This is the mail archive of the libc-alpha@sourceware.cygnus.com mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

UCS-2 conversion



Hi Ulrich,

With your newest patches, it is not clear what "UCS-2" means. If you
look at gconv_builtin.h, it has the endianness of the current CPU. But
if you look at gconv_simple.c, it is big-endian (as it was until last
week, and like UCS-4).

Below are two patches. If you want UCS-2 be big endian, you need the
first one. If you want UCS-2 be machine dependent, you need the second one.

Bruno


=========================== UCS-2 = UCS-2BE ===================================

2000-03-23  Bruno Haible  <haible@clisp.cons.org>

	* iconv/gconv_builtin.h: Define UNICODEBIG and UCS-2BE as aliases
	for UCS-2.
	* iconv/gconv_int.h (__gconv_transform_ucs2reverse_internal,
	__gconv_transform_internal_ucs2reverse): Remove declarations.
	* iconv/gconv_simple.h (__gconv_transform_ucs2little_internal):
	Renamed from __gconv_transform_ucs2reverse_internal.
	(__gconv_transform_internal_ucs2little):
	Renamed from __gconv_transform_internal_ucs2reverse.

*** iconv/gconv_builtin.h.bak	Wed Mar 22 22:09:48 2000
--- iconv/gconv_builtin.h	Thu Mar 23 11:07:32 2000
***************
*** 66,71 ****
--- 66,73 ----
  
  BUILTIN_ALIAS ("UCS2//", "ISO-10646/UCS2/")
  BUILTIN_ALIAS ("UCS-2//", "ISO-10646/UCS2/")
+ BUILTIN_ALIAS ("UNICODEBIG//", "ISO-10646/UCS2/")
+ BUILTIN_ALIAS ("UCS-2BE//", "ISO-10646/UCS2/")
  BUILTIN_ALIAS ("OSF00010100//", "ISO-10646/UCS2/") /* level 1 */
  BUILTIN_ALIAS ("OSF00010101//", "ISO-10646/UCS2/") /* level 2 */
  BUILTIN_ALIAS ("OSF00010102//", "ISO-10646/UCS2/") /* level 3 */
***************
*** 80,114 ****
  			__gconv_transform_internal_ucs2, NULL, NULL,
  			4, 4, 2, 2)
  
- 
- #if BYTE_ORDER == BIG_ENDIAN
- BUILTIN_ALIAS ("UNICODEBIG//", "ISO-10646/UCS2/")
- BUILTIN_ALIAS ("UCS-2BE//", "ISO-10646/UCS2/")
- 
  BUILTIN_ALIAS ("UCS-2LE//", "UNICODELITTLE//")
  
  BUILTIN_TRANSFORMATION (NULL, "UNICODELITTLE//", 15, "INTERNAL",
! 			1, "=ucs2reverse->INTERNAL",
! 			__gconv_transform_ucs2reverse_internal, NULL, NULL,
  			2, 2, 4, 4)
  
  BUILTIN_TRANSFORMATION (NULL, "INTERNAL", 8, "UNICODELITTLE//",
! 			1, "=INTERNAL->ucs2reverse",
! 			__gconv_transform_internal_ucs2reverse, NULL, NULL,
! 			4, 4, 2, 2)
! #else
! BUILTIN_ALIAS ("UNICODELITTLE//", "ISO-10646/UCS2/")
! BUILTIN_ALIAS ("UCS-2LE//", "ISO-10646/UCS2/")
! 
! BUILTIN_ALIAS ("UCS-2BE//", "UNICODEBIG//")
! 
! BUILTIN_TRANSFORMATION (NULL, "UNICODEBIG//", 15, "INTERNAL",
! 			1, "=ucs2reverse->INTERNAL",
! 			__gconv_transform_ucs2reverse_internal, NULL, NULL,
! 			2, 2, 4, 4)
! 
! BUILTIN_TRANSFORMATION (NULL, "INTERNAL", 8, "UNICODEBIG//",
! 			1, "=INTERNAL->ucs2reverse",
! 			__gconv_transform_internal_ucs2reverse, NULL, NULL,
  			4, 4, 2, 2)
- #endif
--- 82,95 ----
  			__gconv_transform_internal_ucs2, NULL, NULL,
  			4, 4, 2, 2)
  
  BUILTIN_ALIAS ("UCS-2LE//", "UNICODELITTLE//")
  
  BUILTIN_TRANSFORMATION (NULL, "UNICODELITTLE//", 15, "INTERNAL",
! 			1, "=ucs2little->INTERNAL",
! 			__gconv_transform_ucs2little_internal, NULL, NULL,
  			2, 2, 4, 4)
  
