|
| 1 | +#pragma once |
| 2 | + |
| 3 | +#include "rune.h" |
| 4 | + |
| 5 | +namespace Karm { |
| 6 | + |
| 7 | +/* --- Check Identity ------------------------------------------------------- */ |
| 8 | + |
| 9 | +constexpr bool isAscii(Rune rune) { |
| 10 | + return rune <= 0x7F; |
| 11 | +} |
| 12 | + |
| 13 | +constexpr bool isAsciiDigit(Rune rune) { |
| 14 | + return rune >= '0' and rune <= '9'; |
| 15 | +} |
| 16 | + |
| 17 | +constexpr bool isAsciiLower(Rune rune) { |
| 18 | + return rune >= 'a' and rune <= 'z'; |
| 19 | +} |
| 20 | + |
| 21 | +constexpr bool isAsciiUpper(Rune rune) { |
| 22 | + return rune >= 'A' and rune <= 'Z'; |
| 23 | +} |
| 24 | + |
| 25 | +constexpr bool isAsciiAlpha(Rune rune) { |
| 26 | + return isAsciiLower(rune) or isAsciiUpper(rune); |
| 27 | +} |
| 28 | + |
| 29 | +constexpr bool isAsciiAlphaNum(Rune rune) { |
| 30 | + return isAsciiAlpha(rune) or isAsciiDigit(rune); |
| 31 | +} |
| 32 | + |
| 33 | +constexpr bool isBinDigit(Rune rune) { |
| 34 | + return rune == '0' or rune == '1'; |
| 35 | +} |
| 36 | + |
| 37 | +constexpr bool isOctDigit(Rune rune) { |
| 38 | + return rune >= '0' and rune <= '7'; |
| 39 | +} |
| 40 | + |
| 41 | +constexpr bool isDecDigit(Rune rune) { |
| 42 | + return rune >= '0' and rune <= '9'; |
| 43 | +} |
| 44 | + |
| 45 | +constexpr bool isHexDigit(Rune rune) { |
| 46 | + return isDecDigit(rune) or (rune >= 'a' and rune <= 'f') or (rune >= 'A' and rune <= 'F'); |
| 47 | +} |
| 48 | + |
| 49 | +constexpr bool isAsciiBlank(Rune rune) { |
| 50 | + return rune == ' ' or rune == '\t'; |
| 51 | +} |
| 52 | + |
| 53 | +constexpr bool isAsciiSpace(Rune rune) { |
| 54 | + return rune == ' ' or (rune >= '\t' and rune <= '\r'); |
| 55 | +} |
| 56 | + |
| 57 | +constexpr bool isAsciiPunct(Rune rune) { |
| 58 | + return (rune >= '!' and rune <= '/') or (rune >= ':' and rune <= '@') or |
| 59 | + (rune >= '[' and rune <= '`') or (rune >= '{' and rune <= '~'); |
| 60 | +} |
| 61 | + |
| 62 | +constexpr bool isAsciiGraph(Rune rune) { |
| 63 | + return rune >= '!' and rune <= '~'; |
| 64 | +} |
| 65 | + |
| 66 | +constexpr bool isAsciiPrint(Rune rune) { |
| 67 | + return rune >= ' ' and rune <= '~'; |
| 68 | +} |
| 69 | + |
| 70 | +constexpr bool isAsciiC0Control(Rune rune) { |
| 71 | + return rune <= 0x20; |
| 72 | +} |
| 73 | + |
| 74 | +constexpr bool isAsciiControl(Rune rune) { |
| 75 | + return isAsciiC0Control(rune) or rune == 0x7F; |
| 76 | +} |
| 77 | + |
| 78 | +constexpr bool isUnicode(Rune rune) { |
| 79 | + return rune <= 0x10FFFF; |
| 80 | +} |
| 81 | + |
| 82 | +constexpr bool isUnicodeControl(Rune rune) { |
| 83 | + return isAsciiC0Control(rune) or (rune >= 0x7E and rune <= 0x9F); |
| 84 | +} |
| 85 | + |
| 86 | +constexpr bool isUnicodeSurrogate(Rune rune) { |
| 87 | + return rune >= 0xD800 and rune <= 0xDFFF; |
| 88 | +} |
| 89 | + |
| 90 | +constexpr bool isUnicodeScalar(Rune rune) { |
| 91 | + return isUnicode(rune) and not isUnicodeSurrogate(rune); |
| 92 | +} |
| 93 | + |
| 94 | +constexpr bool isUnicodeNonCharacter(Rune rune) { |
| 95 | + return isUnicode(rune) and |
| 96 | + ((rune >= 0xFDD0 and rune <= 0xFDEF) or |
| 97 | + ((rune & 0xFFFE) == 0xFFFE) or |
| 98 | + ((rune & 0xFFFF) == 0xFFFF)); |
| 99 | +} |
| 100 | + |
| 101 | +/* --- Rune Conversion ------------------------------------------------------ */ |
| 102 | + |
| 103 | +constexpr Rune toAsciiLower(Rune rune) { |
| 104 | + if (isAsciiUpper(rune)) |
| 105 | + return rune + 0x20; |
| 106 | + |
| 107 | + return rune; |
| 108 | +} |
| 109 | + |
| 110 | +constexpr Rune toAsciiUpper(Rune rune) { |
| 111 | + if (isAsciiLower(rune)) |
| 112 | + return rune - 0x20; |
| 113 | + |
| 114 | + return rune; |
| 115 | +} |
| 116 | + |
| 117 | +/* --- Rune Parsing --------------------------------------------------------- */ |
| 118 | + |
| 119 | +constexpr Rune parseAsciiDecDigit(Rune rune) { |
| 120 | + if (isDecDigit(rune)) |
| 121 | + return rune - '0'; |
| 122 | + |
| 123 | + panic("invalid decimal digit"); |
| 124 | +} |
| 125 | + |
| 126 | +constexpr Rune parseAsciiHexDigit(Rune rune) { |
| 127 | + rune = toAsciiLower(rune); |
| 128 | + |
| 129 | + if (isDecDigit(rune)) |
| 130 | + return rune - '0'; |
| 131 | + |
| 132 | + if (rune >= 'a' and rune <= 'f') |
| 133 | + return rune - 'a' + 10; |
| 134 | + |
| 135 | + panic("invalid hexadecimal digit"); |
| 136 | +} |
| 137 | + |
| 138 | +} // namespace Karm |
0 commit comments