26 constexpr size_t HI_SURROGATE_MIN = 0xd800;
27 constexpr size_t HI_SURROGATE_MAX = 0xdbff;
28 constexpr size_t LO_SURROGATE_MIN = 0xdc00;
29 constexpr size_t LO_SURROGATE_MAX = 0xdfff;
31 constexpr size_t LO_SUPPLEMENTS_MIN = 0x10000;
36 static constexpr size_t MASK1 = 0x80;
37 static constexpr size_t MASK2 = 0x20;
38 static constexpr size_t MASK3 = 0x10;
39 static constexpr size_t LOW_3BITS = 0x7;
40 static constexpr size_t LOW_4BITS = 0xF;
41 static constexpr size_t LOW_5BITS = 0x1F;
42 static constexpr size_t LOW_6BITS = 0x3F;
43 static constexpr size_t L_SURROGATE_START = 0xDC00;
44 static constexpr size_t H_SURROGATE_START = 0xD800;
45 static constexpr size_t SURROGATE_RAIR_START = 0x10000;
50 static constexpr uint16_t DECODE_LEAD_LOW = 0xD800;
51 static constexpr uint16_t DECODE_LEAD_HIGH = 0xDBFF;
52 static constexpr uint16_t DECODE_TRAIL_LOW = 0xDC00;
53 static constexpr uint16_t DECODE_TRAIL_HIGH = 0xDFFF;
54 static constexpr uint32_t DECODE_FIRST_FACTOR = 0x400;
55 static constexpr uint32_t DECODE_SECOND_FACTOR = 0x10000;
58 static constexpr uint16_t SURROGATE_MASK = 0xF800;
59 static constexpr uint16_t UTF16_REPLACEMENT_CHARACTER = 0xFFFD;
64 constexpr size_t U16_LEAD = 0xd7c0;
65 constexpr size_t U16_TAIL = 0xdc00;
67 static constexpr uint8_t BIT_MASK_1 = 0x80;
68 static constexpr uint8_t BIT_MASK_2 = 0xC0;
69 static constexpr uint8_t BIT_MASK_3 = 0xE0;
70 static constexpr uint8_t BIT_MASK_4 = 0xF0;
71 static constexpr uint8_t BIT_MASK_5 = 0xF8;
72 static constexpr uint8_t BIT_MASK_FF = 0xFF;
73 static constexpr uint16_t BIT16_MASK = 0x3FF;
75 constexpr size_t MASK_4BIT = 0x0f;
76 constexpr size_t MASK_5BIT = 0x1f;
77 constexpr size_t MASK_6BIT = 0x3f;
78 constexpr size_t MASK_10BIT = 0x03ff;
79 constexpr size_t MASK_16BIT = 0xffff;
81 static constexpr uint8_t UTF8_1B_MAX = 0x7f;
87 static constexpr uint8_t UTF8_NUL = 0x00U;
88 static constexpr uint16_t UTF8_2B_MAX = 0x7ff;
89 static constexpr uint8_t UTF8_2B_FIRST = 0xc0;
90 static constexpr uint8_t UTF8_2B_SECOND = 0x80;
91 static constexpr uint8_t UTF8_2B_THIRD = 0x3f;
92 static constexpr uint8_t UTF8_2B_FIRST_MIN = 0xc2;  // the minimum for 2 bytes is 128, which is 0xc…
94 static constexpr uint16_t UTF8_3B_MAX = 0xffff;
95 static constexpr uint8_t UTF8_3B_FIRST = 0xe0;
96 static constexpr uint8_t UTF8_3B_SECOND = 0x80;
97 static constexpr uint8_t UTF8_3B_THIRD = 0x80;
98 static constexpr uint8_t UTF8_3B_SECOND_MIN = 0xa0;  // the minimum for 3 bytes is 2048, which is 0…
99 static constexpr uint8_t UTF8_3B_RESERVED_FIRST = 0xED;
100 static constexpr uint8_t UTF8_3B_RESERVED_SECOND_MIN = 0xA0;
101 static constexpr uint8_t UTF8_3B_RESERVED_SECOND_MAX = 0xBF; // U+D800~U+DFFF is reserved for UTF-1…
103 static constexpr uint8_t UTF8_4B_FIRST = 0xf0;
104 static constexpr uint8_t UTF8_4B_SECOND_MIN = 0x90;  // the minimum for 4 bytes is 65536, which is …
105 static constexpr uint8_t UTF8_4B_FIRST_MAX = 0xF4; // the maximum for 4 bytes is 1114111, which is …
106 static constexpr uint8_t UTF8_4B_SECOND_MAX = 0x8F;
108 static constexpr uint8_t byteMask = 0xbf;
109 static constexpr uint8_t byteMark = 0x80;
111 static constexpr uint8_t latin1Limit = 0xFF;
124 static const unsigned char firstByteMark[7] = {0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC};
184     return (ch >= '0' && ch <= '9') || (ch >= 'A' && ch <= 'F') || (ch >= 'a' && ch <= 'f');  in IsHexDigits()
194     // | '0'  | 0x30  |   0   |  in HexChar16Value()
196     // | '9'  | 0x39  |   9   |  in HexChar16Value()
197     // | 'A'  | 0x41  |  10   |  in HexChar16Value()
199     // | 'F'  | 0x46  |  15   |  in HexChar16Value()
200     // | 'a'  | 0x61  |  10   |  in HexChar16Value()
202     // | 'f'  | 0x66  |  15   |  in HexChar16Value()
205     size_t res = ch - '0'; // res in [0x0, 0x9], [0x11, 0x16], [0x31, 0x36]  in HexChar16Value()
207     if (res > 9) { // 9: res in [0x11, 0x16], [0x31, 0x36], which means ch in ['A', 'F'], ['a', 'f']  in HexChar16Value()
208 …res |= 0x20; // 0x20: [0x11, 0x16] -> [0x31, 0x36], converting ['A' - '0', 'F' - '0'] to ['a' - '0…  in HexChar16Value()
209         // res = [0x31, 0x36]  in HexChar16Value()
210         res -= ('a' - '0'); // res = [0x0, 0x5]  in HexChar16Value()
224     // |   0   | 0x30  | '0'  |  in GetHexChar16()
226     // |   9   | 0x39  | '9'  |  in GetHexChar16()
227     // |  10   | 0x41  | 'A'  |  in GetHexChar16()
229     // |  15   | 0x46  | 'F'  |  in GetHexChar16()
231     if (val < 10) { // 10: val in [0x0, 0x9], convert to ['0', '9']  in GetHexChar16()
232         return val + '0';  in GetHexChar16()
234     return val - 0xA + 'A'; // 0xA: val in [0xA, 0xF], convert to ['A', 'F']  in GetHexChar16()