  BUILTIN_TRANSFORMATION (NULL, "INTERNAL", 8, "UNICODELITTLE//",
! 			1, "=INTERNAL->ucs2little",
! 			__gconv_transform_internal_ucs2little, NULL, NULL,
  			4, 4, 2, 2)
*** iconv/gconv_int.h.bak	Wed Mar 22 22:09:49 2000
--- iconv/gconv_int.h	Thu Mar 23 11:05:02 2000
***************
*** 169,176 ****
  __BUILTIN_TRANS (__gconv_transform_internal_utf8);
  __BUILTIN_TRANS (__gconv_transform_ucs2_internal);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs2);
- __BUILTIN_TRANS (__gconv_transform_ucs2reverse_internal);
- __BUILTIN_TRANS (__gconv_transform_internal_ucs2reverse);
  __BUILTIN_TRANS (__gconv_transform_ucs2little_internal);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs2little);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs4);
--- 169,174 ----
*** iconv/gconv_simple.c.bak	Thu Mar 23 01:07:23 2000
--- iconv/gconv_simple.c	Thu Mar 23 11:04:27 2000
***************
*** 457,463 ****
  #define FROM_DIRECTION		1
  #define FROM_LOOP		ucs2little_internal_loop
  #define TO_LOOP			ucs2little_internal_loop /* This is not used.*/
! #define FUNCTION_NAME		__gconv_transform_ucs2reverse_internal
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
--- 457,463 ----
  #define FROM_DIRECTION		1
  #define FROM_LOOP		ucs2little_internal_loop
  #define TO_LOOP			ucs2little_internal_loop /* This is not used.*/
! #define FUNCTION_NAME		__gconv_transform_ucs2little_internal
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
***************
*** 482,488 ****
  #define FROM_DIRECTION		1
  #define FROM_LOOP		internal_ucs2little_loop
  #define TO_LOOP			internal_ucs2little_loop /* This is not used.*/
! #define FUNCTION_NAME		__gconv_transform_internal_ucs2reverse
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
--- 482,488 ----
  #define FROM_DIRECTION		1
  #define FROM_LOOP		internal_ucs2little_loop
  #define TO_LOOP			internal_ucs2little_loop /* This is not used.*/
! #define FUNCTION_NAME		__gconv_transform_internal_ucs2little
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO

=========================== UCS-2 with machine endianness =====================

2000-03-23  Bruno Haible  <haible@clisp.cons.org>

	* iconv/gconv_int.h (__gconv_transform_ucs2little_internal,
	__gconv_transform_internal_ucs2little): Remove declarations.
	* iconv/gconv_simple.c (__gconv_transform_ucs2_internal,
	__gconv_transform_internal_ucs2): Never	need to bswap_16.
	(__gconv_transform_ucs2reverse_internal,
	__gconv_transform_internal_ucs2reverse): Always need to bswap_16.

*** iconv/gconv_int.h.bak	Wed Mar 22 22:09:49 2000
--- iconv/gconv_int.h	Thu Mar 23 11:11:08 2000
***************
*** 171,178 ****
  __BUILTIN_TRANS (__gconv_transform_internal_ucs2);
  __BUILTIN_TRANS (__gconv_transform_ucs2reverse_internal);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs2reverse);
- __BUILTIN_TRANS (__gconv_transform_ucs2little_internal);
- __BUILTIN_TRANS (__gconv_transform_internal_ucs2little);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs4);
  __BUILTIN_TRANS (__gconv_transform_internal_ucs4le);
  __BUILTIN_TRANS (__gconv_transform_internal_utf16);
--- 171,176 ----
*** iconv/gconv_simple.c.bak	Thu Mar 23 01:07:23 2000
--- iconv/gconv_simple.c	Thu Mar 23 11:18:17 2000
***************
*** 396,409 ****
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
- #if __BYTE_ORDER == __LITTLE_ENDIAN
- # define BODY \
-   *((uint32_t *) outptr)++ = bswap_16 (*(uint16_t *) inptr);		      \
-   inptr += 2;
- #else
  # define BODY \
    *((uint32_t *) outptr)++ = *((uint16_t *) inptr)++;
- #endif
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
  
--- 396,403 ----
***************
*** 421,440 ****
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
- #if __BYTE_ORDER == __LITTLE_ENDIAN
- # define BODY \
-   {									      \
-     if (*((uint32_t *) inptr) >= 0x10000)				      \
-       {									      \
- 	result = __GCONV_ILLEGAL_INPUT;					      \
- 	break;								      \
-       }									      \
-     /* Please note that we use the `uint32_t' from-pointer as an `uint16_t'   \
-        pointer which works since we are on a little endian machine.  */	      \
-     *((uint16_t *) outptr)++ = bswap_16 (*((uint16_t *) inptr));	      \
-     inptr += 4;								      \
-   }
- #else
  # define BODY \
    {									      \
      if (*((uint32_t *) inptr) >= 0x10000)				      \
--- 415,420 ----
***************
*** 444,503 ****
        }									      \
      *((uint16_t *) outptr)++ = *((uint32_t *) inptr)++;			      \
    }
- #endif
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
  
