update.

[chise/xemacs-chise.git] / src / mule-charset.c

diff --git a/src/mule-charset.c b/src/mule-charset.c
index be5d4da..57b702b 100644 (file)
--- a/src/mule-charset.c
+++ b/src/mule-charset.c
@@ -66,6 +66,8 @@ Lisp_Object Vcharset_chinese_cns11643_2;
 #ifdef UTF2000
 Lisp_Object Vcharset_ucs;
 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;
@@ -102,6 +104,7 @@ 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_mojikyo;
 Lisp_Object Vcharset_mojikyo_2022_1;
@@ -281,7 +284,7 @@ decoding_table_put_char (Lisp_Object v, int dim, int byte_offset,
       if (dim > 0)
        {
          if (!VECTORP (nv))
-           nv = (XVECTOR_DATA(v)[i] = make_older_vector (ccs_len, Qnil));
+           nv = (XVECTOR_DATA(v)[i] = make_vector (ccs_len, Qnil));
          v = nv;
        }
       else
@@ -294,9 +297,8 @@ Lisp_Object
 put_char_ccs_code_point (Lisp_Object character,
                         Lisp_Object ccs, Lisp_Object value)
 {
-  Lisp_Object encoding_table;
-
   if (!EQ (XCHARSET_NAME (ccs), Qucs)
+      || !INTP (value)
       || (XCHAR (character) != XINT (value)))
     {
       Lisp_Object v = XCHARSET_DECODING_TABLE (ccs);
@@ -350,7 +352,7 @@ put_char_ccs_code_point (Lisp_Object character,
       if (VECTORP (v))
        {
          Lisp_Object cpos = Fget_char_attribute (character, ccs, Qnil);
-         if (!NILP (cpos))
+         if (INTP (cpos))
            {
              decoding_table_remove_char (v, dim, byte_offset, XINT (cpos));
            }
@@ -358,18 +360,12 @@ put_char_ccs_code_point (Lisp_Object character,
       else
        {
          XCHARSET_DECODING_TABLE (ccs)
-           = v = make_older_vector (ccs_len, Qnil);
+           = v = make_vector (ccs_len, Qnil);
        }
 
       decoding_table_put_char (v, dim, byte_offset, code_point, character);
     }
-  if (NILP (encoding_table = XCHARSET_ENCODING_TABLE (ccs)))
-    {
-      XCHARSET_ENCODING_TABLE (ccs)
-       = encoding_table = make_char_id_table (Qnil);
-    }
-  put_char_id_table (XCHAR_TABLE(encoding_table), character, value);
-  return Qt;
+  return value;
 }
 
