update.

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

diff --git a/src/text-coding.c b/src/text-coding.c
index 4b8f2d2..79c9dcb 100644 (file)
--- a/src/text-coding.c
+++ b/src/text-coding.c
@@ -1,7 +1,7 @@
 /* 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 MORIOKA Tomohiko
+   Copyright (C) 1999,2000,2001,2002,2003,2004,2005,2008 MORIOKA Tomohiko
 
 This file is part of XEmacs.
 
@@ -107,6 +107,7 @@ Lisp_Object Qshort, Qno_ascii_eol, Qno_ascii_cntl, Qseven, Qlock_shift;
 #ifdef UTF2000
 Lisp_Object Qutf_8_mcs;
 Lisp_Object Qdisable_composition;
+Lisp_Object Qccs_priority_list;
 Lisp_Object Quse_entity_reference;
 Lisp_Object Qd, Qx, QX;
 #endif
@@ -1052,6 +1053,12 @@ if TYPE is 'ccl:
                parse_charset_conversion_specs (codesys->iso2022.output_conv,
                                                value);
              }
+#ifdef UTF2000
+           else if (EQ (key, Qccs_priority_list))
+             {
+               codesys->ccs_priority_list = value;
+             }
+#endif
            else
              signal_simple_error ("Unrecognized property", key);
          }
@@ -1484,6 +1491,8 @@ Return the PROP property of CODING-SYSTEM.
     return XCODING_SYSTEM_DISABLE_COMPOSITION (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))
+    return XCODING_SYSTEM_CCS_PRIORITY_LIST (coding_system);
 #endif
   else if (type == CODESYS_ISO2022)
     {
@@ -2311,6 +2320,7 @@ struct decoding_stream
   unsigned char counter;
 #endif
 #ifdef UTF2000
+  char bom_flag;
   unsigned char er_counter;
   unsigned char er_buf[ER_BUF_SIZE];
 
@@ -2455,6 +2465,7 @@ reset_decoding_stream (struct decoding_stream *str)
   str->counter = 0;
 #endif /* MULE */
 #ifdef UTF2000
+  str->bom_flag = 0;
   str->er_counter = 0;
   str->combined_char_count = 0;
   str->combining_table = Qnil;
@@ -3246,6 +3257,8 @@ decode_flush_er_chars (struct decoding_stream *str, unsigned_char_dynarr* dst)
     }
 }
 
+EXFUN (Fregexp_quote, 1);
+
 void decode_add_er_char (struct decoding_stream *str, Emchar character,
                         unsigned_char_dynarr* dst);
 void
@@ -3267,7 +3280,7 @@ decode_add_er_char (struct decoding_stream *str, Emchar c,
     {
       Lisp_Object string = make_string (str->er_buf,
                                        str->er_counter);
-      Lisp_Object rest = Vcoded_charset_entity_reference_alist;
+      Lisp_Object rest;
       Lisp_Object cell;
       Lisp_Object ret;
       Lisp_Object pat;
@@ -3275,7 +3288,8 @@ decode_add_er_char (struct decoding_stream *str, Emchar c,
       Lisp_Object char_type;
       int base;
 
-      while (!NILP (rest))
+      for ( rest = Vcoded_charset_entity_reference_alist;
+           !NILP (rest); rest = Fcdr (rest) )
        {                     
          cell = Fcar (rest);
          ccs = Fcar (cell);
@@ -3295,6 +3309,7 @@ decode_add_er_char (struct decoding_stream *str, Emchar c,
            pat = ret;
          else
            continue;
+         pat = Fregexp_quote (pat);
 
          cell = Fcdr (cell);
          cell = Fcdr (cell);
@@ -3333,10 +3348,16 @@ decode_add_er_char (struct decoding_stream *str, Emchar c,
                ? DECODE_CHAR (ccs, code, 0)
                : decode_builtin_char (ccs, code);
 
