...
Run Format

Source file src/unicode/utf8/utf8.go

Documentation: unicode/utf8

  // Copyright 2009 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
  
  // Package utf8 implements functions and constants to support text encoded in
  // UTF-8. It includes functions to translate between runes and UTF-8 byte sequences.
  package utf8
  
  // The conditions RuneError==unicode.ReplacementChar and
  // MaxRune==unicode.MaxRune are verified in the tests.
  // Defining them locally avoids this package depending on package unicode.
  
  // Numbers fundamental to the encoding.
  const (
  	RuneError = '\uFFFD'     // the "error" Rune or "Unicode replacement character"
  	RuneSelf  = 0x80         // characters below Runeself are represented as themselves in a single byte.
  	MaxRune   = '\U0010FFFF' // Maximum valid Unicode code point.
  	UTFMax    = 4            // maximum number of bytes of a UTF-8 encoded Unicode character.
  )
  
  // Code points in the surrogate range are not valid for UTF-8.
  const (
  	surrogateMin = 0xD800
  	surrogateMax = 0xDFFF
  )
  
  const (
  	t1 = 0x00 // 0000 0000
  	tx = 0x80 // 1000 0000
  	t2 = 0xC0 // 1100 0000
  	t3 = 0xE0 // 1110 0000
  	t4 = 0xF0 // 1111 0000
  	t5 = 0xF8 // 1111 1000
  
  	maskx = 0x3F // 0011 1111
  	mask2 = 0x1F // 0001 1111
  	mask3 = 0x0F // 0000 1111
  	mask4 = 0x07 // 0000 0111
  
  	rune1Max = 1<<7 - 1
  	rune2Max = 1<<11 - 1
  	rune3Max = 1<<16 - 1
  
  	// The default lowest and highest continuation byte.
  	locb = 0x80 // 1000 0000
  	hicb = 0xBF // 1011 1111
  
  	// These names of these constants are chosen to give nice alignment in the
  	// table below. The first nibble is an index into acceptRanges or F for
  	// special one-byte cases. The second nibble is the Rune length or the
  	// Status for the special one-byte case.
  	xx = 0xF1 // invalid: size 1
  	as = 0xF0 // ASCII: size 1
  	s1 = 0x02 // accept 0, size 2
  	s2 = 0x13 // accept 1, size 3
  	s3 = 0x03 // accept 0, size 3
  	s4 = 0x23 // accept 2, size 3
  	s5 = 0x34 // accept 3, size 4
  	s6 = 0x04 // accept 0, size 4
  	s7 = 0x44 // accept 4, size 4
  )
  
  // first is information about the first byte in a UTF-8 sequence.
  var first = [256]uint8{
  	//   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x00-0x0F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x10-0x1F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x20-0x2F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x30-0x3F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x40-0x4F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x50-0x5F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x60-0x6F
  	as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x70-0x7F
  	//   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x80-0x8F
  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x90-0x9F
  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xA0-0xAF
  	xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xB0-0xBF
  	xx, xx, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xC0-0xCF
  	s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xD0-0xDF
  	s2, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s4, s3, s3, // 0xE0-0xEF
  	s5, s6, s6, s6, s7, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xF0-0xFF
  }
  
  // acceptRange gives the range of valid values for the second byte in a UTF-8
  // sequence.
  type acceptRange struct {
  	lo uint8 // lowest value for second byte.
  	hi uint8 // highest value for second byte.
  }
  
  var acceptRanges = [...]acceptRange{
  	0: {locb, hicb},
  	1: {0xA0, hicb},
  	2: {locb, 0x9F},
  	3: {0x90, hicb},
  	4: {locb, 0x8F},
  }
  
  // FullRune reports whether the bytes in p begin with a full UTF-8 encoding of a rune.
  // An invalid encoding is considered a full Rune since it will convert as a width-1 error rune.
  func FullRune(p []byte) bool {
  	n := len(p)
  	if n == 0 {
  		return false
  	}
  	x := first[p[0]]
  	if n >= int(x&7) {
  		return true // ASCII, invalid or valid.
  	}
  	// Must be short or invalid.
  	accept := acceptRanges[x>>4]
  	if n > 1 {
  		if c := p[1]; c < accept.lo || accept.hi < c {
  			return true
  		} else if n > 2 && (p[2] < locb || hicb < p[2]) {
  			return true
  		}
  	}
  	return false
  }
  
