X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=src%2Ftext-coding.c;h=e9749b131195b742fd930598d163f8345ed79ccd;hb=a7018caf28cfca1998adae78b74f2d513da83838;hp=b0f987625e1e680b17f60a4349b684fcbc2dbcdf;hpb=7c800d4821261afacd6326a55941484246ad6de6;p=chise%2Fxemacs-chise.git

diff --git a/src/text-coding.c b/src/text-coding.c
index b0f9876..e9749b1 100644
--- 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 MORIOKA Tomohiko
+   Copyright (C) 1999,2000,2001 MORIOKA Tomohiko
 
 This file is part of XEmacs.
 
@@ -47,17 +47,17 @@ Lisp_Object Vcoding_system_for_write;
 Lisp_Object Vfile_name_coding_system;
 
 /* Table of symbols identifying each coding category. */
-Lisp_Object coding_category_symbol[CODING_CATEGORY_LAST + 1];
+Lisp_Object coding_category_symbol[CODING_CATEGORY_LAST];
 
 
 
 struct file_coding_dump {
   /* Coding system currently associated with each coding category. */
-  Lisp_Object coding_category_system[CODING_CATEGORY_LAST + 1];
+  Lisp_Object coding_category_system[CODING_CATEGORY_LAST];
 
   /* Table of all coding categories in decreasing order of priority.
      This describes a permutation of the possible coding categories. */
-  int coding_category_by_priority[CODING_CATEGORY_LAST + 1];
+  int coding_category_by_priority[CODING_CATEGORY_LAST];
 
 #if defined(MULE) && !defined(UTF2000)
   Lisp_Object ucs_to_mule_table[65536];
@@ -65,7 +65,7 @@ struct file_coding_dump {
 } *fcd;
 
 static const struct lrecord_description fcd_description_1[] = {
-  { XD_LISP_OBJECT_ARRAY, offsetof (struct file_coding_dump, coding_category_system), CODING_CATEGORY_LAST + 1 },
+  { XD_LISP_OBJECT_ARRAY, offsetof (struct file_coding_dump, coding_category_system), CODING_CATEGORY_LAST },
 #if defined(MULE) && !defined(UTF2000)
   { XD_LISP_OBJECT_ARRAY, offsetof (struct file_coding_dump, ucs_to_mule_table), countof (fcd->ucs_to_mule_table) },
 #endif
@@ -197,8 +197,10 @@ static int detect_coding_big5 (struct detection_state *st,
 			       const Extbyte *src, size_t n);
 static void decode_coding_big5 (Lstream *decoding, const Extbyte *src,
 				unsigned_char_dynarr *dst, size_t n);
-static void encode_coding_big5 (Lstream *encoding, const Bufbyte *src,
-				unsigned_char_dynarr *dst, size_t n);
+void char_encode_big5 (struct encoding_stream *str, Emchar c,
+		       unsigned_char_dynarr *dst, unsigned int *flags);
+void char_finish_big5 (struct encoding_stream *str,
+		       unsigned_char_dynarr *dst, unsigned int *flags);
 
 static int detect_coding_ucs4 (struct detection_state *st,
 			       const Extbyte *src, size_t n);
@@ -887,7 +889,6 @@ if TYPE is 'ccl:
        (name, type, doc_string, props))
 {
   Lisp_Coding_System *codesys;
-  Lisp_Object rest, key, value;
   enum coding_system_type ty;
   int need_to_setup_eol_systems = 1;
 
@@ -919,122 +920,124 @@ if TYPE is 'ccl:
     CHECK_STRING (doc_string);
   CODING_SYSTEM_DOC_STRING (codesys) = doc_string;
 
-  EXTERNAL_PROPERTY_LIST_LOOP (rest, key, value, props)
-    {
-      if (EQ (key, Qmnemonic))
-	{
-          if (!NILP (value))
-	    CHECK_STRING (value);
-	  CODING_SYSTEM_MNEMONIC (codesys) = value;
-	}
+  {
+    EXTERNAL_PROPERTY_LIST_LOOP_3 (key, value, props)
+      {
+	if (EQ (key, Qmnemonic))
+	  {
+	    if (!NILP (value))
+	      CHECK_STRING (value);
+	    CODING_SYSTEM_MNEMONIC (codesys) = value;
+	  }
 
