/* Functions to handle multilingual characters.
Copyright (C) 1992, 1995 Free Software Foundation, Inc.
Copyright (C) 1995 Sun Microsystems, Inc.
- Copyright (C) 1999,2000,2001,2002 MORIOKA Tomohiko
+ Copyright (C) 1999,2000,2001,2002,2003 MORIOKA Tomohiko
This file is part of XEmacs.
Lisp_Object Vcharset_ucs_bmp;
Lisp_Object Vcharset_ucs_smp;
Lisp_Object Vcharset_ucs_sip;
-Lisp_Object Vcharset_ucs_cns;
-Lisp_Object Vcharset_ucs_jis;
-Lisp_Object Vcharset_ucs_ks;
-Lisp_Object Vcharset_ucs_big5;
Lisp_Object Vcharset_latin_viscii;
Lisp_Object Vcharset_latin_tcvn5712;
Lisp_Object Vcharset_latin_viscii_lower;
Lisp_Object Vcharset_latin_viscii_upper;
+Lisp_Object Vcharset_jis_x0208;
Lisp_Object Vcharset_chinese_big5;
-/* Lisp_Object Vcharset_chinese_big5_cdp; */
-Lisp_Object Vcharset_ideograph_hanziku_1;
-Lisp_Object Vcharset_ideograph_hanziku_2;
-Lisp_Object Vcharset_ideograph_hanziku_3;
-Lisp_Object Vcharset_ideograph_hanziku_4;
-Lisp_Object Vcharset_ideograph_hanziku_5;
-Lisp_Object Vcharset_ideograph_hanziku_6;
-Lisp_Object Vcharset_ideograph_hanziku_7;
-Lisp_Object Vcharset_ideograph_hanziku_8;
-Lisp_Object Vcharset_ideograph_hanziku_9;
-Lisp_Object Vcharset_ideograph_hanziku_10;
-Lisp_Object Vcharset_ideograph_hanziku_11;
-Lisp_Object Vcharset_ideograph_hanziku_12;
-Lisp_Object Vcharset_china3_jef;
-Lisp_Object Vcharset_ideograph_cbeta;
-Lisp_Object Vcharset_ideograph_gt;
-Lisp_Object Vcharset_ideograph_gt_pj_1;
-Lisp_Object Vcharset_ideograph_gt_pj_2;
-Lisp_Object Vcharset_ideograph_gt_pj_3;
-Lisp_Object Vcharset_ideograph_gt_pj_4;
-Lisp_Object Vcharset_ideograph_gt_pj_5;
-Lisp_Object Vcharset_ideograph_gt_pj_6;
-Lisp_Object Vcharset_ideograph_gt_pj_7;
-Lisp_Object Vcharset_ideograph_gt_pj_8;
-Lisp_Object Vcharset_ideograph_gt_pj_9;
-Lisp_Object Vcharset_ideograph_gt_pj_10;
-Lisp_Object Vcharset_ideograph_gt_pj_11;
-Lisp_Object Vcharset_ideograph_daikanwa_2;
-Lisp_Object Vcharset_ideograph_daikanwa;
Lisp_Object Vcharset_ethiopic_ucs;
#endif
Lisp_Object Vcharset_chinese_big5_1;
static const struct lrecord_description charset_lookup_description_1[] = {
{ XD_LISP_OBJECT_ARRAY, offsetof (struct charset_lookup, charset_by_leading_byte),
#ifdef UTF2000
- 128+4*128
+ NUM_LEADING_BYTES+4*128
#else
128+4*128*2
#endif
#ifdef UTF2000
-INLINE_HEADER int CHARSET_BYTE_SIZE (Lisp_Charset* cs);
-INLINE_HEADER int
-CHARSET_BYTE_SIZE (Lisp_Charset* cs)
-{
- /* ad-hoc method for `ascii' */
- if ((CHARSET_CHARS (cs) == 94) &&
- (CHARSET_BYTE_OFFSET (cs) != 33))
- return 128 - CHARSET_BYTE_OFFSET (cs);
- else
- return CHARSET_CHARS (cs);
-}
-
-#define XCHARSET_BYTE_SIZE(ccs) CHARSET_BYTE_SIZE (XCHARSET (ccs))
-
int decoding_table_check_elements (Lisp_Object v, int dim, int ccs_len);
int
decoding_table_check_elements (Lisp_Object v, int dim, int ccs_len)
return 0;
}
-INLINE_HEADER void
-decoding_table_remove_char (Lisp_Object v, int dim, int byte_offset,
- int code_point);
-INLINE_HEADER void
-decoding_table_remove_char (Lisp_Object v, int dim, int byte_offset,
- int code_point)
-{
- int i = -1;
-
- while (dim > 0)
- {
- Lisp_Object nv;
-
- dim--;
- i = ((code_point >> (8 * dim)) & 255) - byte_offset;
- nv = XVECTOR_DATA(v)[i];
- if (!VECTORP (nv))
- break;
- v = nv;
- }
- if (i >= 0)
- XVECTOR_DATA(v)[i] = Qnil;
-}
-
-INLINE_HEADER void
-decoding_table_put_char (Lisp_Object v, int dim, int byte_offset,
- int code_point, Lisp_Object character);