  // FullRuneInString is like FullRune but its input is a string.
  func FullRuneInString(s string) bool {
  	n := len(s)
  	if n == 0 {
  		return false
  	}
  	x := first[s[0]]
  	if n >= int(x&7) {
  		return true // ASCII, invalid, or valid.
  	}
  	// Must be short or invalid.
  	accept := acceptRanges[x>>4]
  	if n > 1 {
  		if c := s[1]; c < accept.lo || accept.hi < c {
  			return true
  		} else if n > 2 && (s[2] < locb || hicb < s[2]) {
  			return true
  		}
  	}
  	return false
  }
  
  // DecodeRune unpacks the first UTF-8 encoding in p and returns the rune and
  // its width in bytes. If p is empty it returns (RuneError, 0). Otherwise, if
  // the encoding is invalid, it returns (RuneError, 1). Both are impossible
  // results for correct, non-empty UTF-8.
  //
  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
  // out of range, or is not the shortest possible UTF-8 encoding for the
  // value. No other validation is performed.
  func DecodeRune(p []byte) (r rune, size int) {
  	n := len(p)
  	if n < 1 {
  		return RuneError, 0
  	}
  	p0 := p[0]
  	x := first[p0]
  	if x >= as {
  		// The following code simulates an additional check for x == xx and
  		// handling the ASCII and invalid cases accordingly. This mask-and-or
  		// approach prevents an additional branch.
  		mask := rune(x) << 31 >> 31 // Create 0x0000 or 0xFFFF.
  		return rune(p[0])&^mask | RuneError&mask, 1
  	}
  	sz := x & 7
  	accept := acceptRanges[x>>4]
  	if n < int(sz) {
  		return RuneError, 1
  	}
  	b1 := p[1]
  	if b1 < accept.lo || accept.hi < b1 {
  		return RuneError, 1
  	}
  	if sz == 2 {
  		return rune(p0&mask2)<<6 | rune(b1&maskx), 2
  	}
  	b2 := p[2]
  	if b2 < locb || hicb < b2 {
  		return RuneError, 1
  	}
  	if sz == 3 {
  		return rune(p0&mask3)<<12 | rune(b1&maskx)<<6 | rune(b2&maskx), 3
  	}
  	b3 := p[3]
  	if b3 < locb || hicb < b3 {
  		return RuneError, 1
  	}
  	return rune(p0&mask4)<<18 | rune(b1&maskx)<<12 | rune(b2&maskx)<<6 | rune(b3&maskx), 4
  }
  
  // DecodeRuneInString is like DecodeRune but its input is a string. If s is
  // empty it returns (RuneError, 0). Otherwise, if the encoding is invalid, it
  // returns (RuneError, 1). Both are impossible results for correct, non-empty
  // UTF-8.
  //
  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
  // out of range, or is not the shortest possible UTF-8 encoding for the
  // value. No other validation is performed.
  func DecodeRuneInString(s string) (r rune, size int) {
  	n := len(s)
  	if n < 1 {
  		return RuneError, 0
  	}
  	s0 := s[0]
  	x := first[s0]
  	if x >= as {
  		// The following code simulates an additional check for x == xx and
  		// handling the ASCII and invalid cases accordingly. This mask-and-or
  		// approach prevents an additional branch.
  		mask := rune(x) << 31 >> 31 // Create 0x0000 or 0xFFFF.
  		return rune(s[0])&^mask | RuneError&mask, 1
  	}
  	sz := x & 7
  	accept := acceptRanges[x>>4]
  	if n < int(sz) {
  		return RuneError, 1
  	}
  	s1 := s[1]
  	if s1 < accept.lo || accept.hi < s1 {
  		return RuneError, 1
  	}
  	if sz == 2 {
  		return rune(s0&mask2)<<6 | rune(s1&maskx), 2
  	}
  	s2 := s[2]
  	if s2 < locb || hicb < s2 {
  		return RuneError, 1
  	}
  	if sz == 3 {
  		return rune(s0&mask3)<<12 | rune(s1&maskx)<<6 | rune(s2&maskx), 3
  	}
  	s3 := s[3]
  	if s3 < locb || hicb < s3 {
  		return RuneError, 1
  	}
  	return rune(s0&mask4)<<18 | rune(s1&maskx)<<12 | rune(s2&maskx)<<6 | rune(s3&maskx), 4
  }
  