-      else if (EQ (key, Qeol_type))
-	{
-	  need_to_setup_eol_systems = NILP (value);
-	  if (EQ (value, Qt))
-	    value = Qnil;
-	  CODING_SYSTEM_EOL_TYPE (codesys) = symbol_to_eol_type (value);
-	}
+	else if (EQ (key, Qeol_type))
+	  {
+	    need_to_setup_eol_systems = NILP (value);
+	    if (EQ (value, Qt))
+	      value = Qnil;
+	    CODING_SYSTEM_EOL_TYPE (codesys) = symbol_to_eol_type (value);
+	  }
 
-      else if (EQ (key, Qpost_read_conversion))
-	CODING_SYSTEM_POST_READ_CONVERSION (codesys) = value;
-      else if (EQ (key, Qpre_write_conversion))
-	CODING_SYSTEM_PRE_WRITE_CONVERSION (codesys) = value;
+	else if (EQ (key, Qpost_read_conversion))
+	  CODING_SYSTEM_POST_READ_CONVERSION (codesys) = value;
+	else if (EQ (key, Qpre_write_conversion))
+	  CODING_SYSTEM_PRE_WRITE_CONVERSION (codesys) = value;
 #ifdef UTF2000
-      else if (EQ (key, Qdisable_composition))
-	CODING_SYSTEM_DISABLE_COMPOSITION (codesys) = !NILP (value);
+	else if (EQ (key, Qdisable_composition))
+	  CODING_SYSTEM_DISABLE_COMPOSITION (codesys) = !NILP (value);
 #endif
 #ifdef MULE
-      else if (ty == CODESYS_ISO2022)
-	{
+	else if (ty == CODESYS_ISO2022)
+	  {
 #define FROB_INITIAL_CHARSET(charset_num) \
   CODING_SYSTEM_ISO2022_INITIAL_CHARSET (codesys, charset_num) = \
     ((EQ (value, Qt) || EQ (value, Qnil)) ? value : Fget_charset (value))
 
-	  if      (EQ (key, Qcharset_g0)) FROB_INITIAL_CHARSET (0);
-	  else if (EQ (key, Qcharset_g1)) FROB_INITIAL_CHARSET (1);
-	  else if (EQ (key, Qcharset_g2)) FROB_INITIAL_CHARSET (2);
-	  else if (EQ (key, Qcharset_g3)) FROB_INITIAL_CHARSET (3);
+	    if      (EQ (key, Qcharset_g0)) FROB_INITIAL_CHARSET (0);
+	    else if (EQ (key, Qcharset_g1)) FROB_INITIAL_CHARSET (1);
+	    else if (EQ (key, Qcharset_g2)) FROB_INITIAL_CHARSET (2);
+	    else if (EQ (key, Qcharset_g3)) FROB_INITIAL_CHARSET (3);
 
 #define FROB_FORCE_CHARSET(charset_num) \
   CODING_SYSTEM_ISO2022_FORCE_CHARSET_ON_OUTPUT (codesys, charset_num) = !NILP (value)
 
-	  else if (EQ (key, Qforce_g0_on_output)) FROB_FORCE_CHARSET (0);
-	  else if (EQ (key, Qforce_g1_on_output)) FROB_FORCE_CHARSET (1);
-	  else if (EQ (key, Qforce_g2_on_output)) FROB_FORCE_CHARSET (2);
-	  else if (EQ (key, Qforce_g3_on_output)) FROB_FORCE_CHARSET (3);
+	    else if (EQ (key, Qforce_g0_on_output)) FROB_FORCE_CHARSET (0);
+	    else if (EQ (key, Qforce_g1_on_output)) FROB_FORCE_CHARSET (1);
+	    else if (EQ (key, Qforce_g2_on_output)) FROB_FORCE_CHARSET (2);
+	    else if (EQ (key, Qforce_g3_on_output)) FROB_FORCE_CHARSET (3);
 