-INLINE_HEADER void
-decoding_table_put_char (Lisp_Object v, int dim, int byte_offset,
- int code_point, Lisp_Object character)
-{
- int i = -1;
- Lisp_Object nv;
- int ccs_len = XVECTOR_LENGTH (v);
-
- while (dim > 0)
- {
- dim--;
- i = ((code_point >> (8 * dim)) & 255) - byte_offset;
- nv = XVECTOR_DATA(v)[i];
- if (dim > 0)
- {
- if (!VECTORP (nv))
- nv = (XVECTOR_DATA(v)[i] = make_vector (ccs_len, Qnil));
- v = nv;
- }
- else
- break;
- }
- XVECTOR_DATA(v)[i] = character;
-}
-
Lisp_Object
put_char_ccs_code_point (Lisp_Object character,
Lisp_Object ccs, Lisp_Object value)
{
- if (!EQ (XCHARSET_NAME (ccs), Qucs)
+ if (!EQ (XCHARSET_NAME (ccs), Qmap_ucs)
|| !INTP (value)
|| (XCHAR (character) != XINT (value)))
{
Lisp_Object v = XCHARSET_DECODING_TABLE (ccs);
- int dim = XCHARSET_DIMENSION (ccs);
- int ccs_len = XCHARSET_BYTE_SIZE (ccs);
- int byte_offset = XCHARSET_BYTE_OFFSET (ccs);
int code_point;
if (CONSP (value))
Lisp_Object cpos = Fget_char_attribute (character, ccs, Qnil);
if (INTP (cpos))
{
- decoding_table_remove_char (v, dim, byte_offset, XINT (cpos));
+ decoding_table_remove_char (ccs, XINT (cpos));
}
}
- else
- {
- XCHARSET_DECODING_TABLE (ccs)
- = v = make_vector (ccs_len, Qnil);
- }
-
- decoding_table_put_char (v, dim, byte_offset, code_point, character);
+ decoding_table_put_char (ccs, code_point, character);
}
return value;
}
if (!NILP (cpos))
{
- decoding_table_remove_char (decoding_table,
- XCHARSET_DIMENSION (ccs),
- XCHARSET_BYTE_OFFSET (ccs),
- XINT (cpos));
+ decoding_table_remove_char (ccs, XINT (cpos));
}
}
if (CHAR_TABLEP (encoding_table))
{
- put_char_id_table (XCHAR_TABLE(encoding_table), character, Qnil);
+ put_char_id_table (XCHAR_TABLE(encoding_table), character, Qunbound);
}
return Qt;
}
Qchinese_gb2312,
Qchinese_gb12345,
Qjapanese_jisx0208,
- Qjapanese_jisx0208_1990,
Qkorean_ksc5601,
Qjapanese_jisx0212,
Qchinese_cns11643_1,
Qchinese_cns11643_2,
#ifdef UTF2000
- Qucs,
+ Qmap_ucs, Qucs,
Qucs_bmp,
Qucs_smp,
Qucs_sip,
- Qucs_cns,
- Qucs_jis,
- Qucs_ks,
- Qucs_big5,
Qlatin_viscii,
Qlatin_tcvn5712,
Qlatin_viscii_lower,
Qlatin_viscii_upper,
Qvietnamese_viscii_lower,
Qvietnamese_viscii_upper,
+ Qmap_jis_x0208,
+ Qmap_jis_x0208_1990,
Qchinese_big5,
- /* Qchinese_big5_cdp, */
- Qideograph_hanziku_1,
- Qideograph_hanziku_2,
- Qideograph_hanziku_3,
- Qideograph_hanziku_4,
- Qideograph_hanziku_5,
- Qideograph_hanziku_6,
- Qideograph_hanziku_7,
- Qideograph_hanziku_8,
- Qideograph_hanziku_9,
- Qideograph_hanziku_10,
- Qideograph_hanziku_11,
- Qideograph_hanziku_12,
- Qchina3_jef,
- Qideograph_cbeta,
- Qideograph_daikanwa_2,
- Qideograph_daikanwa,
- Qideograph_gt,
- Qideograph_gt_pj_1,
- Qideograph_gt_pj_2,
- Qideograph_gt_pj_3,
- Qideograph_gt_pj_4,
- Qideograph_gt_pj_5,
- Qideograph_gt_pj_6,
- Qideograph_gt_pj_7,
- Qideograph_gt_pj_8,
- Qideograph_gt_pj_9,
- Qideograph_gt_pj_10,
- Qideograph_gt_pj_11,
Qethiopic_ucs,
#endif
Qchinese_big5_1,
CHARSET_CCL_PROGRAM (cs) = Qnil;
CHARSET_REVERSE_DIRECTION_CHARSET (cs) = Qnil;
#ifdef UTF2000
- CHARSET_DECODING_TABLE(cs) = Qnil;
+ CHARSET_DECODING_TABLE(cs) = Qunbound;
CHARSET_MIN_CODE (cs) = min_code;
CHARSET_MAX_CODE (cs) = max_code;
CHARSET_CODE_OFFSET (cs) = code_offset;
#define BIG5_SAME_ROW (0xFF - 0xA1 + 0x7F - 0x40)
Emchar
-decode_builtin_char (Lisp_Object charset, int code_point)
+decode_defined_char (Lisp_Object ccs, int code_point)
{
- Lisp_Object mother = XCHARSET_MOTHER (charset);
- int final;
+ int dim = XCHARSET_DIMENSION (ccs);
+ Lisp_Object decoding_table = XCHARSET_DECODING_TABLE (ccs);
+ Emchar char_id = -1;
+ Lisp_Object mother;
- if ( CHARSETP (mother) )
+ while (dim > 0)
{
- int code = code_point;
-
- if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x60 )
- {
- int row = code_point >> 8;
- int cell = code_point & 255;
-
- if (row < 16 + 32)
- return -1;
- else if (row < 16 + 32 + 30)
- code = (row - (16 + 32)) * 94 + cell - 33;
- else if (row < 18 + 32 + 30)
- return -1;
- else if (row < 18 + 32 + 60)
- code = (row - (18 + 32)) * 94 + cell - 33;
- }
- else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94x60 )
+ dim--;
+ decoding_table
+ = get_ccs_octet_table (decoding_table, ccs,
+ (code_point >> (dim * 8)) & 255);
+ }
+ if (CHARP (decoding_table))
+ return XCHAR (decoding_table);
+ if (char_id >= 0)
+ return char_id;
+ else if ( CHARSETP (mother = XCHARSET_MOTHER (ccs)) )
+ {
+ if ( XCHARSET_CONVERSION (ccs) == CONVERSION_IDENTICAL )
{
- int plane = code_point >> 16;
- int row = (code_point >> 8) & 255;
- int cell = code_point & 255;
-
- if (row < 16 + 32)
- return -1;
- else if (row < 16 + 32 + 30)
- code
- = (plane - 33) * 94 * 60
- + (row - (16 + 32)) * 94
- + cell - 33;
- else if (row < 18 + 32 + 30)
- return -1;
- else if (row < 18 + 32 + 60)
- code
- = (plane - 33) * 94 * 60
- + (row - (18 + 32)) * 94
- + cell - 33;
+ if ( EQ (mother, Vcharset_ucs) )
+ return DECODE_CHAR (mother, code_point);
+ else
+ return decode_defined_char (mother, code_point);
}
- return DECODE_CHAR (mother, code + XCHARSET_CODE_OFFSET(charset));
}
- else if (EQ (charset, Vcharset_chinese_big5))
+ return -1;
+}
+
+Emchar
+decode_builtin_char (Lisp_Object charset, int code_point)
+{
+ Lisp_Object mother = XCHARSET_MOTHER (charset);
+ int final;
+
+ if ( XCHARSET_MAX_CODE (charset) > 0 )
{
- int c1 = code_point >> 8;
- int c2 = code_point & 0xFF;
- unsigned int I;
-
- if ( ( (0xA1 <= c1) && (c1 <= 0xFE) )
- &&
- ( ((0x40 <= c2) && (c2 <= 0x7E)) ||
- ((0xA1 <= c2) && (c2 <= 0xFE)) ) )
+ if ( CHARSETP (mother) )
{
- I = (c1 - 0xA1) * BIG5_SAME_ROW
- + c2 - (c2 < 0x7F ? 0x40 : 0x62);
+ int code = code_point;
- if (c1 < 0xC9)
+ if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x60 )
{
- charset = Vcharset_chinese_big5_1;
+ int row = code_point >> 8;
+ int cell = code_point & 255;
+
+ if (row < 16 + 32)
+ return -1;
+ else if (row < 16 + 32 + 30)
+ code = (row - (16 + 32)) * 94 + cell - 33;
+ else if (row < 18 + 32 + 30)
+ return -1;
+ else if (row < 18 + 32 + 60)
+ code = (row - (18 + 32)) * 94 + cell - 33;
}
- else
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94x60 )
{
- charset = Vcharset_chinese_big5_2;
- I -= (BIG5_SAME_ROW) * (0xC9 - 0xA1);
+ int plane = code_point >> 16;
+ int row = (code_point >> 8) & 255;
+ int cell = code_point & 255;
+
+ if (row < 16 + 32)
+ return -1;
+ else if (row < 16 + 32 + 30)
+ code
+ = (plane - 33) * 94 * 60
+ + (row - (16 + 32)) * 94
+ + cell - 33;
+ else if (row < 18 + 32 + 30)
+ return -1;
+ else if (row < 18 + 32 + 60)
+ code
+ = (plane - 33) * 94 * 60
+ + (row - (18 + 32)) * 94
+ + cell - 33;
}
- code_point = ((I / 94 + 33) << 8) | (I % 94 + 33);
+ return
+ decode_builtin_char (mother, code + XCHARSET_CODE_OFFSET(charset));
+ }
+ else
+ {
+ Emchar cid
+ = (XCHARSET_DIMENSION (charset) == 1
+ ?
+ code_point - XCHARSET_BYTE_OFFSET (charset)
+ :
+ ((code_point >> 8) - XCHARSET_BYTE_OFFSET (charset))
+ * XCHARSET_CHARS (charset)
+ + (code_point & 0xFF) - XCHARSET_BYTE_OFFSET (charset))
+ + XCHARSET_CODE_OFFSET (charset);
+ if ((cid < XCHARSET_MIN_CODE (charset))
+ || (XCHARSET_MAX_CODE (charset) < cid))
+ return -1;
+ return cid;
}
}
- if ((final = XCHARSET_FINAL (charset)) >= '0')
+ else if ((final = XCHARSET_FINAL (charset)) >= '0')
{
if (XCHARSET_DIMENSION (charset) == 1)
{
}
}
}
- else if (XCHARSET_MAX_CODE (charset))
- {
- Emchar cid
- = (XCHARSET_DIMENSION (charset) == 1
- ?
- code_point - XCHARSET_BYTE_OFFSET (charset)
- :
- ((code_point >> 8) - XCHARSET_BYTE_OFFSET (charset))
- * XCHARSET_CHARS (charset)
- + (code_point & 0xFF) - XCHARSET_BYTE_OFFSET (charset))
- + XCHARSET_CODE_OFFSET (charset);
- if ((cid < XCHARSET_MIN_CODE (charset))
- || (XCHARSET_MAX_CODE (charset) < cid))
- return -1;
- return cid;
- }
else
return -1;
}
int
-charset_code_point (Lisp_Object charset, Emchar ch)
+charset_code_point (Lisp_Object charset, Emchar ch, int defined_only)
{
Lisp_Object encoding_table = XCHARSET_ENCODING_TABLE (charset);
Lisp_Object ret;
Lisp_Object mother = XCHARSET_MOTHER (charset);
int min = XCHARSET_MIN_CODE (charset);
int max = XCHARSET_MAX_CODE (charset);
- int code;
+ int code = -1;
if ( CHARSETP (mother) )
- code = charset_code_point (mother, ch);
- else
+ {
+ if (XCHARSET_FINAL (charset) >= '0')
+ code = charset_code_point (mother, ch, 1);
+ else
+ code = charset_code_point (mother, ch, defined_only);
+ }
+ else if (defined_only)
+ return -1;
+ else if ( ((max == 0) && CHARSETP (mother)
+ && (XCHARSET_FINAL (charset) == 0))
+ || ((min <= ch) && (ch <= max)) )
code = ch;
- if ( (min <= code) && (code <= max) )
+ if ( ((max == 0) && CHARSETP (mother) && (code >= 0))
+ || ((min <= code) && (code <= max)) )
{
int d = code - XCHARSET_CODE_OFFSET (charset);
- if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x60 )
+ if ( XCHARSET_CONVERSION (charset) == CONVERSION_IDENTICAL )
+ return d;
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94 )
+ return d + 33;
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_96 )
+ return d + 32;
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x60 )
{
int row = d / 94;
int cell = d % 94 + 33;
row += 18 + 32;
return (row << 8) | cell;
}
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94 )
+ return ((d / 94 + 33) << 8) | (d % 94 + 33);
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_96x96 )
+ return ((d / 96 + 32) << 8) | (d % 96 + 32);
else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94x60 )
{
int plane = d / (94 * 60) + 33;
row += 18 + 32;
return (plane << 16) | (row << 8) | cell;
}
- else if (XCHARSET_CHARS (charset) == 94)
- {
- if (XCHARSET_DIMENSION (charset) == 1)
- return d + 33;
- else if (XCHARSET_DIMENSION (charset) == 2)
- return ((d / 94 + 33) << 8) | (d % 94 + 33);
- else if (XCHARSET_DIMENSION (charset) == 3)
- return
- ( (d / (94 * 94) + 33) << 16)
- | ((d / 94 % 94 + 33) << 8)
- | (d % 94 + 33);
- else /* if (XCHARSET_DIMENSION (charset) == 4) */
- return
- ( (d / (94 * 94 * 94) + 33) << 24)
- | ((d / (94 * 94) % 94 + 33) << 16)
- | ((d / 94 % 94 + 33) << 8)
- | (d % 94 + 33);
- }
- else if (XCHARSET_CHARS (charset) == 96)
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94x94 )
+ return
+ ( (d / (94 * 94) + 33) << 16)
+ | ((d / 94 % 94 + 33) << 8)
+ | (d % 94 + 33);
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_96x96x96 )
+ return
+ ( (d / (96 * 96) + 32) << 16)
+ | ((d / 96 % 96 + 32) << 8)
+ | (d % 96 + 32);
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_94x94x94x94 )
+ return
+ ( (d / (94 * 94 * 94) + 33) << 24)
+ | ((d / (94 * 94) % 94 + 33) << 16)
+ | ((d / 94 % 94 + 33) << 8)
+ | (d % 94 + 33);
+ else if ( XCHARSET_CONVERSION (charset) == CONVERSION_96x96x96x96 )
+ return
+ ( (d / (96 * 96 * 96) + 32) << 24)
+ | ((d / (96 * 96) % 96 + 32) << 16)
+ | ((d / 96 % 96 + 32) << 8)
+ | (d % 96 + 32);
+ else
{
- if (XCHARSET_DIMENSION (charset) == 1)
- return d + 32;
- else if (XCHARSET_DIMENSION (charset) == 2)
- return ((d / 96 + 32) << 8) | (d % 96 + 32);
- else if (XCHARSET_DIMENSION (charset) == 3)
- return
- ( (d / (96 * 96) + 32) << 16)
- | ((d / 96 % 96 + 32) << 8)
- | (d % 96 + 32);
- else /* if (XCHARSET_DIMENSION (charset) == 4) */
- return
- ( (d / (96 * 96 * 96) + 32) << 24)
- | ((d / (96 * 96) % 96 + 32) << 16)
- | ((d / 96 % 96 + 32) << 8)
- | (d % 96 + 32);
+ printf ("Unknown CCS-conversion %d is specified!",
+ XCHARSET_CONVERSION (charset));
+ exit (-1);
}
- else
- return code - XCHARSET_CODE_OFFSET (charset);
}
- else if ( (XCHARSET_CODE_OFFSET (charset) == 0) ||
+ else if ( ( XCHARSET_FINAL (charset) >= '0' ) &&
+ ( XCHARSET_MIN_CODE (charset) == 0 )
+ /*
+ (XCHARSET_CODE_OFFSET (charset) == 0) ||
(XCHARSET_CODE_OFFSET (charset)
- == XCHARSET_MIN_CODE (charset)) )
+ == XCHARSET_MIN_CODE (charset))
+ */ )
{
int d;
*charset = Vcharset_ucs_sip;
return c - MIN_CHAR_SIP;
}
- else if (c < MIN_CHAR_DAIKANWA)
- {
- *charset = Vcharset_ucs;
- return c;
- }
- else if (c <= MAX_CHAR_DAIKANWA)
- {
- *charset = Vcharset_ideograph_daikanwa;
- return c - MIN_CHAR_DAIKANWA;
- }
else if (c < MIN_CHAR_94)
{
*charset = Vcharset_ucs;
(charset, table))
{
struct Lisp_Charset *cs;
- size_t i;
+ int i;
int byte_offset;
charset = Fget_charset (charset);
if (VECTORP (v))
{
- size_t j;
+ int j;
for (j = 0; j < XVECTOR_LENGTH (v); j++)
{
if (NILP (defined_only))
c = DECODE_CHAR (charset, c);
else
- c = DECODE_DEFINED_CHAR (charset, c);
+ c = decode_defined_char (charset, c);
return c >= 0 ? make_char (c) : Qnil;
}
}
#ifdef UTF2000
-DEFUN ("encode-char", Fencode_char, 2, 2, 0, /*
+DEFUN ("encode-char", Fencode_char, 2, 3, 0, /*
Return code-point of CHARACTER in specified CHARSET.
*/
- (character, charset))
+ (character, charset, defined_only))
{
int code_point;
CHECK_CHAR_COERCE_INT (character);
charset = Fget_charset (charset);
- code_point = charset_code_point (charset, XCHAR (character));
+ code_point = charset_code_point (charset, XCHAR (character),
+ !NILP (defined_only));
if (code_point >= 0)
return make_int (code_point);
else
defsymbol (&Qchinese_gb2312, "chinese-gb2312");
defsymbol (&Qchinese_gb12345, "chinese-gb12345");
defsymbol (&Qjapanese_jisx0208, "japanese-jisx0208");
- defsymbol (&Qjapanese_jisx0208_1990, "japanese-jisx0208-1990");
defsymbol (&Qkorean_ksc5601, "korean-ksc5601");
defsymbol (&Qjapanese_jisx0212, "japanese-jisx0212");
defsymbol (&Qchinese_cns11643_1, "chinese-cns11643-1");
defsymbol (&Qchinese_cns11643_2, "chinese-cns11643-2");
#ifdef UTF2000
+ defsymbol (&Qmap_ucs, "=ucs");
defsymbol (&Qucs, "ucs");
defsymbol (&Qucs_bmp, "ucs-bmp");
defsymbol (&Qucs_smp, "ucs-smp");
defsymbol (&Qucs_sip, "ucs-sip");
- defsymbol (&Qucs_cns, "ucs-cns");
- defsymbol (&Qucs_jis, "ucs-jis");
- defsymbol (&Qucs_ks, "ucs-ks");
- defsymbol (&Qucs_big5, "ucs-big5");
defsymbol (&Qlatin_viscii, "latin-viscii");
defsymbol (&Qlatin_tcvn5712, "latin-tcvn5712");
defsymbol (&Qlatin_viscii_lower, "latin-viscii-lower");
defsymbol (&Qlatin_viscii_upper, "latin-viscii-upper");
defsymbol (&Qvietnamese_viscii_lower, "vietnamese-viscii-lower");
defsymbol (&Qvietnamese_viscii_upper, "vietnamese-viscii-upper");
- defsymbol (&Qideograph_gt, "ideograph-gt");
- defsymbol (&Qideograph_gt_pj_1, "ideograph-gt-pj-1");
- defsymbol (&Qideograph_gt_pj_2, "ideograph-gt-pj-2");
- defsymbol (&Qideograph_gt_pj_3, "ideograph-gt-pj-3");
- defsymbol (&Qideograph_gt_pj_4, "ideograph-gt-pj-4");
- defsymbol (&Qideograph_gt_pj_5, "ideograph-gt-pj-5");
- defsymbol (&Qideograph_gt_pj_6, "ideograph-gt-pj-6");
- defsymbol (&Qideograph_gt_pj_7, "ideograph-gt-pj-7");
- defsymbol (&Qideograph_gt_pj_8, "ideograph-gt-pj-8");
- defsymbol (&Qideograph_gt_pj_9, "ideograph-gt-pj-9");
- defsymbol (&Qideograph_gt_pj_10, "ideograph-gt-pj-10");
- defsymbol (&Qideograph_gt_pj_11, "ideograph-gt-pj-11");
- defsymbol (&Qideograph_daikanwa_2, "ideograph-daikanwa-2");
- defsymbol (&Qideograph_daikanwa, "ideograph-daikanwa");
+ defsymbol (&Qmap_jis_x0208, "=jis-x0208");
+ defsymbol (&Qmap_jis_x0208_1990, "=jis-x0208-1990");
defsymbol (&Qchinese_big5, "chinese-big5");
- /* defsymbol (&Qchinese_big5_cdp, "chinese-big5-cdp"); */
- defsymbol (&Qideograph_hanziku_1, "ideograph-hanziku-1");
- defsymbol (&Qideograph_hanziku_2, "ideograph-hanziku-2");
- defsymbol (&Qideograph_hanziku_3, "ideograph-hanziku-3");
- defsymbol (&Qideograph_hanziku_4, "ideograph-hanziku-4");
- defsymbol (&Qideograph_hanziku_5, "ideograph-hanziku-5");
- defsymbol (&Qideograph_hanziku_6, "ideograph-hanziku-6");
- defsymbol (&Qideograph_hanziku_7, "ideograph-hanziku-7");
- defsymbol (&Qideograph_hanziku_8, "ideograph-hanziku-8");
- defsymbol (&Qideograph_hanziku_9, "ideograph-hanziku-9");
- defsymbol (&Qideograph_hanziku_10, "ideograph-hanziku-10");
- defsymbol (&Qideograph_hanziku_11, "ideograph-hanziku-11");
- defsymbol (&Qideograph_hanziku_12, "ideograph-hanziku-12");
- defsymbol (&Qchina3_jef, "china3-jef");
- defsymbol (&Qideograph_cbeta, "ideograph-cbeta");
defsymbol (&Qethiopic_ucs, "ethiopic-ucs");
#endif
defsymbol (&Qchinese_big5_1, "chinese-big5-1");
#ifdef UTF2000
staticpro (&Vcharset_ucs);
Vcharset_ucs =
- make_charset (LEADING_BYTE_UCS, Qucs, 256, 4,
+ make_charset (LEADING_BYTE_UCS, Qmap_ucs, 256, 4,
1, 2, 0, CHARSET_LEFT_TO_RIGHT,
build_string ("UCS"),
build_string ("UCS"),
build_string ("ISO/IEC 10646"),
build_string (""),
- Qnil, 0, 0xFFFFFFF, 0, 0, Qnil, CONVERSION_IDENTICAL);
+ Qnil, 0, 0x7FFFFFFF, 0, 0, Qnil, CONVERSION_IDENTICAL);
staticpro (&Vcharset_ucs_bmp);
Vcharset_ucs_bmp =
make_charset (LEADING_BYTE_UCS_BMP, Qucs_bmp, 256, 2,
build_string ("\\(ISO10646.*-2\\|UCS00-2\\)"),
Qnil, MIN_CHAR_SIP, MAX_CHAR_SIP,
MIN_CHAR_SIP, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ucs_cns);
- Vcharset_ucs_cns =
- make_charset (LEADING_BYTE_UCS_CNS, Qucs_cns, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("UCS for CNS"),
- build_string ("UCS for CNS 11643"),
- build_string ("ISO/IEC 10646 for CNS 11643"),
- build_string (""),
- Qnil, 0, 0, 0, 0,
- Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ucs_jis);
- Vcharset_ucs_jis =
- make_charset (LEADING_BYTE_UCS_JIS, Qucs_jis, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("UCS for JIS"),
- build_string ("UCS for JIS X 0208, 0212 and 0213"),
- build_string ("ISO/IEC 10646 for JIS X 0208, 0212 and 0213"),
- build_string (""),
- Qnil, 0, 0, 0, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ucs_ks);
- Vcharset_ucs_ks =
- make_charset (LEADING_BYTE_UCS_KS, Qucs_ks, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("UCS for KS"),
- build_string ("UCS for CCS defined by KS"),
- build_string ("ISO/IEC 10646 for Korean Standards"),
- build_string (""),
- Qnil, 0, 0, 0, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ucs_big5);
- Vcharset_ucs_big5 =
- make_charset (LEADING_BYTE_UCS_BIG5, Qucs_big5, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("UCS for Big5"),
- build_string ("UCS for Big5"),
- build_string ("ISO/IEC 10646 for Big5"),
- build_string (""),
- Qnil, 0, 0, 0, 0, Qnil, CONVERSION_IDENTICAL);
#else
# define MIN_CHAR_THAI 0
# define MAX_CHAR_THAI 0
build_string ("TIS620 (Thai)"),
build_string ("TIS620.2529 (Thai)"),
build_string ("tis620"),
- Qnil, MIN_CHAR_THAI, MAX_CHAR_THAI,
- MIN_CHAR_THAI, 32, Qnil, CONVERSION_IDENTICAL);
+ Qnil, 0, 0, 0, 32, Qnil, CONVERSION_IDENTICAL);
staticpro (&Vcharset_greek_iso8859_7);
Vcharset_greek_iso8859_7 =
make_charset (LEADING_BYTE_GREEK_ISO8859_7, Qgreek_iso8859_7, 96, 1,
build_string ("ISO8859-9 (Latin-5)"),
build_string ("iso8859-9"),
Qnil, 0, 0, 0, 32, Qnil, CONVERSION_IDENTICAL);
+#ifdef UTF2000
+ staticpro (&Vcharset_jis_x0208);
+ Vcharset_jis_x0208 =
+ make_charset (LEADING_BYTE_JIS_X0208,
+ Qmap_jis_x0208, 94, 2,
+ 2, 0, 'B', CHARSET_LEFT_TO_RIGHT,
+ build_string ("JIS X0208"),
+ build_string ("JIS X0208 Common"),
+ build_string ("JIS X0208 Common part"),
+ build_string ("jisx0208\\.1990"),
+ Qnil,
+ MIN_CHAR_JIS_X0208_1990,
+ MAX_CHAR_JIS_X0208_1990, MIN_CHAR_JIS_X0208_1990, 33,
+ Qnil, CONVERSION_94x94);
+#endif
staticpro (&Vcharset_japanese_jisx0208_1978);
Vcharset_japanese_jisx0208_1978 =
make_charset (LEADING_BYTE_JAPANESE_JISX0208_1978,
build_string
("JIS X0208:1978 Japanese Kanji (so called \"old JIS\")"),
build_string ("\\(jisx0208\\|jisc6226\\)\\.1978"),
- Qnil, 0, 0, 0, 33, Qnil, CONVERSION_IDENTICAL);
+ Qnil, 0, 0, 0, 33,
+#ifdef UTF2000
+ Vcharset_jis_x0208,
+#else
+ Qnil,
+#endif
+ CONVERSION_IDENTICAL);
staticpro (&Vcharset_chinese_gb2312);
Vcharset_chinese_gb2312 =
make_charset (LEADING_BYTE_CHINESE_GB2312, Qchinese_gb2312, 94, 2,
build_string ("JIS X0208:1983 (Japanese)"),
build_string ("JIS X0208:1983 Japanese Kanji"),
build_string ("jisx0208\\.1983"),
- Qnil, 0, 0, 0, 33, Qnil, CONVERSION_IDENTICAL);
+ Qnil, 0, 0, 0, 33,
+#ifdef UTF2000
+ Vcharset_jis_x0208,
+#else
+ Qnil,
+#endif
+ CONVERSION_IDENTICAL);
#ifdef UTF2000
staticpro (&Vcharset_japanese_jisx0208_1990);
Vcharset_japanese_jisx0208_1990 =
make_charset (LEADING_BYTE_JAPANESE_JISX0208_1990,
- Qjapanese_jisx0208_1990, 94, 2,
+ Qmap_jis_x0208_1990, 94, 2,
2, 0, 0, CHARSET_LEFT_TO_RIGHT,
build_string ("JISX0208-1990"),
build_string ("JIS X0208:1990 (Japanese)"),
build_string ("JIS X0208:1990 Japanese Kanji"),
build_string ("jisx0208\\.1990"),
Qnil,
- MIN_CHAR_JIS_X0208_1990,
- MAX_CHAR_JIS_X0208_1990, MIN_CHAR_JIS_X0208_1990, 33,
- Qnil, CONVERSION_IDENTICAL);
+ 0x2121 /* MIN_CHAR_JIS_X0208_1990 */,
+ 0x7426 /* MAX_CHAR_JIS_X0208_1990 */,
+ 0 /* MIN_CHAR_JIS_X0208_1990 */, 33,
+ Vcharset_jis_x0208 /* Qnil */,
+ CONVERSION_IDENTICAL /* CONVERSION_94x94 */);
#endif
staticpro (&Vcharset_korean_ksc5601);
Vcharset_korean_ksc5601 =
build_string ("Big5"),
build_string ("Big5"),
build_string ("Big5 Chinese traditional"),
- build_string ("big5"),
+ build_string ("big5-0"),
Qnil,
- 0 /* MIN_CHAR_BIG5_CDP */,
- 0 /* MAX_CHAR_BIG5_CDP */, 0, 0,
- Qnil, CONVERSION_IDENTICAL);
-#if 0
- staticpro (&Vcharset_chinese_big5_cdp);
- Vcharset_chinese_big5_cdp =
- make_charset (LEADING_BYTE_CHINESE_BIG5_CDP, Qchinese_big5_cdp, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("Big5-CDP"),
- build_string ("Big5 + CDP extension"),
- build_string ("Big5 with CDP extension"),
- build_string ("big5\\.cdp-0"),
- Qnil, MIN_CHAR_BIG5_CDP, MAX_CHAR_BIG5_CDP,
+ MIN_CHAR_BIG5_CDP, MAX_CHAR_BIG5_CDP,
MIN_CHAR_BIG5_CDP, 0, Qnil, CONVERSION_IDENTICAL);
-#endif
-#define DEF_HANZIKU(n) \
- staticpro (&Vcharset_ideograph_hanziku_##n); \
- Vcharset_ideograph_hanziku_##n = \
- make_charset (LEADING_BYTE_HANZIKU_##n, Qideograph_hanziku_##n, 256, 2, \
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT, \
- build_string ("HZK-"#n), \
- build_string ("HANZIKU-"#n), \
- build_string ("HANZIKU (pseudo BIG5 encoding) part "#n), \
- build_string \
- ("hanziku-"#n"$"), \
- Qnil, MIN_CHAR_HANZIKU_##n, MAX_CHAR_HANZIKU_##n, \
- MIN_CHAR_HANZIKU_##n, 0, Qnil, CONVERSION_IDENTICAL);
- DEF_HANZIKU (1);
- DEF_HANZIKU (2);
- DEF_HANZIKU (3);
- DEF_HANZIKU (4);
- DEF_HANZIKU (5);
- DEF_HANZIKU (6);
- DEF_HANZIKU (7);
- DEF_HANZIKU (8);
- DEF_HANZIKU (9);
- DEF_HANZIKU (10);
- DEF_HANZIKU (11);
- DEF_HANZIKU (12);
- staticpro (&Vcharset_china3_jef);
- Vcharset_china3_jef =
- make_charset (LEADING_BYTE_CHINA3_JEF, Qchina3_jef, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("JC3"),
- build_string ("JEF + CHINA3"),
- build_string ("JEF + CHINA3 private characters"),
- build_string ("china3jef-0"),
- Qnil, MIN_CHAR_CHINA3_JEF, MAX_CHAR_CHINA3_JEF,
- MIN_CHAR_CHINA3_JEF, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ideograph_cbeta);
- Vcharset_ideograph_cbeta =
- make_charset (LEADING_BYTE_CBETA, Qideograph_cbeta, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("CB"),
- build_string ("CBETA"),
- build_string ("CBETA private characters"),
- build_string ("cbeta-0"),
- Qnil, MIN_CHAR_CBETA, MAX_CHAR_CBETA,
- MIN_CHAR_CBETA, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ideograph_gt);
- Vcharset_ideograph_gt =
- make_charset (LEADING_BYTE_GT, Qideograph_gt, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("GT"),
- build_string ("GT"),
- build_string ("GT"),
- build_string (""),
- Qnil, MIN_CHAR_GT, MAX_CHAR_GT,
- MIN_CHAR_GT, 0, Qnil, CONVERSION_IDENTICAL);
-#define DEF_GT_PJ(n) \
- staticpro (&Vcharset_ideograph_gt_pj_##n); \
- Vcharset_ideograph_gt_pj_##n = \
- make_charset (LEADING_BYTE_GT_PJ_##n, Qideograph_gt_pj_##n, 94, 2, \
- 2, 0, 0, CHARSET_LEFT_TO_RIGHT, \
- build_string ("GT-PJ-"#n), \
- build_string ("GT (pseudo JIS encoding) part "#n), \
- build_string ("GT 2000 (pseudo JIS encoding) part "#n), \
- build_string \
- ("\\(GTpj-"#n "\\|jisx0208\\.GT-"#n "\\)$"), \
- Qnil, 0, 0, 0, 33, Qnil, CONVERSION_IDENTICAL);
- DEF_GT_PJ (1);
- DEF_GT_PJ (2);
- DEF_GT_PJ (3);
- DEF_GT_PJ (4);
- DEF_GT_PJ (5);
- DEF_GT_PJ (6);
- DEF_GT_PJ (7);
- DEF_GT_PJ (8);
- DEF_GT_PJ (9);
- DEF_GT_PJ (10);
- DEF_GT_PJ (11);
-
- staticpro (&Vcharset_ideograph_daikanwa_2);
- Vcharset_ideograph_daikanwa_2 =
- make_charset (LEADING_BYTE_DAIKANWA_2, Qideograph_daikanwa_2, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("Daikanwa Rev."),
- build_string ("Morohashi's Daikanwa Rev."),
- build_string
- ("Daikanwa dictionary (revised version)"),
- build_string ("Daikanwa\\(\\.[0-9]+\\)?-2"),
- Qnil, 0, 0, 0, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ideograph_daikanwa);
- Vcharset_ideograph_daikanwa =
- make_charset (LEADING_BYTE_DAIKANWA_3, Qideograph_daikanwa, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("Daikanwa"),
- build_string ("Morohashi's Daikanwa Rev.2"),
- build_string
- ("Daikanwa dictionary (second revised version)"),
- build_string ("Daikanwa\\(\\.[0-9]+\\)?-3"),
- Qnil, MIN_CHAR_DAIKANWA, MAX_CHAR_DAIKANWA,
- MIN_CHAR_DAIKANWA, 0, Qnil, CONVERSION_IDENTICAL);
staticpro (&Vcharset_ethiopic_ucs);
Vcharset_ethiopic_ucs =
projects / chise / xemacs-chise.git- / blobdiff