Lisp_Object Vcharset_chinese_cns11643_5;
Lisp_Object Vcharset_chinese_cns11643_6;
Lisp_Object Vcharset_chinese_cns11643_7;
+Lisp_Object Vcharset_ucs_bmp;
+Lisp_Object Vcharset_latin_viscii_lower;
+Lisp_Object Vcharset_latin_viscii_upper;
#endif
Lisp_Object Vcharset_chinese_big5_1;
Lisp_Object Vcharset_chinese_big5_2;
Lisp_Object charset_by_leading_byte[NUM_LEADING_BYTES];
/* Table of charsets indexed by type/final-byte/direction. */
+#ifdef UTF2000
+Lisp_Object charset_by_attributes[4][128];
+#else
Lisp_Object charset_by_attributes[4][128][2];
+#endif
#ifndef UTF2000
/* Table of number of bytes in the string representation of a character
};
#endif
+#ifdef UTF2000
+Emchar_to_byte_table*
+make_byte_from_character_table ()
+{
+ Emchar_to_byte_table* table
+ = (Emchar_to_byte_table*) xmalloc (sizeof (Emchar_to_byte_table));
+
+ table->base = NULL;
+ return table;
+}
+
+void
+put_byte_from_character_table (Emchar ch, unsigned char val,
+ Emchar_to_byte_table* table)
+{
+ if (table->base == NULL)
+ {
+ table->base = xmalloc (128);
+ table->offset = ch - (ch % 128);
+ table->size = 128;
+ table->base[ch - table->offset] = val;
+ }
+ else
+ {
+ int i = ch - table->offset;
+
+ if (i < 0)
+ {
+ size_t new_size = table->size - i;
+ size_t j;
+
+ new_size += 128 - (new_size % 128);
+ table->base = xrealloc (table->base, new_size);
+ memmove (table->base + (new_size - table->size), table->base,
+ table->size);
+ for (j = 0; j < (new_size - table->size); j++)
+ table->base[j] = 0;
+ table->offset -= (new_size - table->size);
+ table->base[ch - table->offset] = val;
+ table->size = new_size;
+ }
+ else if (i >= table->size)
+ {
+ size_t new_size = i + 1;
+ size_t j;
+
+ new_size += 128 - (new_size % 128);
+ table->base = xrealloc (table->base, new_size);
+ for (j = table->size; j < new_size; j++)
+ table->base[j] = 0;
+ table->base[i] = val;
+ table->size = new_size;
+ }
+ else
+ {
+ table->base[i] = val;
+ }
+ }
+}
+
+unsigned char
+get_byte_from_character_table (Emchar ch, Emchar_to_byte_table* table)
+{
+ size_t i = ch - table->offset;
+ if (i < table->size)
+ return table->base[i];
+ else
+ return 0;
+}
+
+#define CHAR96(ft,b) (MIN_CHAR_96 + (ft - '0') * 96 + (b & 0x7f) - 32)
+
+Emchar latin_jisx0201_to_ucs[94] =
+{
+ 0x0021 /* 0x21 EXCLAMATION MARK */,
+ 0x0022 /* 0x22 QUOTATION MARK */,
+ 0x0023 /* 0x23 NUMBER SIGN */,
+ 0x0024 /* 0x24 DOLLAR SIGN */,
+ 0x0025 /* 0x25 PERCENT SIGN */,
+ 0x0026 /* 0x26 AMPERSAND */,
+ 0x0027 /* 0x27 APOSTROPHE */,
+ 0x0028 /* 0x28 LEFT PARENTHESIS */,
+ 0x0029 /* 0x29 RIGHT PARENTHESIS */,
+ 0x002A /* 0x2A ASTERISK */,
+ 0x002B /* 0x2B PLUS SIGN */,
+ 0x002C /* 0x2C COMMA */,
+ 0x002D /* 0x2D HYPHEN-MINUS */,
+ 0x002E /* 0x2E FULL STOP */,
+ 0x002F /* 0x2F SOLIDUS */,
+ 0x0030 /* 0x30 DIGIT ZERO */,
+ 0x0031 /* 0x31 DIGIT ONE */,
+ 0x0032 /* 0x32 DIGIT TWO */,
+ 0x0033 /* 0x33 DIGIT THREE */,
+ 0x0034 /* 0x34 DIGIT FOUR */,
+ 0x0035 /* 0x35 DIGIT FIVE */,
+ 0x0036 /* 0x36 DIGIT SIX */,
+ 0x0037 /* 0x37 DIGIT SEVEN */,
+ 0x0038 /* 0x38 DIGIT EIGHT */,
+ 0x0039 /* 0x39 DIGIT NINE */,
+ 0x003A /* 0x3A COLON */,
+ 0x003B /* 0x3B SEMICOLON */,
+ 0x003C /* 0x3C LESS-THAN SIGN */,
+ 0x003D /* 0x3D EQUALS SIGN */,
+ 0x003E /* 0x3E GREATER-THAN SIGN */,
+ 0x003F /* 0x3F QUESTION MARK */,
+ 0x0040 /* 0x40 COMMERCIAL AT */,
+ 0x0041 /* 0x41 LATIN CAPITAL LETTER A */,
+ 0x0042 /* 0x42 LATIN CAPITAL LETTER B */,
+ 0x0043 /* 0x43 LATIN CAPITAL LETTER C */,
+ 0x0044 /* 0x44 LATIN CAPITAL LETTER D */,
+ 0x0045 /* 0x45 LATIN CAPITAL LETTER E */,
+ 0x0046 /* 0x46 LATIN CAPITAL LETTER F */,
+ 0x0047 /* 0x47 LATIN CAPITAL LETTER G */,
+ 0x0048 /* 0x48 LATIN CAPITAL LETTER H */,
+ 0x0049 /* 0x49 LATIN CAPITAL LETTER I */,
+ 0x004A /* 0x4A LATIN CAPITAL LETTER J */,
+ 0x004B /* 0x4B LATIN CAPITAL LETTER K */,
+ 0x004C /* 0x4C LATIN CAPITAL LETTER L */,
+ 0x004D /* 0x4D LATIN CAPITAL LETTER M */,
+ 0x004E /* 0x4E LATIN CAPITAL LETTER N */,
+ 0x004F /* 0x4F LATIN CAPITAL LETTER O */,
+ 0x0050 /* 0x50 LATIN CAPITAL LETTER P */,
+ 0x0051 /* 0x51 LATIN CAPITAL LETTER Q */,
+ 0x0052 /* 0x52 LATIN CAPITAL LETTER R */,
+ 0x0053 /* 0x53 LATIN CAPITAL LETTER S */,
+ 0x0054 /* 0x54 LATIN CAPITAL LETTER T */,
+ 0x0055 /* 0x55 LATIN CAPITAL LETTER U */,
+ 0x0056 /* 0x56 LATIN CAPITAL LETTER V */,
+ 0x0057 /* 0x57 LATIN CAPITAL LETTER W */,
+ 0x0058 /* 0x58 LATIN CAPITAL LETTER X */,
+ 0x0059 /* 0x59 LATIN CAPITAL LETTER Y */,
+ 0x005A /* 0x5A LATIN CAPITAL LETTER Z */,
+ 0x005B /* 0x5B LEFT SQUARE BRACKET */,
+ 0x00A5 /* 0x5C YEN SIGN */,
+ 0x005D /* 0x5D RIGHT SQUARE BRACKET */,
+ 0x005E /* 0x5E CIRCUMFLEX ACCENT */,
+ 0x005F /* 0x5F LOW LINE */,
+ 0x0060 /* 0x60 GRAVE ACCENT */,
+ 0x0061 /* 0x61 LATIN SMALL LETTER A */,
+ 0x0062 /* 0x62 LATIN SMALL LETTER B */,
+ 0x0063 /* 0x63 LATIN SMALL LETTER C */,
+ 0x0064 /* 0x64 LATIN SMALL LETTER D */,
+ 0x0065 /* 0x65 LATIN SMALL LETTER E */,
+ 0x0066 /* 0x66 LATIN SMALL LETTER F */,
+ 0x0067 /* 0x67 LATIN SMALL LETTER G */,
+ 0x0068 /* 0x68 LATIN SMALL LETTER H */,
+ 0x0069 /* 0x69 LATIN SMALL LETTER I */,
+ 0x006A /* 0x6A LATIN SMALL LETTER J */,
+ 0x006B /* 0x6B LATIN SMALL LETTER K */,
+ 0x006C /* 0x6C LATIN SMALL LETTER L */,
+ 0x006D /* 0x6D LATIN SMALL LETTER M */,
+ 0x006E /* 0x6E LATIN SMALL LETTER N */,
+ 0x006F /* 0x6F LATIN SMALL LETTER O */,
+ 0x0070 /* 0x70 LATIN SMALL LETTER P */,
+ 0x0071 /* 0x71 LATIN SMALL LETTER Q */,
+ 0x0072 /* 0x72 LATIN SMALL LETTER R */,
+ 0x0073 /* 0x73 LATIN SMALL LETTER S */,
+ 0x0074 /* 0x74 LATIN SMALL LETTER T */,
+ 0x0075 /* 0x75 LATIN SMALL LETTER U */,
+ 0x0076 /* 0x76 LATIN SMALL LETTER V */,
+ 0x0077 /* 0x77 LATIN SMALL LETTER W */,
+ 0x0078 /* 0x78 LATIN SMALL LETTER X */,
+ 0x0079 /* 0x79 LATIN SMALL LETTER Y */,
+ 0x007A /* 0x7A LATIN SMALL LETTER Z */,
+ 0x007B /* 0x7B LEFT CURLY BRACKET */,
+ 0x007C /* 0x7C VERTICAL LINE */,
+ 0x007D /* 0x7D RIGHT CURLY BRACKET */,
+ 0x203E /* 0x7E OVERLINE */
+};
+
+Emchar latin_iso8859_2_to_ucs[96] =
+{
+ 0x00A0 /* 0xA0 NO-BREAK SPACE */,
+ 0x0104 /* 0xA1 LATIN CAPITAL LETTER A WITH OGONEK */,
+ 0x02D8 /* 0xA2 BREVE */,
+ 0x0141 /* 0xA3 LATIN CAPITAL LETTER L WITH STROKE */,
+ 0x00A4 /* 0xA4 CURRENCY SIGN */,
+ 0x013D /* 0xA5 LATIN CAPITAL LETTER L WITH CARON */,
+ 0x015A /* 0xA6 LATIN CAPITAL LETTER S WITH ACUTE */,
+ 0x00A7 /* 0xA7 SECTION SIGN */,
+ 0x00A8 /* 0xA8 DIAERESIS */,
+ 0x0160 /* 0xA9 LATIN CAPITAL LETTER S WITH CARON */,
+ 0x015E /* 0xAA LATIN CAPITAL LETTER S WITH CEDILLA */,
+ 0x0164 /* 0xAB LATIN CAPITAL LETTER T WITH CARON */,
+ 0x0179 /* 0xAC LATIN CAPITAL LETTER Z WITH ACUTE */,
+ 0x00AD /* 0xAD SOFT HYPHEN */,
+ 0x017D /* 0xAE LATIN CAPITAL LETTER Z WITH CARON */,
+ 0x017B /* 0xAF LATIN CAPITAL LETTER Z WITH DOT ABOVE */,
+ 0x00B0 /* 0xB0 DEGREE SIGN */,
+ 0x0105 /* 0xB1 LATIN SMALL LETTER A WITH OGONEK */,
+ 0x02DB /* 0xB2 OGONEK */,
+ 0x0142 /* 0xB3 LATIN SMALL LETTER L WITH STROKE */,
+ 0x00B4 /* 0xB4 ACUTE ACCENT */,
+ 0x013E /* 0xB5 LATIN SMALL LETTER L WITH CARON */,
+ 0x015B /* 0xB6 LATIN SMALL LETTER S WITH ACUTE */,
+ 0x02C7 /* 0xB7 CARON */,
+ 0x00B8 /* 0xB8 CEDILLA */,
+ 0x0161 /* 0xB9 LATIN SMALL LETTER S WITH CARON */,
+ 0x015F /* 0xBA LATIN SMALL LETTER S WITH CEDILLA */,
+ 0x0165 /* 0xBB LATIN SMALL LETTER T WITH CARON */,
+ 0x017A /* 0xBC LATIN SMALL LETTER Z WITH ACUTE */,
