mirror of
https://github.com/ultimatepp/ultimatepp.git
synced 2026-09-10 03:54:54 -06:00
Core: Unicode optimisations
git-svn-id: svn://ultimatepp.org/upp/trunk@14831 f0d560ea-af0d-0410-9eb7-867de7ffcac7
This commit is contained in:
parent
b5c6926048
commit
4aca1f3cb1
4 changed files with 47 additions and 189 deletions
|
|
@ -2010,169 +2010,8 @@ static word sComb(const word *s, word chr)
|
|||
|
||||
word UnicodeCombine(word chr, word combine)
|
||||
{
|
||||
static word comb300[] = { 0x41, 0xC0, 0x45, 0xC8, 0x49, 0xCC, 0x4E, 0x1F8, 0x4F, 0xD2, 0x55, 0xD9, 0x57, 0x1E80, 0x59, 0x1EF2, 0x61, 0xE0, 0x65, 0xE8, 0x69, 0xEC, 0x6E, 0x1F9, 0x6F, 0xF2, 0x75, 0xF9, 0x77, 0x1E81, 0x79, 0x1EF3, 0xA8, 0x1FED, 0xC2, 0x1EA6, 0xCA, 0x1EC0, 0xD4, 0x1ED2, 0xDC, 0x1DB, 0xE2, 0x1EA7, 0xEA, 0x1EC1, 0xF4, 0x1ED3, 0xFC, 0x1DC, 0x102, 0x1EB0, 0x103, 0x1EB1, 0x112, 0x1E14, 0x113, 0x1E15, 0x14C, 0x1E50, 0x14D, 0x1E51, 0x1A0, 0x1EDC, 0x1A1, 0x1EDD, 0x1AF, 0x1EEA, 0x1B0, 0x1EEB, 0x391, 0x1FBA, 0x395, 0x1FC8, 0x397, 0x1FCA, 0x399, 0x1FDA, 0x39F, 0x1FF8, 0x3A5, 0x1FEA, 0x3A9, 0x1FFA, 0x3B1, 0x1F70, 0x3B5, 0x1F72, 0x3B7, 0x1F74, 0x3B9, 0x1F76, 0x3BF, 0x1F78, 0x3C5, 0x1F7A, 0x3C9, 0x1F7C, 0x3CA, 0x1FD2, 0x3CB, 0x1FE2, 0x415, 0x400, 0x418, 0x40D, 0x435, 0x450, 0x438, 0x45D, 0x1F00, 0x1F02, 0x1F01, 0x1F03, 0x1F08, 0x1F0A, 0x1F09, 0x1F0B, 0x1F10, 0x1F12, 0x1F11, 0x1F13, 0x1F18, 0x1F1A, 0x1F19, 0x1F1B, 0x1F20, 0x1F22, 0x1F21, 0x1F23, 0x1F28, 0x1F2A, 0x1F29, 0x1F2B, 0x1F30, 0x1F32, 0x1F31, 0x1F33, 0x1F38, 0x1F3A, 0x1F39, 0x1F3B, 0x1F40, 0x1F42, 0x1F41, 0x1F43, 0x1F48, 0x1F4A, 0x1F49, 0x1F4B, 0x1F50, 0x1F52, 0x1F51, 0x1F53, 0x1F59, 0x1F5B, 0x1F60, 0x1F62, 0x1F61, 0x1F63, 0x1F68, 0x1F6A, 0x1F69, 0x1F6B, 0x1FBF, 0x1FCD, 0x1FFE, 0x1FDD, 0 };
|
||||
static word comb301[] = { 0x41, 0xC1, 0x43, 0x106, 0x45, 0xC9, 0x47, 0x1F4, 0x49, 0xCD, 0x4B, 0x1E30, 0x4C, 0x139, 0x4D, 0x1E3E, 0x4E, 0x143, 0x4F, 0xD3, 0x50, 0x1E54, 0x52, 0x154, 0x53, 0x15A, 0x55, 0xDA, 0x57, 0x1E82, 0x59, 0xDD, 0x5A, 0x179, 0x61, 0xE1, 0x63, 0x107, 0x65, 0xE9, 0x67, 0x1F5, 0x69, 0xED, 0x6B, 0x1E31, 0x6C, 0x13A, 0x6D, 0x1E3F, 0x6E, 0x144, 0x6F, 0xF3, 0x70, 0x1E55, 0x72, 0x155, 0x73, 0x15B, 0x75, 0xFA, 0x77, 0x1E83, 0x79, 0xFD, 0x7A, 0x17A, 0xA8, 0x385, 0xC2, 0x1EA4, 0xC5, 0x1FA, 0xC6, 0x1FC, 0xC7, 0x1E08, 0xCA, 0x1EBE, 0xCF, 0x1E2E, 0xD4, 0x1ED0, 0xD5, 0x1E4C, 0xD8, 0x1FE, 0xDC, 0x1D7, 0xE2, 0x1EA5, 0xE5, 0x1FB, 0xE6, 0x1FD, 0xE7, 0x1E09, 0xEA, 0x1EBF, 0xEF, 0x1E2F, 0xF4, 0x1ED1, 0xF5, 0x1E4D, 0xF8, 0x1FF, 0xFC, 0x1D8, 0x102, 0x1EAE, 0x103, 0x1EAF, 0x112, 0x1E16, 0x113, 0x1E17, 0x14C, 0x1E52, 0x14D, 0x1E53, 0x168, 0x1E78, 0x169, 0x1E79, 0x1A0, 0x1EDA, 0x1A1, 0x1EDB, 0x1AF, 0x1EE8, 0x1B0, 0x1EE9, 0x308, 0x344, 0x391, 0x386, 0x395, 0x388, 0x397, 0x389, 0x399, 0x38A, 0x39F, 0x38C, 0x3A5, 0x38E, 0x3A9, 0x38F, 0x3B1, 0x3AC, 0x3B5, 0x3AD, 0x3B7, 0x3AE, 0x3B9, 0x3AF, 0x3BF, 0x3CC, 0x3C5, 0x3CD, 0x3C9, 0x3CE, 0x3CA, 0x390, 0x3CB, 0x3B0, 0x3D2, 0x3D3, 0x413, 0x403, 0x41A, 0x40C, 0x433, 0x453, 0x43A, 0x45C, 0x1F00, 0x1F04, 0x1F01, 0x1F05, 0x1F08, 0x1F0C, 0x1F09, 0x1F0D, 0x1F10, 0x1F14, 0x1F11, 0x1F15, 0x1F18, 0x1F1C, 0x1F19, 0x1F1D, 0x1F20, 0x1F24, 0x1F21, 0x1F25, 0x1F28, 0x1F2C, 0x1F29, 0x1F2D, 0x1F30, 0x1F34, 0x1F31, 0x1F35, 0x1F38, 0x1F3C, 0x1F39, 0x1F3D, 0x1F40, 0x1F44, 0x1F41, 0x1F45, 0x1F48, 0x1F4C, 0x1F49, 0x1F4D, 0x1F50, 0x1F54, 0x1F51, 0x1F55, 0x1F59, 0x1F5D, 0x1F60, 0x1F64, 0x1F61, 0x1F65, 0x1F68, 0x1F6C, 0x1F69, 0x1F6D, 0x1FBF, 0x1FCE, 0x1FFE, 0x1FDE, 0 };
