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_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_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;
Qucs_bmp,
Qucs_smp,
Qucs_sip,
+ Qucs_gb,
Qucs_cns,
Qucs_jis,
Qucs_ks,
Qlatin_viscii_upper,
Qvietnamese_viscii_lower,
Qvietnamese_viscii_upper,
+ Qjis_x0208,
Qchinese_big5,
/* Qchinese_big5_cdp, */
Qideograph_hanziku_1,
Qideograph_hanziku_10,
Qideograph_hanziku_11,
Qideograph_hanziku_12,
- Qchina3_jef,
Qideograph_cbeta,
Qideograph_daikanwa_2,
Qideograph_daikanwa,
Lisp_Object mother = XCHARSET_MOTHER (charset);
int final;
- if ( CHARSETP (mother) && (XCHARSET_MAX_CODE (charset) > 0) )
+ if ( XCHARSET_MAX_CODE (charset) > 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 )
- {
- 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;
- }
- return
- decode_builtin_char (mother, code + XCHARSET_CODE_OFFSET(charset));
- }
-#if 0
- else if (EQ (charset, Vcharset_chinese_big5))
- {
- 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;
}
}
-#endif
- 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 ( ((max == 0) && CHARSETP (mother)) ||
- ((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;
}
#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 (&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 (&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_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
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,
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,
+ Qjis_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 =
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 =
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,
projects / chise / xemacs-chise.git- / blobdiff