#include <stdlib.h>
#include "fuzzy/common.h"
extern const unsigned char midClass[];
extern const unsigned char initClass[];
extern const unsigned char className[];
static const unsigned char translit_utf8_lookup[] = {
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x00, 0x00,
};
static int utf8Read(const unsigned char* z, int n, int* pSize) {
int c, i;
if (NEVER(n == 0)) {
c = i = 0;
} else {
c = z[0];
i = 1;
if (c >= 0xc0) {
c = translit_utf8_lookup[c - 0xc0];
while (i < n && (z[i] & 0xc0) == 0x80) {
c = (c << 6) + (0x3f & z[i++]);
}
}
}
*pSize = i;
return c;
}
typedef struct Transliteration Transliteration;
struct Transliteration {
unsigned short int cFrom;
unsigned char cTo0, cTo1, cTo2, cTo3;
};
static const Transliteration translit[] = {
{0x00A0, 0x20, 0x00, 0x00, 0x00},
{0x00B5, 0x75, 0x00, 0x00, 0x00},
{0x00C0, 0x41, 0x00, 0x00, 0x00},
{0x00C1, 0x41, 0x00, 0x00, 0x00},
{0x00C2, 0x41, 0x00, 0x00, 0x00},
{0x00C3, 0x41, 0x00, 0x00, 0x00},
{0x00C4, 0x41, 0x65, 0x00, 0x00},
{0x00C5, 0x41, 0x61, 0x00, 0x00},
{0x00C6, 0x41, 0x45, 0x00, 0x00},
{0x00C7, 0x43, 0x00, 0x00, 0x00},
{0x00C8, 0x45, 0x00, 0x00, 0x00},
{0x00C9, 0x45, 0x00, 0x00, 0x00},
{0x00CA, 0x45, 0x00, 0x00, 0x00},
{0x00CB, 0x45, 0x00, 0x00, 0x00},
{0x00CC, 0x49, 0x00, 0x00, 0x00},
{0x00CD, 0x49, 0x00, 0x00, 0x00},
{0x00CE, 0x49, 0x00, 0x00, 0x00},
{0x00CF, 0x49, 0x00, 0x00, 0x00},
{0x00D0, 0x44, 0x00, 0x00, 0x00},
{0x00D1, 0x4E, 0x00, 0x00, 0x00},
{0x00D2, 0x4F, 0x00, 0x00, 0x00},
{0x00D3, 0x4F, 0x00, 0x00, 0x00},
{0x00D4, 0x4F, 0x00, 0x00, 0x00},
{0x00D5, 0x4F, 0x00, 0x00, 0x00},
{0x00D6, 0x4F, 0x65, 0x00, 0x00},
{0x00D7, 0x78, 0x00, 0x00, 0x00},
{0x00D8, 0x4F, 0x00, 0x00, 0x00},
{0x00D9, 0x55, 0x00, 0x00, 0x00},
{0x00DA, 0x55, 0x00, 0x00, 0x00},
{0x00DB, 0x55, 0x00, 0x00, 0x00},
{0x00DC, 0x55, 0x65, 0x00, 0x00},
{0x00DD, 0x59, 0x00, 0x00, 0x00},
{0x00DE, 0x54, 0x68, 0x00, 0x00},
{0x00DF, 0x73, 0x73, 0x00, 0x00},
{0x00E0, 0x61, 0x00, 0x00, 0x00},
{0x00E1, 0x61, 0x00, 0x00, 0x00},
{0x00E2, 0x61, 0x00, 0x00, 0x00},
{0x00E3, 0x61, 0x00, 0x00, 0x00},