|
||||
static word comb302[] = { 0x41, 0xC2, 0x43, 0x108, 0x45, 0xCA, 0x47, 0x11C, 0x48, 0x124, 0x49, 0xCE, 0x4A, 0x134, 0x4F, 0xD4, 0x53, 0x15C, 0x55, 0xDB, 0x57, 0x174, 0x59, 0x176, 0x5A, 0x1E90, 0x61, 0xE2, 0x63, 0x109, 0x65, 0xEA, 0x67, 0x11D, 0x68, 0x125, 0x69, 0xEE, 0x6A, 0x135, 0x6F, 0xF4, 0x73, 0x15D, 0x75, 0xFB, 0x77, 0x175, 0x79, 0x177, 0x7A, 0x1E91, 0x1EA0, 0x1EAC, 0x1EA1, 0x1EAD, 0x1EB8, 0x1EC6, 0x1EB9, 0x1EC7, 0x1ECC, 0x1ED8, 0x1ECD, 0x1ED9, 0 };
|
||||
static word comb303[] = { 0x41, 0xC3, 0x45, 0x1EBC, 0x49, 0x128, 0x4E, 0xD1, 0x4F, 0xD5, 0x55, 0x168, 0x56, 0x1E7C, 0x59, 0x1EF8, 0x61, 0xE3, 0x65, 0x1EBD, 0x69, 0x129, 0x6E, 0xF1, 0x6F, 0xF5, 0x75, 0x169, 0x76, 0x1E7D, 0x79, 0x1EF9, 0xC2, 0x1EAA, 0xCA, 0x1EC4, 0xD4, 0x1ED6, 0xE2, 0x1EAB, 0xEA, 0x1EC5, 0xF4, 0x1ED7, 0x102, 0x1EB4, 0x103, 0x1EB5, 0x1A0, 0x1EE0, 0x1A1, 0x1EE1, 0x1AF, 0x1EEE, 0x1B0, 0x1EEF, 0 };
|
||||
static word comb304[] = { 0x41, 0x100, 0x45, 0x112, 0x47, 0x1E20, 0x49, 0x12A, 0x4F, 0x14C, 0x55, 0x16A, 0x59, 0x232, 0x61, 0x101, 0x65, 0x113, 0x67, 0x1E21, 0x69, 0x12B, 0x6F, 0x14D, 0x75, 0x16B, 0x79, 0x233, 0xC4, 0x1DE, 0xC6, 0x1E2, 0xD5, 0x22C, 0xD6, 0x22A, 0xDC, 0x1D5, 0xE4, 0x1DF, 0xE6, 0x1E3, 0xF5, 0x22D, 0xF6, 0x22B, 0xFC, 0x1D6, 0x1EA, 0x1EC, 0x1EB, 0x1ED, 0x226, 0x1E0, 0x227, 0x1E1, 0x22E, 0x230, 0x22F, 0x231, 0x391, 0x1FB9, 0x399, 0x1FD9, 0x3A5, 0x1FE9, 0x3B1, 0x1FB1, 0x3B9, 0x1FD1, 0x3C5, 0x1FE1, 0x418, 0x4E2, 0x423, 0x4EE, 0x438, 0x4E3, 0x443, 0x4EF, 0x1E36, 0x1E38, 0x1E37, 0x1E39, 0x1E5A, 0x1E5C, 0x1E5B, 0x1E5D, 0 };
|
||||
static word comb305[] = { 0 };
|
||||
static word comb306[] = { 0x41, 0x102, 0x45, 0x114, 0x47, 0x11E, 0x49, 0x12C, 0x4F, 0x14E, 0x55, 0x16C, 0x61, 0x103, 0x65, 0x115, 0x67, 0x11F, 0x69, 0x12D, 0x6F, 0x14F, 0x75, 0x16D, 0x228, 0x1E1C, 0x229, 0x1E1D, 0x391, 0x1FB8, 0x399, 0x1FD8, 0x3A5, 0x1FE8, 0x3B1, 0x1FB0, 0x3B9, 0x1FD0, 0x3C5, 0x1FE0, 0x410, 0x4D0, 0x415, 0x4D6, 0x416, 0x4C1, 0x418, 0x419, 0x423, 0x40E, 0x430, 0x4D1, 0x435, 0x4D7, 0x436, 0x4C2, 0x438, 0x439, 0x443, 0x45E, 0x1EA0, 0x1EB6, 0x1EA1, 0x1EB7, 0 };
|
||||
static word comb307[] = { 0x41, 0x226, 0x42, 0x1E02, 0x43, 0x10A, 0x44, 0x1E0A, 0x45, 0x116, 0x46, 0x1E1E, 0x47, 0x120, 0x48, 0x1E22, 0x49, 0x130, 0x4D, 0x1E40, 0x4E, 0x1E44, 0x4F, 0x22E, 0x50, 0x1E56, 0x52, 0x1E58, 0x53, 0x1E60, 0x54, 0x1E6A, 0x57, 0x1E86, 0x58, 0x1E8A, 0x59, 0x1E8E, 0x5A, 0x17B, 0x61, 0x227, 0x62, 0x1E03, 0x63, 0x10B, 0x64, 0x1E0B, 0x65, 0x117, 0x66, 0x1E1F, 0x67, 0x121, 0x68, 0x1E23, 0x6D, 0x1E41, 0x6E, 0x1E45, 0x6F, 0x22F, 0x70, 0x1E57, 0x72, 0x1E59, 0x73, 0x1E61, 0x74, 0x1E6B, 0x77, 0x1E87, 0x78, 0x1E8B, 0x79, 0x1E8F, 0x7A, 0x17C, 0x15A, 0x1E64, 0x15B, 0x1E65, 0x160, 0x1E66, 0x161, 0x1E67, 0x17F, 0x1E9B, 0x1E62, 0x1E68, 0x1E63, 0x1E69, 0 };