-             DECODE_ADD_UCS_CHAR (chr, dst);
+             if ( chr >= 0 )
+               DECODE_ADD_UCS_CHAR (chr, dst);
+             else
+               {
+                 Dynarr_add_many (dst, str->er_buf, str->er_counter);
+                 Dynarr_add (dst, ';');
+               }
+
              goto decoded;
            }
-         rest = Fcdr (rest);
        }
       if (!NILP (Fstring_match (build_string ("^&MCS-\\([0-9A-F]+\\)$"),
                                string, Qnil, Qnil)))
@@ -4485,6 +4506,7 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
   eol_type_t eol_type  = str->eol_type;
   unsigned char counter        = str->counter;
 #ifdef UTF2000
+  int bom_flag = str->bom_flag;
   Lisp_Object ccs
     = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (DECODING_STREAM_DATA
                                             (decoding)->codesys, 0);
@@ -4500,11 +4522,20 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
              COMPOSE_FLUSH_CHARS (str, dst);
              decode_flush_er_chars (str, dst);
              DECODE_HANDLE_EOL_TYPE (eol_type, c, flags, dst);
+
+             if ( bom_flag == 0 )
+               bom_flag = -1;
+
              DECODE_ADD_UCS_CHAR (c, dst);
            }
          else if ( c < 0xC0 )
-           /* decode_add_er_char (str, c, dst); */
-           COMPOSE_ADD_CHAR (str, c, dst);
+           {         
+             if ( bom_flag == 0 )
+               bom_flag = -1;
+
+             /* decode_add_er_char (str, c, dst); */
+             COMPOSE_ADD_CHAR (str, c, dst);
+           }
          else
            {
              /* decode_flush_er_chars (str, dst); */
@@ -4542,6 +4573,17 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
            {
              Emchar char_id;
 
+             if ( bom_flag == 0 )
+               {
+                 if ( cpos == 0xFEFF )
+                   {
+                     bom_flag = 1;
+                     goto decoded;
+                   }
+                 else
+                   bom_flag = -1;
+               }
+
              if (!NILP (ccs))
                {
                  char_id = decode_defined_char (ccs, cpos, 0);
@@ -4552,6 +4594,7 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
              else
                char_id = cpos;
              COMPOSE_ADD_CHAR (str, char_id, dst);
+           decoded:
              cpos = 0;
              counter = 0;
            }
@@ -4584,6 +4627,9 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
   str->flags   = flags;
   str->cpos    = cpos;
   str->counter = counter;
+#ifdef UTF2000
+  str->bom_flag = bom_flag;
+#endif
 }
 
 void
@@ -5970,7 +6016,19 @@ char_encode_iso2022 (struct encoding_stream *str, Emchar ch,
        {
          Lisp_Object original_default_coded_charset_priority_list
            = Vdefault_coded_charset_priority_list;
-
+         Vdefault_coded_charset_priority_list
+           = CODING_SYSTEM_CCS_PRIORITY_LIST (codesys);
+         while (!EQ (Vdefault_coded_charset_priority_list, Qnil))
+           {
+             code_point = ENCODE_CHAR (ch, charset);
+             if (XCHARSET_FINAL (charset))
+               goto found;
+             Vdefault_coded_charset_priority_list
+               = Fcdr (Fmemq (XCHARSET_NAME (charset),
+                              Vdefault_coded_charset_priority_list));
+           }
+         Vdefault_coded_charset_priority_list
+           = original_default_coded_charset_priority_list;
          while (!EQ (Vdefault_coded_charset_priority_list, Qnil))
            {
              code_point = ENCODE_CHAR (ch, charset);
@@ -6347,6 +6405,7 @@ syms_of_file_coding (void)
 #ifdef UTF2000
   defsymbol (&Qutf_8_mcs, "utf-8-mcs");
   defsymbol (&Qdisable_composition, "disable-composition");
+  defsymbol (&Qccs_priority_list, "ccs-priority-list");
   defsymbol (&Quse_entity_reference, "use-entity-reference");
   defsymbol (&Qd, "d");
   defsymbol (&Qx, "x");
@@ -6520,6 +6579,9 @@ complex_vars_of_file_coding (void)
   DEFINE_CODESYS_PROP (CODESYS_PROP_ISO2022, Qescape_quoted);
   DEFINE_CODESYS_PROP (CODESYS_PROP_ISO2022, Qinput_charset_conversion);
   DEFINE_CODESYS_PROP (CODESYS_PROP_ISO2022, Qoutput_charset_conversion);
+#ifdef UTF2000
+  DEFINE_CODESYS_PROP (CODESYS_PROP_ISO2022, Qccs_priority_list);
+#endif
 
   DEFINE_CODESYS_PROP (CODESYS_PROP_CCL,     Qencode);
   DEFINE_CODESYS_PROP (CODESYS_PROP_CCL,     Qdecode);