{0x00E4, 0x61, 0x65, 0x00, 0x00},
{0x00E5, 0x61, 0x61, 0x00, 0x00},
{0x00E6, 0x61, 0x65, 0x00, 0x00},
{0x00E7, 0x63, 0x00, 0x00, 0x00},
{0x00E8, 0x65, 0x00, 0x00, 0x00},
{0x00E9, 0x65, 0x00, 0x00, 0x00},
{0x00EA, 0x65, 0x00, 0x00, 0x00},
{0x00EB, 0x65, 0x00, 0x00, 0x00},
{0x00EC, 0x69, 0x00, 0x00, 0x00},
{0x00ED, 0x69, 0x00, 0x00, 0x00},
{0x00EE, 0x69, 0x00, 0x00, 0x00},
{0x00EF, 0x69, 0x00, 0x00, 0x00},
{0x00F0, 0x64, 0x00, 0x00, 0x00},
{0x00F1, 0x6E, 0x00, 0x00, 0x00},
{0x00F2, 0x6F, 0x00, 0x00, 0x00},
{0x00F3, 0x6F, 0x00, 0x00, 0x00},
{0x00F4, 0x6F, 0x00, 0x00, 0x00},
{0x00F5, 0x6F, 0x00, 0x00, 0x00},
{0x00F6, 0x6F, 0x65, 0x00, 0x00},
{0x00F7, 0x3A, 0x00, 0x00, 0x00},
{0x00F8, 0x6F, 0x00, 0x00, 0x00},
{0x00F9, 0x75, 0x00, 0x00, 0x00},
{0x00FA, 0x75, 0x00, 0x00, 0x00},
{0x00FB, 0x75, 0x00, 0x00, 0x00},
{0x00FC, 0x75, 0x65, 0x00, 0x00},
{0x00FD, 0x79, 0x00, 0x00, 0x00},
{0x00FE, 0x74, 0x68, 0x00, 0x00},
{0x00FF, 0x79, 0x00, 0x00, 0x00},
{0x0100, 0x41, 0x00, 0x00, 0x00},
{0x0101, 0x61, 0x00, 0x00, 0x00},
{0x0102, 0x41, 0x00, 0x00, 0x00},
{0x0103, 0x61, 0x00, 0x00, 0x00},
{0x0104, 0x41, 0x00, 0x00, 0x00},
{0x0105, 0x61, 0x00, 0x00, 0x00},
{0x0106, 0x43, 0x00, 0x00, 0x00},
{0x0107, 0x63, 0x00, 0x00, 0x00},
{0x0108, 0x43, 0x68, 0x00, 0x00},
{0x0109, 0x63, 0x68, 0x00, 0x00},
{0x010A, 0x43, 0x00, 0x00, 0x00},
{0x010B, 0x63, 0x00, 0x00, 0x00},
{0x010C, 0x43, 0x00, 0x00, 0x00},
{0x010D, 0x63, 0x00, 0x00, 0x00},
{0x010E, 0x44, 0x00, 0x00, 0x00},
{0x010F, 0x64, 0x00, 0x00, 0x00},
{0x0110, 0x44, 0x00, 0x00, 0x00},
{0x0111, 0x64, 0x00, 0x00, 0x00},
{0x0112, 0x45, 0x00, 0x00, 0x00},
{0x0113, 0x65, 0x00, 0x00, 0x00},
{0x0114, 0x45, 0x00, 0x00, 0x00},
{0x0115, 0x65, 0x00, 0x00, 0x00},
{0x0116, 0x45, 0x00, 0x00, 0x00},
{0x0117, 0x65, 0x00, 0x00, 0x00},
{0x0118, 0x45, 0x00, 0x00, 0x00},
{0x0119, 0x65, 0x00, 0x00, 0x00},
{0x011A, 0x45, 0x00, 0x00, 0x00},
{0x011B, 0x65, 0x00, 0x00, 0x00},
{0x011C, 0x47, 0x68, 0x00, 0x00},
{0x011D, 0x67, 0x68, 0x00, 0x00},
{0x011E, 0x47, 0x00, 0x00, 0x00},
{0x011F, 0x67, 0x00, 0x00, 0x00},
{0x0120, 0x47, 0x00, 0x00, 0x00},