|
||||
static word comb308[] = { 0x41, 0xC4, 0x45, 0xCB, 0x48, 0x1E26, 0x49, 0xCF, 0x4F, 0xD6, 0x55, 0xDC, 0x57, 0x1E84, 0x58, 0x1E8C, 0x59, 0x178, 0x61, 0xE4, 0x65, 0xEB, 0x68, 0x1E27, 0x69, 0xEF, 0x6F, 0xF6, 0x74, 0x1E97, 0x75, 0xFC, 0x77, 0x1E85, 0x78, 0x1E8D, 0x79, 0xFF, 0xD5, 0x1E4E, 0xF5, 0x1E4F, 0x16A, 0x1E7A, 0x16B, 0x1E7B, 0x399, 0x3AA, 0x3A5, 0x3AB, 0x3B9, 0x3CA, 0x3C5, 0x3CB, 0x3D2, 0x3D4, 0x406, 0x407, 0x410, 0x4D2, 0x415, 0x401, 0x416, 0x4DC, 0x417, 0x4DE, 0x418, 0x4E4, 0x41E, 0x4E6, 0x423, 0x4F0, 0x427, 0x4F4, 0x42B, 0x4F8, 0x42D, 0x4EC, 0x430, 0x4D3, 0x435, 0x451, 0x436, 0x4DD, 0x437, 0x4DF, 0x438, 0x4E5, 0x43E, 0x4E7, 0x443, 0x4F1, 0x447, 0x4F5, 0x44B, 0x4F9, 0x44D, 0x4ED, 0x456, 0x457, 0x4D8, 0x4DA, 0x4D9, 0x4DB, 0x4E8, 0x4EA, 0x4E9, 0x4EB, 0 };
|
||||
static word comb309[] = { 0x41, 0x1EA2, 0x45, 0x1EBA, 0x49, 0x1EC8, 0x4F, 0x1ECE, 0x55, 0x1EE6, 0x59, 0x1EF6, 0x61, 0x1EA3, 0x65, 0x1EBB, 0x69, 0x1EC9, 0x6F, 0x1ECF, 0x75, 0x1EE7, 0x79, 0x1EF7, 0xC2, 0x1EA8, 0xCA, 0x1EC2, 0xD4, 0x1ED4, 0xE2, 0x1EA9, 0xEA, 0x1EC3, 0xF4, 0x1ED5, 0x102, 0x1EB2, 0x103, 0x1EB3, 0x1A0, 0x1EDE, 0x1A1, 0x1EDF, 0x1AF, 0x1EEC, 0x1B0, 0x1EED, 0 };
|
||||
static word comb30a[] = { 0x41, 0xC5, 0x55, 0x16E, 0x61, 0xE5, 0x75, 0x16F, 0x77, 0x1E98, 0x79, 0x1E99, 0 };
|
||||
static word comb30b[] = { 0x4F, 0x150, 0x55, 0x170, 0x6F, 0x151, 0x75, 0x171, 0x423, 0x4F2, 0x443, 0x4F3, 0 };
|
||||
static word comb30c[] = { 0x41, 0x1CD, 0x43, 0x10C, 0x44, 0x10E, 0x45, 0x11A, 0x47, 0x1E6, 0x48, 0x21E, 0x49, 0x1CF, 0x4B, 0x1E8, 0x4C, 0x13D, 0x4E, 0x147, 0x4F, 0x1D1, 0x52, 0x158, 0x53, 0x160, 0x54, 0x164, 0x55, 0x1D3, 0x5A, 0x17D, 0x61, 0x1CE, 0x63, 0x10D, 0x64, 0x10F, 0x65, 0x11B, 0x67, 0x1E7, 0x68, 0x21F, 0x69, 0x1D0, 0x6A, 0x1F0, 0x6B, 0x1E9, 0x6C, 0x13E, 0x6E, 0x148, 0x6F, 0x1D2, 0x72, 0x159, 0x73, 0x161, 0x74, 0x165, 0x75, 0x1D4, 0x7A, 0x17E, 0xDC, 0x1D9, 0xFC, 0x1DA, 0x1B7, 0x1EE, 0x292, 0x1EF, 0 };
|
||||
static word comb30d[] = { 0 };
|
||||
static word comb30e[] = { 0 };
|
||||
static word comb30f[] = { 0x41, 0x200, 0x45, 0x204, 0x49, 0x208, 0x4F, 0x20C, 0x52, 0x210, 0x55, 0x214, 0x61, 0x201, 0x65, 0x205, 0x69, 0x209, 0x6F, 0x20D, 0x72, 0x211, 0x75, 0x215, 0x474, 0x476, 0x475, 0x477, 0 };
|
||||
|
||||
static const word *comb[] = {
|
||||
comb300, comb301, comb302, comb303, comb304, comb305, comb306, comb307,
|
||||
comb308, comb309, comb30a, comb30b, comb30c, comb30d, comb30e, comb30f,
|
||||
};
|
||||
if(combine >= 0x300 && combine <= 0x30F)
|
||||
return sComb(comb[combine - 0x300], chr);
|
||||
|
||||
static word comb311[] = { 0x41, 0x202, 0x45, 0x206, 0x49, 0x20A, 0x4F, 0x20E, 0x52, 0x212, 0x55, 0x216, 0x61, 0x203, 0x65, 0x207, 0x69, 0x20B, 0x6F, 0x20F, 0x72, 0x213, 0x75, 0x217, 0 };
|
||||