  
! /* Convert from UCS2 in little endian to the internal (UCS4-like) format.  */
  #define DEFINE_INIT		0
  #define DEFINE_FINI		0
  #define MIN_NEEDED_FROM		2
  #define MIN_NEEDED_TO		4
  #define FROM_DIRECTION		1
! #define FROM_LOOP		ucs2little_internal_loop
! #define TO_LOOP			ucs2little_internal_loop /* This is not used.*/
  #define FUNCTION_NAME		__gconv_transform_ucs2reverse_internal
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
- #if __BYTE_ORDER == __LITTLE_ENDIAN
- # define BODY \
-   *((uint32_t *) outptr)++ = *((uint16_t *) inptr)++;
- #else
  # define BODY \
    *((uint32_t *) outptr)++ = bswap_16 (*(uint16_t *) inptr);		      \
    inptr += 2;
- #endif
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
  
  
! /* Convert from the internal (UCS4-like) format to UCS2 in little endian.  */
  #define DEFINE_INIT		0
  #define DEFINE_FINI		0
  #define MIN_NEEDED_FROM		4
  #define MIN_NEEDED_TO		2
  #define FROM_DIRECTION		1
! #define FROM_LOOP		internal_ucs2little_loop
! #define TO_LOOP			internal_ucs2little_loop /* This is not used.*/
  #define FUNCTION_NAME		__gconv_transform_internal_ucs2reverse
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
- #if __BYTE_ORDER == __LITTLE_ENDIAN
- # define BODY \
-   {									      \
-     if (*((uint32_t *) inptr) >= 0x10000)				      \
-       {									      \
- 	result = __GCONV_ILLEGAL_INPUT;					      \
- 	break;								      \
-       }									      \
-     *((uint16_t *) outptr)++ = *((uint32_t *) inptr)++;			      \
-   }
- #else
  # define BODY \
    {									      \
      if (*((uint32_t *) inptr) >= 0x10000)				      \
--- 424,466 ----
        }									      \
      *((uint16_t *) outptr)++ = *((uint32_t *) inptr)++;			      \
    }
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
  
  
! /* Convert from UCS2 in other endianness to the internal (UCS4-like) format.  */
  #define DEFINE_INIT		0
  #define DEFINE_FINI		0
  #define MIN_NEEDED_FROM		2
  #define MIN_NEEDED_TO		4
  #define FROM_DIRECTION		1
! #define FROM_LOOP		ucs2reverse_internal_loop
! #define TO_LOOP			ucs2reverse_internal_loop /* This is not used.*/
  #define FUNCTION_NAME		__gconv_transform_ucs2reverse_internal
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
  # define BODY \
    *((uint32_t *) outptr)++ = bswap_16 (*(uint16_t *) inptr);		      \
    inptr += 2;
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
  
  
! /* Convert from the internal (UCS4-like) format to UCS2 in other endianness.  */
  #define DEFINE_INIT		0
  #define DEFINE_FINI		0
  #define MIN_NEEDED_FROM		4
  #define MIN_NEEDED_TO		2
  #define FROM_DIRECTION		1
! #define FROM_LOOP		internal_ucs2reverse_loop
! #define TO_LOOP			internal_ucs2reverse_loop /* This is not used.*/
  #define FUNCTION_NAME		__gconv_transform_internal_ucs2reverse
  
  #define MIN_NEEDED_INPUT	MIN_NEEDED_FROM
  #define MIN_NEEDED_OUTPUT	MIN_NEEDED_TO
  #define LOOPFCT			FROM_LOOP
  # define BODY \
    {									      \
      if (*((uint32_t *) inptr) >= 0x10000)				      \
***************
*** 505,513 ****
  	result = __GCONV_ILLEGAL_INPUT;					      \
  	break;								      \
        }									      \
!     *((uint16_t *) outptr)++ = bswap_16 (((uint16_t *) inptr)[1]);	      \
      inptr += 4;								      \
    }
- #endif
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>
--- 468,475 ----
  	result = __GCONV_ILLEGAL_INPUT;					      \
  	break;								      \
        }									      \
!     *((uint16_t *) outptr)++ = bswap_16 (*((uint32_t *) inptr));	      \
      inptr += 4;								      \
    }
  #include <iconv/loop.c>
  #include <iconv/skeleton.c>

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]