{0x0121, 0x67, 0x00, 0x00, 0x00},
{0x0122, 0x47, 0x00, 0x00, 0x00},
{0x0123, 0x67, 0x00, 0x00, 0x00},
{0x0124, 0x48, 0x68, 0x00, 0x00},
{0x0125, 0x68, 0x68, 0x00, 0x00},
{0x0126, 0x48, 0x00, 0x00, 0x00},
{0x0127, 0x68, 0x00, 0x00, 0x00},
{0x0128, 0x49, 0x00, 0x00, 0x00},
{0x0129, 0x69, 0x00, 0x00, 0x00},
{0x012A, 0x49, 0x00, 0x00, 0x00},
{0x012B, 0x69, 0x00, 0x00, 0x00},
{0x012C, 0x49, 0x00, 0x00, 0x00},
{0x012D, 0x69, 0x00, 0x00, 0x00},
{0x012E, 0x49, 0x00, 0x00, 0x00},
{0x012F, 0x69, 0x00, 0x00, 0x00},
{0x0130, 0x49, 0x00, 0x00, 0x00},
{0x0131, 0x69, 0x00, 0x00, 0x00},
{0x0132, 0x49, 0x4A, 0x00, 0x00},
{0x0133, 0x69, 0x6A, 0x00, 0x00},
{0x0134, 0x4A, 0x68, 0x00, 0x00},
{0x0135, 0x6A, 0x68, 0x00, 0x00},
{0x0136, 0x4B, 0x00, 0x00, 0x00},
{0x0137, 0x6B, 0x00, 0x00, 0x00},
{0x0138, 0x6B, 0x00, 0x00, 0x00},
{0x0139, 0x4C, 0x00, 0x00, 0x00},
{0x013A, 0x6C, 0x00, 0x00, 0x00},
{0x013B, 0x4C, 0x00, 0x00, 0x00},
{0x013C, 0x6C, 0x00, 0x00, 0x00},
{0x013D, 0x4C, 0x00, 0x00, 0x00},
{0x013E, 0x6C, 0x00, 0x00, 0x00},
{0x013F, 0x4C, 0x2E, 0x00, 0x00},
{0x0140, 0x6C, 0x2E, 0x00, 0x00},
{0x0141, 0x4C, 0x00, 0x00, 0x00},
{0x0142, 0x6C, 0x00, 0x00, 0x00},
{0x0143, 0x4E, 0x00, 0x00, 0x00},
{0x0144, 0x6E, 0x00, 0x00, 0x00},
{0x0145, 0x4E, 0x00, 0x00, 0x00},
{0x0146, 0x6E, 0x00, 0x00, 0x00},
{0x0147, 0x4E, 0x00, 0x00, 0x00},
{0x0148, 0x6E, 0x00, 0x00, 0x00},
{0x0149, 0x27, 0x6E, 0x00, 0x00},
{0x014A, 0x4E, 0x47, 0x00, 0x00},
{0x014B, 0x6E, 0x67, 0x00, 0x00},
{0x014C, 0x4F, 0x00, 0x00, 0x00},
{0x014D, 0x6F, 0x00, 0x00, 0x00},
{0x014E, 0x4F, 0x00, 0x00, 0x00},
{0x014F, 0x6F, 0x00, 0x00, 0x00},
{0x0150, 0x4F, 0x00, 0x00, 0x00},
{0x0151, 0x6F, 0x00, 0x00, 0x00},
{0x0152, 0x4F, 0x45, 0x00, 0x00},
{0x0153, 0x6F, 0x65, 0x00, 0x00},
{0x0154, 0x52, 0x00, 0x00, 0x00},
{0x0155, 0x72, 0x00, 0x00, 0x00},
{0x0156, 0x52, 0x00, 0x00, 0x00},
{0x0157, 0x72, 0x00, 0x00, 0x00},
{0x0158, 0x52, 0x00, 0x00, 0x00},
{0x0159, 0x72, 0x00, 0x00, 0x00},
{0x015A, 0x53, 0x00, 0x00, 0x00},
{0x015B, 0x73, 0x00, 0x00, 0x00},
{0x015C, 0x53, 0x68, 0x00, 0x00},