 #define FROB_BOOLEAN_PROPERTY(prop) \
   CODING_SYSTEM_ISO2022_##prop (codesys) = !NILP (value)
 
-	  else if (EQ (key, Qshort))         FROB_BOOLEAN_PROPERTY (SHORT);
-	  else if (EQ (key, Qno_ascii_eol))  FROB_BOOLEAN_PROPERTY (NO_ASCII_EOL);
-	  else if (EQ (key, Qno_ascii_cntl)) FROB_BOOLEAN_PROPERTY (NO_ASCII_CNTL);
-	  else if (EQ (key, Qseven))         FROB_BOOLEAN_PROPERTY (SEVEN);
-	  else if (EQ (key, Qlock_shift))    FROB_BOOLEAN_PROPERTY (LOCK_SHIFT);
-	  else if (EQ (key, Qno_iso6429))    FROB_BOOLEAN_PROPERTY (NO_ISO6429);
-	  else if (EQ (key, Qescape_quoted)) FROB_BOOLEAN_PROPERTY (ESCAPE_QUOTED);
+	    else if (EQ (key, Qshort))         FROB_BOOLEAN_PROPERTY (SHORT);
+	    else if (EQ (key, Qno_ascii_eol))  FROB_BOOLEAN_PROPERTY (NO_ASCII_EOL);
+	    else if (EQ (key, Qno_ascii_cntl)) FROB_BOOLEAN_PROPERTY (NO_ASCII_CNTL);
+	    else if (EQ (key, Qseven))         FROB_BOOLEAN_PROPERTY (SEVEN);
+	    else if (EQ (key, Qlock_shift))    FROB_BOOLEAN_PROPERTY (LOCK_SHIFT);
+	    else if (EQ (key, Qno_iso6429))    FROB_BOOLEAN_PROPERTY (NO_ISO6429);
+	    else if (EQ (key, Qescape_quoted)) FROB_BOOLEAN_PROPERTY (ESCAPE_QUOTED);
 
-	  else if (EQ (key, Qinput_charset_conversion))
-	    {
-	      codesys->iso2022.input_conv =
-		Dynarr_new (charset_conversion_spec);
-	      parse_charset_conversion_specs (codesys->iso2022.input_conv,
-					      value);
-	    }
-	  else if (EQ (key, Qoutput_charset_conversion))
-	    {
-	      codesys->iso2022.output_conv =
-		Dynarr_new (charset_conversion_spec);
-	      parse_charset_conversion_specs (codesys->iso2022.output_conv,
-					      value);
-	    }
-	  else
-	    signal_simple_error ("Unrecognized property", key);
-	}
-      else if (EQ (type, Qccl))
-	{
-	  Lisp_Object sym;
-	  struct ccl_program test_ccl;
-	  Extbyte *suffix;
-
-	  /* Check key first.  */
-	  if (EQ (key, Qdecode))
-	    suffix = "-ccl-decode";
-	  else if (EQ (key, Qencode))
-	    suffix = "-ccl-encode";
-	  else
-	    signal_simple_error ("Unrecognized property", key);
+	    else if (EQ (key, Qinput_charset_conversion))
+	      {
+		codesys->iso2022.input_conv =
+		  Dynarr_new (charset_conversion_spec);
+		parse_charset_conversion_specs (codesys->iso2022.input_conv,
+						value);
+	      }
+	    else if (EQ (key, Qoutput_charset_conversion))
+	      {
+		codesys->iso2022.output_conv =
+		  Dynarr_new (charset_conversion_spec);
+		parse_charset_conversion_specs (codesys->iso2022.output_conv,
+						value);
+	      }
+	    else
+	      signal_simple_error ("Unrecognized property", key);
+	  }
+	else if (EQ (type, Qccl))
+	  {
+	    Lisp_Object sym;
+	    struct ccl_program test_ccl;
+	    Extbyte *suffix;
+
+	    /* Check key first.  */
+	    if (EQ (key, Qdecode))
+	      suffix = "-ccl-decode";
+	    else if (EQ (key, Qencode))
+	      suffix = "-ccl-encode";
+	    else
+	      signal_simple_error ("Unrecognized property", key);
 
