Update.

2002-11-30  Bruno Haible  <bruno@clisp.org>

	* iconv/gconv.h (__gconv_btowc_fct): New typedef.
	(struct __gconv_step): New field __btowc_fct.
	* wcsmbs/btowc.c (__btowc): Use the __btowc_fct shortcut if possible.
	* iconv/gconv_int.h (__BUILTIN_TRANSFORM): Renamed from
	__BUILTIN_TRANS.
	(__gconv_btwoc_ascii): New declaration.
	* iconv/gconv_simple.c (BUILTIN_TRANSFORMATION): Add BtowcFct argument.
	(__gconv_btwoc_ascii): New function.
	* iconv/gconv_builtin.h: Add BtowcFct argument to all
	BUILTIN_TRANSFORMATION invocations.
	* iconv/gconv_conf.c (BUILTIN_TRANSFORMATION): Add BtowcFct argument.
	* iconv/iconvconfig.c (BUILTIN_TRANSFORMATION): Likewise.
	* iconv/gconv_builtin.c (map): New field btowc_fct.
	(BUILTIN_TRANSFORMATION): Add BtowcFct argument. Use it to initialize
	btowc_fct field.
	(__gconv_get_builtin_trans): Initialize __btowc_fct field.
	* iconv/gconv_cache.c (find_module): Initialize __btowc_fct field.
	* iconv/gconv_db.c (gen_steps, increment_counter): Likewise.
	* wcsmbs/wcsmbsload.c (to_wc, to_mb): Likewise.
	* iconv/skeleton.c: Document STORE_REST and FROM_ONEBYTE.
	(gconv_init): Initialize __btowc_fct field.
	Undefine EXTRA_LOOP_ARGS and FROM_ONEBYTE at the end.
	* iconv/loop.c: Document ONEBYTE_BODY.
	(gconv_btowc, FROM_ONEBYTE): Define if ONEBYTE_BODY is defined.
	Undefine ONEBYTE_BODY at the end.
	* iconvdata/8bit-generic.c (ONEBYTE_BODY): New macro.
	* iconvdata/8bit-gap.c (NONNUL): New macro.
	(BODY for FROM_LOOP): Use it.
	(ONEBYTE_BODY): New macro.
	* iconvdata/isiri-3342.c (HAS_HOLES): Set to 1.
	(NONNUL): New macro.
	* iconvdata/ansi_x3.110.c (ONEBYTE_BODY): New macro.
	* iconvdata/armscii-8.c (ONEBYTE_BODY): New macro.
	* iconvdata/cp1255.c (ONEBYTE_BODY): New macro.
	* iconvdata/cp1258.c (ONEBYTE_BODY): New macro.
	* iconvdata/tcvn5712-1.c (ONEBYTE_BODY): New macro.
	* iconvdata/big5.c (ONEBYTE_BODY): New macro.
	* iconvdata/big5hkscs.c (ONEBYTE_BODY): New macro.
	* iconvdata/euc-cn.c (ONEBYTE_BODY): New macro.
	* iconvdata/euc-jp.c (ONEBYTE_BODY): New macro.
	* iconvdata/euc-jisx0213.c (ONEBYTE_BODY): New macro.
	* iconvdata/euc-kr.c (ONEBYTE_BODY): New macro.
	* iconvdata/euc-tw.c (ONEBYTE_BODY): New macro.
	* iconvdata/gbk.c (ONEBYTE_BODY): New macro.
	* iconvdata/gb18030.c (ONEBYTE_BODY): New macro.
	* iconvdata/ibm932.c: Include <stdbool.h>.
	(TRUE, FALSE): Remove macros.
	(BODY for FROM_LOOP): Remove unused variable rp1.
	(ONEBYTE_BODY): New macro.
	(BODY for TO_LOOP): Use bool.
	* iconvdata/ibm932.h (__ibm932sb_to_ucs4_idx): Remove array.
	* iconvdata/ibm943.c: Include <stdbool.h>.
	(TRUE, FALSE): Remove macros.
	(BODY for FROM_LOOP): Remove unused variable rp1.
	(ONEBYTE_BODY): New macro.
	(BODY for TO_LOOP): Use bool.
	* iconvdata/ibm943.h (__ibm943sb_to_ucs4_idx): Remove array.
	* iconvdata/iso8859-1.c (ONEBYTE_BODY): New macro.
	* iconvdata/iso_6937-2.c (ONEBYTE_BODY): New macro.
	* iconvdata/iso_6937.c (ONEBYTE_BODY): New macro.
	* iconvdata/johab.c (ONEBYTE_BODY): New macro.
	* iconvdata/sjis.c (ONEBYTE_BODY): New macro.
	* iconvdata/shift_jisx0213.c (ONEBYTE_BODY): New macro.
	* iconvdata/t.61.c (ONEBYTE_BODY): New macro.
	* iconvdata/uhc.c (ONEBYTE_BODY): New macro.
	* iconvdata/gbbig5.c: Tweak comment.