+ 0x02DD /* 0xBD DOUBLE ACUTE ACCENT */,
+ 0x017E /* 0xBE LATIN SMALL LETTER Z WITH CARON */,
+ 0x017C /* 0xBF LATIN SMALL LETTER Z WITH DOT ABOVE */,
+ 0x0154 /* 0xC0 LATIN CAPITAL LETTER R WITH ACUTE */,
+ 0x00C1 /* 0xC1 LATIN CAPITAL LETTER A WITH ACUTE */,
+ 0x00C2 /* 0xC2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX */,
+ 0x0102 /* 0xC3 LATIN CAPITAL LETTER A WITH BREVE */,
+ 0x00C4 /* 0xC4 LATIN CAPITAL LETTER A WITH DIAERESIS */,
+ 0x0139 /* 0xC5 LATIN CAPITAL LETTER L WITH ACUTE */,
+ 0x0106 /* 0xC6 LATIN CAPITAL LETTER C WITH ACUTE */,
+ 0x00C7 /* 0xC7 LATIN CAPITAL LETTER C WITH CEDILLA */,
+ 0x010C /* 0xC8 LATIN CAPITAL LETTER C WITH CARON */,
+ 0x00C9 /* 0xC9 LATIN CAPITAL LETTER E WITH ACUTE */,
+ 0x0118 /* 0xCA LATIN CAPITAL LETTER E WITH OGONEK */,
+ 0x00CB /* 0xCB LATIN CAPITAL LETTER E WITH DIAERESIS */,
+ 0x011A /* 0xCC LATIN CAPITAL LETTER E WITH CARON */,
+ 0x00CD /* 0xCD LATIN CAPITAL LETTER I WITH ACUTE */,
+ 0x00CE /* 0xCE LATIN CAPITAL LETTER I WITH CIRCUMFLEX */,
+ 0x010E /* 0xCF LATIN CAPITAL LETTER D WITH CARON */,
+ 0x0110 /* 0xD0 LATIN CAPITAL LETTER D WITH STROKE */,
+ 0x0143 /* 0xD1 LATIN CAPITAL LETTER N WITH ACUTE */,
+ 0x0147 /* 0xD2 LATIN CAPITAL LETTER N WITH CARON */,
+ 0x00D3 /* 0xD3 LATIN CAPITAL LETTER O WITH ACUTE */,
+ 0x00D4 /* 0xD4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX */,
+ 0x0150 /* 0xD5 LATIN CAPITAL LETTER O WITH DOUBLE ACUTE */,
+ 0x00D6 /* 0xD6 LATIN CAPITAL LETTER O WITH DIAERESIS */,
+ 0x00D7 /* 0xD7 MULTIPLICATION SIGN */,
+ 0x0158 /* 0xD8 LATIN CAPITAL LETTER R WITH CARON */,
+ 0x016E /* 0xD9 LATIN CAPITAL LETTER U WITH RING ABOVE */,
+ 0x00DA /* 0xDA LATIN CAPITAL LETTER U WITH ACUTE */,
+ 0x0170 /* 0xDB LATIN CAPITAL LETTER U WITH DOUBLE ACUTE */,
+ 0x00DC /* 0xDC LATIN CAPITAL LETTER U WITH DIAERESIS */,
+ 0x00DD /* 0xDD LATIN CAPITAL LETTER Y WITH ACUTE */,
+ 0x0162 /* 0xDE LATIN CAPITAL LETTER T WITH CEDILLA */,
+ 0x00DF /* 0xDF LATIN SMALL LETTER SHARP S */,
+ 0x0155 /* 0xE0 LATIN SMALL LETTER R WITH ACUTE */,
+ 0x00E1 /* 0xE1 LATIN SMALL LETTER A WITH ACUTE */,
+ 0x00E2 /* 0xE2 LATIN SMALL LETTER A WITH CIRCUMFLEX */,
+ 0x0103 /* 0xE3 LATIN SMALL LETTER A WITH BREVE */,
+ 0x00E4 /* 0xE4 LATIN SMALL LETTER A WITH DIAERESIS */,
+ 0x013A /* 0xE5 LATIN SMALL LETTER L WITH ACUTE */,
+ 0x0107 /* 0xE6 LATIN SMALL LETTER C WITH ACUTE */,
+ 0x00E7 /* 0xE7 LATIN SMALL LETTER C WITH CEDILLA */,
+ 0x010D /* 0xE8 LATIN SMALL LETTER C WITH CARON */,
+ 0x00E9 /* 0xE9 LATIN SMALL LETTER E WITH ACUTE */,
+ 0x0119 /* 0xEA LATIN SMALL LETTER E WITH OGONEK */,
+ 0x00EB /* 0xEB LATIN SMALL LETTER E WITH DIAERESIS */,
+ 0x011B /* 0xEC LATIN SMALL LETTER E WITH CARON */,
+ 0x00ED /* 0xED LATIN SMALL LETTER I WITH ACUTE */,
+ 0x00EE /* 0xEE LATIN SMALL LETTER I WITH CIRCUMFLEX */,
+ 0x010F /* 0xEF LATIN SMALL LETTER D WITH CARON */,
+ 0x0111 /* 0xF0 LATIN SMALL LETTER D WITH STROKE */,
+ 0x0144 /* 0xF1 LATIN SMALL LETTER N WITH ACUTE */,
+ 0x0148 /* 0xF2 LATIN SMALL LETTER N WITH CARON */,
+ 0x00F3 /* 0xF3 LATIN SMALL LETTER O WITH ACUTE */,
+ 0x00F4 /* 0xF4 LATIN SMALL LETTER O WITH CIRCUMFLEX */,
+ 0x0151 /* 0xF5 LATIN SMALL LETTER O WITH DOUBLE ACUTE */,
+ 0x00F6 /* 0xF6 LATIN SMALL LETTER O WITH DIAERESIS */,
+ 0x00F7 /* 0xF7 DIVISION SIGN */,
+ 0x0159 /* 0xF8 LATIN SMALL LETTER R WITH CARON */,
+ 0x016F /* 0xF9 LATIN SMALL LETTER U WITH RING ABOVE */,
+ 0x00FA /* 0xFA LATIN SMALL LETTER U WITH ACUTE */,
+ 0x0171 /* 0xFB LATIN SMALL LETTER U WITH DOUBLE ACUTE */,
+ 0x00FC /* 0xFC LATIN SMALL LETTER U WITH DIAERESIS */,
+ 0x00FD /* 0xFD LATIN SMALL LETTER Y WITH ACUTE */,
+ 0x0163 /* 0xFE LATIN SMALL LETTER T WITH CEDILLA */,
+ 0x02D9 /* 0xFF DOT ABOVE */
+};
+
+Emchar latin_iso8859_3_to_ucs[96] =
+{
+ 0x00A0 /* 0xA0 NO-BREAK SPACE */,
+ 0x0126 /* 0xA1 LATIN CAPITAL LETTER H WITH STROKE */,
+ 0x02D8 /* 0xA2 BREVE */,
+ 0x00A3 /* 0xA3 POUND SIGN */,
+ 0x00A4 /* 0xA4 CURRENCY SIGN */,
+ CHAR96('C', 0xA5),
+ 0x0124 /* 0xA6 LATIN CAPITAL LETTER H WITH CIRCUMFLEX */,
+ 0x00A7 /* 0xA7 SECTION SIGN */,
+ 0x00A8 /* 0xA8 DIAERESIS */,
+ 0x0130 /* 0xA9 LATIN CAPITAL LETTER I WITH DOT ABOVE */,
+ 0x015E /* 0xAA LATIN CAPITAL LETTER S WITH CEDILLA */,
+ 0x011E /* 0xAB LATIN CAPITAL LETTER G WITH BREVE */,
+ 0x0134 /* 0xAC LATIN CAPITAL LETTER J WITH CIRCUMFLEX */,
+ 0x00AD /* 0xAD SOFT HYPHEN */,
+ CHAR96('C', 0xAE),
+ 0x017B /* 0xAF LATIN CAPITAL LETTER Z WITH DOT ABOVE */,
+ 0x00B0 /* 0xB0 DEGREE SIGN */,
+ 0x0127 /* 0xB1 LATIN SMALL LETTER H WITH STROKE */,
+ 0x00B2 /* 0xB2 SUPERSCRIPT TWO */,
+ 0x00B3 /* 0xB3 SUPERSCRIPT THREE */,
+ 0x00B4 /* 0xB4 ACUTE ACCENT */,
+ 0x00B5 /* 0xB5 MICRO SIGN */,
+ 0x0125 /* 0xB6 LATIN SMALL LETTER H WITH CIRCUMFLEX */,
+ 0x00B7 /* 0xB7 MIDDLE DOT */,
+ 0x00B8 /* 0xB8 CEDILLA */,
+ 0x0131 /* 0xB9 LATIN SMALL LETTER DOTLESS I */,
+ 0x015F /* 0xBA LATIN SMALL LETTER S WITH CEDILLA */,
+ 0x011F /* 0xBB LATIN SMALL LETTER G WITH BREVE */,
+ 0x0135 /* 0xBC LATIN SMALL LETTER J WITH CIRCUMFLEX */,
+ 0x00BD /* 0xBD VULGAR FRACTION ONE HALF */,
+ CHAR96('C', 0xBE),
+ 0x017C /* 0xBF LATIN SMALL LETTER Z WITH DOT ABOVE */,
+ 0x00C0 /* 0xC0 LATIN CAPITAL LETTER A WITH GRAVE */,
+ 0x00C1 /* 0xC1 LATIN CAPITAL LETTER A WITH ACUTE */,
+ 0x00C2 /* 0xC2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX */,
+ CHAR96('C', 0xC3),
+ 0x00C4 /* 0xC4 LATIN CAPITAL LETTER A WITH DIAERESIS */,
+ 0x010A /* 0xC5 LATIN CAPITAL LETTER C WITH DOT ABOVE */,
+ 0x0108 /* 0xC6 LATIN CAPITAL LETTER C WITH CIRCUMFLEX */,
+ 0x00C7 /* 0xC7 LATIN CAPITAL LETTER C WITH CEDILLA */,
+ 0x00C8 /* 0xC8 LATIN CAPITAL LETTER E WITH GRAVE */,
+ 0x00C9 /* 0xC9 LATIN CAPITAL LETTER E WITH ACUTE */,
+ 0x00CA /* 0xCA LATIN CAPITAL LETTER E WITH CIRCUMFLEX */,
+ 0x00CB /* 0xCB LATIN CAPITAL LETTER E WITH DIAERESIS */,
+ 0x00CC /* 0xCC LATIN CAPITAL LETTER I WITH GRAVE */,
+ 0x00CD /* 0xCD LATIN CAPITAL LETTER I WITH ACUTE */,
+ 0x00CE /* 0xCE LATIN CAPITAL LETTER I WITH CIRCUMFLEX */,
+ 0x00CF /* 0xCF LATIN CAPITAL LETTER I WITH DIAERESIS */,
+ CHAR96('C', 0xD0),
+ 0x00D1 /* 0xD1 LATIN CAPITAL LETTER N WITH TILDE */,
+ 0x00D2 /* 0xD2 LATIN CAPITAL LETTER O WITH GRAVE */,
+ 0x00D3 /* 0xD3 LATIN CAPITAL LETTER O WITH ACUTE */,
+ 0x00D4 /* 0xD4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX */,
+ 0x0120 /* 0xD5 LATIN CAPITAL LETTER G WITH DOT ABOVE */,
+ 0x00D6 /* 0xD6 LATIN CAPITAL LETTER O WITH DIAERESIS */,
+ 0x00D7 /* 0xD7 MULTIPLICATION SIGN */,
+ 0x011C /* 0xD8 LATIN CAPITAL LETTER G WITH CIRCUMFLEX */,
+ 0x00D9 /* 0xD9 LATIN CAPITAL LETTER U WITH GRAVE */,
+ 0x00DA /* 0xDA LATIN CAPITAL LETTER U WITH ACUTE */,
+ 0x00DB /* 0xDB LATIN CAPITAL LETTER U WITH CIRCUMFLEX */,
+ 0x00DC /* 0xDC LATIN CAPITAL LETTER U WITH DIAERESIS */,
+ 0x016C /* 0xDD LATIN CAPITAL LETTER U WITH BREVE */,
+ 0x015C /* 0xDE LATIN CAPITAL LETTER S WITH CIRCUMFLEX */,
+ 0x00DF /* 0xDF LATIN SMALL LETTER SHARP S */,
+ 0x00E0 /* 0xE0 LATIN SMALL LETTER A WITH GRAVE */,
+ 0x00E1 /* 0xE1 LATIN SMALL LETTER A WITH ACUTE */,
+ 0x00E2 /* 0xE2 LATIN SMALL LETTER A WITH CIRCUMFLEX */,
+ CHAR96('C', 0xE3),
+ 0x00E4 /* 0xE4 LATIN SMALL LETTER A WITH DIAERESIS */,
+ 0x010B /* 0xE5 LATIN SMALL LETTER C WITH DOT ABOVE */,
+ 0x0109 /* 0xE6 LATIN SMALL LETTER C WITH CIRCUMFLEX */,
+ 0x00E7 /* 0xE7 LATIN SMALL LETTER C WITH CEDILLA */,
