it returns (RuneError, 1). Both are impossible
   240  // results for correct, non-empty UTF-8.
   241  //
   242  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
   243  // out of range, or is not the shortest possible UTF-8 encoding for the
   244  // value. No other validation is performed.
   245  func DecodeLastRune(p []byte) (r rune, size int) {
   246  	end := len(p)
   247  	if end == 0 {
   248  		return RuneError, 0
   249  	}
   250  	start := end - 1
   251  	r = rune(p[start])
   252  	if r < RuneSelf {
   253  		return r, 1
   254  	}
   255  	// guard against O(n^2) behavior when traversing
   256  	// backwards through strings with long sequences of
   257  	// invalid UTF-8.
   258  	lim := end - UTFMax
   259  	if lim < 0 {
   260  		lim = 0
   261  	}
   262  	for start--; start >= lim; start-- {
   263  		if RuneStart(p[start]) {
   264  			break
   265  		}
   266  	}
   267  	if start < 0 {
   268  		start = 0
   269  	}
   270  	r, size = DecodeRune(p[start:end])
   271  	if start+size != end {
   272  		return RuneError, 1
   273  	}
   274  	return r, size
   275  }
   276  
   277  // DecodeLastRuneInString is like DecodeLastRune but its input is a string. If
   278  // s is empty it returns (RuneError, 0). Otherwise, if the encoding is invalid,
   279  // it returns (RuneError, 1). Both are impossible results for correct,
   280  // non-empty UTF-8.
   281  //
   282  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
   283  // out of range, or is not the shortest possible UTF-8 encoding for the
   284  // value. No other validation is performed.
   285  func DecodeLastRuneInString(s string) (r rune, size int) {
   286  	end := len(s)
   287  	if end == 0 {
   288  		return RuneError, 0
   289  	}
   290  	start := end - 1
   291  	r = rune(s[start])
   292  	if r < RuneSelf {
   293  		return r, 1
   294  	}
   295  	// guard against O(n^2) behavior when traversing
   296  	// backwards through strings with long sequences of
   297  	// invalid UTF-8.
   298  	lim := end - UTFMax
   299  	if lim < 0 {
   300  		lim = 0
   301  	}
   302  	for start--; start >= lim; start-- {
   303  		if RuneStart(s[start]) {
   304  			break
   305  		}
   306  	}
   307  	if start < 0 {
   308  		start = 0
   309  	}
   310  	r, size = DecodeRuneInString(s[start:end])
   311  	if start+size != end {
   312  		return RuneError, 1
   313  	}
   314  	return r, size
   315  }
   316  
   317  // RuneLen returns the number of bytes required to encode the rune.
   318  // It returns -1 if the rune is not a valid value to encode in UTF-8.
   319  func RuneLen(r rune) int {
   320  	switch {
   321  	case r < 0:
   322  		return -1
   323  	case r <= rune1Max:
   324  		return 1
   325  	case r <= rune2Max:
   326  		return 2
   327  	case surrogateMin <= r && r <= surrogateMax:
   328  		return -1
   329  	case r <= rune3Max:
   330  		return 3
   331  	case r <= MaxRune:
   332  		return 4
   333  	}
   334  	return -1
   335  }
   336  
   337  // EncodeRune writes into p (which must be large enough) the UTF-8 encoding of the rune.
   338  // It returns the number of bytes written.
   339  func EncodeRune(p []byte, r rune) int {
   340  	// Negative values are erroneous. Making it unsigned addresses the problem.
   341  	switch i := uint32(r); {
   342  	case i <= rune1Max:
   343  		p[0] = byte(r)
   344  		return 1
   345  	case i <= rune2Max:
   346  		_ = p[1] // eliminate bounds checks
   347  		p[0] = t2 | byte(r>>6)
   348  		p[1] = tx | byte(r)&maskx
   349  		return 2
   350  	case i > MaxRune, surrogateMin <= i && i <= surrogateMax:
   351  		r = RuneError
   352  		fallthrough
   353  	case i <= rune3Max:
   354  		_ = p[2] // eliminate bounds checks
   355  		p[0] = t3 | byte(r>>12)
   356  		p[1] = tx | byte(r>>6)&maskx
   357  		p[2] = tx | byte(r)&maskx
   358  		return 3
   359  	default:
   360  		_ = p[3] // eliminate bounds checks
   361  		p[0] = t4 | byte(r>>18)
   362  		p[1] = tx | byte(r>>12)&maskx
   363  		p[2] = tx | byte(r>>6)&maskx
   364  		p[3] = tx | byte(r)&maskx
   365  		return 4
   366  	}
   367  }
   368  
   369  // RuneCount returns the number of runes in p. Erroneous and short
   370  // encodings are treated as single runes of width 1 byte.