30 files changed, 292 insertions, 87 deletions

diff --git a/iconvdata/8bit-gap.c b/iconvdata/8bit-gap.c
index 2ddc49eb64..9d48d170ba 100644
--- a/iconvdata/8bit-gap.c
+++ b/iconvdata/8bit-gap.c
@@ -32,6 +32,10 @@ struct gap
 /* Now we can include the tables.  */
 #include TABLES
 
+#ifndef NONNUL
+# define NONNUL(c)	((c) != '\0')
+#endif
+
 
 #define FROM_LOOP		from_gap
 #define TO_LOOP			to_gap
@@ -49,7 +53,7 @@ struct gap
   {									      \
     uint32_t ch = to_ucs4[*inptr];					      \
 									      \
-    if (HAS_HOLES && __builtin_expect (ch, L'\1') == L'\0' && *inptr != '\0') \
+    if (HAS_HOLES && __builtin_expect (ch == L'\0', 0) && NONNUL (*inptr))    \
       {									      \
 	/* This is an illegal character.  */				      \
 	STANDARD_FROM_LOOP_ERR_HANDLER (1);				      \
@@ -63,6 +67,15 @@ struct gap
     ++inptr;								      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    uint32_t ch = to_ucs4[c];						      \
+									      \
+    if (HAS_HOLES && __builtin_expect (ch == L'\0', 0) && NONNUL (c))	      \
+      return WEOF;							      \
+    else								      \
+      return ch;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/8bit-generic.c b/iconvdata/8bit-generic.c
index 6ed5532693..373289f84d 100644
--- a/iconvdata/8bit-generic.c
+++ b/iconvdata/8bit-generic.c
@@ -47,6 +47,15 @@
     ++inptr;								      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    uint32_t ch = to_ucs4[c];						      \
+									      \
+    if (HAS_HOLES && __builtin_expect (ch == L'\0', 0) && c != '\0')	      \
+      return WEOF;							      \
+    else								      \
+      return ch;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/ansi_x3.110.c b/iconvdata/ansi_x3.110.c
index 08746cf5f7..b20e12a79c 100644
--- a/iconvdata/ansi_x3.110.c
+++ b/iconvdata/ansi_x3.110.c
@@ -404,7 +404,7 @@ static const char from_ucs4[][2] =
 									      \
     if (__builtin_expect (ch >= 0xc1, 0) && ch <= 0xcf)			      \
       {									      \
-	/* Composed character.  First test whether the next character	      \
+	/* Composed character.  First test whether the next byte	      \
 	   is also available.  */					      \
 	uint32_t ch2;							      \
 									      \
@@ -449,6 +449,15 @@ static const char from_ucs4[][2] =
     inptr += incr;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    uint32_t ch = to_ucs4[c];						      \
+									      \
+    if (__builtin_expect (ch == 0, 0) && c != '\0')			      \
+      return WEOF;							      \
+    else								      \
+      return ch;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/armscii-8.c b/iconvdata/armscii-8.c
index b3db9a99a8..dbaf8be2ee 100644
--- a/iconvdata/armscii-8.c
+++ b/iconvdata/armscii-8.c
@@ -71,6 +71,17 @@ static const uint16_t map_from_armscii_8[0xfe - 0xa2 + 1] =
     ++inptr;								      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c <= 0xa0)							      \
+      /* Upto and including 0xa0 the ARMSCII-8 corresponds to Unicode.  */    \
+      return c;								      \
+    else if (c >= 0xa2 && c <= 0xfe)					      \
+      /* Use the table.  */						      \
+      return map_from_armscii_8[c - 0xa2];				      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/big5.c b/iconvdata/big5.c
index 22c3ab6ddd..cdf8a76c5d 100644
--- a/iconvdata/big5.c
+++ b/iconvdata/big5.c
@@ -8397,7 +8397,7 @@ static const char from_ucs4_tab15[][2] =
 									      \
     if (ch >= 0xa1 && ch <= 0xf9)					      \
       {									      \
-	/* Two-byte character.  First test whether the next character	      \