+ 0x00E8 /* 0xE8 LATIN SMALL LETTER E WITH GRAVE */,
+ 0x00E9 /* 0xE9 LATIN SMALL LETTER E WITH ACUTE */,
+ 0x00EA /* 0xEA LATIN SMALL LETTER E WITH CIRCUMFLEX */,
+ 0x00EB /* 0xEB LATIN SMALL LETTER E WITH DIAERESIS */,
+ 0x00EC /* 0xEC LATIN SMALL LETTER I WITH GRAVE */,
+ 0x00ED /* 0xED LATIN SMALL LETTER I WITH ACUTE */,
+ 0x00EE /* 0xEE LATIN SMALL LETTER I WITH CIRCUMFLEX */,
+ 0x00EF /* 0xEF LATIN SMALL LETTER I WITH DIAERESIS */,
+ CHAR96('C', 0xF0),
+ 0x00F1 /* 0xF1 LATIN SMALL LETTER N WITH TILDE */,
+ 0x00F2 /* 0xF2 LATIN SMALL LETTER O WITH GRAVE */,
+ 0x00F3 /* 0xF3 LATIN SMALL LETTER O WITH ACUTE */,
+ 0x00F4 /* 0xF4 LATIN SMALL LETTER O WITH CIRCUMFLEX */,
+ 0x0121 /* 0xF5 LATIN SMALL LETTER G WITH DOT ABOVE */,
+ 0x00F6 /* 0xF6 LATIN SMALL LETTER O WITH DIAERESIS */,
+ 0x00F7 /* 0xF7 DIVISION SIGN */,
+ 0x011D /* 0xF8 LATIN SMALL LETTER G WITH CIRCUMFLEX */,
+ 0x00F9 /* 0xF9 LATIN SMALL LETTER U WITH GRAVE */,
+ 0x00FA /* 0xFA LATIN SMALL LETTER U WITH ACUTE */,
+ 0x00FB /* 0xFB LATIN SMALL LETTER U WITH CIRCUMFLEX */,
+ 0x00FC /* 0xFC LATIN SMALL LETTER U WITH DIAERESIS */,
+ 0x016D /* 0xFD LATIN SMALL LETTER U WITH BREVE */,
+ 0x015D /* 0xFE LATIN SMALL LETTER S WITH CIRCUMFLEX */,
+ 0x02D9 /* 0xFF DOT ABOVE */
+};
+
+Emchar latin_iso8859_4_to_ucs[96] =
+{
+ 0x00A0 /* 0xA0 NO-BREAK SPACE */,
+ 0x0104 /* 0xA1 LATIN CAPITAL LETTER A WITH OGONEK */,
+ 0x0138 /* 0xA2 LATIN SMALL LETTER KRA */,
+ 0x0156 /* 0xA3 LATIN CAPITAL LETTER R WITH CEDILLA */,
+ 0x00A4 /* 0xA4 CURRENCY SIGN */,
+ 0x0128 /* 0xA5 LATIN CAPITAL LETTER I WITH TILDE */,
+ 0x013B /* 0xA6 LATIN CAPITAL LETTER L WITH CEDILLA */,
+ 0x00A7 /* 0xA7 SECTION SIGN */,
+ 0x00A8 /* 0xA8 DIAERESIS */,
+ 0x0160 /* 0xA9 LATIN CAPITAL LETTER S WITH CARON */,
+ 0x0112 /* 0xAA LATIN CAPITAL LETTER E WITH MACRON */,
+ 0x0122 /* 0xAB LATIN CAPITAL LETTER G WITH CEDILLA */,
+ 0x0166 /* 0xAC LATIN CAPITAL LETTER T WITH STROKE */,
+ 0x00AD /* 0xAD SOFT HYPHEN */,
+ 0x017D /* 0xAE LATIN CAPITAL LETTER Z WITH CARON */,
+ 0x00AF /* 0xAF MACRON */,
+ 0x00B0 /* 0xB0 DEGREE SIGN */,
+ 0x0105 /* 0xB1 LATIN SMALL LETTER A WITH OGONEK */,
+ 0x02DB /* 0xB2 OGONEK */,
+ 0x0157 /* 0xB3 LATIN SMALL LETTER R WITH CEDILLA */,
+ 0x00B4 /* 0xB4 ACUTE ACCENT */,
+ 0x0129 /* 0xB5 LATIN SMALL LETTER I WITH TILDE */,
+ 0x013C /* 0xB6 LATIN SMALL LETTER L WITH CEDILLA */,
+ 0x02C7 /* 0xB7 CARON */,
+ 0x00B8 /* 0xB8 CEDILLA */,
+ 0x0161 /* 0xB9 LATIN SMALL LETTER S WITH CARON */,
+ 0x0113 /* 0xBA LATIN SMALL LETTER E WITH MACRON */,
+ 0x0123 /* 0xBB LATIN SMALL LETTER G WITH CEDILLA */,
+ 0x0167 /* 0xBC LATIN SMALL LETTER T WITH STROKE */,
+ 0x014A /* 0xBD LATIN CAPITAL LETTER ENG */,
+ 0x017E /* 0xBE LATIN SMALL LETTER Z WITH CARON */,
+ 0x014B /* 0xBF LATIN SMALL LETTER ENG */,
+ 0x0100 /* 0xC0 LATIN CAPITAL LETTER A WITH MACRON */,
+ 0x00C1 /* 0xC1 LATIN CAPITAL LETTER A WITH ACUTE */,
+ 0x00C2 /* 0xC2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX */,
+ 0x00C3 /* 0xC3 LATIN CAPITAL LETTER A WITH TILDE */,
+ 0x00C4 /* 0xC4 LATIN CAPITAL LETTER A WITH DIAERESIS */,
+ 0x00C5 /* 0xC5 LATIN CAPITAL LETTER A WITH RING ABOVE */,
+ 0x00C6 /* 0xC6 LATIN CAPITAL LETTER AE */,
+ 0x012E /* 0xC7 LATIN CAPITAL LETTER I WITH OGONEK */,
+ 0x010C /* 0xC8 LATIN CAPITAL LETTER C WITH CARON */,
+ 0x00C9 /* 0xC9 LATIN CAPITAL LETTER E WITH ACUTE */,
+ 0x0118 /* 0xCA LATIN CAPITAL LETTER E WITH OGONEK */,
+ 0x00CB /* 0xCB LATIN CAPITAL LETTER E WITH DIAERESIS */,
+ 0x0116 /* 0xCC LATIN CAPITAL LETTER E WITH DOT ABOVE */,
+ 0x00CD /* 0xCD LATIN CAPITAL LETTER I WITH ACUTE */,
+ 0x00CE /* 0xCE LATIN CAPITAL LETTER I WITH CIRCUMFLEX */,
+ 0x012A /* 0xCF LATIN CAPITAL LETTER I WITH MACRON */,
+ 0x0110 /* 0xD0 LATIN CAPITAL LETTER D WITH STROKE */,
+ 0x0145 /* 0xD1 LATIN CAPITAL LETTER N WITH CEDILLA */,
+ 0x014C /* 0xD2 LATIN CAPITAL LETTER O WITH MACRON */,
+ 0x0136 /* 0xD3 LATIN CAPITAL LETTER K WITH CEDILLA */,
+ 0x00D4 /* 0xD4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX */,
+ 0x00D5 /* 0xD5 LATIN CAPITAL LETTER O WITH TILDE */,
+ 0x00D6 /* 0xD6 LATIN CAPITAL LETTER O WITH DIAERESIS */,
+ 0x00D7 /* 0xD7 MULTIPLICATION SIGN */,
+ 0x00D8 /* 0xD8 LATIN CAPITAL LETTER O WITH STROKE */,
+ 0x0172 /* 0xD9 LATIN CAPITAL LETTER U WITH OGONEK */,
+ 0x00DA /* 0xDA LATIN CAPITAL LETTER U WITH ACUTE */,
+ 0x00DB /* 0xDB LATIN CAPITAL LETTER U WITH CIRCUMFLEX */,
+ 0x00DC /* 0xDC LATIN CAPITAL LETTER U WITH DIAERESIS */,
+ 0x0168 /* 0xDD LATIN CAPITAL LETTER U WITH TILDE */,
+ 0x016A /* 0xDE LATIN CAPITAL LETTER U WITH MACRON */,
+ 0x00DF /* 0xDF LATIN SMALL LETTER SHARP S */,
+ 0x0101 /* 0xE0 LATIN SMALL LETTER A WITH MACRON */,
+ 0x00E1 /* 0xE1 LATIN SMALL LETTER A WITH ACUTE */,
+ 0x00E2 /* 0xE2 LATIN SMALL LETTER A WITH CIRCUMFLEX */,
+ 0x00E3 /* 0xE3 LATIN SMALL LETTER A WITH TILDE */,
+ 0x00E4 /* 0xE4 LATIN SMALL LETTER A WITH DIAERESIS */,
+ 0x00E5 /* 0xE5 LATIN SMALL LETTER A WITH RING ABOVE */,
+ 0x00E6 /* 0xE6 LATIN SMALL LETTER AE */,
+ 0x012F /* 0xE7 LATIN SMALL LETTER I WITH OGONEK */,
+ 0x010D /* 0xE8 LATIN SMALL LETTER C WITH CARON */,
+ 0x00E9 /* 0xE9 LATIN SMALL LETTER E WITH ACUTE */,
+ 0x0119 /* 0xEA LATIN SMALL LETTER E WITH OGONEK */,
+ 0x00EB /* 0xEB LATIN SMALL LETTER E WITH DIAERESIS */,
+ 0x0117 /* 0xEC LATIN SMALL LETTER E WITH DOT ABOVE */,
+ 0x00ED /* 0xED LATIN SMALL LETTER I WITH ACUTE */,
+ 0x00EE /* 0xEE LATIN SMALL LETTER I WITH CIRCUMFLEX */,
+ 0x012B /* 0xEF LATIN SMALL LETTER I WITH MACRON */,
+ 0x0111 /* 0xF0 LATIN SMALL LETTER D WITH STROKE */,
+ 0x0146 /* 0xF1 LATIN SMALL LETTER N WITH CEDILLA */,
+ 0x014D /* 0xF2 LATIN SMALL LETTER O WITH MACRON */,
+ 0x0137 /* 0xF3 LATIN SMALL LETTER K WITH CEDILLA */,
+ 0x00F4 /* 0xF4 LATIN SMALL LETTER O WITH CIRCUMFLEX */,
+ 0x00F5 /* 0xF5 LATIN SMALL LETTER O WITH TILDE */,
+ 0x00F6 /* 0xF6 LATIN SMALL LETTER O WITH DIAERESIS */,
+ 0x00F7 /* 0xF7 DIVISION SIGN */,
+ 0x00F8 /* 0xF8 LATIN SMALL LETTER O WITH STROKE */,
+ 0x0173 /* 0xF9 LATIN SMALL LETTER U WITH OGONEK */,
+ 0x00FA /* 0xFA LATIN SMALL LETTER U WITH ACUTE */,
+ 0x00FB /* 0xFB LATIN SMALL LETTER U WITH CIRCUMFLEX */,
+ 0x00FC /* 0xFC LATIN SMALL LETTER U WITH DIAERESIS */,
+ 0x0169 /* 0xFD LATIN SMALL LETTER U WITH TILDE */,
+ 0x016B /* 0xFE LATIN SMALL LETTER U WITH MACRON */,
+ 0x02D9 /* 0xFF DOT ABOVE */
+};
+
+Emchar latin_iso8859_9_to_ucs[96] =
+{
+ 0x00A0 /* 0xA0 NO-BREAK SPACE */,
+ 0x00A1 /* 0xA1 INVERTED EXCLAMATION MARK */,
+ 0x00A2 /* 0xA2 CENT SIGN */,
+ 0x00A3 /* 0xA3 POUND SIGN */,
+ 0x00A4 /* 0xA4 CURRENCY SIGN */,
+ 0x00A5 /* 0xA5 YEN SIGN */,
+ 0x00A6 /* 0xA6 BROKEN BAR */,
+ 0x00A7 /* 0xA7 SECTION SIGN */,
+ 0x00A8 /* 0xA8 DIAERESIS */,
+ 0x00A9 /* 0xA9 COPYRIGHT SIGN */,
+ 0x00AA /* 0xAA FEMININE ORDINAL INDICATOR */,
+ 0x00AB /* 0xAB LEFT-POINTING DOUBLE ANGLE QUOTATION MARK */,
+ 0x00AC /* 0xAC NOT SIGN */,
+ 0x00AD /* 0xAD SOFT HYPHEN */,
+ 0x00AE /* 0xAE REGISTERED SIGN */,
+ 0x00AF /* 0xAF MACRON */,
+ 0x00B0 /* 0xB0 DEGREE SIGN */,
+ 0x00B1 /* 0xB1 PLUS-MINUS SIGN */,
+ 0x00B2 /* 0xB2 SUPERSCRIPT TWO */,
+ 0x00B3 /* 0xB3 SUPERSCRIPT THREE */,
+ 0x00B4 /* 0xB4 ACUTE ACCENT */,
+ 0x00B5 /* 0xB5 MICRO SIGN */,
+ 0x00B6 /* 0xB6 PILCROW SIGN */,
+ 0x00B7 /* 0xB7 MIDDLE DOT */,
+ 0x00B8 /* 0xB8 CEDILLA */,
+ 0x00B9 /* 0xB9 SUPERSCRIPT ONE */,
+ 0x00BA /* 0xBA MASCULINE ORDINAL INDICATOR */,