static word comb313[] = { 0x391, 0x1F08, 0x395, 0x1F18, 0x397, 0x1F28, 0x399, 0x1F38, 0x39F, 0x1F48, 0x3A9, 0x1F68, 0x3B1, 0x1F00, 0x3B5, 0x1F10, 0x3B7, 0x1F20, 0x3B9, 0x1F30, 0x3BF, 0x1F40, 0x3C1, 0x1FE4, 0x3C5, 0x1F50, 0x3C9, 0x1F60, 0 };
|
||||
static word comb314[] = { 0x391, 0x1F09, 0x395, 0x1F19, 0x397, 0x1F29, 0x399, 0x1F39, 0x39F, 0x1F49, 0x3A1, 0x1FEC, 0x3A5, 0x1F59, 0x3A9, 0x1F69, 0x3B1, 0x1F01, 0x3B5, 0x1F11, 0x3B7, 0x1F21, 0x3B9, 0x1F31, 0x3BF, 0x1F41, 0x3C1, 0x1FE5, 0x3C5, 0x1F51, 0x3C9, 0x1F61, 0 };
|
||||
static word comb31b[] = { 0x4F, 0x1A0, 0x55, 0x1AF, 0x6F, 0x1A1, 0x75, 0x1B0, 0 };
|
||||
static word comb323[] = { 0x41, 0x1EA0, 0x42, 0x1E04, 0x44, 0x1E0C, 0x45, 0x1EB8, 0x48, 0x1E24, 0x49, 0x1ECA, 0x4B, 0x1E32, 0x4C, 0x1E36, 0x4D, 0x1E42, 0x4E, 0x1E46, 0x4F, 0x1ECC, 0x52, 0x1E5A, 0x53, 0x1E62, 0x54, 0x1E6C, 0x55, 0x1EE4, 0x56, 0x1E7E, 0x57, 0x1E88, 0x59, 0x1EF4, 0x5A, 0x1E92, 0x61, 0x1EA1, 0x62, 0x1E05, 0x64, 0x1E0D, 0x65, 0x1EB9, 0x68, 0x1E25, 0x69, 0x1ECB, 0x6B, 0x1E33, 0x6C, 0x1E37, 0x6D, 0x1E43, 0x6E, 0x1E47, 0x6F, 0x1ECD, 0x72, 0x1E5B, 0x73, 0x1E63, 0x74, 0x1E6D, 0x75, 0x1EE5, 0x76, 0x1E7F, 0x77, 0x1E89, 0x79, 0x1EF5, 0x7A, 0x1E93, 0x1A0, 0x1EE2, 0x1A1, 0x1EE3, 0x1AF, 0x1EF0, 0x1B0, 0x1EF1, 0 };
|
||||
static word comb324[] = { 0x55, 0x1E72, 0x75, 0x1E73, 0 };
|
||||
static word comb325[] = { 0x41, 0x1E00, 0x61, 0x1E01, 0 };
|
||||
static word comb326[] = { 0x53, 0x218, 0x54, 0x21A, 0x73, 0x219, 0x74, 0x21B, 0 };
|
||||
static word comb327[] = { 0x43, 0xC7, 0x44, 0x1E10, 0x45, 0x228, 0x47, 0x122, 0x48, 0x1E28, 0x4B, 0x136, 0x4C, 0x13B, 0x4E, 0x145, 0x52, 0x156, 0x53, 0x15E, 0x54, 0x162, 0x63, 0xE7, 0x64, 0x1E11, 0x65, 0x229, 0x67, 0x123, 0x68, 0x1E29, 0x6B, 0x137, 0x6C, 0x13C, 0x6E, 0x146, 0x72, 0x157, 0x73, 0x15F, 0x74, 0x163, 0 };
|
||||
static word comb328[] = { 0x41, 0x104, 0x45, 0x118, 0x49, 0x12E, 0x4F, 0x1EA, 0x55, 0x172, 0x61, 0x105, 0x65, 0x119, 0x69, 0x12F, 0x6F, 0x1EB, 0x75, 0x173, 0 };
|
||||
static word comb32d[] = { 0x44, 0x1E12, 0x45, 0x1E18, 0x4C, 0x1E3C, 0x4E, 0x1E4A, 0x54, 0x1E70, 0x55, 0x1E76, 0x64, 0x1E13, 0x65, 0x1E19, 0x6C, 0x1E3D, 0x6E, 0x1E4B, 0x74, 0x1E71, 0x75, 0x1E77, 0 };
|
||||
static word comb32e[] = { 0x48, 0x1E2A, 0x68, 0x1E2B, 0 };
|
||||
static word comb330[] = { 0x45, 0x1E1A, 0x49, 0x1E2C, 0x55, 0x1E74, 0x65, 0x1E1B, 0x69, 0x1E2D, 0x75, 0x1E75, 0 };
|
||||
static word comb331[] = { 0x42, 0x1E06, 0x44, 0x1E0E, 0x4B, 0x1E34, 0x4C, 0x1E3A, 0x4E, 0x1E48, 0x52, 0x1E5E, 0x54, 0x1E6E, 0x5A, 0x1E94, 0x62, 0x1E07, 0x64, 0x1E0F, 0x68, 0x1E96, 0x6B, 0x1E35, 0x6C, 0x1E3B, 0x6E, 0x1E49, 0x72, 0x1E5F, 0x74, 0x1E6F, 0x7A, 0x1E95, 0 };
|
||||
static word comb338[] = { 0x3C, 0x226E, 0x3D, 0x2260, 0x3E, 0x226F, 0x2190, 0x219A, 0x2192, 0x219B, 0x2194, 0x21AE, 0x21D0, 0x21CD, 0x21D2, 0x21CF, 0x21D4, 0x21CE, 0x2203, 0x2204, 0x2208, 0x2209, 0x220B, 0x220C, 0x2223, 0x2224, 0x2225, 0x2226, 0x223C, 0x2241, 0x2243, 0x2244, 0x2245, 0x2247, 0x2248, 0x2249, 0x224D, 0x226D, 0x2261, 0x2262, 0x2264, 0x2270, 0x2265, 0x2271, 0x2272, 0x2274, 0x2273, 0x2275, 0x2276, 0x2278, 0x2277, 0x2279, 0x227A, 0x2280, 0x227B, 0x2281, 0x227C, 0x22E0, 0x227D, 0x22E1, 0x2282, 0x2284, 0x2283, 0x2285, 0x2286, 0x2288, 0x2287, 0x2289, 0x2291, 0x22E2, 0x2292, 0x22E3, 0x22A2, 0x22AC, 0x22A8, 0x22AD, 0x22A9, 0x22AE, 0x22AB, 0x22AF, 0x22B2, 0x22EA, 0x22B3, 0x22EB, 0x22B4, 0x22EC, 0x22B5, 0x22ED, 0x2ADD, 0x2ADC, 0 };
