update.

[chise/xemacs-chise.git.1] / src / text-coding.c

diff --git a/src/text-coding.c b/src/text-coding.c
index 79c9dcb..6df01da 100644 (file)
--- a/src/text-coding.c
+++ b/src/text-coding.c
@@ -1,7 +1,8 @@
 /* Code conversion functions.
    Copyright (C) 1991, 1995 Free Software Foundation, Inc.
    Copyright (C) 1995 Sun Microsystems, Inc.
-   Copyright (C) 1999,2000,2001,2002,2003,2004,2005,2008 MORIOKA Tomohiko
+   Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008, 2011,
+     2012 MORIOKA Tomohiko
 
 This file is part of XEmacs.
 
@@ -106,10 +107,11 @@ Lisp_Object Qshort, Qno_ascii_eol, Qno_ascii_cntl, Qseven, Qlock_shift;
 #endif
 #ifdef UTF2000
 Lisp_Object Qutf_8_mcs;
-Lisp_Object Qdisable_composition;
+Lisp_Object Qdisable_composition, Qenable_decomposition;
 Lisp_Object Qccs_priority_list;
 Lisp_Object Quse_entity_reference;
 Lisp_Object Qd, Qx, QX;
+Lisp_Object Vdecomposition_feature_list;
 #endif
 Lisp_Object Qencode, Qdecode;
 
@@ -847,8 +849,10 @@ character set.  Recognized properties are:
                `coding-system-property' will return nil.)
 
 'disable-composition
-     If non-nil, composition/decomposition for combining characters
-     are disabled.
+     If non-nil, composition for combining characters is disabled.
+
+'enable-decomposition
+     If non-nil, decomposition for combining characters is enabled.
 
 'use-entity-reference
      If non-nil, SGML style entity-reference is used for non-system-characters.
@@ -1005,6 +1009,8 @@ if TYPE is 'ccl:
 #ifdef UTF2000
        else if (EQ (key, Qdisable_composition))
          CODING_SYSTEM_DISABLE_COMPOSITION (codesys) = !NILP (value);
+       else if (EQ (key, Qenable_decomposition))
+         CODING_SYSTEM_ENABLE_DECOMPOSITION (codesys) = !NILP (value);
        else if (EQ (key, Quse_entity_reference))
          CODING_SYSTEM_USE_ENTITY_REFERENCE (codesys) = !NILP (value);
 #endif
@@ -1489,6 +1495,8 @@ Return the PROP property of CODING-SYSTEM.
 #ifdef UTF2000
   else if (EQ (prop, Qdisable_composition))
     return XCODING_SYSTEM_DISABLE_COMPOSITION (coding_system) ? Qt : Qnil;
+  else if (EQ (prop, Qenable_decomposition))
+    return XCODING_SYSTEM_ENABLE_DECOMPOSITION (coding_system) ? Qt : Qnil;
   else if (EQ (prop, Quse_entity_reference))
     return XCODING_SYSTEM_USE_ENTITY_REFERENCE (coding_system) ? Qt : Qnil;
   else if (EQ (prop, Qccs_priority_list))
@@ -2195,8 +2203,8 @@ do {                                              \
     }                                          \
 } while (0)
 
-INLINE_HEADER void DECODE_ADD_UCS_CHAR(Emchar c, unsigned_char_dynarr* dst);
-INLINE_HEADER void
+static void DECODE_ADD_UCS_CHAR(Emchar c, unsigned_char_dynarr* dst);
+static void
 DECODE_ADD_UCS_CHAR(Emchar c, unsigned_char_dynarr* dst)
 {
   if ( c <= 0x7f )
@@ -3413,11 +3421,22 @@ char_encode_as_entity_reference (Emchar ch, char* buf)
        char_type = Qnil;
       if (!NILP (ccs = Ffind_charset (ccs)))
        {
-         int code_point = charset_code_point (ccs, ch, 0);
+         int code_point
+           = charset_code_point (ccs, ch,
+                                 NILP (char_type) ?
+                                 CHAR_ALL : CHAR_ISOLATED_ONLY );
 
          if ( (code_point >= 0)
-              && (NILP (char_type)
-                  || DECODE_CHAR (ccs, code_point, 0) != ch) )
+              && ( NILP (char_type)
+#if 1
+                   || ( charset_code_point (ccs, ch, CHAR_DEFINED_ONLY)
+                        == -1 )
+#endif
+#if 0
+                   || ( DECODE_CHAR (ccs, code_point, 0) != ch )
+#endif
+                   )
+              )
            {
              Lisp_Object ret;
 