+ 0x00BB /* 0xBB RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK */,
+ 0x00BC /* 0xBC VULGAR FRACTION ONE QUARTER */,
+ 0x00BD /* 0xBD VULGAR FRACTION ONE HALF */,
+ 0x00BE /* 0xBE VULGAR FRACTION THREE QUARTERS */,
+ 0x00BF /* 0xBF INVERTED QUESTION MARK */,
+ 0x00C0 /* 0xC0 LATIN CAPITAL LETTER A WITH GRAVE */,
+ 0x00C1 /* 0xC1 LATIN CAPITAL LETTER A WITH ACUTE */,
+ 0x00C2 /* 0xC2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX */,
+ 0x00C3 /* 0xC3 LATIN CAPITAL LETTER A WITH TILDE */,
+ 0x00C4 /* 0xC4 LATIN CAPITAL LETTER A WITH DIAERESIS */,
+ 0x00C5 /* 0xC5 LATIN CAPITAL LETTER A WITH RING ABOVE */,
+ 0x00C6 /* 0xC6 LATIN CAPITAL LETTER AE */,
+ 0x00C7 /* 0xC7 LATIN CAPITAL LETTER C WITH CEDILLA */,
+ 0x00C8 /* 0xC8 LATIN CAPITAL LETTER E WITH GRAVE */,
+ 0x00C9 /* 0xC9 LATIN CAPITAL LETTER E WITH ACUTE */,
+ 0x00CA /* 0xCA LATIN CAPITAL LETTER E WITH CIRCUMFLEX */,
+ 0x00CB /* 0xCB LATIN CAPITAL LETTER E WITH DIAERESIS */,
+ 0x00CC /* 0xCC LATIN CAPITAL LETTER I WITH GRAVE */,
+ 0x00CD /* 0xCD LATIN CAPITAL LETTER I WITH ACUTE */,
+ 0x00CE /* 0xCE LATIN CAPITAL LETTER I WITH CIRCUMFLEX */,
+ 0x00CF /* 0xCF LATIN CAPITAL LETTER I WITH DIAERESIS */,
+ 0x011E /* 0xD0 LATIN CAPITAL LETTER G WITH BREVE */,
+ 0x00D1 /* 0xD1 LATIN CAPITAL LETTER N WITH TILDE */,
+ 0x00D2 /* 0xD2 LATIN CAPITAL LETTER O WITH GRAVE */,
+ 0x00D3 /* 0xD3 LATIN CAPITAL LETTER O WITH ACUTE */,
+ 0x00D4 /* 0xD4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX */,
+ 0x00D5 /* 0xD5 LATIN CAPITAL LETTER O WITH TILDE */,
+ 0x00D6 /* 0xD6 LATIN CAPITAL LETTER O WITH DIAERESIS */,
+ 0x00D7 /* 0xD7 MULTIPLICATION SIGN */,
+ 0x00D8 /* 0xD8 LATIN CAPITAL LETTER O WITH STROKE */,
+ 0x00D9 /* 0xD9 LATIN CAPITAL LETTER U WITH GRAVE */,
+ 0x00DA /* 0xDA LATIN CAPITAL LETTER U WITH ACUTE */,
+ 0x00DB /* 0xDB LATIN CAPITAL LETTER U WITH CIRCUMFLEX */,
+ 0x00DC /* 0xDC LATIN CAPITAL LETTER U WITH DIAERESIS */,
+ 0x0130 /* 0xDD LATIN CAPITAL LETTER I WITH DOT ABOVE */,
+ 0x015E /* 0xDE LATIN CAPITAL LETTER S WITH CEDILLA */,
+ 0x00DF /* 0xDF LATIN SMALL LETTER SHARP S */,
+ 0x00E0 /* 0xE0 LATIN SMALL LETTER A WITH GRAVE */,
+ 0x00E1 /* 0xE1 LATIN SMALL LETTER A WITH ACUTE */,
+ 0x00E2 /* 0xE2 LATIN SMALL LETTER A WITH CIRCUMFLEX */,
+ 0x00E3 /* 0xE3 LATIN SMALL LETTER A WITH TILDE */,
+ 0x00E4 /* 0xE4 LATIN SMALL LETTER A WITH DIAERESIS */,
+ 0x00E5 /* 0xE5 LATIN SMALL LETTER A WITH RING ABOVE */,
+ 0x00E6 /* 0xE6 LATIN SMALL LETTER AE */,
+ 0x00E7 /* 0xE7 LATIN SMALL LETTER C WITH CEDILLA */,
+ 0x00E8 /* 0xE8 LATIN SMALL LETTER E WITH GRAVE */,
+ 0x00E9 /* 0xE9 LATIN SMALL LETTER E WITH ACUTE */,
+ 0x00EA /* 0xEA LATIN SMALL LETTER E WITH CIRCUMFLEX */,
+ 0x00EB /* 0xEB LATIN SMALL LETTER E WITH DIAERESIS */,
+ 0x00EC /* 0xEC LATIN SMALL LETTER I WITH GRAVE */,
+ 0x00ED /* 0xED LATIN SMALL LETTER I WITH ACUTE */,
+ 0x00EE /* 0xEE LATIN SMALL LETTER I WITH CIRCUMFLEX */,
+ 0x00EF /* 0xEF LATIN SMALL LETTER I WITH DIAERESIS */,
+ 0x011F /* 0xF0 LATIN SMALL LETTER G WITH BREVE */,
+ 0x00F1 /* 0xF1 LATIN SMALL LETTER N WITH TILDE */,
+ 0x00F2 /* 0xF2 LATIN SMALL LETTER O WITH GRAVE */,
+ 0x00F3 /* 0xF3 LATIN SMALL LETTER O WITH ACUTE */,
+ 0x00F4 /* 0xF4 LATIN SMALL LETTER O WITH CIRCUMFLEX */,
+ 0x00F5 /* 0xF5 LATIN SMALL LETTER O WITH TILDE */,
+ 0x00F6 /* 0xF6 LATIN SMALL LETTER O WITH DIAERESIS */,
+ 0x00F7 /* 0xF7 DIVISION SIGN */,
+ 0x00F8 /* 0xF8 LATIN SMALL LETTER O WITH STROKE */,
+ 0x00F9 /* 0xF9 LATIN SMALL LETTER U WITH GRAVE */,
+ 0x00FA /* 0xFA LATIN SMALL LETTER U WITH ACUTE */,
+ 0x00FB /* 0xFB LATIN SMALL LETTER U WITH CIRCUMFLEX */,
+ 0x00FC /* 0xFC LATIN SMALL LETTER U WITH DIAERESIS */,
+ 0x0131 /* 0xFD LATIN SMALL LETTER DOTLESS I */,
+ 0x015F /* 0xFE LATIN SMALL LETTER S WITH CEDILLA */,
+ 0x00FF /* 0xFF LATIN SMALL LETTER Y WITH DIAERESIS */,
+};
+
+Emchar latin_viscii_lower_to_ucs[96] =
+{
+ CHAR96('1', 0x20),
+ 0x1eaf /* 0x21 */,
+ 0x1eb1 /* 0x22 */,
+ 0x1eb7 /* 0x23 */,
+ 0x1ea5 /* 0x24 */,
+ 0x1ea7 /* 0x25 */,
+ 0x1ea9 /* 0x26 */,
+ 0x1ead /* 0x27 */,
+ 0x1ebd /* 0x28 */,
+ 0x1eb9 /* 0x29 */,
+ 0x1ebf /* 0x2a */,
+ 0x1ec1 /* 0x2b */,
+ 0x1ec3 /* 0x2c */,
+ 0x1ec5 /* 0x2d */,
+ 0x1ec7 /* 0x2e */,
+ 0x1ed1 /* 0x2f */,
+ 0x1ed3 /* 0x30 */,
+ 0x1ed5 /* 0x31 */,
+ 0x1ed7 /* 0x32 */,
+ CHAR96('1', 0x33),
+ CHAR96('1', 0x34),
+ 0x1ed9 /* 0x35 */,
+ 0x1edd /* 0x36 */,
+ 0x1edf /* 0x37 */,
+ 0x1ecb /* 0x38 */,
+ CHAR96('1', 0x39),
+ CHAR96('1', 0x3A),
+ CHAR96('1', 0x3B),
+ CHAR96('1', 0x3C),
+ 0x01a1 /* 0x3d */,
+ 0x1edb /* 0x3e */,
+ CHAR96('1', 0x3F),
+ CHAR96('1', 0x40),
+ CHAR96('1', 0x41),
+ CHAR96('1', 0x42),
+ CHAR96('1', 0x43),
+ CHAR96('1', 0x44),
+ CHAR96('1', 0x45),
+ 0x1eb3 /* 0x46 */,
+ 0x1eb5 /* 0x47 */,
+ CHAR96('1', 0x48),
+ CHAR96('1', 0x49),
+ CHAR96('1', 0x4A),
+ CHAR96('1', 0x4B),
+ CHAR96('1', 0x4C),
+ CHAR96('1', 0x4D),
+ CHAR96('1', 0x4E),
+ 0x1ef3 /* 0x4f */,
+ CHAR96('1', 0x50),
+ 0x1ee9 /* 0x51 */,
+ CHAR96('1', 0x52),
+ CHAR96('1', 0x53),
+ CHAR96('1', 0x54),
+ 0x1ea1 /* 0x55 */,
+ 0x1ef7 /* 0x56 */,
+ 0x1eeb /* 0x57 */,
+ 0x1eed /* 0x58 */,
+ CHAR96('1', 0x59),
+ CHAR96('1', 0x5A),
+ 0x1ef9 /* 0x5b */,
+ 0x1ef5 /* 0x5c */,
+ CHAR96('1', 0x5D),
+ 0x1ee1 /* 0x5e */,
+ 0x01b0 /* 0x5f */,
+ 0x00e0 /* 0x60 */,
+ 0x00e1 /* 0x61 */,
+ 0x00e2 /* 0x62 */,
+ 0x00e3 /* 0x63 */,
+ 0x1ea3 /* 0x64 */,
+ 0x0103 /* 0x65 */,
+ 0x1eef /* 0x66 */,
+ 0x1eab /* 0x67 */,
+ 0x00e8 /* 0x68 */,
+ 0x00e9 /* 0x69 */,
+ 0x00ea /* 0x6a */,
+ 0x1ebb /* 0x6b */,
+ 0x00ec /* 0x6c */,
+ 0x00ed /* 0x6d */,
+ 0x0129 /* 0x6e */,
+ 0x1ec9 /* 0x6f */,
+ 0x0111 /* 0x70 */,
+ 0x1ef1 /* 0x71 */,
+ 0x00f2 /* 0x72 */,
+ 0x00f3 /* 0x73 */,
+ 0x00f4 /* 0x74 */,
+ 0x00f5 /* 0x75 */,
+ 0x1ecf /* 0x76 */,
+ 0x1ecd /* 0x77 */,
+ 0x1ee5 /* 0x78 */,
+ 0x00f9 /* 0x79 */,
+ 0x00fa /* 0x7a */,
+ 0x0169 /* 0x7b */,
+ 0x1ee7 /* 0x7c */,
+ 0x00fd /* 0x7d */,
+ 0x1ee3 /* 0x7e */,
+ CHAR96('1', 0x7F)
+};
+
+Emchar latin_viscii_upper_to_ucs[96] =
+{
+ CHAR96('2', 0x20),
+ 0x1eae /* 0x21 */,
+ 0x1eb0 /* 0x22 */,
+ 0x1eb6 /* 0x23 */,
+ 0x1ea4 /* 0x24 */,
+ 0x1ea6 /* 0x25 */,
+ 0x1ea8 /* 0x26 */,
+ 0x1eac /* 0x27 */,
+ 0x1ebc /* 0x28 */,
+ 0x1eb8 /* 0x29 */,
+ 0x1ebe /* 0x2a */,
+ 0x1ec0 /* 0x2b */,
+ 0x1ec2 /* 0x2c */,
+ 0x1ec4 /* 0x2d */,
+ 0x1ec6 /* 0x2e */,
+ 0x1ed0 /* 0x2f */,
+ 0x1ed2 /* 0x30 */,
+ 0x1ed4 /* 0x31 */,
+ 0x1ed6 /* 0x32 */,
+ CHAR96('2', 0x33),
+ CHAR96('2', 0x34),
+ 0x1ed8 /* 0x35 */,
+ 0x1edc /* 0x36 */,
+ 0x1ede /* 0x37 */,
+ 0x1eca /* 0x38 */,
+ CHAR96('2', 0x39),
+ CHAR96('2', 0x3a),
+ CHAR96('2', 0x3b),
+ CHAR96('2', 0x3c),
+ 0x01a0 /* 0x3d */,
+ 0x1eda /* 0x3e */,
+ CHAR96('2', 0x3f),
+ CHAR96('2', 0x40),
+ CHAR96('2', 0x41),
+ CHAR96('2', 0x42),
+ CHAR96('2', 0x43),
+ CHAR96('2', 0x44),
+ CHAR96('2', 0x45),
+ 0x1eb2 /* 0x46 */,
+ 0x1eb4 /* 0x47 */,
+ CHAR96('2', 0x48),
+ CHAR96('2', 0x49),
+ CHAR96('2', 0x4a),
+ CHAR96('2', 0x4b),
+ CHAR96('2', 0x4c),
+ CHAR96('2', 0x4d),
+ CHAR96('2', 0x4e),
+ 0x1ef2 /* 0x4f */,
+ CHAR96('2', 0x50),
+ 0x1ee8 /* 0x51 */,
+ CHAR96('2', 0x52),
+ CHAR96('2', 0x53),
+ CHAR96('2', 0x54),
+ 0x1ea0 /* 0x55 */,
+ 0x1ef6 /* 0x56 */,
+ 0x1eea /* 0x57 */,
+ 0x1eec /* 0x58 */,
+ CHAR96('2', 0x59),
+ CHAR96('2', 0x5a),
+ 0x1ef8 /* 0x5b */,
+ 0x1ef4 /* 0x5c */,
+ CHAR96('2', 0x5d),
+ 0x1ee0 /* 0x5e */,
+ 0x01af /* 0x5f */,