-	  /* If value is vector, register it as a ccl program
-	     associated with an newly created symbol for
-	     backward compatibility.  */
-	  if (VECTORP (value))
-	    {
-	      sym = Fintern (concat2 (Fsymbol_name (name),
-				      build_string (suffix)),
-			     Qnil);
-	      Fregister_ccl_program (sym, value);
-	    }
-	  else
-	    {
-	      CHECK_SYMBOL (value);
-	      sym = value;
-	    }
-	  /* check if the given ccl programs are valid.  */
-	  if (setup_ccl_program (&test_ccl, sym) < 0)
-	    signal_simple_error ("Invalid CCL program", value);
+	    /* If value is vector, register it as a ccl program
+	       associated with an newly created symbol for
+	       backward compatibility.  */
+	    if (VECTORP (value))
+	      {
+		sym = Fintern (concat2 (Fsymbol_name (name),
+					build_string (suffix)),
+			       Qnil);
+		Fregister_ccl_program (sym, value);
+	      }
+	    else
+	      {
+		CHECK_SYMBOL (value);
+		sym = value;
+	      }
+	    /* check if the given ccl programs are valid.  */
+	    if (setup_ccl_program (&test_ccl, sym) < 0)
+	      signal_simple_error ("Invalid CCL program", value);
 
-	  if (EQ (key, Qdecode))
-	    CODING_SYSTEM_CCL_DECODE (codesys) = sym;
-	  else if (EQ (key, Qencode))
-	    CODING_SYSTEM_CCL_ENCODE (codesys) = sym;
+	    if (EQ (key, Qdecode))
+	      CODING_SYSTEM_CCL_DECODE (codesys) = sym;
+	    else if (EQ (key, Qencode))
+	      CODING_SYSTEM_CCL_ENCODE (codesys) = sym;
 
-	}
+	  }
 #endif /* MULE */
-      else
-	signal_simple_error ("Unrecognized property", key);
-    }
+	else
+	  signal_simple_error ("Unrecognized property", key);
+      }
+  }
 
   if (need_to_setup_eol_systems)
     setup_eol_coding_systems (codesys);
@@ -1464,7 +1467,7 @@ decode_coding_category (Lisp_Object symbol)
   int i;
 
   CHECK_SYMBOL (symbol);
-  for (i = 0; i <= CODING_CATEGORY_LAST; i++)
+  for (i = 0; i < CODING_CATEGORY_LAST; i++)
     if (EQ (coding_category_symbol[i], symbol))
       return i;
 
@@ -1480,7 +1483,7 @@ Return a list of all recognized coding categories.
   int i;
   Lisp_Object list = Qnil;
 
-  for (i = CODING_CATEGORY_LAST; i >= 0; i--)
+  for (i = CODING_CATEGORY_LAST - 1; i >= 0; i--)
     list = Fcons (coding_category_symbol[i], list);
   return list;
 }
@@ -1494,13 +1497,13 @@ previously.
 */
        (list))
 {
-  int category_to_priority[CODING_CATEGORY_LAST + 1];
+  int category_to_priority[CODING_CATEGORY_LAST];
   int i, j;
   Lisp_Object rest;
 
   /* First generate a list that maps coding categories to priorities. */
 
-  for (i = 0; i <= CODING_CATEGORY_LAST; i++)
+  for (i = 0; i < CODING_CATEGORY_LAST; i++)
     category_to_priority[i] = -1;
 
   /* Highest priority comes from the specified list. */
@@ -1517,7 +1520,7 @@ previously.
   /* Now go through the existing categories by priority to retrieve
      the categories not yet specified and preserve their priority
      order. */
-  for (j = 0; j <= CODING_CATEGORY_LAST; j++)
+  for (j = 0; j < CODING_CATEGORY_LAST; j++)
     {
       int cat = fcd->coding_category_by_priority[j];
       if (category_to_priority[cat] < 0)
@@ -1527,7 +1530,7 @@ previously.
   /* Now we need to construct the inverse of the mapping we just
      constructed. */
 