@@ -3476,7 +3495,7 @@ char_encode_as_entity_reference (Emchar ch, char* buf)
 /************************************************************************/
 /*                          character composition                       */
 /************************************************************************/
-extern Lisp_Object Qcomposition;
+extern Lisp_Object Qcomposition, Qrep_decomposition;
 
 INLINE_HEADER void
 COMPOSE_FLUSH_CHARS (struct decoding_stream *str, unsigned_char_dynarr* dst);
@@ -4657,10 +4676,47 @@ char_encode_utf8 (struct encoding_stream *str, Emchar ch,
 
       if ( (code_point < 0) || (code_point > 0xEFFFF) )
        {
-         Lisp_Object map
-           = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (str->codesys, 1);
-         Lisp_Object ret;
+         Lisp_Object map, ret;
+
+         if ( CODING_SYSTEM_ENABLE_DECOMPOSITION (str->codesys) )
+           {
+             Lisp_Object rest = Vdecomposition_feature_list;
+             Lisp_Object decomp_f;
+             Lisp_Object seq = Qnil;
+             struct gcpro gcpro1;
+
+             while ( CONSP (rest) )
+               {
+                 decomp_f = XCAR (rest);
+                 GCPRO1 (rest);
+                 seq = Fchar_feature (make_char (ch), decomp_f, Qnil,
+                                      Qnil, Qnil);
+                 UNGCPRO;
+                 if ( !NILP (seq) )
+                   break;
+                 rest = XCDR (rest);
+               }
+
+             if ( CONSP (seq) )
+               {
+                 Lisp_Object base = Fcar (seq);
+
+                 seq = Fcdr (seq);
+                 if ( CHARP (base) && CONSP (seq) )
+                   {
+                     Lisp_Object comb = Fcar (seq);
 
+                     if ( CHARP (comb) )
+                       {
+                         char_encode_utf8 (str, XCHAR (base), dst, flags);
+                         char_encode_utf8 (str, XCHAR (comb), dst, flags);
+                         return;
+                       }
+                   }
+               }
+           }
+
+         map = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (str->codesys, 1);
          if ( !NILP (map)
               && INTP (ret = Fchar_feature (make_char (ch),
                                             map, Qnil,
@@ -6405,6 +6461,7 @@ syms_of_file_coding (void)
 #ifdef UTF2000
   defsymbol (&Qutf_8_mcs, "utf-8-mcs");
   defsymbol (&Qdisable_composition, "disable-composition");
+  defsymbol (&Qenable_decomposition, "enable-decomposition");
   defsymbol (&Qccs_priority_list, "ccs-priority-list");
   defsymbol (&Quse_entity_reference, "use-entity-reference");
   defsymbol (&Qd, "d");
@@ -6534,6 +6591,14 @@ and behaviors of various editing commands.
 Setting this to nil does not do anything.
 */ );
   enable_multibyte_characters = 1;
+
+#ifdef UTF2000
+  DEFVAR_LISP ("decomposition-feature-list",
+              &Vdecomposition_feature_list /*
+List of `=decomposition@FOO' feature to encode characters as IVS.
+*/ );
+  Vdecomposition_feature_list = Qnil;
+#endif
 }
 
 void
@@ -6587,6 +6652,7 @@ complex_vars_of_file_coding (void)
   DEFINE_CODESYS_PROP (CODESYS_PROP_CCL,     Qdecode);
 #ifdef UTF2000
   DEFINE_CODESYS_PROP (CODESYS_PROP_ALL_OK,  Qdisable_composition);
+  DEFINE_CODESYS_PROP (CODESYS_PROP_ALL_OK,  Qenable_decomposition);
   DEFINE_CODESYS_PROP (CODESYS_PROP_ALL_OK,  Quse_entity_reference);
 #endif
 #endif /* MULE */