 Lisp_Object
@@ -438,6 +434,8 @@ Lisp_Object Qascii,
 #ifdef UTF2000
   Qucs,
   Qucs_bmp,
+  Qucs_smp,
+  Qucs_sip,
   Qucs_cns,
   Qucs_jis,
   Qucs_ks,
@@ -464,6 +462,7 @@ Lisp_Object Qascii,
   Qideograph_hanziku_12,
   Qchina3_jef,
   Qideograph_cbeta,
+  Qideograph_daikanwa_2,
   Qideograph_daikanwa,
   Qideograph_gt,
   Qideograph_gt_pj_1,
@@ -844,8 +843,7 @@ mark_charset (Lisp_Object obj)
   mark_object (cs->registry);
   mark_object (cs->ccl_program);
 #ifdef UTF2000
-  mark_object (cs->encoding_table);
-  /* mark_object (cs->decoding_table); */
+  mark_object (cs->decoding_table);
 #endif
   return cs->name;
 }
@@ -892,7 +890,6 @@ static const struct lrecord_description charset_description[] = {
   { XD_LISP_OBJECT, offsetof (Lisp_Charset, ccl_program) },
 #ifdef UTF2000
   { XD_LISP_OBJECT, offsetof (Lisp_Charset, decoding_table) },
-  { XD_LISP_OBJECT, offsetof (Lisp_Charset, encoding_table) },
 #endif
   { XD_END }
 };
@@ -938,7 +935,6 @@ make_charset (Charset_ID id, Lisp_Object name,
   CHARSET_REVERSE_DIRECTION_CHARSET (cs) = Qnil;
 #ifdef UTF2000
   CHARSET_DECODING_TABLE(cs) = Qnil;
-  CHARSET_ENCODING_TABLE(cs) = Qnil;
   CHARSET_UCS_MIN(cs) = ucs_min;
   CHARSET_UCS_MAX(cs) = ucs_max;
   CHARSET_CODE_OFFSET(cs) = code_offset;
@@ -1254,6 +1250,16 @@ encode_builtin_char_1 (Emchar c, Lisp_Object* charset)
       *charset = Vcharset_ucs_bmp;
       return c;
     }
+  else if (c <= MAX_CHAR_SMP)
+    {
+      *charset = Vcharset_ucs_smp;
+      return c - MIN_CHAR_SMP;
+    }
+  else if (c <= MAX_CHAR_SIP)
+    {
+      *charset = Vcharset_ucs_sip;
+      return c - MIN_CHAR_SIP;
+    }
   else if (c < MIN_CHAR_DAIKANWA)
     {
       *charset = Vcharset_ucs;
@@ -1850,7 +1856,8 @@ Recognized properties are those listed in `make-charset', as well as
   if (EQ (prop, Qdimension))   return make_int (CHARSET_DIMENSION (cs));
   if (EQ (prop, Qcolumns))     return make_int (CHARSET_COLUMNS (cs));
   if (EQ (prop, Qgraphic))     return make_int (CHARSET_GRAPHIC (cs));
-  if (EQ (prop, Qfinal))       return make_char (CHARSET_FINAL (cs));
+  if (EQ (prop, Qfinal))       return CHARSET_FINAL (cs) == 0 ?
+                                Qnil : make_char (CHARSET_FINAL (cs));
   if (EQ (prop, Qchars))       return make_int (CHARSET_CHARS (cs));
   if (EQ (prop, Qregistry))    return CHARSET_REGISTRY (cs);
   if (EQ (prop, Qccl_program)) return CHARSET_CCL_PROGRAM (cs);
@@ -1941,8 +1948,6 @@ Set mapping-table of CHARSET to TABLE.
 
   if (NILP (table))
     {
-      if (VECTORP (CHARSET_DECODING_TABLE(cs)))
-       make_vector_newer (CHARSET_DECODING_TABLE(cs));
       CHARSET_DECODING_TABLE(cs) = Qnil;
       return table;
     }
@@ -1977,8 +1982,8 @@ Set mapping-table of CHARSET to TABLE.
          Lisp_Object c = XVECTOR_DATA(table)[i];
 
          if (CHARP (c))
-           put_char_ccs_code_point (c, charset,
-                                    make_int (i + byte_offset));
+           Fput_char_attribute (c, XCHARSET_NAME (charset),
+                                make_int (i + byte_offset));
        }
       break;
     case 2:
@@ -1995,16 +2000,16 @@ Set mapping-table of CHARSET to TABLE.
                  Lisp_Object c = XVECTOR_DATA(v)[j];
 