-  for (i = 0; i <= CODING_CATEGORY_LAST; i++)
+  for (i = 0; i < CODING_CATEGORY_LAST; i++)
     fcd->coding_category_by_priority[category_to_priority[i]] = i;
 
   /* Phew!  That was confusing. */
@@ -1542,7 +1545,7 @@ Return a list of coding categories in descending order of priority.
   int i;
   Lisp_Object list = Qnil;
 
-  for (i = CODING_CATEGORY_LAST; i >= 0; i--)
+  for (i = CODING_CATEGORY_LAST - 1; i >= 0; i--)
     list = Fcons (coding_category_symbol[fcd->coding_category_by_priority[i]],
 		  list);
   return list;
@@ -1792,7 +1795,7 @@ coding_system_from_mask (int mask)
 #endif
       /* Look through the coding categories by priority and find
 	 the first one that is allowed. */
-      for (i = 0; i <= CODING_CATEGORY_LAST; i++)
+      for (i = 0; i < CODING_CATEGORY_LAST; i++)
 	{
 	  cat = fcd->coding_category_by_priority[i];
 	  if ((mask & (1 << cat)) &&
@@ -1990,7 +1993,7 @@ type.  Optional arg BUFFER defaults to the current buffer.
 #ifdef MULE
       decst.mask = postprocess_iso2022_mask (decst.mask);
 #endif
-      for (i = CODING_CATEGORY_LAST; i >= 0; i--)
+      for (i = CODING_CATEGORY_LAST - 1; i >= 0; i--)
 	{
 	  int sys = fcd->coding_category_by_priority[i];
 	  if (decst.mask & (1 << sys))
@@ -2226,7 +2229,8 @@ COMPOSE_ADD_CHAR(struct decoding_stream *str,
   else if (!CHAR_ID_TABLE_P (str->combining_table))
     {
       Lisp_Object ret
-	= get_char_id_table (character, Vcharacter_composition_table);
+	= get_char_id_table (XCHAR_ID_TABLE(Vcharacter_composition_table),
+			     character);
 
       if (NILP (ret))
 	DECODE_ADD_UCS_CHAR (character, dst);
@@ -2240,12 +2244,15 @@ COMPOSE_ADD_CHAR(struct decoding_stream *str,
   else
     {
       Lisp_Object ret
-	= get_char_id_table (character, str->combining_table);
+	= get_char_id_table (XCHAR_ID_TABLE(str->combining_table),
+			     character);
 
       if (CHARP (ret))
 	{
 	  Emchar char2 = XCHARVAL (ret);
-	  ret = get_char_id_table (char2, Vcharacter_composition_table);
+	  ret =
+	    get_char_id_table (XCHAR_ID_TABLE(Vcharacter_composition_table),
+			       char2);
 	  if (NILP (ret))
 	    {
 	      DECODE_ADD_UCS_CHAR (char2, dst);
@@ -2900,6 +2907,10 @@ reset_encoding_stream (struct encoding_stream *str)
       str->encode_char = &char_encode_shift_jis;
       str->finish = &char_finish_shift_jis;
       break;
+    case CODESYS_BIG5:
+      str->encode_char = &char_encode_big5;
+      str->finish = &char_finish_big5;
+      break;
     default:
       break;
     }
@@ -3015,9 +3026,6 @@ mule_encode (Lstream *encoding, const Bufbyte *src,
       encode_coding_no_conversion (encoding, src, dst, n);
       break;
 #ifdef MULE
-    case CODESYS_BIG5:
-      encode_coding_big5 (encoding, src, dst, n);
-      break;
     case CODESYS_CCL:
       str->ccl.last_block = str->flags & CODING_STATE_END;
       /* When applying ccl program to stream, MUST NOT set NULL
@@ -3445,8 +3453,13 @@ Return the corresponding character code in SHIFT-JIS as a cons of two bytes.
    contains frequently used characters and the latter contains less
    frequently used characters.  */
 