+ 0x00c0 /* 0x60 */,
+ 0x00c1 /* 0x61 */,
+ 0x00c2 /* 0x62 */,
+ 0x00c3 /* 0x63 */,
+ 0x1ea2 /* 0x64 */,
+ 0x0102 /* 0x65 */,
+ 0x1eee /* 0x66 */,
+ 0x1eaa /* 0x67 */,
+ 0x00c8 /* 0x68 */,
+ 0x00c9 /* 0x69 */,
+ 0x00ca /* 0x6a */,
+ 0x1eba /* 0x6b */,
+ 0x00cc /* 0x6c */,
+ 0x00cd /* 0x6d */,
+ 0x0128 /* 0x6e */,
+ 0x1ec8 /* 0x6f */,
+ 0x0110 /* 0x70 */,
+ 0x1ef0 /* 0x71 */,
+ 0x00d2 /* 0x72 */,
+ 0x00d3 /* 0x73 */,
+ 0x00d4 /* 0x74 */,
+ 0x00d5 /* 0x75 */,
+ 0x1ece /* 0x76 */,
+ 0x1ecc /* 0x77 */,
+ 0x1ee4 /* 0x78 */,
+ 0x00d9 /* 0x79 */,
+ 0x00da /* 0x7a */,
+ 0x0168 /* 0x7b */,
+ 0x1ee6 /* 0x7c */,
+ 0x00dd /* 0x7d */,
+ 0x1ee2 /* 0x7e */,
+ CHAR96('2', 0x7f)
+};
+
+Emchar latin_tcvn5712_to_ucs[96] =
+{
+ 0x00A0 /* 0xA0 NO-BREAK SPACE */,
+ 0x0102 /* 0xA1 LATIN CAPITAL LETTER A WITH BREVE */,
+ 0x00C2 /* 0xA2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX */,
+ 0x00CA /* 0xA3 LATIN CAPITAL LETTER E WITH CIRCUMFLEX */,
+ 0x00D4 /* 0xA4 LATIN CAPITAL LETTER O WITH CIRCUMFLEX */,
+ 0x01A0 /* 0xA5 LATIN CAPITAL LETTER O WITH HORN */,
+ 0x01AF /* 0xA6 LATIN CAPITAL LETTER U WITH HORN */,
+ 0x0110 /* 0xA7 LATIN CAPITAL LETTER D WITH STROKE */,
+ 0x0103 /* 0xA8 LATIN SMALL LETTER A WITH BREVE */,
+ 0x00E2 /* 0xA9 LATIN SMALL LETTER A WITH CIRCUMFLEX */,
+ 0x00EA /* 0xAA LATIN SMALL LETTER E WITH CIRCUMFLEX */,
+ 0x00F4 /* 0xAB LATIN SMALL LETTER O WITH CIRCUMFLEX */,
+ 0x01A1 /* 0xAC LATIN SMALL LETTER O WITH HORN */,
+ 0x01B0 /* 0xAD LATIN SMALL LETTER U WITH HORN */,
+ 0x0111 /* 0xAE LATIN SMALL LETTER D WITH STROKE */,
+ 0x1EB0 /* 0xAF LATIN CAPITAL LETTER A WITH BREVE AND GRAVE */,
+ 0x0300 /* 0xB0 COMBINING GRAVE ACCENT */,
+ 0x0309 /* 0xB1 COMBINING HOOK ABOVE */,
+ 0x0303 /* 0xB2 COMBINING TILDE */,
+ 0x0301 /* 0xB3 COMBINING ACUTE ACCENT */,
+ 0x0323 /* 0xB4 COMBINING DOT BELOW */,
+ 0x00E0 /* 0xB5 LATIN SMALL LETTER A WITH GRAVE */,
+ 0x1EA3 /* 0xB6 LATIN SMALL LETTER A WITH HOOK ABOVE */,
+ 0x00E3 /* 0xB7 LATIN SMALL LETTER A WITH TILDE */,
+ 0x00E1 /* 0xB8 LATIN SMALL LETTER A WITH ACUTE */,
+ 0x1EA1 /* 0xB9 LATIN SMALL LETTER A WITH DOT BELOW */,
+ 0x1EB2 /* 0xBA LATIN CAPITAL LETTER A WITH BREVE AND HOOK ABOVE */,
+ 0x1EB1 /* 0xBB LATIN SMALL LETTER A WITH BREVE AND GRAVE */,
+ 0x1EB3 /* 0xBC LATIN SMALL LETTER A WITH BREVE AND HOOK ABOVE */,
+ 0x1EB5 /* 0xBD LATIN SMALL LETTER A WITH BREVE AND TILDE */,
+ 0x1EAF /* 0xBE LATIN SMALL LETTER A WITH BREVE AND ACUTE */,
+ 0x1EB4 /* 0xBF LATIN CAPITAL LETTER A WITH BREVE AND TILDE */,
+ 0x1EAE /* 0xC0 LATIN CAPITAL LETTER A WITH BREVE AND ACUTE */,
+ 0x1EA6 /* 0xC1 LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND GRAVE */,
+ 0x1EA8 /* 0xC2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1EAA /* 0xC3 LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND TILDE */,
+ 0x1EA4 /* 0xC4 LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND ACUTE */,
+ 0x1EC0 /* 0xC5 LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND GRAVE */,
+ 0x1EB7 /* 0xC6 LATIN SMALL LETTER A WITH BREVE AND DOT BELOW */,
+ 0x1EA7 /* 0xC7 LATIN SMALL LETTER A WITH CIRCUMFLEX AND GRAVE */,
+ 0x1EA9 /* 0xC8 LATIN SMALL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1EAB /* 0xC9 LATIN SMALL LETTER A WITH CIRCUMFLEX AND TILDE */,
+ 0x1EA5 /* 0xCA LATIN SMALL LETTER A WITH CIRCUMFLEX AND ACUTE */,
+ 0x1EAD /* 0xCB LATIN SMALL LETTER A WITH CIRCUMFLEX AND DOT BELOW */,
+ 0x00E8 /* 0xCC LATIN SMALL LETTER E WITH GRAVE */,
+ 0x1EC2 /* 0xCD LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1EBB /* 0xCE LATIN SMALL LETTER E WITH HOOK ABOVE */,
+ 0x1EBD /* 0xCF LATIN SMALL LETTER E WITH TILDE */,
+ 0x00E9 /* 0xD0 LATIN SMALL LETTER E WITH ACUTE */,
+ 0x1EB9 /* 0xD1 LATIN SMALL LETTER E WITH DOT BELOW */,
+ 0x1EC1 /* 0xD2 LATIN SMALL LETTER E WITH CIRCUMFLEX AND GRAVE */,
+ 0x1EC3 /* 0xD3 LATIN SMALL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1EC5 /* 0xD4 LATIN SMALL LETTER E WITH CIRCUMFLEX AND TILDE */,
+ 0x1EBF /* 0xD5 LATIN SMALL LETTER E WITH CIRCUMFLEX AND ACUTE */,
+ 0x1EC7 /* 0xD6 LATIN SMALL LETTER E WITH CIRCUMFLEX AND DOT BELOW */,
+ 0x00EC /* 0xD7 LATIN SMALL LETTER I WITH GRAVE */,
+ 0x1EC9 /* 0xD8 LATIN SMALL LETTER I WITH HOOK ABOVE */,
+ 0x1EC4 /* 0xD9 LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND TILDE */,
+ 0x1EBE /* 0xDA LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND ACUTE */,
+ 0x1ED2 /* 0xDB LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND GRAVE */,
+ 0x0129 /* 0xDC LATIN SMALL LETTER I WITH TILDE */,
+ 0x00ED /* 0xDD LATIN SMALL LETTER I WITH ACUTE */,
+ 0x1ECB /* 0xDE LATIN SMALL LETTER I WITH DOT BELOW */,
+ 0x00F2 /* 0xDF LATIN SMALL LETTER O WITH GRAVE */,
+ 0x1ED4 /* 0xE0 LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1ECF /* 0xE1 LATIN SMALL LETTER O WITH HOOK ABOVE */,
+ 0x00F5 /* 0xE2 LATIN SMALL LETTER O WITH TILDE */,
+ 0x00F3 /* 0xE3 LATIN SMALL LETTER O WITH ACUTE */,
+ 0x1ECD /* 0xE4 LATIN SMALL LETTER O WITH DOT BELOW */,
+ 0x1ED3 /* 0xE5 LATIN SMALL LETTER O WITH CIRCUMFLEX AND GRAVE */,
+ 0x1ED5 /* 0xE6 LATIN SMALL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE */,
+ 0x1ED7 /* 0xE7 LATIN SMALL LETTER O WITH CIRCUMFLEX AND TILDE */,
+ 0x1ED1 /* 0xE8 LATIN SMALL LETTER O WITH CIRCUMFLEX AND ACUTE */,
+ 0x1ED9 /* 0xE9 LATIN SMALL LETTER O WITH CIRCUMFLEX AND DOT BELOW */,
+ 0x1EDD /* 0xEA LATIN SMALL LETTER O WITH HORN AND GRAVE */,
+ 0x1EDF /* 0xEB LATIN SMALL LETTER O WITH HORN AND HOOK ABOVE */,
+ 0x1EE1 /* 0xEC LATIN SMALL LETTER O WITH HORN AND TILDE */,
+ 0x1EDB /* 0xED LATIN SMALL LETTER O WITH HORN AND ACUTE */,
+ 0x1EE3 /* 0xEE LATIN SMALL LETTER O WITH HORN AND DOT BELOW */,
+ 0x00F9 /* 0xEF LATIN SMALL LETTER U WITH GRAVE */,
+ 0x1ED6 /* 0xF0 LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND TILDE */,
+ 0x1EE7 /* 0xF1 LATIN SMALL LETTER U WITH HOOK ABOVE */,
+ 0x0169 /* 0xF2 LATIN SMALL LETTER U WITH TILDE */,
+ 0x00FA /* 0xF3 LATIN SMALL LETTER U WITH ACUTE */,
+ 0x1EE5 /* 0xF4 LATIN SMALL LETTER U WITH DOT BELOW */,
+ 0x1EEB /* 0xF5 LATIN SMALL LETTER U WITH HORN AND GRAVE */,
+ 0x1EED /* 0xF6 LATIN SMALL LETTER U WITH HORN AND HOOK ABOVE */,
+ 0x1EEF /* 0xF7 LATIN SMALL LETTER U WITH HORN AND TILDE */,
+ 0x1EE9 /* 0xF8 LATIN SMALL LETTER U WITH HORN AND ACUTE */,
+ 0x1EF1 /* 0xF9 LATIN SMALL LETTER U WITH HORN AND DOT BELOW */,
+ 0x1EF3 /* 0xFA LATIN SMALL LETTER Y WITH GRAVE */,
+ 0x1EF7 /* 0xFB LATIN SMALL LETTER Y WITH HOOK ABOVE */,
+ 0x1EF9 /* 0xFC LATIN SMALL LETTER Y WITH TILDE */,
+ 0x00FD /* 0xFD LATIN SMALL LETTER Y WITH ACUTE */,
+ 0x1EF5 /* 0xFE LATIN SMALL LETTER Y WITH DOT BELOW */,
+ 0x1ED0 /* 0xFF LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND ACUTE */
+};
+
Lisp_Object Vutf_2000_version;
+#endif
+#ifndef UTF2000
int leading_code_private_11;
+#endif
Lisp_Object Qcharsetp;
Qchinese_cns11643_5,
Qchinese_cns11643_6,
Qchinese_cns11643_7,
+ Qucs_bmp,
+ Qlatin_viscii_lower,
+ Qlatin_viscii_upper,
#endif
Qchinese_big5_1,
Qchinese_big5_2,
Lisp_Object Vcharset_hash_table;
-static Bufbyte next_allocated_1_byte_leading_byte;
-static Bufbyte next_allocated_2_byte_leading_byte;
+static Charset_ID next_allocated_1_byte_leading_byte;
+static Charset_ID next_allocated_2_byte_leading_byte;
/* Composite characters are characters constructed by overstriking two
or more regular characters.