{0x015D, 0x73, 0x68, 0x00, 0x00},
{0x015E, 0x53, 0x00, 0x00, 0x00},
{0x015F, 0x73, 0x00, 0x00, 0x00},
{0x0160, 0x53, 0x00, 0x00, 0x00},
{0x0161, 0x73, 0x00, 0x00, 0x00},
{0x0162, 0x54, 0x00, 0x00, 0x00},
{0x0163, 0x74, 0x00, 0x00, 0x00},
{0x0164, 0x54, 0x00, 0x00, 0x00},
{0x0165, 0x74, 0x00, 0x00, 0x00},
{0x0166, 0x54, 0x00, 0x00, 0x00},
{0x0167, 0x74, 0x00, 0x00, 0x00},
{0x0168, 0x55, 0x00, 0x00, 0x00},
{0x0169, 0x75, 0x00, 0x00, 0x00},
{0x016A, 0x55, 0x00, 0x00, 0x00},
{0x016B, 0x75, 0x00, 0x00, 0x00},
{0x016C, 0x55, 0x00, 0x00, 0x00},
{0x016D, 0x75, 0x00, 0x00, 0x00},
{0x016E, 0x55, 0x00, 0x00, 0x00},
{0x016F, 0x75, 0x00, 0x00, 0x00},
{0x0170, 0x55, 0x00, 0x00, 0x00},
{0x0171, 0x75, 0x00, 0x00, 0x00},
{0x0172, 0x55, 0x00, 0x00, 0x00},
{0x0173, 0x75, 0x00, 0x00, 0x00},
{0x0174, 0x57, 0x00, 0x00, 0x00},
{0x0175, 0x77, 0x00, 0x00, 0x00},
{0x0176, 0x59, 0x00, 0x00, 0x00},
{0x0177, 0x79, 0x00, 0x00, 0x00},
{0x0178, 0x59, 0x00, 0x00, 0x00},
{0x0179, 0x5A, 0x00, 0x00, 0x00},
{0x017A, 0x7A, 0x00, 0x00, 0x00},
{0x017B, 0x5A, 0x00, 0x00, 0x00},
{0x017C, 0x7A, 0x00, 0x00, 0x00},
{0x017D, 0x5A, 0x00, 0x00, 0x00},
{0x017E, 0x7A, 0x00, 0x00, 0x00},
{0x017F, 0x73, 0x00, 0x00, 0x00},
{0x0192, 0x66, 0x00, 0x00, 0x00},
{0x0218, 0x53, 0x00, 0x00, 0x00},
{0x0219, 0x73, 0x00, 0x00, 0x00},
{0x021A, 0x54, 0x00, 0x00, 0x00},
{0x021B, 0x74, 0x00, 0x00, 0x00},
{0x0386, 0x41, 0x00, 0x00, 0x00},
{0x0388, 0x45, 0x00, 0x00, 0x00},
{0x0389, 0x49, 0x00, 0x00, 0x00},
{0x038A, 0x49, 0x00, 0x00, 0x00},
{0x038C, 0x4f, 0x00, 0x00, 0x00},
{0x038E, 0x59, 0x00, 0x00, 0x00},
{0x038F, 0x4f, 0x00, 0x00, 0x00},
{0x0390, 0x69, 0x00, 0x00, 0x00},
{0x0391, 0x41, 0x00, 0x00, 0x00},
{0x0392, 0x42, 0x00, 0x00, 0x00},
{0x0393, 0x47, 0x00, 0x00, 0x00},
{0x0394, 0x44, 0x00, 0x00, 0x00},
{0x0395, 0x45, 0x00, 0x00, 0x00},
{0x0396, 0x5a, 0x00, 0x00, 0x00},
{0x0397, 0x49, 0x00, 0x00, 0x00},
{0x0398, 0x54, 0x68, 0x00, 0x00},
{0x0399, 0x49, 0x00, 0x00, 0x00},
{0x039A, 0x4b, 0x00, 0x00, 0x00},
{0x039B, 0x4c, 0x00, 0x00, 0x00},
{0x039C, 0x4d, 0x00, 0x00, 0x00},
{0x039D, 0x4e, 0x00, 0x00, 0x00},