+#ifdef UTF2000
+#define BYTE_BIG5_TWO_BYTE_1_P(c) \
+  ((c) >= 0x81 && (c) <= 0xFE)
+#else
 #define BYTE_BIG5_TWO_BYTE_1_P(c) \
   ((c) >= 0xA1 && (c) <= 0xFE)
+#endif
 
 /* Is this the second byte of a Shift-JIS two-byte char? */
 
@@ -3532,8 +3545,11 @@ detect_coding_big5 (struct detection_state *st, const Extbyte *src, size_t n)
   while (n--)
     {
       unsigned char c = *(unsigned char *)src++;
-      if (c == ISO_CODE_ESC || c == ISO_CODE_SI || c == ISO_CODE_SO ||
-	  (c >= 0x80 && c <= 0xA0))
+      if (c == ISO_CODE_ESC || c == ISO_CODE_SI || c == ISO_CODE_SO
+#ifndef UTF2000
+	  || (c >= 0x80 && c <= 0xA0)
+#endif
+	  )
 	return 0;
       if (st->big5.in_second_byte)
 	{
@@ -3541,7 +3557,13 @@ detect_coding_big5 (struct detection_state *st, const Extbyte *src, size_t n)
 	  if (c < 0x40 || (c >= 0x80 && c <= 0xA0))
 	    return 0;
 	}
-      else if (c >= 0xA1)
+      else if (
+#ifdef UTF2000
+	       c >= 0x81
+#else
+	       c >= 0xA1
+#endif
+	       )
 	st->big5.in_second_byte = 1;
     }
   return CODING_CATEGORY_BIG5_MASK;
@@ -3604,66 +3626,71 @@ decode_coding_big5 (Lstream *decoding, const Extbyte *src,
 
 /* Convert internally-formatted data to Big5. */
 
-static void
-encode_coding_big5 (Lstream *encoding, const Bufbyte *src,
-		    unsigned_char_dynarr *dst, size_t n)
+void
+char_encode_big5 (struct encoding_stream *str, Emchar ch,
+		  unsigned_char_dynarr *dst, unsigned int *flags)
 {
-#ifndef UTF2000
-  unsigned char c;
-  struct encoding_stream *str = ENCODING_STREAM_DATA (encoding);
-  unsigned int flags  = str->flags;
-  unsigned int ch     = str->ch;
   eol_type_t eol_type = CODING_SYSTEM_EOL_TYPE (str->codesys);
 
-  while (n--)
+  if (ch == '\n')
     {
-      c = *src++;
-      if (c == '\n')
-	{
-	  if (eol_type != EOL_LF && eol_type != EOL_AUTODETECT)
-	    Dynarr_add (dst, '\r');
-	  if (eol_type != EOL_CR)
-	    Dynarr_add (dst, '\n');
-	}
-      else if (BYTE_ASCII_P (c))
-	{
-	  /* ASCII. */
-	  Dynarr_add (dst, c);
-	}
-      else if (BUFBYTE_LEADING_BYTE_P (c))
+      if (eol_type != EOL_LF && eol_type != EOL_AUTODETECT)
+	Dynarr_add (dst, '\r');
+      if (eol_type != EOL_CR)
+	Dynarr_add (dst, ch);
+    }
+  else
+    {
+#ifdef UTF2000
+      int code_point;
+
+      if ((code_point = charset_code_point (Vcharset_ascii, ch)) >= 0)
+	Dynarr_add (dst, code_point);
+      else if ((code_point
+		= charset_code_point (Vcharset_chinese_big5, ch)) >= 0)
 	{
-	  if (c == LEADING_BYTE_CHINESE_BIG5_1 ||
-	      c == LEADING_BYTE_CHINESE_BIG5_2)
-	    {
-	      /* A recognized leading byte. */
-	      ch = c;
-	      continue; /* not done with this character. */
-	    }
-	  /* otherwise just ignore this character. */
+	  Dynarr_add (dst, code_point >> 8);
+	  Dynarr_add (dst, code_point & 0xFF);
 	}
-      else if (ch == LEADING_BYTE_CHINESE_BIG5_1 ||
-	       ch == LEADING_BYTE_CHINESE_BIG5_2)
+      else if ((code_point
+		= charset_code_point (Vcharset_chinese_big5_1, ch)) >= 0)
 	{
-	  /* Previous char was a recognized leading byte. */
-	  ch = (ch << 8) | c;
-	  continue; /* not done with this character. */
+	  unsigned int I
+	    = ((code_point >> 8) - 33) * (0xFF - 0xA1)
+	    + ((code_point & 0xFF) - 33);
+	  unsigned char b1 = I / BIG5_SAME_ROW + 0xA1;
+	  unsigned char b2 = I % BIG5_SAME_ROW;
+
+	  b2 += b2 < 0x3F ? 0x40 : 0x62;
+	  Dynarr_add (dst, b1);
+	  Dynarr_add (dst, b2);
 	}
-      else if (ch)
+      else if ((code_point
+		= charset_code_point (Vcharset_chinese_big5_2, ch)) >= 0)
 	{
-	  /* Encountering second byte of a Big5 character. */
+	  unsigned int I
+	    = ((code_point >> 8) - 33) * (0xFF - 0xA1)
+	    + ((code_point & 0xFF) - 33);
 	  unsigned char b1, b2;
 