{
Bufbyte *p;
#ifndef UTF2000
- Bufbyte lb;
+ Charset_ID lb;
int c1, c2;
Lisp_Object charset;
#endif
};
DEFINE_LRECORD_IMPLEMENTATION ("charset", charset,
- mark_charset, print_charset, 0, 0, 0, charset_description,
+ mark_charset, print_charset, 0, 0, 0,
+ charset_description,
struct Lisp_Charset);
/* Make a new charset. */
static Lisp_Object
-make_charset (int id, Lisp_Object name, unsigned char rep_bytes,
+make_charset (Charset_ID id, Lisp_Object name,
unsigned char type, unsigned char columns, unsigned char graphic,
- Bufbyte final, unsigned char direction, Lisp_Object short_name,
+ Bufbyte final, unsigned char direction, Lisp_Object short_name,
Lisp_Object long_name, Lisp_Object doc,
- Lisp_Object reg)
+ Lisp_Object reg,
+ Emchar* decoding_table,
+ Emchar ucs_min, Emchar ucs_max, Emchar code_offset)
{
Lisp_Object obj;
struct Lisp_Charset *cs =
CHARSET_NAME (cs) = name;
CHARSET_SHORT_NAME (cs) = short_name;
CHARSET_LONG_NAME (cs) = long_name;
- CHARSET_REP_BYTES (cs) = rep_bytes;
CHARSET_DIRECTION (cs) = direction;
CHARSET_TYPE (cs) = type;
CHARSET_COLUMNS (cs) = columns;
CHARSET_REGISTRY (cs) = reg;
CHARSET_CCL_PROGRAM (cs) = Qnil;
CHARSET_REVERSE_DIRECTION_CHARSET (cs) = Qnil;
+#ifdef UTF2000
+ CHARSET_DECODING_TABLE(cs) = decoding_table;
+ CHARSET_UCS_MIN(cs) = ucs_min;
+ CHARSET_UCS_MAX(cs) = ucs_max;
+ CHARSET_CODE_OFFSET(cs) = code_offset;
+#endif
+
+ switch ( CHARSET_TYPE (cs) )
+ {
+ case CHARSET_TYPE_94:
+ CHARSET_DIMENSION (cs) = 1;
+ CHARSET_CHARS (cs) = 94;
+#ifdef UTF2000
+ if (decoding_table != NULL)
+ {
+ size_t i;
+ CHARSET_TO_BYTE1_TABLE(cs) = make_byte_from_character_table();
+ for (i = 0; i < 94; i++)
+ {
+ Emchar c = decoding_table[i];
+
+ if (c <= 0xffff)
+ put_byte_from_character_table (c, i + 33,
+ CHARSET_TO_BYTE1_TABLE(cs));
+ }
+ }
+ else
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+ case CHARSET_TYPE_96:
+ CHARSET_DIMENSION (cs) = 1;
+ CHARSET_CHARS (cs) = 96;
+#ifdef UTF2000
+ if (decoding_table != NULL)
+ {
+ size_t i;
+ CHARSET_TO_BYTE1_TABLE(cs) = make_byte_from_character_table();
+ for (i = 0; i < 96; i++)
+ {
+ Emchar c = decoding_table[i];
+
+ if (c <= 0xffff)
+ put_byte_from_character_table (c, i + 32,
+ CHARSET_TO_BYTE1_TABLE(cs));
+ }
+ }
+ else
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+ case CHARSET_TYPE_94X94:
+ CHARSET_DIMENSION (cs) = 2;
+ CHARSET_CHARS (cs) = 94;
+#ifdef UTF2000
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+ case CHARSET_TYPE_96X96:
+ CHARSET_DIMENSION (cs) = 2;
+ CHARSET_CHARS (cs) = 96;
+#ifdef UTF2000
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+#ifdef UTF2000
+ case CHARSET_TYPE_128X128:
+ CHARSET_DIMENSION (cs) = 2;
+ CHARSET_CHARS (cs) = 128;
+#ifdef UTF2000
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+ case CHARSET_TYPE_256X256:
+ CHARSET_DIMENSION (cs) = 2;
+ CHARSET_CHARS (cs) = 256;
+#ifdef UTF2000
+ CHARSET_TO_BYTE1_TABLE(cs) = NULL;
+ CHARSET_TO_BYTE2_TABLE(cs) = NULL;
+#endif
+ break;
+#endif
+ }
- CHARSET_DIMENSION (cs) = (CHARSET_TYPE (cs) == CHARSET_TYPE_94 ||
- CHARSET_TYPE (cs) == CHARSET_TYPE_96) ? 1 : 2;
- CHARSET_CHARS (cs) = (CHARSET_TYPE (cs) == CHARSET_TYPE_94 ||
- CHARSET_TYPE (cs) == CHARSET_TYPE_94X94) ? 94 : 96;
-
+#ifndef UTF2000
+ if (id == LEADING_BYTE_ASCII)
+ CHARSET_REP_BYTES (cs) = 1;
+ else if (id < 0xA0)
+ CHARSET_REP_BYTES (cs) = CHARSET_DIMENSION (cs) + 1;
+ else
+ CHARSET_REP_BYTES (cs) = CHARSET_DIMENSION (cs) + 2;
+#endif
+
if (final)
{
/* some charsets do not have final characters. This includes
ASCII, Control-1, Composite, and the two faux private
charsets. */
+#if UTF2000
+ assert (NILP (charset_by_attributes[type][final]));
+ charset_by_attributes[type][final] = obj;
+#else
assert (NILP (charset_by_attributes[type][final][direction]));
charset_by_attributes[type][final][direction] = obj;
+#endif
}
assert (NILP (charset_by_leading_byte[id - MIN_LEADING_BYTE]));
#ifndef UTF2000
if (id < 0xA0)
/* official leading byte */
- rep_bytes_by_first_byte[id] = rep_bytes;
+ rep_bytes_by_first_byte[id] = CHARSET_REP_BYTES (cs);
#endif
/* Some charsets are "faux" and don't have names or really exist at
static int
get_unallocated_leading_byte (int dimension)
{
- int lb;
+ Charset_ID lb;
if (dimension == 1)
{
return lb;
}
+#ifdef UTF2000
+unsigned char
+charset_get_byte1 (Lisp_Object charset, Emchar ch)
+{
+ Emchar_to_byte_table* table;
+ int d;
+
+ if ((table = XCHARSET_TO_BYTE1_TABLE (charset)) != NULL)
+ return get_byte_from_character_table (ch, table);
+ else if ((CHARSET_UCS_MIN (XCHARSET (charset)) <= ch)
+ && (ch <= CHARSET_UCS_MAX (XCHARSET (charset))))
+ return ch - CHARSET_UCS_MIN (XCHARSET (charset))
+ + CHARSET_CODE_OFFSET (XCHARSET (charset));
+ else if (XCHARSET_DIMENSION (charset) == 1)
+ {
+ if (XCHARSET_CHARS (charset) == 94)
+ {
+ if (((d = ch - (MIN_CHAR_94
+ + (XCHARSET_FINAL (charset) - '0') * 94)) >= 0)
+ && (d < 94))
+ return d + 32;
+ }
+ else if (XCHARSET_CHARS (charset) == 96)
+ {
+ if (((d = ch - (MIN_CHAR_96
+ + (XCHARSET_FINAL (charset) - '0') * 96)) >= 0)
+ && (d < 96))
+ return d + 33;
+ }
+ else
+ return 0;
+ }
+ else if (XCHARSET_DIMENSION (charset) == 2)
+ {
+ if (XCHARSET_CHARS (charset) == 94)
+ {
+ if (((d = ch - (MIN_CHAR_94x94
+ + (XCHARSET_FINAL (charset) - '0') * 94 * 94)) >= 0)
+ && (d < 94 * 94))
+ return (d / 94) + 33;
+ }
+ else if (XCHARSET_CHARS (charset) == 96)
+ {
+ if (((d = ch - (MIN_CHAR_96x96
+ + (XCHARSET_FINAL (charset) - '0') * 96 * 96)) >= 0)
+ && (d < 96 * 96))
+ return (d / 96) + 32;
+ }
+ }
+ return 0;
+}
+
+unsigned char
+charset_get_byte2 (Lisp_Object charset, Emchar ch)
+{
+ if (XCHARSET_DIMENSION (charset) == 1)
+ return 0;
+ else
+ {
+ Emchar_to_byte_table* table;
+
+ if ((table = XCHARSET_TO_BYTE2_TABLE (charset)) != NULL)
+ return get_byte_from_character_table (ch, table);
+ else if (EQ (charset, Vcharset_ucs_bmp))
+ return (ch >> 8) & 0xff;
+ else if (XCHARSET_CHARS (charset) == 94)
+ return (MIN_CHAR_94x94
+ + (XCHARSET_FINAL (charset) - '0') * 94 * 94 <= ch)
+ && (ch < MIN_CHAR_94x94
+ + (XCHARSET_FINAL (charset) - '0' + 1) * 94 * 94) ?
+ ((ch - MIN_CHAR_94x94) % 94) + 33 : 0;
+ else /* if (XCHARSET_CHARS (charset) == 96) */
+ return (MIN_CHAR_96x96
+ + (XCHARSET_FINAL (charset) - '0') * 96 * 96 <= ch)
+ && (ch < MIN_CHAR_96x96
+ + (XCHARSET_FINAL (charset) - '0' + 1) * 96 * 96) ?
