decode_rune.go (3767B)
1 package encoder 2 3 import "unicode/utf8" 4 5 const ( 6 // The default lowest and highest continuation byte. 7 locb = 128 //0b10000000 8 hicb = 191 //0b10111111 9 10 // These names of these constants are chosen to give nice alignment in the 11 // table below. The first nibble is an index into acceptRanges or F for 12 // special one-byte cases. The second nibble is the Rune length or the 13 // Status for the special one-byte case. 14 xx = 0xF1 // invalid: size 1 15 as = 0xF0 // ASCII: size 1 16 s1 = 0x02 // accept 0, size 2 17 s2 = 0x13 // accept 1, size 3 18 s3 = 0x03 // accept 0, size 3 19 s4 = 0x23 // accept 2, size 3 20 s5 = 0x34 // accept 3, size 4 21 s6 = 0x04 // accept 0, size 4 22 s7 = 0x44 // accept 4, size 4 23 ) 24 25 // first is information about the first byte in a UTF-8 sequence. 26 var first = [256]uint8{ 27 // 1 2 3 4 5 6 7 8 9 A B C D E F 28 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x00-0x0F 29 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x10-0x1F 30 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x20-0x2F 31 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x30-0x3F 32 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x40-0x4F 33 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x50-0x5F 34 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x60-0x6F 35 as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, as, // 0x70-0x7F 36 // 1 2 3 4 5 6 7 8 9 A B C D E F 37 xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x80-0x8F 38 xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0x90-0x9F 39 xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xA0-0xAF 40 xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xB0-0xBF 41 xx, xx, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xC0-0xCF 42 s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, s1, // 0xD0-0xDF 43 s2, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s3, s4, s3, s3, // 0xE0-0xEF 44 s5, s6, s6, s6, s7, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, xx, // 0xF0-0xFF 45 } 46 47 const ( 48 lineSep = byte(168) //'\u2028' 49 paragraphSep = byte(169) //'\u2029' 50 ) 51 52 type decodeRuneState int 53 54 const ( 55 validUTF8State decodeRuneState = iota 56 runeErrorState 57 lineSepState 58 paragraphSepState 59 ) 60 61 func decodeRuneInString(s string) (decodeRuneState, int) { 62 n := len(s) 63 s0 := s[0] 64 x := first[s0] 65 if x >= as { 66 // The following code simulates an additional check for x == xx and 67 // handling the ASCII and invalid cases accordingly. This mask-and-or 68 // approach prevents an additional branch. 69 mask := rune(x) << 31 >> 31 // Create 0x0000 or 0xFFFF. 70 if rune(s[0])&^mask|utf8.RuneError&mask == utf8.RuneError { 71 return runeErrorState, 1 72 } 73 return validUTF8State, 1 74 } 75 sz := int(x & 7) 76 if n < sz { 77 return runeErrorState, 1 78 } 79 s1 := s[1] 80 switch x >> 4 { 81 case 0: 82 if s1 < locb || hicb < s1 { 83 return runeErrorState, 1 84 } 85 case 1: 86 if s1 < 0xA0 || hicb < s1 { 87 return runeErrorState, 1 88 } 89 case 2: 90 if s1 < locb || 0x9F < s1 { 91 return runeErrorState, 1 92 } 93 case 3: 94 if s1 < 0x90 || hicb < s1 { 95 return runeErrorState, 1 96 } 97 case 4: 98 if s1 < locb || 0x8F < s1 { 99 return runeErrorState, 1 100 } 101 } 102 if sz <= 2 { 103 return validUTF8State, 2 104 } 105 s2 := s[2] 106 if s2 < locb || hicb < s2 { 107 return runeErrorState, 1 108 } 109 if sz <= 3 { 110 // separator character prefixes: [2]byte{226, 128} 111 if s0 == 226 && s1 == 128 { 112 switch s2 { 113 case lineSep: 114 return lineSepState, 3 115 case paragraphSep: 116 return paragraphSepState, 3 117 } 118 } 119 return validUTF8State, 3 120 } 121 s3 := s[3] 122 if s3 < locb || hicb < s3 { 123 return runeErrorState, 1 124 } 125 return validUTF8State, 4 126 }