|
||||
static word comb342[] = { 0xA8, 0x1FC1, 0x3B1, 0x1FB6, 0x3B7, 0x1FC6, 0x3B9, 0x1FD6, 0x3C5, 0x1FE6, 0x3C9, 0x1FF6, 0x3CA, 0x1FD7, 0x3CB, 0x1FE7, 0x1F00, 0x1F06, 0x1F01, 0x1F07, 0x1F08, 0x1F0E, 0x1F09, 0x1F0F, 0x1F20, 0x1F26, 0x1F21, 0x1F27, 0x1F28, 0x1F2E, 0x1F29, 0x1F2F, 0x1F30, 0x1F36, 0x1F31, 0x1F37, 0x1F38, 0x1F3E, 0x1F39, 0x1F3F, 0x1F50, 0x1F56, 0x1F51, 0x1F57, 0x1F59, 0x1F5F, 0x1F60, 0x1F66, 0x1F61, 0x1F67, 0x1F68, 0x1F6E, 0x1F69, 0x1F6F, 0x1FBF, 0x1FCF, 0x1FFE, 0x1FDF, 0 };
|
||||
static word comb345[] = { 0x391, 0x1FBC, 0x397, 0x1FCC, 0x3A9, 0x1FFC, 0x3AC, 0x1FB4, 0x3AE, 0x1FC4, 0x3B1, 0x1FB3, 0x3B7, 0x1FC3, 0x3C9, 0x1FF3, 0x3CE, 0x1FF4, 0x1F00, 0x1F80, 0x1F01, 0x1F81, 0x1F02, 0x1F82, 0x1F03, 0x1F83, 0x1F04, 0x1F84, 0x1F05, 0x1F85, 0x1F06, 0x1F86, 0x1F07, 0x1F87, 0x1F08, 0x1F88, 0x1F09, 0x1F89, 0x1F0A, 0x1F8A, 0x1F0B, 0x1F8B, 0x1F0C, 0x1F8C, 0x1F0D, 0x1F8D, 0x1F0E, 0x1F8E, 0x1F0F, 0x1F8F, 0x1F20, 0x1F90, 0x1F21, 0x1F91, 0x1F22, 0x1F92, 0x1F23, 0x1F93, 0x1F24, 0x1F94, 0x1F25, 0x1F95, 0x1F26, 0x1F96, 0x1F27, 0x1F97, 0x1F28, 0x1F98, 0x1F29, 0x1F99, 0x1F2A, 0x1F9A, 0x1F2B, 0x1F9B, 0x1F2C, 0x1F9C, 0x1F2D, 0x1F9D, 0x1F2E, 0x1F9E, 0x1F2F, 0x1F9F, 0x1F60, 0x1FA0, 0x1F61, 0x1FA1, 0x1F62, 0x1FA2, 0x1F63, 0x1FA3, 0x1F64, 0x1FA4, 0x1F65, 0x1FA5, 0x1F66, 0x1FA6, 0x1F67, 0x1FA7, 0x1F68, 0x1FA8, 0x1F69, 0x1FA9, 0x1F6A, 0x1FAA, 0x1F6B, 0x1FAB, 0x1F6C, 0x1FAC, 0x1F6D, 0x1FAD, 0x1F6E, 0x1FAE, 0x1F6F, 0x1FAF, 0x1F70, 0x1FB2, 0x1F74, 0x1FC2, 0x1F7C, 0x1FF2, 0x1FB6, 0x1FB7, 0x1FC6, 0x1FC7, 0x1FF6, 0x1FF7, 0 };
|
||||
static word comb5b4[] = { 0x5D9, 0xFB1D, 0 };
|
||||
static word comb5b7[] = { 0x5D0, 0xFB2E, 0x5F2, 0xFB1F, 0 };
|
||||
static word comb5b8[] = { 0x5D0, 0xFB2F, 0 };
|
||||
static word comb5b9[] = { 0x5D5, 0xFB4B, 0 };
|
||||
static word comb5bc[] = { 0x5D0, 0xFB30, 0x5D1, 0xFB31, 0x5D2, 0xFB32, 0x5D3, 0xFB33, 0x5D4, 0xFB34, 0x5D5, 0xFB35, 0x5D6, 0xFB36, 0x5D8, 0xFB38, 0x5D9, 0xFB39, 0x5DA, 0xFB3A, 0x5DB, 0xFB3B, 0x5DC, 0xFB3C, 0x5DE, 0xFB3E, 0x5E0, 0xFB40, 0x5E1, 0xFB41, 0x5E3, 0xFB43, 0x5E4, 0xFB44, 0x5E6, 0xFB46, 0x5E7, 0xFB47, 0x5E8, 0xFB48, 0x5E9, 0xFB49, 0x5EA, 0xFB4A, 0 };
|
||||
static word comb5bf[] = { 0x5D1, 0xFB4C, 0x5DB, 0xFB4D, 0x5E4, 0xFB4E, 0 };
|
||||
static word comb5c1[] = { 0x5E9, 0xFB2A, 0xFB49, 0xFB2C, 0 };
|
||||
static word comb5c2[] = { 0x5E9, 0xFB2B, 0xFB49, 0xFB2D, 0 };
|
||||
static word comb653[] = { 0x627, 0x622, 0 };
|
||||
static word comb654[] = { 0x627, 0x623, 0x648, 0x624, 0x64A, 0x626, 0x6C1, 0x6C2, 0x6D2, 0x6D3, 0x6D5, 0x6C0, 0 };
|
||||
static word comb655[] = { 0x627, 0x625, 0 };
|
||||
static word comb93c[] = { 0x915, 0x958, 0x916, 0x959, 0x917, 0x95A, 0x91C, 0x95B, 0x921, 0x95C, 0x922, 0x95D, 0x928, 0x929, 0x92B, 0x95E, 0x92F, 0x95F, 0x930, 0x931, 0x933, 0x934, 0 };