-	  ENCODE_BIG5 (ch >> 8, ch & 0xFF, c, b1, b2);
+	  I += BIG5_SAME_ROW * (0xC9 - 0xA1);
+	  b1 = I / BIG5_SAME_ROW + 0xA1;
+	  b2 = I % BIG5_SAME_ROW;
+	  b2 += b2 < 0x3F ? 0x40 : 0x62;
 	  Dynarr_add (dst, b1);
 	  Dynarr_add (dst, b2);
 	}
-
-      ch = 0;
+      else
+	Dynarr_add (dst, '?');
+#else
+#endif
     }
+}
 
-  str->flags = flags;
-  str->ch    = ch;
-#endif
+void
+char_finish_big5 (struct encoding_stream *str, unsigned_char_dynarr *dst,
+		  unsigned int *flags)
+{
 }
 
 
@@ -3837,6 +3864,95 @@ detect_coding_utf8 (struct detection_state *st, const Extbyte *src, size_t n)
 }
 
 static void
+decode_output_utf8_partial_char (unsigned char counter,
+				 unsigned int cpos,
+				 unsigned_char_dynarr *dst)
+{
+  if (counter == 5)
+    DECODE_ADD_BINARY_CHAR ( (cpos|0xFC), dst);
+  else if (counter == 4)
+    {
+      if (cpos < (1 << 6))
+	DECODE_ADD_BINARY_CHAR ( (cpos|0xF8), dst);
+      else
+	{
+	  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xFC), dst);
+	  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
+	}
+    }
+  else if (counter == 3)
+    {
+      if (cpos < (1 << 6))
+	DECODE_ADD_BINARY_CHAR ( (cpos|0xF0), dst);
+      else if (cpos < (1 << 12))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xF8), dst);
+	  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
+	}
+      else
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xFC), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
+	}
+    }
+  else if (counter == 2)
+    {
+      if (cpos < (1 << 6))
+	DECODE_ADD_BINARY_CHAR ( (cpos|0xE0), dst);
+      else if (cpos < (1 << 12))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xF0), dst);
+	  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
+	}
+      else if (cpos < (1 << 18))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xF8), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >> 6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos      &0x3F)|0x80), dst);
+	}
+      else
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 18)|0xFC), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
+	}
+    }
+  else
+    {
+      if (cpos < (1 << 6))
+	DECODE_ADD_BINARY_CHAR ( (cpos|0xC0), dst);
+      else if (cpos < (1 << 12))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xE0), dst);
+	  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
+	}
+      else if (cpos < (1 << 18))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xF0), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
+	}
+      else if (cpos < (1 << 24))
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 18)|0xF8), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
+	}
+      else
+	{
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 24)|0xFC), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >> 18)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
+	  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
+	}
+    }
+}
+
+static void
 decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
 		    unsigned_char_dynarr *dst, size_t n)
 {
@@ -3896,88 +4012,7 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
 	}
       else
 	{
-	  if (counter == 5)
-	    DECODE_ADD_BINARY_CHAR ( (cpos|0xFC), dst);
-	  else if (counter == 4)
-	    {
-	      if (cpos < (1 << 6))
-		DECODE_ADD_BINARY_CHAR ( (cpos|0xF8), dst);
-	      else
-		{
-		  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xFC), dst);
-		  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
-		}
-	    }
-	  else if (counter == 3)