+ ((ch - MIN_CHAR_96x96) % 96) + 32 : 0;
+ }
+}
+
+Lisp_Object Vdefault_coded_charset_priority_list;
+#endif
+
\f
/************************************************************************/
/* Basic charset Lisp functions */
error
("Character set already defined for this DIMENSION/CHARS/FINAL combo");
+#ifdef UTF2000
+ if (dimension == 1)
+ {
+ if (chars == 94)
+ {
+ /* id = CHARSET_ID_OFFSET_94 + final; */
+ id = get_unallocated_leading_byte (dimension);
+ }
+ else if (chars == 96)
+ {
+ id = get_unallocated_leading_byte (dimension);
+ }
+ else
+ {
+ abort ();
+ }
+ }
+ else if (dimension == 2)
+ {
+ if (chars == 94)
+ {
+ id = get_unallocated_leading_byte (dimension);
+ }
+ else if (chars == 96)
+ {
+ id = get_unallocated_leading_byte (dimension);
+ }
+ else
+ {
+ abort ();
+ }
+ }
+ else
+ {
+ abort ();
+ }
+#else
id = get_unallocated_leading_byte (dimension);
+#endif
if (NILP (doc_string))
doc_string = build_string ("");
if (columns == -1)
columns = dimension;
- charset = make_charset (id, name, dimension + 2, type, columns, graphic,
- final, direction, short_name, long_name, doc_string, registry);
+ charset = make_charset (id, name, type, columns, graphic,
+ final, direction, short_name, long_name,
+ doc_string, registry,
+ NULL,
+ 0, 0, 0);
if (!NILP (ccl_program))
XCHARSET_CCL_PROGRAM (charset) = ccl_program;
return charset;
long_name = CHARSET_LONG_NAME (cs);
registry = CHARSET_REGISTRY (cs);
- new_charset = make_charset (id, new_name, dimension + 2, type, columns,
+ new_charset = make_charset (id, new_name, type, columns,
graphic, final, direction, short_name, long_name,
- doc_string, registry);
+ doc_string, registry,
+ NULL,
+ 0, 0, 0);
CHARSET_REVERSE_DIRECTION_CHARSET (cs) = new_charset;
XCHARSET_REVERSE_DIRECTION_CHARSET (new_charset) = charset;
if (EQ (charset, Vcharset_ascii)) lowlim = 0, highlim = 127;
else if (EQ (charset, Vcharset_control_1)) lowlim = 0, highlim = 31;
+#ifdef UTF2000
+ else if (CHARSET_CHARS (cs) == 256) lowlim = 0, highlim = 255;
+#endif
else if (CHARSET_CHARS (cs) == 94) lowlim = 33, highlim = 126;
else /* CHARSET_CHARS (cs) == 96) */ lowlim = 32, highlim = 127;
the 8th bit off ARG1 and ARG2 becaue it allows programmers to
write (make-char 'latin-iso8859-2 CODE) where code is the actual
Latin 2 code of the character. */
- a1 = XINT (arg1) & 0x7f;
+#ifdef UTF2000
+ a1 = XINT (arg1);
+ if (highlim < 128)
+ a1 &= 0x7f;
+#else
+ a1 = XINT (arg1);
+#endif
if (a1 < lowlim || a1 > highlim)
args_out_of_range_3 (arg1, make_int (lowlim), make_int (highlim));
}
CHECK_INT (arg2);
+#ifdef UTF2000
+ a2 = XINT (arg2);
+ if (highlim < 128)
+ a2 &= 0x7f;
+#else
a2 = XINT (arg2) & 0x7f;
+#endif
if (a2 < lowlim || a2 > highlim)
args_out_of_range_3 (arg2, make_int (lowlim), make_int (highlim));
{
CHECK_CHAR_COERCE_INT (ch);
- return XCHARSET_NAME (CHARSET_BY_LEADING_BYTE
- (CHAR_LEADING_BYTE (XCHAR (ch))));
+ return XCHARSET_NAME (CHAR_CHARSET (XCHAR (ch)));
}
DEFUN ("split-char", Fsplit_char, 1, 1, 0, /*
defsymbol (&Qchinese_cns11643_5, "chinese-cns11643-5");
defsymbol (&Qchinese_cns11643_6, "chinese-cns11643-6");
defsymbol (&Qchinese_cns11643_7, "chinese-cns11643-7");
+ defsymbol (&Qucs_bmp, "ucs-bmp");
+ defsymbol (&Qlatin_viscii_lower, "vietnamese-viscii-lower");
+ defsymbol (&Qlatin_viscii_upper, "vietnamese-viscii-upper");
#endif
defsymbol (&Qchinese_big5_1, "chinese-big5-1");
defsymbol (&Qchinese_big5_2, "chinese-big5-2");
defsymbol (&Qcomposite, "composite");
-
-#ifdef UTF2000
- Vutf_2000_version = build_string("0.4 (Shin-Imamiya)");
- DEFVAR_LISP ("utf-2000-version", &Vutf_2000_version /*
-Version number of UTF-2000.
-*/ );
-#endif
}
void
vars_of_mule_charset (void)
{
- int i, j, k;
+ int i, j;
+#ifndef UTF2000
+ int k;
+#endif
/* Table of charsets indexed by leading byte. */
for (i = 0; i < countof (charset_by_leading_byte); i++)
charset_by_leading_byte[i] = Qnil;
+#ifdef UTF2000
+ /* Table of charsets indexed by type/final-byte. */
+ for (i = 0; i < countof (charset_by_attributes); i++)
+ for (j = 0; j < countof (charset_by_attributes[0]); j++)
+ charset_by_attributes[i][j] = Qnil;
+#else
/* Table of charsets indexed by type/final-byte/direction. */
for (i = 0; i < countof (charset_by_attributes); i++)
for (j = 0; j < countof (charset_by_attributes[0]); j++)
for (k = 0; k < countof (charset_by_attributes[0][0]); k++)
charset_by_attributes[i][j][k] = Qnil;
+#endif
next_allocated_1_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_1;
#ifdef UTF2000
next_allocated_2_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_2;
#endif
+#ifndef UTF2000
leading_code_private_11 = PRE_LEADING_BYTE_PRIVATE_1;
DEFVAR_INT ("leading-code-private-11", &leading_code_private_11 /*
Leading-code of private TYPE9N charset of column-width 1.
*/ );
leading_code_private_11 = PRE_LEADING_BYTE_PRIVATE_1;
+#endif
+
+#ifdef UTF2000
+ Vutf_2000_version = build_string("0.7 (Hirano)");
+ DEFVAR_LISP ("utf-2000-version", &Vutf_2000_version /*
+Version number of UTF-2000.
+*/ );
+
+ Vdefault_coded_charset_priority_list = Qnil;
+ DEFVAR_LISP ("default-coded-charset-priority-list",
+ &Vdefault_coded_charset_priority_list /*
+Default order of preferred coded-character-set.
+*/ );
+#endif
}
void
/* Predefined character sets. We store them into variables for
ease of access. */
+#ifdef UTF2000
+ Vcharset_ucs_bmp =
+ make_charset (LEADING_BYTE_UCS_BMP, Qucs_bmp,
+ CHARSET_TYPE_256X256, 1, 0, 0,
+ CHARSET_LEFT_TO_RIGHT,
+ build_string ("BMP"),
+ build_string ("BMP"),
+ build_string ("BMP"),
+ build_string (""),
+ NULL, 0, 0xFFFF, 0);
+#else
+# define latin_iso8859_2_to_ucs NULL
+# define latin_iso8859_3_to_ucs NULL
+# define latin_iso8859_4_to_ucs NULL
+# define latin_iso8859_9_to_ucs NULL
+# define latin_jisx0201_to_ucs NULL
+# define MIN_CHAR_THAI 0
+# define MAX_CHAR_THAI 0
+# define MIN_CHAR_GREEK 0
+# define MAX_CHAR_GREEK 0
+# define MIN_CHAR_HEBREW 0
+# define MAX_CHAR_HEBREW 0
+# define MIN_CHAR_HALFWIDTH_KATAKANA 0
+# define MAX_CHAR_HALFWIDTH_KATAKANA 0
+# define MIN_CHAR_CYRILLIC 0
+# define MAX_CHAR_CYRILLIC 0
+#endif
Vcharset_ascii =
- make_charset (LEADING_BYTE_ASCII, Qascii, 1,
+ make_charset (LEADING_BYTE_ASCII, Qascii,
CHARSET_TYPE_94, 1, 0, 'B',
CHARSET_LEFT_TO_RIGHT,
build_string ("ASCII"),
build_string ("ASCII)"),
build_string ("ASCII (ISO646 IRV)"),
- build_string ("\\(iso8859-[0-9]*\\|-ascii\\)"));
+ build_string ("\\(iso8859-[0-9]*\\|-ascii\\)"),
+ NULL, 0, 0x7F, 0);
Vcharset_control_1 =
- make_charset (LEADING_BYTE_CONTROL_1, Qcontrol_1, 2,
+ make_charset (LEADING_BYTE_CONTROL_1, Qcontrol_1,
CHARSET_TYPE_94, 1, 1, 0,
CHARSET_LEFT_TO_RIGHT,
build_string ("C1"),
build_string ("Control characters"),
build_string ("Control characters 128-191"),
- build_string (""));
+ build_string (""),
+ NULL, 0x80, 0x9F, 0);
Vcharset_latin_iso8859_1 =
- make_charset (LEADING_BYTE_LATIN_ISO8859_1, Qlatin_iso8859_1, 2,
+ make_charset (LEADING_BYTE_LATIN_ISO8859_1, Qlatin_iso8859_1,
CHARSET_TYPE_96, 1, 1, 'A',
CHARSET_LEFT_TO_RIGHT,
build_string ("Latin-1"),
build_string ("ISO8859-1 (Latin-1)"),
build_string ("ISO8859-1 (Latin-1)"),
- build_string ("iso8859-1"));
+ build_string ("iso8859-1"),
+ NULL, 0xA0, 0xFF, 32);
Vcharset_latin_iso8859_2 =
- make_charset (LEADING_BYTE_LATIN_ISO8859_2, Qlatin_iso8859_2, 2,
+ make_charset (LEADING_BYTE_LATIN_ISO8859_2, Qlatin_iso8859_2,
CHARSET_TYPE_96, 1, 1, 'B',
CHARSET_LEFT_TO_RIGHT,
build_string ("Latin-2"),
build_string ("ISO8859-2 (Latin-2)"),
build_string ("ISO8859-2 (Latin-2)"),
- build_string ("iso8859-2"));
+ build_string ("iso8859-2"),
+ latin_iso8859_2_to_ucs, 0, 0, 32);
Vcharset_latin_iso8859_3 =
- make_charset (LEADING_BYTE_LATIN_ISO8859_3, Qlatin_iso8859_3, 2,
+ make_charset (LEADING_BYTE_LATIN_ISO8859_3, Qlatin_iso8859_3,
CHARSET_TYPE_96, 1, 1, 'C',
CHARSET_LEFT_TO_RIGHT,
build_string ("Latin-3"),
build_string ("ISO8859-3 (Latin-3)"),
build_string ("ISO8859-3 (Latin-3)"),
- build_string ("iso8859-3"));
+ build_string ("iso8859-3"),
+ latin_iso8859_3_to_ucs, 0, 0, 32);
Vcharset_latin_iso8859_4 =
- make_charset (LEADING_BYTE_LATIN_ISO8859_4, Qlatin_iso8859_4, 2,
+ make_charset (LEADING_BYTE_LATIN_ISO8859_4, Qlatin_iso8859_4,
CHARSET_TYPE_96, 1, 1, 'D',
CHARSET_LEFT_TO_RIGHT,
build_string ("Latin-4"),
build_string ("ISO8859-4 (Latin-4)"),
build_string ("ISO8859-4 (Latin-4)"),
- build_string ("iso8859-4"));
+ build_string ("iso8859-4"),
+ latin_iso8859_4_to_ucs, 0, 0, 32);
Vcharset_thai_tis620 =
- make_charset (LEADING_BYTE_THAI_TIS620, Qthai_tis620, 2,
+ make_charset (LEADING_BYTE_THAI_TIS620, Qthai_tis620,
CHARSET_TYPE_96, 1, 1, 'T',
CHARSET_LEFT_TO_RIGHT,
build_string ("TIS620"),
build_string ("TIS620 (Thai)"),
build_string ("TIS620.2529 (Thai)"),
- build_string ("tis620"));
+ build_string ("tis620"),
+ NULL, MIN_CHAR_THAI, MAX_CHAR_THAI, 32);
Vcharset_greek_iso8859_7 =
- make_charset (LEADING_BYTE_GREEK_ISO8859_7, Qgreek_iso8859_7, 2,
+ make_charset (LEADING_BYTE_GREEK_ISO8859_7, Qgreek_iso8859_7,
CHARSET_TYPE_96, 1, 1, 'F',
CHARSET_LEFT_TO_RIGHT,
build_string ("ISO8859-7"),
build_string ("ISO8859-7 (Greek)"),
build_string ("ISO8859-7 (Greek)"),
- build_string ("iso8859-7"));
+ build_string ("iso8859-7"),
+ NULL, MIN_CHAR_GREEK, MAX_CHAR_GREEK, 32);
Vcharset_arabic_iso8859_6 =
- make_charset (LEADING_BYTE_ARABIC_ISO8859_6, Qarabic_iso8859_6, 2,
+ make_charset (LEADING_BYTE_ARABIC_ISO8859_6, Qarabic_iso8859_6,
CHARSET_TYPE_96, 1, 1, 'G',
CHARSET_RIGHT_TO_LEFT,
build_string ("ISO8859-6"),
build_string ("ISO8859-6 (Arabic)"),
build_string ("ISO8859-6 (Arabic)"),
- build_string ("iso8859-6"));
+ build_string ("iso8859-6"),
+ NULL, 0, 0, 32);
Vcharset_hebrew_iso8859_8 =
- make_charset (LEADING_BYTE_HEBREW_ISO8859_8, Qhebrew_iso8859_8, 2,