                  if (CHARP (c))
-                   put_char_ccs_code_point
-                     (c, charset,
+                   Fput_char_attribute
+                     (c, XCHARSET_NAME (charset),
                       make_int ( ( (i + byte_offset) << 8 )
                                  | (j + byte_offset)
                                  ) );
                }
            }
          else if (CHARP (v))
-           put_char_ccs_code_point (v, charset,
-                                    make_int (i + byte_offset));
+           Fput_char_attribute (v, XCHARSET_NAME (charset),
+                                make_int (i + byte_offset));
        }
       break;
     }
@@ -2381,6 +2386,8 @@ syms_of_mule_charset (void)
 #ifdef UTF2000
   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");
@@ -2403,6 +2410,7 @@ syms_of_mule_charset (void)
   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 (&Qchinese_big5,           "chinese-big5");
   defsymbol (&Qchinese_big5_cdp,       "chinese-big5-cdp");
@@ -2528,14 +2536,33 @@ complex_vars_of_mule_charset (void)
     make_charset (LEADING_BYTE_UCS_BMP, Qucs_bmp, 256, 2,
                  1, 2, 0, CHARSET_LEFT_TO_RIGHT,
                  build_string ("BMP"),
-                 build_string ("BMP"),
+                 build_string ("UCS-BMP"),
                  build_string ("ISO/IEC 10646 Group 0 Plane 0 (BMP)"),
-                 build_string ("\\(ISO10646.*-1\\|UNICODE[23]?-0\\)"),
+                 build_string
+                 ("\\(ISO10646.*-[01]\\|UCS00-0\\|UNICODE[23]?-0\\)"),
                  Qnil, 0, 0xFFFF, 0, 0);
+  staticpro (&Vcharset_ucs_smp);
+  Vcharset_ucs_smp =
+    make_charset (LEADING_BYTE_UCS_SMP, Qucs_smp, 256, 2,
+                 1, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 build_string ("SMP"),
+                 build_string ("UCS-SMP"),
+                 build_string ("ISO/IEC 10646 Group 0 Plane 1 (SMP)"),
+                 build_string ("UCS00-1"),
+                 Qnil, MIN_CHAR_SMP, MAX_CHAR_SMP, 0, 0);
+  staticpro (&Vcharset_ucs_sip);
+  Vcharset_ucs_sip =
+    make_charset (LEADING_BYTE_UCS_SIP, Qucs_sip, 256, 2,
+                 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 build_string ("SIP"),
+                 build_string ("UCS-SIP"),
+                 build_string ("ISO/IEC 10646 Group 0 Plane 2 (SIP)"),
+                 build_string ("\\(ISO10646.*-2\\|UCS00-2\\)"),
+                 Qnil, MIN_CHAR_SIP, MAX_CHAR_SIP, 0, 0);
   staticpro (&Vcharset_ucs_cns);
   Vcharset_ucs_cns =
     make_charset (LEADING_BYTE_UCS_CNS, Qucs_cns, 256, 3,
-                 1, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 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"),
@@ -2544,7 +2571,7 @@ complex_vars_of_mule_charset (void)
   staticpro (&Vcharset_ucs_jis);
   Vcharset_ucs_jis =
     make_charset (LEADING_BYTE_UCS_JIS, Qucs_jis, 256, 3,
-                 1, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 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"),
@@ -2553,7 +2580,7 @@ complex_vars_of_mule_charset (void)
   staticpro (&Vcharset_ucs_ks);
   Vcharset_ucs_ks =
     make_charset (LEADING_BYTE_UCS_KS, Qucs_ks, 256, 3,
-                 1, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 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"),
@@ -2562,7 +2589,7 @@ complex_vars_of_mule_charset (void)
   staticpro (&Vcharset_ucs_big5);
   Vcharset_ucs_big5 =
     make_charset (LEADING_BYTE_UCS_BIG5, Qucs_big5, 256, 3,
-                 1, 2, 0, CHARSET_LEFT_TO_RIGHT,
+                 2, 2, 0, CHARSET_LEFT_TO_RIGHT,
                  build_string ("UCS for Big5"),
                  build_string ("UCS for Big5"),
                  build_string ("ISO/IEC 10646 for Big5"),
@@ -2842,8 +2869,8 @@ complex_vars_of_mule_charset (void)
                  build_string ("Big5 Chinese traditional"),
                  build_string ("big5"),
                  Qnil,
-                 /* 0 */ MIN_CHAR_BIG5_CDP,
-                 /* 0 */ MAX_CHAR_BIG5_CDP, 0, 0);
+                 0 /* MIN_CHAR_BIG5_CDP */,
+                 0 /* MAX_CHAR_BIG5_CDP */, 0, 0);
   staticpro (&Vcharset_chinese_big5_cdp);
   Vcharset_chinese_big5_cdp =
     make_charset (LEADING_BYTE_CHINESE_BIG5_CDP, Qchinese_big5_cdp, 256, 2,
@@ -2926,14 +2953,25 @@ complex_vars_of_mule_charset (void)
   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);
   staticpro (&Vcharset_ideograph_daikanwa);
   Vcharset_ideograph_daikanwa =
-    make_charset (LEADING_BYTE_DAIKANWA, Qideograph_daikanwa, 256, 2,
+    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"),
-                 build_string ("Daikanwa dictionary by MOROHASHI Tetsuji"),
-                 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, 0, 0);
   staticpro (&Vcharset_mojikyo);
   Vcharset_mojikyo =