|
||||
static word comb9bc[] = { 0x9A1, 0x9DC, 0x9A2, 0x9DD, 0x9AF, 0x9DF, 0 };
|
||||
static word comb9be[] = { 0x9C7, 0x9CB, 0 };
|
||||
static word comb9d7[] = { 0x9C7, 0x9CC, 0 };
|
||||
static word comba3c[] = { 0xA16, 0xA59, 0xA17, 0xA5A, 0xA1C, 0xA5B, 0xA2B, 0xA5E, 0xA32, 0xA33, 0xA38, 0xA36, 0 };
|
||||
static word combb3c[] = { 0xB21, 0xB5C, 0xB22, 0xB5D, 0 };
|
||||
static word combb3e[] = { 0xB47, 0xB4B, 0 };
|
||||
static word combb56[] = { 0xB47, 0xB48, 0 };
|
||||
static word combb57[] = { 0xB47, 0xB4C, 0 };
|
||||
static word combbbe[] = { 0xBC6, 0xBCA, 0xBC7, 0xBCB, 0 };
|
||||
static word combbd7[] = { 0xB92, 0xB94, 0xBC6, 0xBCC, 0 };
|
||||
static word combc56[] = { 0xC46, 0xC48, 0 };
|
||||
static word combcc2[] = { 0xCC6, 0xCCA, 0 };
|
||||
static word combcd5[] = { 0xCBF, 0xCC0, 0xCC6, 0xCC7, 0xCCA, 0xCCB, 0 };
|
||||
static word combcd6[] = { 0xCC6, 0xCC8, 0 };
|
||||
static word combd3e[] = { 0xD46, 0xD4A, 0xD47, 0xD4B, 0 };
|
||||
static word combd57[] = { 0xD46, 0xD4C, 0 };
|
||||
static word combdca[] = { 0xDD9, 0xDDA, 0xDDC, 0xDDD, 0 };
|
||||
static word combdcf[] = { 0xDD9, 0xDDC, 0 };
|
||||
static word combddf[] = { 0xDD9, 0xDDE, 0 };
|
||||
static word combf72[] = { 0xF71, 0xF73, 0 };
|
||||
static word combf74[] = { 0xF71, 0xF75, 0 };
|
||||
static word combf80[] = { 0xF71, 0xF81, 0xFB2, 0xF76, 0xFB3, 0xF78, 0 };
|
||||
static word combfb5[] = { 0xF40, 0xF69, 0xF90, 0xFB9, 0 };
|
||||
static word combfb7[] = { 0xF42, 0xF43, 0xF4C, 0xF4D, 0xF51, 0xF52, 0xF56, 0xF57, 0xF5B, 0xF5C, 0xF92, 0xF93, 0xF9C, 0xF9D, 0xFA1, 0xFA2, 0xFA6, 0xFA7, 0xFAB, 0xFAC, 0 };
|
||||
static word comb102e[] = { 0x1025, 0x1026, 0 };
|
||||
static word comb1b35[] = { 0x1B05, 0x1B06, 0x1B07, 0x1B08, 0x1B09, 0x1B0A, 0x1B0B, 0x1B0C, 0x1B0D, 0x1B0E, 0x1B11, 0x1B12, 0x1B3A, 0x1B3B, 0x1B3C, 0x1B3D, 0x1B3E, 0x1B40, 0x1B3F, 0x1B41, 0x1B42, 0x1B43, 0 };
|
||||
static word comb3099[] = { 0x3046, 0x3094, 0x304B, 0x304C, 0x304D, 0x304E, 0x304F, 0x3050, 0x3051, 0x3052, 0x3053, 0x3054, 0x3055, 0x3056, 0x3057, 0x3058, 0x3059, 0x305A, 0x305B, 0x305C, 0x305D, 0x305E, 0x305F, 0x3060, 0x3061, 0x3062, 0x3064, 0x3065, 0x3066, 0x3067, 0x3068, 0x3069, 0x306F, 0x3070, 0x3072, 0x3073, 0x3075, 0x3076, 0x3078, 0x3079, 0x307B, 0x307C, 0x309D, 0x309E, 0x30A6, 0x30F4, 0x30AB, 0x30AC, 0x30AD, 0x30AE, 0x30AF, 0x30B0, 0x30B1, 0x30B2, 0x30B3, 0x30B4, 0x30B5, 0x30B6, 0x30B7, 0x30B8, 0x30B9, 0x30BA, 0x30BB, 0x30BC, 0x30BD, 0x30BE, 0x30BF, 0x30C0, 0x30C1, 0x30C2, 0x30C4, 0x30C5, 0x30C6, 0x30C7, 0x30C8, 0x30C9, 0x30CF, 0x30D0, 0x30D2, 0x30D3, 0x30D5, 0x30D6, 0x30D8, 0x30D9, 0x30DB, 0x30DC, 0x30EF, 0x30F7, 0x30F0, 0x30F8, 0x30F1, 0x30F9, 0x30F2, 0x30FA, 0x30FD, 0x30FE, 0 };
|
||||
static word comb309a[] = { 0x306F, 0x3071, 0x3072, 0x3074, 0x3075, 0x3077, 0x3078, 0x307A, 0x307B, 0x307D, 0x30CF, 0x30D1, 0x30D2, 0x30D4, 0x30D5, 0x30D7, 0x30D8, 0x30DA, 0x30DB, 0x30DD, 0 };
|
||||
|
||||
|
||||
// Maybe binary search would be better here?