+ make_charset (LEADING_BYTE_HEBREW_ISO8859_8, Qhebrew_iso8859_8,
CHARSET_TYPE_96, 1, 1, 'H',
CHARSET_RIGHT_TO_LEFT,
build_string ("ISO8859-8"),
build_string ("ISO8859-8 (Hebrew)"),
build_string ("ISO8859-8 (Hebrew)"),
- build_string ("iso8859-8"));
+ build_string ("iso8859-8"),
+ NULL, MIN_CHAR_HEBREW, MAX_CHAR_HEBREW, 32);
Vcharset_katakana_jisx0201 =
- make_charset (LEADING_BYTE_KATAKANA_JISX0201, Qkatakana_jisx0201, 2,
+ make_charset (LEADING_BYTE_KATAKANA_JISX0201, Qkatakana_jisx0201,
CHARSET_TYPE_94, 1, 1, 'I',
CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0201 Kana"),
build_string ("JISX0201.1976 (Japanese Kana)"),
build_string ("JISX0201.1976 Japanese Kana"),
- build_string ("jisx0201.1976"));
+ build_string ("jisx0201.1976"),
+ NULL,
+ MIN_CHAR_HALFWIDTH_KATAKANA,
+ MAX_CHAR_HALFWIDTH_KATAKANA, 33);
Vcharset_latin_jisx0201 =
- make_charset (LEADING_BYTE_LATIN_JISX0201, Qlatin_jisx0201, 2,
+ make_charset (LEADING_BYTE_LATIN_JISX0201, Qlatin_jisx0201,
CHARSET_TYPE_94, 1, 0, 'J',
CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0201 Roman"),
build_string ("JISX0201.1976 (Japanese Roman)"),
build_string ("JISX0201.1976 Japanese Roman"),
- build_string ("jisx0201.1976"));
+ build_string ("jisx0201.1976"),
+ latin_jisx0201_to_ucs, 0, 0, 33);
Vcharset_cyrillic_iso8859_5 =
- make_charset (LEADING_BYTE_CYRILLIC_ISO8859_5, Qcyrillic_iso8859_5, 2,
+ make_charset (LEADING_BYTE_CYRILLIC_ISO8859_5, Qcyrillic_iso8859_5,
CHARSET_TYPE_96, 1, 1, 'L',
CHARSET_LEFT_TO_RIGHT,
build_string ("ISO8859-5"),
build_string ("ISO8859-5 (Cyrillic)"),
build_string ("ISO8859-5 (Cyrillic)"),
- build_string ("iso8859-5"));
+ build_string ("iso8859-5"),
+ NULL, MIN_CHAR_CYRILLIC, MAX_CHAR_CYRILLIC, 32);
Vcharset_latin_iso8859_9 =
- make_charset (LEADING_BYTE_LATIN_ISO8859_9, Qlatin_iso8859_9, 2,
+ make_charset (LEADING_BYTE_LATIN_ISO8859_9, Qlatin_iso8859_9,
CHARSET_TYPE_96, 1, 1, 'M',
CHARSET_LEFT_TO_RIGHT,
build_string ("Latin-5"),
build_string ("ISO8859-9 (Latin-5)"),
build_string ("ISO8859-9 (Latin-5)"),
- build_string ("iso8859-9"));
+ build_string ("iso8859-9"),
+ latin_iso8859_9_to_ucs, 0, 0, 32);
Vcharset_japanese_jisx0208_1978 =
- make_charset (LEADING_BYTE_JAPANESE_JISX0208_1978, Qjapanese_jisx0208_1978, 3,
+ make_charset (LEADING_BYTE_JAPANESE_JISX0208_1978, Qjapanese_jisx0208_1978,
CHARSET_TYPE_94X94, 2, 0, '@',
CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0208.1978"),
build_string ("JISX0208.1978 (Japanese)"),
build_string
("JISX0208.1978 Japanese Kanji (so called \"old JIS\")"),
- build_string ("\\(jisx0208\\|jisc6226\\)\\.1978"));
+ build_string ("\\(jisx0208\\|jisc6226\\)\\.1978"),
+ NULL, 0, 0, 33);
Vcharset_chinese_gb2312 =
- make_charset (LEADING_BYTE_CHINESE_GB2312, Qchinese_gb2312, 3,
+ make_charset (LEADING_BYTE_CHINESE_GB2312, Qchinese_gb2312,
CHARSET_TYPE_94X94, 2, 0, 'A',
CHARSET_LEFT_TO_RIGHT,
build_string ("GB2312"),
build_string ("GB2312)"),
build_string ("GB2312 Chinese simplified"),
- build_string ("gb2312"));
+ build_string ("gb2312"),
+ NULL, 0, 0, 33);
Vcharset_japanese_jisx0208 =
- make_charset (LEADING_BYTE_JAPANESE_JISX0208, Qjapanese_jisx0208, 3,
+ make_charset (LEADING_BYTE_JAPANESE_JISX0208, Qjapanese_jisx0208,
CHARSET_TYPE_94X94, 2, 0, 'B',
CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0208"),
build_string ("JISX0208.1983/1990 (Japanese)"),
build_string ("JISX0208.1983/1990 Japanese Kanji"),
- build_string ("jisx0208.19\\(83\\|90\\)"));
+ build_string ("jisx0208.19\\(83\\|90\\)"),
+ NULL, 0, 0, 33);
Vcharset_korean_ksc5601 =
- make_charset (LEADING_BYTE_KOREAN_KSC5601, Qkorean_ksc5601, 3,
+ make_charset (LEADING_BYTE_KOREAN_KSC5601, Qkorean_ksc5601,
CHARSET_TYPE_94X94, 2, 0, 'C',
CHARSET_LEFT_TO_RIGHT,
build_string ("KSC5601"),
build_string ("KSC5601 (Korean"),
build_string ("KSC5601 Korean Hangul and Hanja"),
- build_string ("ksc5601"));
+ build_string ("ksc5601"),
+ NULL, 0, 0, 33);
Vcharset_japanese_jisx0212 =
- make_charset (LEADING_BYTE_JAPANESE_JISX0212, Qjapanese_jisx0212, 3,
+ make_charset (LEADING_BYTE_JAPANESE_JISX0212, Qjapanese_jisx0212,
CHARSET_TYPE_94X94, 2, 0, 'D',
CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0212"),
build_string ("JISX0212 (Japanese)"),
build_string ("JISX0212 Japanese Supplement"),
- build_string ("jisx0212"));
+ build_string ("jisx0212"),
+ NULL, 0, 0, 33);
#define CHINESE_CNS_PLANE_RE(n) "cns11643[.-]\\(.*[.-]\\)?" n "$"
Vcharset_chinese_cns11643_1 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_1, Qchinese_cns11643_1, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_1, Qchinese_cns11643_1,
CHARSET_TYPE_94X94, 2, 0, 'G',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-1"),
build_string ("CNS11643-1 (Chinese traditional)"),
build_string
("CNS 11643 Plane 1 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("1")));
+ build_string (CHINESE_CNS_PLANE_RE("1")),
+ NULL, 0, 0, 33);
Vcharset_chinese_cns11643_2 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_2, Qchinese_cns11643_2, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_2, Qchinese_cns11643_2,
CHARSET_TYPE_94X94, 2, 0, 'H',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-2"),
build_string ("CNS11643-2 (Chinese traditional)"),
build_string
("CNS 11643 Plane 2 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("2")));
+ build_string (CHINESE_CNS_PLANE_RE("2")),
+ NULL, 0, 0, 33);
#ifdef UTF2000
Vcharset_chinese_cns11643_3 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_3, Qchinese_cns11643_3, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_3, Qchinese_cns11643_3,
CHARSET_TYPE_94X94, 2, 0, 'I',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-3"),
build_string ("CNS11643-3 (Chinese traditional)"),
build_string
("CNS 11643 Plane 3 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("3")));
+ build_string (CHINESE_CNS_PLANE_RE("3")),
+ NULL, 0, 0, 33);
Vcharset_chinese_cns11643_4 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_4, Qchinese_cns11643_4, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_4, Qchinese_cns11643_4,
CHARSET_TYPE_94X94, 2, 0, 'J',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-4"),
build_string ("CNS11643-4 (Chinese traditional)"),
build_string
("CNS 11643 Plane 4 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("4")));
+ build_string (CHINESE_CNS_PLANE_RE("4")),
+ NULL, 0, 0, 33);
Vcharset_chinese_cns11643_5 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_5, Qchinese_cns11643_5, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_5, Qchinese_cns11643_5,
CHARSET_TYPE_94X94, 2, 0, 'K',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-5"),
build_string ("CNS11643-5 (Chinese traditional)"),
build_string
("CNS 11643 Plane 5 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("5")));
+ build_string (CHINESE_CNS_PLANE_RE("5")),
+ NULL, 0, 0, 33);
Vcharset_chinese_cns11643_6 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_6, Qchinese_cns11643_6, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_6, Qchinese_cns11643_6,
CHARSET_TYPE_94X94, 2, 0, 'L',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-6"),
build_string ("CNS11643-6 (Chinese traditional)"),
build_string
("CNS 11643 Plane 6 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("6")));
+ build_string (CHINESE_CNS_PLANE_RE("6")),
+ NULL, 0, 0, 33);
Vcharset_chinese_cns11643_7 =
- make_charset (LEADING_BYTE_CHINESE_CNS11643_7, Qchinese_cns11643_7, 3,
+ make_charset (LEADING_BYTE_CHINESE_CNS11643_7, Qchinese_cns11643_7,
CHARSET_TYPE_94X94, 2, 0, 'M',
CHARSET_LEFT_TO_RIGHT,
build_string ("CNS11643-7"),
build_string ("CNS11643-7 (Chinese traditional)"),
build_string
("CNS 11643 Plane 7 Chinese traditional"),
- build_string (CHINESE_CNS_PLANE_RE("7")));
+ build_string (CHINESE_CNS_PLANE_RE("7")),
+ NULL, 0, 0, 33);
+ Vcharset_latin_viscii_lower =
+ make_charset (LEADING_BYTE_LATIN_VISCII_LOWER, Qlatin_viscii_lower,
+ CHARSET_TYPE_96, 1, 1, '1',
+ CHARSET_LEFT_TO_RIGHT,
+ build_string ("VISCII lower"),
+ build_string ("VISCII lower (Vietnamese)"),
+ build_string ("VISCII lower (Vietnamese)"),
+ build_string ("VISCII1.1"),
+ latin_viscii_lower_to_ucs, 0, 0, 32);
+ Vcharset_latin_viscii_upper =
+ make_charset (LEADING_BYTE_LATIN_VISCII_UPPER, Qlatin_viscii_upper,
+ CHARSET_TYPE_96, 1, 1, '2',
+ CHARSET_LEFT_TO_RIGHT,
+ build_string ("VISCII upper"),
+ build_string ("VISCII upper (Vietnamese)"),
+ build_string ("VISCII upper (Vietnamese)"),
+ build_string ("VISCII1.1"),
+ latin_viscii_upper_to_ucs, 0, 0, 32);
#endif
Vcharset_chinese_big5_1 =
- make_charset (LEADING_BYTE_CHINESE_BIG5_1, Qchinese_big5_1, 3,
+ make_charset (LEADING_BYTE_CHINESE_BIG5_1, Qchinese_big5_1,
CHARSET_TYPE_94X94, 2, 0, '0',
CHARSET_LEFT_TO_RIGHT,
build_string ("Big5"),
build_string ("Big5 (Level-1)"),
build_string
("Big5 Level-1 Chinese traditional"),
- build_string ("big5"));
+ build_string ("big5"),
+ NULL, 0, 0, 33);
Vcharset_chinese_big5_2 =
- make_charset (LEADING_BYTE_CHINESE_BIG5_2, Qchinese_big5_2, 3,
+ make_charset (LEADING_BYTE_CHINESE_BIG5_2, Qchinese_big5_2,
CHARSET_TYPE_94X94, 2, 0, '1',
CHARSET_LEFT_TO_RIGHT,
build_string ("Big5"),
build_string ("Big5 (Level-2)"),
build_string
("Big5 Level-2 Chinese traditional"),
- build_string ("big5"));
-
+ build_string ("big5"),
+ NULL, 0, 0, 33);
#ifdef ENABLE_COMPOSITE_CHARS
/* #### For simplicity, we put composite chars into a 96x96 charset.
This is going to lead to problems because you can run out of
room, esp. as we don't yet recycle numbers. */
Vcharset_composite =
- make_charset (LEADING_BYTE_COMPOSITE, Qcomposite, 3,
+ make_charset (LEADING_BYTE_COMPOSITE, Qcomposite,
CHARSET_TYPE_96X96, 2, 0, 0,
CHARSET_LEFT_TO_RIGHT,
build_string ("Composite"),
projects / chise / xemacs-chise.git / blobdiff