+	/* Two-byte character.  First test whether the next byte	      \
 	   is also available.  */					      \
 	uint32_t ch2;							      \
 	int idx;							      \
@@ -8447,6 +8447,13 @@ static const char from_ucs4_tab15[][2] =
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c <= 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/big5hkscs.c b/iconvdata/big5hkscs.c
index 0903968669..b1aedd171b 100644
--- a/iconvdata/big5hkscs.c
+++ b/iconvdata/big5hkscs.c
@@ -16817,7 +16817,7 @@ static struct
 									      \
     if (ch >= 0x81 && ch <= 0xfe)					      \
       {									      \
-	/* Two-byte character.  First test whether the next character	      \
+	/* Two-byte character.  First test whether the next byte	      \
 	   is also available.  */					      \
 	uint32_t ch2;							      \
 	int idx;							      \
@@ -16852,6 +16852,13 @@ static struct
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c <= 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/cp1255.c b/iconvdata/cp1255.c
index 19edb27a66..6cccf526e6 100644
--- a/iconvdata/cp1255.c
+++ b/iconvdata/cp1255.c
@@ -321,6 +321,15 @@ static const struct { unsigned int idx; unsigned int len; } comp_table[8] = {
   }
 #define LOOP_NEED_FLAGS
 #define EXTRA_LOOP_DECLS	, int *statep
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    uint32_t ch = to_ucs4[c - 0x80];					      \
+    if (ch == L'\0' || (ch >= 0x05d0 && ch <= 0x05f2))			      \
+      return WEOF;							      \
+    return ch;								      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/cp1258.c b/iconvdata/cp1258.c
index 853ef63531..553d50b443 100644
--- a/iconvdata/cp1258.c
+++ b/iconvdata/cp1258.c
@@ -480,6 +480,22 @@ static const struct
   }
 #define LOOP_NEED_FLAGS
 #define EXTRA_LOOP_DECLS	, int *statep
+#define ONEBYTE_BODY \
+  {									      \
+    uint32_t ch;							      \
+									      \
+    if (c < 0x80)							      \
+      ch = c;								      \
+    else								      \
+      {									      \
+	ch = to_ucs4[c - 0x80];						      \
+	if (ch == L'\0')						      \
+	  return WEOF;							      \
+      }									      \
+    if (ch >= 0x0041 && ch <= 0x01b0)					      \
+      return WEOF;							      \
+    return ch;								      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/euc-cn.c b/iconvdata/euc-cn.c
index c1bcadb35b..f912317876 100644
--- a/iconvdata/euc-cn.c
+++ b/iconvdata/euc-cn.c
@@ -54,7 +54,7 @@
       else								      \
 	{								      \
 	  /* Two or more byte character.  First test whether the	      \
-	     next character is also available.  */			      \
+	     next byte is also available.  */				      \
 	  const unsigned char *endp;					      \
 									      \
 	  if (__builtin_expect (inptr + 1 >= inend, 0))			      \
@@ -88,6 +88,13 @@
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/euc-jisx0213.c b/iconvdata/euc-jisx0213.c
index 733cbc15ee..dc6ccfd0cc 100644
--- a/iconvdata/euc-jisx0213.c
+++ b/iconvdata/euc-jisx0213.c
@@ -230,6 +230,13 @@
   }
 #define LOOP_NEED_FLAGS
 #define EXTRA_LOOP_DECLS	, int *statep
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/euc-jp.c b/iconvdata/euc-jp.c
index 7faa080322..9cb6a47040 100644
--- a/iconvdata/euc-jp.c
+++ b/iconvdata/euc-jp.c
@@ -117,6 +117,13 @@
     put32 (outptr, ch);							      \
     outptr += 4;							      \
   }
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x8e || (c >= 0x90 && c <= 0x9f))				      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #define LOOP_NEED_FLAGS
 #include <iconv/loop.c>
 