  // DecodeLastRune unpacks the last UTF-8 encoding in p and returns the rune and
  // its width in bytes. If p is empty it returns (RuneError, 0). Otherwise, if
  // the encoding is invalid, it returns (RuneError, 1). Both are impossible
  // results for correct, non-empty UTF-8.
  //
  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
  // out of range, or is not the shortest possible UTF-8 encoding for the
  // value. No other validation is performed.
  func DecodeLastRune(p []byte) (r rune, size int) {
  	end := len(p)
  	if end == 0 {
  		return RuneError, 0
  	}
  	start := end - 1
  	r = rune(p[start])
  	if r < RuneSelf {
  		return r, 1
  	}
  	// guard against O(n^2) behavior when traversing
  	// backwards through strings with long sequences of
  	// invalid UTF-8.
  	lim := end - UTFMax
  	if lim < 0 {
  		lim = 0
  	}
  	for start--; start >= lim; start-- {
  		if RuneStart(p[start]) {
  			break
  		}
  	}
  	if start < 0 {
  		start = 0
  	}
  	r, size = DecodeRune(p[start:end])
  	if start+size != end {
  		return RuneError, 1
  	}
  	return r, size
  }
  
  // DecodeLastRuneInString is like DecodeLastRune but its input is a string. If
  // s is empty it returns (RuneError, 0). Otherwise, if the encoding is invalid,
  // it returns (RuneError, 1). Both are impossible results for correct,
  // non-empty UTF-8.
  //
  // An encoding is invalid if it is incorrect UTF-8, encodes a rune that is
  // out of range, or is not the shortest possible UTF-8 encoding for the
  // value. No other validation is performed.
  func DecodeLastRuneInString(s string) (r rune, size int) {
  	end := len(s)
  	if end == 0 {
  		return RuneError, 0
  	}
  	start := end - 1
  	r = rune(s[start])
  	if r < RuneSelf {
  		return r, 1
  	}
  	// guard against O(n^2) behavior when traversing
  	// backwards through strings with long sequences of
  	// invalid UTF-8.
  	lim := end - UTFMax
  	if lim < 0 {
  		lim = 0
  	}
  	for start--; start >= lim; start-- {
  		if RuneStart(s[start]) {
  			break
  		}
  	}
  	if start < 0 {
  		start = 0
  	}
  	r, size = DecodeRuneInString(s[start:end])
  	if start+size != end {
  		return RuneError, 1
  	}
  	return r, size
  }
  
  // RuneLen returns the number of bytes required to encode the rune.
  // It returns -1 if the rune is not a valid value to encode in UTF-8.
  func RuneLen(r rune) int {
  	switch {
  	case r < 0:
  		return -1
  	case r <= rune1Max:
  		return 1
  	case r <= rune2Max:
  		return 2
  	case surrogateMin <= r && r <= surrogateMax:
  		return -1
  	case r <= rune3Max:
  		return 3
  	case r <= MaxRune:
  		return 4
  	}
  	return -1
  }
  
  // EncodeRune writes into p (which must be large enough) the UTF-8 encoding of the rune.
  // It returns the number of bytes written.
  func EncodeRune(p []byte, r rune) int {
  	// Negative values are erroneous. Making it unsigned addresses the problem.
  	switch i := uint32(r); {
  	case i <= rune1Max:
  		p[0] = byte(r)
  		return 1
  	case i <= rune2Max:
  		_ = p[1] // eliminate bounds checks
  		p[0] = t2 | byte(r>>6)
  		p[1] = tx | byte(r)&maskx
  		return 2
  	case i > MaxRune, surrogateMin <= i && i <= surrogateMax:
  		r = RuneError
  		fallthrough
  	case i <= rune3Max:
  		_ = p[2] // eliminate bounds checks
  		p[0] = t3 | byte(r>>12)
  		p[1] = tx | byte(r>>6)&maskx
  		p[2] = tx | byte(r)&maskx
  		return 3
  	default:
  		_ = p[3] // eliminate bounds checks
  		p[0] = t4 | byte(r>>18)
  		p[1] = tx | byte(r>>12)&maskx
  		p[2] = tx | byte(r>>6)&maskx
  		p[3] = tx | byte(r)&maskx
  		return 4
  	}
  }
  