-	    {
-	      if (cpos < (1 << 6))
-		DECODE_ADD_BINARY_CHAR ( (cpos|0xF0), dst);
-	      else if (cpos < (1 << 12))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xF8), dst);
-		  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
-		}
-	      else
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xFC), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
-		}
-	    }
-	  else if (counter == 2)
-	    {
-	      if (cpos < (1 << 6))
-		DECODE_ADD_BINARY_CHAR ( (cpos|0xE0), dst);
-	      else if (cpos < (1 << 12))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xF0), dst);
-		  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
-		}
-	      else if (cpos < (1 << 18))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xF8), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >> 6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos      &0x3F)|0x80), dst);
-		}
-	      else
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 18)|0xFC), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
-		}
-	    }
-	  else
-	    {
-	      if (cpos < (1 << 6))
-		DECODE_ADD_BINARY_CHAR ( (cpos|0xC0), dst);
-	      else if (cpos < (1 << 12))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ((cpos >> 6)|0xE0), dst);
-		  DECODE_ADD_BINARY_CHAR ( ((cpos&0x3F)|0x80), dst);
-		}
-	      else if (cpos < (1 << 18))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 12)|0xF0), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
-		}
-	      else if (cpos < (1 << 24))
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 18)|0xF8), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
-		}
-	      else
-		{
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos >> 24)|0xFC), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >> 18)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >> 12)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( (((cpos >>  6)&0x3F)|0x80), dst);
-		  DECODE_ADD_BINARY_CHAR ( ( (cpos       &0x3F)|0x80), dst);
-		}
-	    }
+	  decode_output_utf8_partial_char (counter, cpos, dst);
 	  DECODE_ADD_BINARY_CHAR (c, dst);
 	  cpos = 0;
 	  counter = 0;
@@ -3988,7 +4023,7 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src,
   if (flags & CODING_STATE_END)
     if (counter > 0)
       {
-	DECODE_ADD_BINARY_CHAR (cpos, dst);
+	decode_output_utf8_partial_char (counter, cpos, dst);
 	cpos = 0;
 	counter = 0;
       }
@@ -5764,10 +5799,10 @@ vars_of_file_coding (void)
   int i;
 
   fcd = xnew (struct file_coding_dump);
-  dumpstruct (&fcd, &fcd_description);
+  dump_add_root_struct_ptr (&fcd, &fcd_description);
 
   /* Initialize to something reasonable ... */
-  for (i = 0; i <= CODING_CATEGORY_LAST; i++)
+  for (i = 0; i < CODING_CATEGORY_LAST; i++)
     {
       fcd->coding_category_system[i] = Qnil;
       fcd->coding_category_by_priority[i] = i;
@@ -5831,7 +5866,7 @@ complex_vars_of_file_coding (void)
     make_lisp_hash_table (50, HASH_TABLE_NON_WEAK, HASH_TABLE_EQ);
 
   the_codesys_prop_dynarr = Dynarr_new (codesys_prop);
-  dumpstruct (&the_codesys_prop_dynarr, &codesys_prop_dynarr_description);
+  dump_add_root_struct_ptr (&the_codesys_prop_dynarr, &codesys_prop_dynarr_description);
 
 #define DEFINE_CODESYS_PROP(Prop_Type, Sym) do	\
 {						\
@@ -5907,7 +5942,7 @@ complex_vars_of_file_coding (void)
 
 #if defined(MULE) && !defined(UTF2000)
   {
-    unsigned int i;
+    size_t i;
 
     for (i = 0; i < countof (fcd->ucs_to_mule_table); i++)
       fcd->ucs_to_mule_table[i] = Qnil;