|
||||
switch(combine) {
|
||||
case 0x311: return sComb(comb311, chr);
|
||||
case 0x313: return sComb(comb313, chr);
|
||||
case 0x314: return sComb(comb314, chr);
|
||||
case 0x31b: return sComb(comb31b, chr);
|
||||
case 0x323: return sComb(comb323, chr);
|
||||
case 0x324: return sComb(comb324, chr);
|
||||
case 0x325: return sComb(comb325, chr);
|
||||
case 0x326: return sComb(comb326, chr);
|
||||
case 0x327: return sComb(comb327, chr);
|
||||
case 0x328: return sComb(comb328, chr);
|
||||
case 0x32d: return sComb(comb32d, chr);
|
||||
case 0x32e: return sComb(comb32e, chr);
|
||||
case 0x330: return sComb(comb330, chr);
|
||||
case 0x331: return sComb(comb331, chr);
|
||||
case 0x338: return sComb(comb338, chr);
|
||||
case 0x342: return sComb(comb342, chr);
|
||||
case 0x345: return sComb(comb345, chr);
|
||||
case 0x5b4: return sComb(comb5b4, chr);
|
||||
case 0x5b7: return sComb(comb5b7, chr);
|
||||
case 0x5b8: return sComb(comb5b8, chr);
|
||||
case 0x5b9: return sComb(comb5b9, chr);
|
||||
case 0x5bc: return sComb(comb5bc, chr);
|
||||
case 0x5bf: return sComb(comb5bf, chr);
|
||||
case 0x5c1: return sComb(comb5c1, chr);
|
||||
case 0x5c2: return sComb(comb5c2, chr);
|
||||
case 0x653: return sComb(comb653, chr);
|
||||
case 0x654: return sComb(comb654, chr);
|
||||
case 0x655: return sComb(comb655, chr);
|
||||
case 0x93c: return sComb(comb93c, chr);
|
||||
case 0x9bc: return sComb(comb9bc, chr);
|
||||
case 0x9be: return sComb(comb9be, chr);
|
||||
case 0x9d7: return sComb(comb9d7, chr);
|
||||
case 0xa3c: return sComb(comba3c, chr);
|
||||
case 0xb3c: return sComb(combb3c, chr);
|
||||
case 0xb3e: return sComb(combb3e, chr);
|
||||
case 0xb56: return sComb(combb56, chr);
|
||||
case 0xb57: return sComb(combb57, chr);
|
||||
case 0xbbe: return sComb(combbbe, chr);
|
||||
case 0xbd7: return sComb(combbd7, chr);
|
||||
case 0xc56: return sComb(combc56, chr);
|
||||
case 0xcc2: return sComb(combcc2, chr);
|
||||
case 0xcd5: return sComb(combcd5, chr);
|
||||
case 0xcd6: return sComb(combcd6, chr);
|
||||
case 0xd3e: return sComb(combd3e, chr);
|
||||
case 0xd57: return sComb(combd57, chr);
|
||||
case 0xdca: return sComb(combdca, chr);
|
||||
case 0xdcf: return sComb(combdcf, chr);
|
||||
case 0xddf: return sComb(combddf, chr);
|
||||
case 0xf72: return sComb(combf72, chr);
|
||||
case 0xf74: return sComb(combf74, chr);
|
||||
case 0xf80: return sComb(combf80, chr);
|
||||
case 0xfb5: return sComb(combfb5, chr);
|
||||
case 0xfb7: return sComb(combfb7, chr);
|
||||
case 0x102e: return sComb(comb102e, chr);
|
||||
case 0x1b35: return sComb(comb1b35, chr);
|
||||
case 0x3099: return sComb(comb3099, chr);
|
||||
case 0x309a: return sComb(comb309a, chr);
|
||||
default: break;
|
||||
}
|
||||
|
||||
// Unicode > 16-bit canonical precompositions tables:
|
||||
// static dword comb110ba[] = { 0x11099, 0x1109A, 0x1109B, 0x1109C, 0x110A5, 0x110AB, 0 };
|
||||
// static dword comb11127[] = { 0x11131, 0x1112E, 0x11132, 0x1112F, 0 };
|
||||
// static dword comb1133e[] = { 0x11347, 0x1134B, 0 };
|
||||
// static dword comb11357[] = { 0x11347, 0x1134C, 0 };
|
||||
// static dword comb114b0[] = { 0x114B9, 0x114BC, 0 };
|
||||
// static dword comb114ba[] = { 0x114B9, 0x114BB, 0 };
|
||||
// static dword comb114bd[] = { 0x114B9, 0x114BE, 0 };
|
||||
// static dword comb115af[] = { 0x115B8, 0x115BA, 0x115B9, 0x115BB, 0 };
|
||||
// static dword comb11930[] = { 0x11935, 0x11938, 0 };
|
||||
// static dword comb1d165[] = { 0x1D157, 0x1D15E, 0x1D158, 0x1D15F, 0x1D1B9, 0x1D1BB, 0x1D1BA, 0x1D1BC, 0 };
|
||||
// static dword comb1d16e[] = { 0x1D15F, 0x1D160, 0x1D1BB, 0x1D1BD, 0x1D1BC, 0x1D1BE, 0 };
|
||||
// static dword comb1d16f[] = { 0x1D15F, 0x1D161, 0x1D1BB, 0x1D1BF, 0x1D1BC, 0x1D1C0, 0 };
|
||||
// static dword comb1d170[] = { 0x1D15F, 0x1D162, 0 };
|
||||
// static dword comb1d171[] = { 0x1D15F, 0x1D163, 0 };
|
||||
// static dword comb1d172[] = { 0x1D15F, 0x1D164, 0 };
|
||||
|
||||
return 0;
|
||||
dword h[2] = { chr, combine };
|
||||
return UnicodeCompose(h, 2);
|
||||
}
|
||||
|
||||
void ToUpper(char *t, const char *s, int len, byte charset)
|
||||
|
|
|
|||
|
|
@ -49,27 +49,55 @@ byte unicode_fast_info__[] = {
|
|||
};
|
||||
|
||||
struct UnicodeInfo {
|
||||
struct CharBits {
|
||||
Vector<byte *> page;
|
||||
|
||||
void Set(int i) {
|
||||
int ii = i >> 10;
|
||||
byte *b = page.At(ii, NULL);
|
||||
if(!b) {
|
||||
b = page[ii] = new byte[128];
|
||||