{0x039E, 0x58, 0x00, 0x00, 0x00},
{0x039F, 0x4f, 0x00, 0x00, 0x00},
{0x03A0, 0x50, 0x00, 0x00, 0x00},
{0x03A1, 0x52, 0x00, 0x00, 0x00},
{0x03A3, 0x53, 0x00, 0x00, 0x00},
{0x03A4, 0x54, 0x00, 0x00, 0x00},
{0x03A5, 0x59, 0x00, 0x00, 0x00},
{0x03A6, 0x46, 0x00, 0x00, 0x00},
{0x03A7, 0x43, 0x68, 0x00, 0x00},
{0x03A8, 0x50, 0x73, 0x00, 0x00},
{0x03A9, 0x4f, 0x00, 0x00, 0x00},
{0x03AA, 0x49, 0x00, 0x00, 0x00},
{0x03AB, 0x59, 0x00, 0x00, 0x00},
{0x03AC, 0x61, 0x00, 0x00, 0x00},
{0x03AD, 0x65, 0x00, 0x00, 0x00},
{0x03AE, 0x69, 0x00, 0x00, 0x00},
{0x03AF, 0x69, 0x00, 0x00, 0x00},
{0x03B1, 0x61, 0x00, 0x00, 0x00},
{0x03B2, 0x62, 0x00, 0x00, 0x00},
{0x03B3, 0x67, 0x00, 0x00, 0x00},
{0x03B4, 0x64, 0x00, 0x00, 0x00},
{0x03B5, 0x65, 0x00, 0x00, 0x00},
{0x03B6, 0x7a, 0x00, 0x00, 0x00},
{0x03B7, 0x69, 0x00, 0x00, 0x00},
{0x03B8, 0x74, 0x68, 0x00, 0x00},
{0x03B9, 0x69, 0x00, 0x00, 0x00},
{0x03BA, 0x6b, 0x00, 0x00, 0x00},
{0x03BB, 0x6c, 0x00, 0x00, 0x00},
{0x03BC, 0x6d, 0x00, 0x00, 0x00},
{0x03BD, 0x6e, 0x00, 0x00, 0x00},
{0x03BE, 0x78, 0x00, 0x00, 0x00},
{0x03BF, 0x6f, 0x00, 0x00, 0x00},
{0x03C0, 0x70, 0x00, 0x00, 0x00},
{0x03C1, 0x72, 0x00, 0x00, 0x00},
{0x03C3, 0x73, 0x00, 0x00, 0x00},
{0x03C4, 0x74, 0x00, 0x00, 0x00},
{0x03C5, 0x79, 0x00, 0x00, 0x00},
{0x03C6, 0x66, 0x00, 0x00, 0x00},
{0x03C7, 0x63, 0x68, 0x00, 0x00},
{0x03C8, 0x70, 0x73, 0x00, 0x00},
{0x03C9, 0x6f, 0x00, 0x00, 0x00},
{0x03CA, 0x69, 0x00, 0x00, 0x00},
{0x03CB, 0x79, 0x00, 0x00, 0x00},
{0x03CC, 0x6f, 0x00, 0x00, 0x00},
{0x03CD, 0x79, 0x00, 0x00, 0x00},
{0x03CE, 0x69, 0x00, 0x00, 0x00},
{0x0400, 0x45, 0x00, 0x00, 0x00},
{0x0401, 0x45, 0x00, 0x00, 0x00},
{0x0402, 0x44, 0x00, 0x00, 0x00},
{0x0403, 0x47, 0x00, 0x00, 0x00},
{0x0404, 0x45, 0x00, 0x00, 0x00},
{0x0405, 0x5a, 0x00, 0x00, 0x00},
{0x0406, 0x49, 0x00, 0x00, 0x00},
{0x0407, 0x49, 0x00, 0x00, 0x00},
{0x0408, 0x4a, 0x00, 0x00, 0x00},
{0x0409, 0x49, 0x00, 0x00, 0x00},
{0x040A, 0x4e, 0x00, 0x00, 0x00},
{0x040B, 0x44, 0x00, 0x00, 0x00},
{0x040C, 0x4b, 0x00, 0x00, 0x00},
{0x040D, 0x49, 0x00, 0x00, 0x00},
{0x040E, 0x55, 0x00, 0x00, 0x00},
{0x040F, 0x44, 0x00, 0x00, 0x00},
{0x0410, 0x41, 0x00, 0x00, 0x00},
{0x0411, 0x42, 0x00, 0x00, 0x00},
{0x0412, 0x56, 0x00, 0x00, 0x00},
{0x0413, 0x47, 0x00, 0x00, 0x00},
{0x0414, 0x44, 0x00, 0x00, 0x00},
{0x0415, 0x45, 0x00, 0x00, 0x00},
{0x0416, 0x5a, 0x68, 0x00, 0x00},
{0x0417, 0x5a, 0x00, 0x00, 0x00},
{0x0418, 0x49, 0x00, 0x00, 0x00},
{0x0419, 0x49, 0x00, 0x00, 0x00},
{0x041A, 0x4b, 0x00, 0x00, 0x00},
{0x041B, 0x4c, 0x00, 0x00, 0x00},
{0x041C, 0x4d, 0x00, 0x00, 0x00},
{0x041D, 0x4e, 0x00, 0x00, 0x00},
{0x041E, 0x4f, 0x00, 0x00, 0x00},
{0x041F, 0x50, 0x00, 0x00, 0x00},
{0x0420, 0x52, 0x00, 0x00, 0x00},
{0x0421, 0x53, 0x00, 0x00, 0x00},
{0x0422, 0x54, 0x00, 0x00, 0x00},
{0x0423, 0x55, 0x00, 0x00, 0x00},
{0x0424, 0x46, 0x00, 0x00, 0x00},
{0x0425, 0x4b, 0x68, 0x00, 0x00},
{0x0426, 0x54, 0x63, 0x00, 0x00},
{0x0427, 0x43, 0x68, 0x00, 0x00},
{0x0428, 0x53, 0x68, 0x00, 0x00},
{0x0429, 0x53, 0x68, 0x63, 0x68},
{0x042A, 0x61, 0x00, 0x00, 0x00},
{0x042B, 0x59, 0x00, 0x00, 0x00},
{0x042C, 0x59, 0x00, 0x00, 0x00},
{0x042D, 0x45, 0x00, 0x00, 0x00},
{0x042E, 0x49, 0x75, 0x00, 0x00},
{0x042F, 0x49, 0x61, 0x00, 0x00},
{0x0430, 0x61, 0x00, 0x00, 0x00},
{0x0431, 0x62, 0x00, 0x00, 0x00},
{0x0432, 0x76, 0x00, 0x00, 0x00},
{0x0433, 0x67, 0x00, 0x00, 0x00},
{0x0434, 0x64, 0x00, 0x00, 0x00},
{0x0435, 0x65, 0x00, 0x00, 0x00},
{0x0436, 0x7a, 0x68, 0x00, 0x00},
{0x0437, 0x7a, 0x00, 0x00, 0x00},
{0x0438, 0x69, 0x00, 0x00, 0x00},
{0x0439, 0x69, 0x00, 0x00, 0x00},
{0x043A, 0x6b, 0x00, 0x00, 0x00},
{0x043B, 0x6c, 0x00, 0x00, 0x00},
{0x043C, 0x6d, 0x00, 0x00, 0x00},
{0x043D, 0x6e, 0x00, 0x00, 0x00},
{0x043E, 0x6f, 0x00, 0x00, 0x00},
{0x043F, 0x70, 0x00, 0x00, 0x00},
{0x0440, 0x72, 0x00, 0x00, 0x00},
{0x0441, 0x73, 0x00, 0x00, 0x00},
{0x0442, 0x74, 0x00, 0x00, 0x00},
{0x0443, 0x75, 0x00, 0x00, 0x00},
{0x0444, 0x66, 0x00, 0x00, 0x00},
{0x0445, 0x6b, 0x68, 0x00, 0x00},
{0x0446, 0x74, 0x63, 0x00, 0x00},
{0x0447, 0x63, 0x68, 0x00, 0x00},
{0x0448, 0x73, 0x68, 0x00, 0x00},
{0x0449, 0x73, 0x68, 0x63, 0x68},
{0x044A, 0x61, 0x00, 0x00, 0x00},
{0x044B, 0x79, 0x00, 0x00, 0x00},
{0x044C, 0x79, 0x00, 0x00, 0x00},
{0x044D, 0x65, 0x00, 0x00, 0x00},
{0x044E, 0x69, 0x75, 0x00, 0x00},
{0x044F, 0x69, 0x61, 0x00, 0x00},
{0x0450, 0x65, 0x00, 0x00, 0x00},
{0x0451, 0x65, 0x00, 0x00, 0x00},
{0x0452, 0x64, 0x00, 0x00, 0x00},
{0x0453, 0x67, 0x00, 0x00, 0x00},
{0x0454, 0x65, 0x00, 0x00, 0x00},
{0x0455, 0x7a, 0x00, 0x00, 0x00},
{0x0456, 0x69, 0x00, 0x00, 0x00},
{0x0457, 0x69, 0x00, 0x00, 0x00},
{0x0458, 0x6a, 0x00, 0x00, 0x00},
{0x0459, 0x69, 0x00, 0x00, 0x00},
{0x045A, 0x6e, 0x00, 0x00, 0x00},
{0x045B, 0x64, 0x00, 0x00, 0x00},
{0x045C, 0x6b, 0x00, 0x00, 0x00},
{0x045D, 0x69, 0x00, 0x00, 0x00},
{0x045E, 0x75, 0x00, 0x00, 0x00},
{0x045F, 0x64, 0x00, 0x00, 0x00},
{0x1E02, 0x42, 0x00, 0x00, 0x00},
{0x1E03, 0x62, 0x00, 0x00, 0x00},
{0x1E0A, 0x44, 0x00, 0x00, 0x00},
{0x1E0B, 0x64, 0x00, 0x00, 0x00},
{0x1E1E, 0x46, 0x00, 0x00, 0x00},
{0x1E1F, 0x66, 0x00, 0x00, 0x00},
{0x1E40, 0x4D, 0x00, 0x00, 0x00},
{0x1E41, 0x6D, 0x00, 0x00, 0x00},
{0x1E56, 0x50, 0x00, 0x00, 0x00},
{0x1E57, 0x70, 0x00, 0x00, 0x00},
{0x1E60, 0x53, 0x00, 0x00, 0x00},
{0x1E61, 0x73, 0x00, 0x00, 0x00},
{0x1E6A, 0x54, 0x00, 0x00, 0x00},
{0x1E6B, 0x74, 0x00, 0x00, 0x00},
{0x1E80, 0x57, 0x00, 0x00, 0x00},
{0x1E81, 0x77, 0x00, 0x00, 0x00},
{0x1E82, 0x57, 0x00, 0x00, 0x00},
{0x1E83, 0x77, 0x00, 0x00, 0x00},
{0x1E84, 0x57, 0x00, 0x00, 0x00},
{0x1E85, 0x77, 0x00, 0x00, 0x00},
{0x1EF2, 0x59, 0x00, 0x00, 0x00},
{0x1EF3, 0x79, 0x00, 0x00, 0x00},
{0xFB00, 0x66, 0x66, 0x00, 0x00},
{0xFB01, 0x66, 0x69, 0x00, 0x00},
{0xFB02, 0x66, 0x6C, 0x00, 0x00},
{0xFB05, 0x73, 0x74, 0x00, 0x00},
{0xFB06, 0x73, 0x74, 0x00, 0x00},
};
static const Transliteration* spellfixFindTranslit(int c, int* pxTop) {
*pxTop = (sizeof(translit) / sizeof(translit[0])) - 1;
return translit;
}
unsigned char* transliterate(const unsigned char* zIn, int nIn) {
unsigned char* zOut = malloc(nIn * 4 + 1);
int c, sz, nOut;
if (zOut == 0)
return 0;
nOut = 0;
while (nIn > 0) {
c = utf8Read(zIn, nIn, &sz);
zIn += sz;
nIn -= sz;
if (c <= 127) {
zOut[nOut++] = (unsigned char)c;
