1// Copyright 2016 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 5package runtime 6 7import _ "unsafe" // For go:linkname. 8 9// For gccgo, use go:linkname to export compiler-called functions. 10// 11//go:linkname decoderune 12 13// Numbers fundamental to the encoding. 14const ( 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) 19 20// Code points in the surrogate range are not valid for UTF-8. 21const ( 22 surrogateMin = 0xD800 23 surrogateMax = 0xDFFF 24) 25 26const ( 27 t1 = 0x00 // 0000 0000 28 tx = 0x80 // 1000 0000 29 t2 = 0xC0 // 1100 0000 30 t3 = 0xE0 // 1110 0000 31 t4 = 0xF0 // 1111 0000 32 t5 = 0xF8 // 1111 1000 33 34 maskx = 0x3F // 0011 1111 35 mask2 = 0x1F // 0001 1111 36 mask3 = 0x0F // 0000 1111 37 mask4 = 0x07 // 0000 0111 38 39 rune1Max = 1<<7 - 1 40 rune2Max = 1<<11 - 1 41 rune3Max = 1<<16 - 1 42 43 // The default lowest and highest continuation byte. 44 locb = 0x80 // 1000 0000 45 hicb = 0xBF // 1011 1111 46) 47 48// countrunes returns the number of runes in s. 49func countrunes(s string) int { 50 n := 0 51 for range s { 52 n++ 53 } 54 return n 55} 56 57// decoderune returns the non-ASCII rune at the start of 58// s[k:] and the index after the rune in s. 59// 60// decoderune assumes that caller has checked that 61// the to be decoded rune is a non-ASCII rune. 62// 63// If the string appears to be incomplete or decoding problems 64// are encountered (runeerror, k + 1) is returned to ensure 65// progress when decoderune is used to iterate over a string. 66func decoderune(s string, k int) (r rune, pos int) { 67 pos = k 68 69 if k >= len(s) { 70 return runeError, k + 1 71 } 72 73 s = s[k:] 74 75 switch { 76 case t2 <= s[0] && s[0] < t3: 77 // 0080-07FF two byte sequence 78 if len(s) > 1 && (locb <= s[1] && s[1] <= hicb) { 79 r = rune(s[0]&mask2)<<6 | rune(s[1]&maskx) 80 pos += 2 81 if rune1Max < r { 82 return 83 } 84 } 85 case t3 <= s[0] && s[0] < t4: 86 // 0800-FFFF three byte sequence 87 if len(s) > 2 && (locb <= s[1] && s[1] <= hicb) && (locb <= s[2] && s[2] <= hicb) { 88 r = rune(s[0]&mask3)<<12 | rune(s[1]&maskx)<<6 | rune(s[2]&maskx) 89 pos += 3 90 if rune2Max < r && !(surrogateMin <= r && r <= surrogateMax) { 91 return 92 } 93 } 94 case t4 <= s[0] && s[0] < t5: 95 // 10000-1FFFFF four byte sequence 96 if len(s) > 3 && (locb <= s[1] && s[1] <= hicb) && (locb <= s[2] && s[2] <= hicb) && (locb <= s[3] && s[3] <= hicb) { 97 r = rune(s[0]&mask4)<<18 | rune(s[1]&maskx)<<12 | rune(s[2]&maskx)<<6 | rune(s[3]&maskx) 98 pos += 4 99 if rune3Max < r && r <= maxRune { 100 return 101 } 102 } 103 } 104 105 return runeError, k + 1 106} 107 108// encoderune writes into p (which must be large enough) the UTF-8 encoding of the rune. 109// It returns the number of bytes written. 110func encoderune(p []byte, r rune) int { 111 // Negative values are erroneous. Making it unsigned addresses the problem. 112 switch i := uint32(r); { 113 case i <= rune1Max: 114 p[0] = byte(r) 115 return 1 116 case i <= rune2Max: 117 _ = p[1] // eliminate bounds checks 118 p[0] = t2 | byte(r>>6) 119 p[1] = tx | byte(r)&maskx 120 return 2 121 case i > maxRune, surrogateMin <= i && i <= surrogateMax: 122 r = runeError 123 fallthrough 124 case i <= rune3Max: 125 _ = p[2] // eliminate bounds checks 126 p[0] = t3 | byte(r>>12) 127 p[1] = tx | byte(r>>6)&maskx 128 p[2] = tx | byte(r)&maskx 129 return 3 130 default: 131 _ = p[3] // eliminate bounds checks 132 p[0] = t4 | byte(r>>18) 133 p[1] = tx | byte(r>>12)&maskx 134 p[2] = tx | byte(r>>6)&maskx 135 p[3] = tx | byte(r)&maskx 136 return 4 137 } 138} 139