  // RuneCount returns the number of runes in p. Erroneous and short
  // encodings are treated as single runes of width 1 byte.
  func RuneCount(p []byte) int {
  	np := len(p)
  	var n int
  	for i := 0; i < np; {
  		n++
  		c := p[i]
  		if c < RuneSelf {
  			// ASCII fast path
  			i++
  			continue
  		}
  		x := first[c]
  		if x == xx {
  			i++ // invalid.
  			continue
  		}
  		size := int(x & 7)
  		if i+size > np {
  			i++ // Short or invalid.
  			continue
  		}
  		accept := acceptRanges[x>>4]
  		if c := p[i+1]; c < accept.lo || accept.hi < c {
  			size = 1
  		} else if size == 2 {
  		} else if c := p[i+2]; c < locb || hicb < c {
  			size = 1
  		} else if size == 3 {
  		} else if c := p[i+3]; c < locb || hicb < c {
  			size = 1
  		}
  		i += size
  	}
  	return n
  }
  
  // RuneCountInString is like RuneCount but its input is a string.
  func RuneCountInString(s string) (n int) {
  	ns := len(s)
  	for i := 0; i < ns; n++ {
  		c := s[i]
  		if c < RuneSelf {
  			// ASCII fast path
  			i++
  			continue
  		}
  		x := first[c]
  		if x == xx {
  			i++ // invalid.
  			continue
  		}
  		size := int(x & 7)
  		if i+size > ns {
  			i++ // Short or invalid.
  			continue
  		}
  		accept := acceptRanges[x>>4]
  		if c := s[i+1]; c < accept.lo || accept.hi < c {
  			size = 1
  		} else if size == 2 {
  		} else if c := s[i+2]; c < locb || hicb < c {
  			size = 1
  		} else if size == 3 {
  		} else if c := s[i+3]; c < locb || hicb < c {
  			size = 1
  		}
  		i += size
  	}
  	return n
  }
  
  // RuneStart reports whether the byte could be the first byte of an encoded,
  // possibly invalid rune. Second and subsequent bytes always have the top two
  // bits set to 10.
  func RuneStart(b byte) bool { return b&0xC0 != 0x80 }
  
  // Valid reports whether p consists entirely of valid UTF-8-encoded runes.
  func Valid(p []byte) bool {
  	n := len(p)
  	for i := 0; i < n; {
  		pi := p[i]
  		if pi < RuneSelf {
  			i++
  			continue
  		}
  		x := first[pi]
  		if x == xx {
  			return false // Illegal starter byte.
  		}
  		size := int(x & 7)
  		if i+size > n {
  			return false // Short or invalid.
  		}
  		accept := acceptRanges[x>>4]
  		if c := p[i+1]; c < accept.lo || accept.hi < c {
  			return false
  		} else if size == 2 {
  		} else if c := p[i+2]; c < locb || hicb < c {
  			return false
  		} else if size == 3 {
  		} else if c := p[i+3]; c < locb || hicb < c {
  			return false
  		}
  		i += size
  	}
  	return true
  }
  
  // ValidString reports whether s consists entirely of valid UTF-8-encoded runes.
  func ValidString(s string) bool {
  	n := len(s)
  	for i := 0; i < n; {
  		si := s[i]
  		if si < RuneSelf {
  			i++
  			continue
  		}
  		x := first[si]
  		if x == xx {
  			return false // Illegal starter byte.
  		}
  		size := int(x & 7)
  		if i+size > n {
  			return false // Short or invalid.
  		}
  		accept := acceptRanges[x>>4]
  		if c := s[i+1]; c < accept.lo || accept.hi < c {
  			return false
  		} else if size == 2 {
  		} else if c := s[i+2]; c < locb || hicb < c {
  			return false
  		} else if size == 3 {
  		} else if c := s[i+3]; c < locb || hicb < c {
  			return false
  		}
  		i += size
  	}
  	return true
  }
  
  // ValidRune reports whether r can be legally encoded as UTF-8.
  // Code points that are out of range or a surrogate half are illegal.
  func ValidRune(r rune) bool {
  	switch {
  	case 0 <= r && r < surrogateMin:
  		return true
  	case surrogateMax < r && r <= MaxRune:
  		return true
  	}
  	return false
  }
  

View as plain text