memset(b, 0, 128);
|
||||
}
|
||||
b[(i >> 3) & 127] |= (1 << (i & 7));
|
||||
}
|
||||
|
||||
void Shrink() { page.Shrink(); }
|
||||
|
||||
bool Get(int i) {
|
||||
int ii = i >> 10;
|
||||
return ii >= 0 && ii < page.GetCount() && page[ii] && (page[ii][(i >> 3) & 127] & (1 << (i & 7)));
|
||||
}
|
||||
|
||||
~CharBits() {
|
||||
for(byte *b : page)
|
||||
if(b)
|
||||
delete[] b;
|
||||
}
|
||||
};
|
||||
|
||||
Index<dword> composed;
|
||||
Index<String> decomposed; // using String as the vessel for dwords
|
||||
int canonical_count;
|
||||
Index<dword> lower;
|
||||
Index<dword> upper;
|
||||
Index<dword> rtl;
|
||||
Index<dword> letter;
|
||||
Index<dword> islower;
|
||||
Index<dword> isupper;
|
||||
Index<dword> ismark;
|
||||
|
||||
Index<dword> lower;
|
||||
Index<dword> upper;
|
||||
CharBits rtl;
|
||||
CharBits letter;
|
||||
CharBits islower;
|
||||
CharBits isupper;
|
||||
CharBits ismark;
|
||||
|
||||
void Load(const Vector<dword>& data, Index<dword>& x, int& offset);
|
||||
void Load(const Vector<dword>& data, CharBits& x, int& offset);
|
||||
|
||||
UnicodeInfo();
|
||||
};
|
||||
|
||||
void UnicodeInfo::Load(const Vector<dword>& data, Index<dword>& x, int& offset)
|
||||
void UnicodeInfo::Load(const Vector<dword>& data, CharBits& x, int& offset)
|
||||
{
|
||||
int count = data[offset];
|
||||
for(int i = 0; i < count; i++)
|
||||
x.Add(data[i + 1 + offset]);
|
||||
x.Set(data[i + 1 + offset]);
|
||||
x.Shrink();
|
||||
offset += count + 1;
|
||||
}
|
||||
|
|
@ -97,7 +125,6 @@ UnicodeInfo::UnicodeInfo()
|
|||
data.Add(val);
|
||||
prev = val;
|
||||
}
|
||||
|
||||
|
||||
int count = canonical_count = data[0];
|
||||
count += data[1];
|
||||
|
|
@ -182,7 +209,6 @@ dword UnicodeCompose(const dword *t, int count)
|
|||
return q >= 0 && q < f.canonical_count ? f.composed[q] : 0;
|
||||
}
|
||||
|
||||
|
||||
dword ToUpperRest_(dword c)
|
||||
{
|
||||
const UnicodeInfo& f = Single<UnicodeInfo>();
|
||||
|
|
@ -214,32 +240,32 @@ char ToAsciiRest_(dword c)
|
|||
if(q < 0)
|
||||
return ' ';
|
||||
int ch = *(const dword *)~f.decomposed[q];
|
||||
return ch < 128 ? ch : ' ';
|
||||
return ch >= ' ' && ch < 128 ? ch : ' ';
|
||||
}
|
||||
|
||||
bool IsRTL_(dword c)
|
||||
{
|
||||
return Single<UnicodeInfo>().rtl.Find(c) >= 0;
|
||||
return Single<UnicodeInfo>().rtl.Get(c);
|
||||
}
|
||||
|
||||
bool IsLetter_(dword c)
|
||||
{
|
||||
return Single<UnicodeInfo>().letter.Find(c) >= 0;
|
||||
return Single<UnicodeInfo>().letter.Get(c);
|
||||
}
|
||||
|
||||
bool IsLower_(dword c)
|
||||
{
|
||||
return c != (dword)ToUpper((int)c) || Single<UnicodeInfo>().islower.Find(c) >= 0;
|
||||
return c != (dword)ToUpper((int)c) || Single<UnicodeInfo>().islower.Get(c);
|
||||
}
|
||||
|
||||
bool IsUpper_(dword c)
|
||||
{
|
||||
return c != (dword)ToLower((int)c) || Single<UnicodeInfo>().isupper.Find(c) >= 0;
|
||||
return c != (dword)ToLower((int)c) || Single<UnicodeInfo>().isupper.Get(c);
|
||||
}
|
||||
|
||||
bool IsMark_(dword c)
|
||||
{
|
||||
return Single<UnicodeInfo>().ismark.Find(c) >= 0;
|
||||
return Single<UnicodeInfo>().ismark.Get(c);
|
||||
}
|
||||
|
||||
};
|
||||
|
|
|
|||
|
|
@ -404,13 +404,6 @@ character.&]
|
|||
[s2;%% Returns true if [%-*@3 c] is a double`-width UNICODE character
|
||||
(like CJK ideograph).&]
|
||||
[s3;%% &]
|
||||
[s4;%% &]
|
||||
[s5;:UnicodeCombine`(word`,word`): [_^word^ word]_[* UnicodeCombine]([_^word^ word]_[*@3 chr],
|
||||
[_^word^ word]_[*@3 combine])&]
|
||||
[s2;%% Combines [%-*@3 chr] with diacritical marking [%-*@3 combine],
|
||||
returns the result as UNICODE character. Not all combinations
|
||||
are covered...&]
|
||||
[s3;%% &]
|
||||
[s4; &]
|
||||
[s5;:Upp`:`:Utf8ToAscii`(const Upp`:`:String`&`): [_^Upp`:`:String^ String]_[* Utf8ToAsci
|
||||
i]([@(0.0.255) const]_[_^Upp`:`:String^ String][@(0.0.255) `&]_[*@3 src])&]
|
||||
|
|
|
|||
|
|
@ -228,7 +228,7 @@ word]>_[* UnicodeDecompose]([_^Upp`:`:dword^ dword]_[*@3 codepoint],
|
|||
[@(0.0.255) bool]_[*@3 only`_canonical])&]
|
||||
[s2;%% Returns UNICODE decomposition of given [%-*@3 codepoint] into
|
||||
base and combining characters. If [%-*@3 only`_canonical] is true,
|
||||
only canonical decomposition is allowed..&]
|
||||
only canonical decomposition is allowed.&]
|
||||
[s3;%% &]
|
||||
[s4; &]
|
||||
[s5;:Upp`:`:UnicodeCompose`(const Upp`:`:dword`*`,int`): [_^Upp`:`:dword^ dword]_[* Unico
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue