/* 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_gb;
-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_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_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
Qchinese_gb2312,
Qchinese_gb12345,
Qjapanese_jisx0208,
- Qjapanese_jisx0208_1990,
Qkorean_ksc5601,
Qjapanese_jisx0212,
Qchinese_cns11643_1,
Qucs_bmp,
Qucs_smp,
Qucs_sip,
- Qucs_gb,
- Qucs_cns,
- Qucs_jis,
- Qucs_ks,
- Qucs_big5,
Qlatin_viscii,
Qlatin_tcvn5712,
Qlatin_viscii_lower,
Qlatin_viscii_upper,
Qvietnamese_viscii_lower,
Qvietnamese_viscii_upper,
- Qjis_x0208,
+ 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,
- 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,
}
if (CHARP (decoding_table))
return XCHAR (decoding_table);
-#ifdef HAVE_DATABASE
+#ifdef HAVE_CHISE_CLIENT
if (EQ (decoding_table, Qunloaded))
{
char_id = load_char_decoding_entry_maybe (ccs, code_point);
*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++)
{
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 (&Qucs_bmp, "ucs-bmp");
defsymbol (&Qucs_smp, "ucs-smp");
defsymbol (&Qucs_sip, "ucs-sip");
- defsymbol (&Qucs_gb, "ucs-gb");
- 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 (&Qjis_x0208, "=jis-x0208");
- 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 (&Qideograph_cbeta, "ideograph-cbeta");
defsymbol (&Qethiopic_ucs, "ethiopic-ucs");
#endif
defsymbol (&Qchinese_big5_1, "chinese-big5-1");
build_string ("\\(ISO10646.*-2\\|UCS00-2\\)"),
Qnil, MIN_CHAR_SIP, MAX_CHAR_SIP,
MIN_CHAR_SIP, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ucs_gb);
- Vcharset_ucs_gb =
- make_charset (LEADING_BYTE_UCS_GB, Qucs_gb, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("UCS for GB"),
- build_string ("UCS for GB"),
- build_string ("ISO/IEC 10646 for GB"),
- build_string (""),
- Qnil, 0, 0, 0, 0, Vcharset_ucs, 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, Vcharset_ucs, 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, Vcharset_ucs, 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, Vcharset_ucs, 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, Vcharset_ucs, CONVERSION_IDENTICAL);
#else
# define MIN_CHAR_THAI 0
# define MAX_CHAR_THAI 0
staticpro (&Vcharset_jis_x0208);
Vcharset_jis_x0208 =
make_charset (LEADING_BYTE_JIS_X0208,
- Qjis_x0208, 94, 2,
+ Qmap_jis_x0208, 94, 2,
2, 0, 'B', CHARSET_LEFT_TO_RIGHT,
build_string ("JIS X0208"),
build_string ("JIS X0208 Common"),
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)"),
Qnil,
MIN_CHAR_BIG5_CDP, MAX_CHAR_BIG5_CDP,
MIN_CHAR_BIG5_CDP, 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, 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"$"), \
- Qnil, MIN_CHAR_HANZIKU_##n, MAX_CHAR_HANZIKU_##n, \
- MIN_CHAR_HANZIKU_##n, 0, Qnil, CONVERSION_IDENTICAL);
- DEF_HANZIKU (1);
- DEF_HANZIKU (2);
- DEF_HANZIKU (3);
- DEF_HANZIKU (4);
- DEF_HANZIKU (5);
- DEF_HANZIKU (6);
- DEF_HANZIKU (7);
- DEF_HANZIKU (8);
- DEF_HANZIKU (9);
- DEF_HANZIKU (10);
- DEF_HANZIKU (11);
- DEF_HANZIKU (12);
- staticpro (&Vcharset_ideograph_cbeta);
- Vcharset_ideograph_cbeta =
- make_charset (LEADING_BYTE_CBETA, Qideograph_cbeta, 256, 2,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("CB"),
- build_string ("CBETA"),
- build_string ("CBETA private characters"),
- build_string ("cbeta-0"),
- Qnil, MIN_CHAR_CBETA, MAX_CHAR_CBETA,
- MIN_CHAR_CBETA, 0, Qnil, CONVERSION_IDENTICAL);
- staticpro (&Vcharset_ideograph_gt);
- Vcharset_ideograph_gt =
- make_charset (LEADING_BYTE_GT, Qideograph_gt, 256, 3,
- 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
- build_string ("GT"),
- build_string ("GT"),
- build_string ("GT"),
- build_string (""),
- Qnil, MIN_CHAR_GT, MAX_CHAR_GT,
- MIN_CHAR_GT, 0, Qnil, CONVERSION_IDENTICAL);
-#define DEF_GT_PJ(n) \
- staticpro (&Vcharset_ideograph_gt_pj_##n); \
- Vcharset_ideograph_gt_pj_##n = \
- make_charset (LEADING_BYTE_GT_PJ_##n, Qideograph_gt_pj_##n, 94, 2, \
- 2, 0, 0, CHARSET_LEFT_TO_RIGHT, \
- build_string ("GT-PJ-"#n), \
- build_string ("GT (pseudo JIS encoding) part "#n), \
- build_string ("GT 2000 (pseudo JIS encoding) part "#n), \
- build_string \
- ("\\(GTpj-"#n "\\|jisx0208\\.GT-"#n "\\)$"), \
- 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