} else {
int xTop, xBtm, x;
const Transliteration* tbl = spellfixFindTranslit(c, &xTop);
xBtm = 0;
while (xTop >= xBtm) {
x = (xTop + xBtm) / 2;
if (tbl[x].cFrom == c) {
zOut[nOut++] = tbl[x].cTo0;
if (tbl[x].cTo1) {
zOut[nOut++] = tbl[x].cTo1;
if (tbl[x].cTo2) {
zOut[nOut++] = tbl[x].cTo2;
if (tbl[x].cTo3) {
zOut[nOut++] = tbl[x].cTo3;
}
}
}
c = 0;
break;
} else if (tbl[x].cFrom > c) {
xTop = x - 1;
} else {
xBtm = x + 1;
}
}
if (c)
zOut[nOut++] = '?';
}
}
zOut[nOut] = 0;
return zOut;
}
int translen_to_charlen(const char* zIn, int nIn, int nTrans) {
int i, c, sz, nOut;
int nChar;
i = nOut = 0;
for (nChar = 0; i < nIn && nOut < nTrans; nChar++) {
c = utf8Read((const unsigned char*)&zIn[i], nIn - i, &sz);
i += sz;
nOut++;
if (c >= 128) {
int xTop, xBtm, x;
const Transliteration* tbl = spellfixFindTranslit(c, &xTop);
xBtm = 0;
while (xTop >= xBtm) {
x = (xTop + xBtm) / 2;
if (tbl[x].cFrom == c) {
if (tbl[x].cTo1) {
nOut++;
if (tbl[x].cTo2) {
nOut++;
if (tbl[x].cTo3) {
nOut++;
}
}
}
break;
} else if (tbl[x].cFrom > c) {
xTop = x - 1;
} else {
xBtm = x + 1;
}
}
}
}
return nChar;
}
int script_code(const unsigned char* zIn, int nIn) {
int c, sz;
int scriptMask = 0;
int res;
int seenDigit = 0;
while (nIn > 0) {
c = utf8Read(zIn, nIn, &sz);
zIn += sz;
nIn -= sz;
if (c < 0x02af) {
if (c >= 0x80 || midClass[c & 0x7f] < CCLASS_DIGIT) {
scriptMask |= SCRIPT_LATIN;
} else if (c >= '0' && c <= '9') {
seenDigit = 1;
}
} else if (c >= 0x0400 && c <= 0x04ff) {
scriptMask |= SCRIPT_CYRILLIC;
} else if (c >= 0x0386 && c <= 0x03ce) {
scriptMask |= SCRIPT_GREEK;
} else if (c >= 0x0590 && c <= 0x05ff) {
scriptMask |= SCRIPT_HEBREW;
} else if (c >= 0x0600 && c <= 0x06ff) {
scriptMask |= SCRIPT_ARABIC;
}
}
if (scriptMask == 0 && seenDigit)
scriptMask = SCRIPT_LATIN;
switch (scriptMask) {
case 0:
res = 999;
break;
case SCRIPT_LATIN:
res = 215;
break;
case SCRIPT_CYRILLIC:
res = 220;
break;
case SCRIPT_GREEK:
res = 200;
break;
case SCRIPT_HEBREW:
res = 125;
break;
case SCRIPT_ARABIC:
res = 160;
break;
default:
res = 998;
break;
}
return res;
}