diff --git a/iconvdata/euc-kr.c b/iconvdata/euc-kr.c
index 37ed71fd60..7774720bd6 100644
--- a/iconvdata/euc-kr.c
+++ b/iconvdata/euc-kr.c
@@ -90,7 +90,7 @@ euckr_from_ucs4 (uint32_t ch, unsigned char *cp)
       }									      \
     else								      \
       {									      \
-	/* Two-byte character.  First test whether the next character	      \
+	/* Two-byte character.  First test whether the next byte	      \
 	   is also available.  */					      \
 	ch = ksc5601_to_ucs4 (&inptr, inend - inptr, 0x80);		      \
 	if (__builtin_expect (ch == 0, 0))				      \
@@ -108,6 +108,13 @@ euckr_from_ucs4 (uint32_t ch, unsigned char *cp)
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c <= 0x9f)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/euc-tw.c b/iconvdata/euc-tw.c
index 8d3b1aa1b5..6674dad397 100644
--- a/iconvdata/euc-tw.c
+++ b/iconvdata/euc-tw.c
@@ -112,6 +112,13 @@
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/gb18030.c b/iconvdata/gb18030.c
index f6b3576416..31f3c20939 100644
--- a/iconvdata/gb18030.c
+++ b/iconvdata/gb18030.c
@@ -25772,6 +25772,13 @@ static const unsigned char __ucs_to_gb18030_tab2[8192][2] =
     *((uint32_t *) outptr)++ = ch;					      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/gbbig5.c b/iconvdata/gbbig5.c
index 695ffb2c07..b2b4743e24 100644
--- a/iconvdata/gbbig5.c
+++ b/iconvdata/gbbig5.c
@@ -4802,7 +4802,7 @@ const char __from_big5_to_gb2312 [13973][2] =
     else if (ch >= 0xa1 && ch <= 0xf7)					      \
       {									      \
 	/* Two-byte character.  First test whether the		      	      \
-	   next character is also available.  */			      \
+	   next byte is also available.  */				      \
 	const char *cp;							      \
 	int idx;							      \
 									      \
@@ -4891,7 +4891,7 @@ const char __from_big5_to_gb2312 [13973][2] =
     else if (ch >= 0xa1 && ch <= 0xf9)					      \
       {									      \
 	/* Two byte character.  First test whether the			      \
-	   next character is also available.  */			      \
+	   next byte is also available.  */				      \
 	const char *cp;							      \
 	int idx;							      \
 									      \
diff --git a/iconvdata/gbk.c b/iconvdata/gbk.c
index a738a9eb9d..ad8a37412a 100644
--- a/iconvdata/gbk.c
+++ b/iconvdata/gbk.c
@@ -13154,7 +13154,7 @@ static const char __gbk_from_ucs4_tab12[][2] =
       else								      \
 	{								      \
 	  /* Two or more byte character.  First test whether the	      \
-	     next character is also available.  */			      \
+	     next byte is also available.  */				      \
 	  uint32_t ch2;							      \
 	  int idx;							      \
 									      \
@@ -13195,6 +13195,13 @@ static const char __gbk_from_ucs4_tab12[][2] =
     outptr += 4;							      \
   }
 #define LOOP_NEED_FLAGS
+#define ONEBYTE_BODY \
+  {									      \
+    if (c < 0x80)							      \
+      return c;								      \
+    else								      \
+      return WEOF;							      \
+  }
 #include <iconv/loop.c>
 
 
diff --git a/iconvdata/ibm932.c b/iconvdata/ibm932.c
index 1e87067fef..420b19ab80 100644
--- a/iconvdata/ibm932.c
+++ b/iconvdata/ibm932.c
@@ -20,13 +20,9 @@
 
 #include <dlfcn.h>
 #include <stdint.h>
+#include <stdbool.h>
 #include "ibm932.h"
 
-#ifndef TRUE
-#define TRUE  1
-#define FALSE 0
-#endif
-
 #define FROM	0
 #define TO	1
 
@@ -50,38 +46,24 @@
 #define LOOPFCT			FROM_LOOP
 #define BODY \
   {									      \
-    const struct gap *rp1 = __ibm932sb_to_ucs4_idx;			      \
     const struct gap *rp2 = __ibm932db_to_ucs4_idx;			      \
     uint32_t ch = *inptr;						      \
     uint32_t res;							      \
 									      \
-    if (__builtin_expect (ch >= 0xffff, 0))				      \
-      {									      \
-	rp1 = NULL;							      \