   371  func RuneCount(p []byte) int {
   372  	np := len(p)
   373  	var n int
   374  	for i := 0; i < np; {
   375  		n++
   376  		c := p[i]
   377  		if c < RuneSelf {
   378  			// ASCII fast path
   379  			i++
   380  			continue
   381  		}
   382  		x := first[c]
   383  		if x == xx {
   384  			i++ // invalid.
   385  			continue
   386  		}
   387  		size := int(x & 7)
   388  		if i+size > np {
   389  			i++ // Short or invalid.
   390  			continue
   391  		}
   392  		accept := acceptRanges[x>>4]
   393  		if c := p[i+1]; c < accept.lo || accept.hi < c {
   394  			size = 1
   395  		} else if size == 2 {
   396  		} else if c := p[i+2]; c < locb || hicb < c {
   397  			size = 1
   398  		} else if size == 3 {
   399  		} else if c := p[i+3]; c < locb || hicb < c {
   400  			size = 1
   401  		}
   402  		i += size
   403  	}
   404  	return n
   405  }
   406  
   407  // RuneCountInString is like RuneCount but its input is a string.
   408  func RuneCountInString(s string) (n int) {
   409  	ns := len(s)
   410  	for i := 0; i < ns; n++ {
   411  		c := s[i]
   412  		if c < RuneSelf {
   413  			// ASCII fast path
   414  			i++
   415  			continue
   416  		}
   417  		x := first[c]
   418  		if x == xx {
   419  			i++ // invalid.
   420  			continue
   421  		}
   422  		size := int(x & 7)
   423  		if i+size > ns {
   424  			i++ // Short or invalid.
   425  			continue
   426  		}
   427  		accept := acceptRanges[x>>4]
   428  		if c := s[i+1]; c < accept.lo || accept.hi < c {
   429  			size = 1
   430  		} else if size == 2 {
   431  		} else if c := s[i+2]; c < locb || hicb < c {
   432  			size = 1
   433  		} else if size == 3 {
   434  		} else if c := s[i+3]; c < locb || hicb < c {
   435  			size = 1
   436  		}
   437  		i += size
   438  	}
   439  	return n
   440  }
   441  
   442  // RuneStart reports whether the byte could be the first byte of an encoded,
   443  // possibly invalid rune. Second and subsequent bytes always have the top two
   444  // bits set to 10.
   445  func RuneStart(b byte) bool { return b&0xC0 != 0x80 }
   446  
   447  // Valid reports whether p consists entirely of valid UTF-8-encoded runes.
   448  func Valid(p []byte) bool {
   449  	n := len(p)
   450  	for i := 0; i < n; {
   451  		pi := p[i]
   452  		if pi < RuneSelf {
   453  			i++
   454  			continue
   455  		}
   456  		x := first[pi]
   457  		if x == xx {
   458  			return false // Illegal starter byte.
   459  		}
   460  		size := int(x & 7)
   461  		if i+size > n {
   462  			return false // Short or invalid.
   463  		}
   464  		accept := acceptRanges[x>>4]
   465  		if c := p[i+1]; c < accept.lo || accept.hi < c {
   466  			return false
   467  		} else if size == 2 {
   468  		} else if c := p[i+2]; c < locb || hicb < c {
   469  			return false
   470  		} else if size == 3 {
   471  		} else if c := p[i+3]; c < locb || hicb < c {
   472  			return false
   473  		}
   474  		i += size
   475  	}
   476  	return true
   477  }
   478  
   479  // ValidString reports whether s consists entirely of valid UTF-8-encoded runes.
   480  func ValidString(s string) bool {
   481  	n := len(s)
   482  	for i := 0; i < n; {
   483  		si := s[i]
   484  		if si < RuneSelf {
   485  			i++
   486  			continue
   487  		}
   488  		x := first[si]
   489  		if x == xx {
   490  			return false // Illegal starter byte.
   491  		}
   492  		size := int(x & 7)
   493  		if i+size > n {
   494  			return false // Short or invalid.
   495  		}
   496  		accept := acceptRanges[x>>4]
   497  		if c := s[i+1]; c < accept.lo || accept.hi < c {
   498  			return false
   499  		} else if size == 2 {
   500  		} else if c := s[i+2]; c < locb || hicb < c {
   501  			return false
   502  		} else if size == 3 {
   503  		} else if c := s[i+3]; c < locb || hicb < c {
   504  			return false
   505  		}
   506  		i += size
   507  	}
   508  	return true
   509  }
   510  
   511  // ValidRune reports whether r can be legally encoded as UTF-8.
   512  // Code points that are out of range or a surrogate half are illegal.
   513  func ValidRune(r rune) bool {
   514  	switch {
   515  	case 0 <= r && r < surrogateMin:
   516  		return true
   517  	case surrogateMax < r && r <= MaxRune:
   518  		return true
   519  	}
   520  	return false
   521  }
   522
View as plain text