[thirdparty/glibc.git] / iconvdata / ibm939.c

/* Conversion to and from IBM939.
   Copyright (C) 2000-2015 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by Masahide Washizawa <washi@yamato.ibm.co.jp>, 2000.

   The GNU C Library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
   License as published by the Free Software Foundation; either
   version 2.1 of the License, or (at your option) any later version.

   The GNU C Library is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public
   License along with the GNU C Library; if not, see
   <http://www.gnu.org/licenses/>.  */

/* IBM939 is designed for the representation of Japanese Latin/Kanji
   using a stateful EBCDIC encoding scheme.  It is also known as
   CCSID 939 or CP939. See:
   https://www-01.ibm.com/software/globalization/ccsid/ccsid939.html */

#include <dlfcn.h>
#include <stdint.h>
#include <wchar.h>
#include <byteswap.h>
#include "ibm939.h"

/* The shift sequences for this charset (it does not use ESC).  */
#define SI 		0x0F  /* Shift In, host code to turn DBCS off.  */
#define SO 		0x0E  /* Shift Out, host code to turn DBCS on.  */

/* Definitions used in the body of the `gconv' function.  */
#define CHARSET_NAME	"IBM939//"
#define FROM_LOOP	from_ibm939
#define TO_LOOP		to_ibm939
#define ONE_DIRECTION			0
#define FROM_LOOP_MIN_NEEDED_FROM	1
#define FROM_LOOP_MAX_NEEDED_FROM	2
#define FROM_LOOP_MIN_NEEDED_TO		4
#define FROM_LOOP_MAX_NEEDED_TO		4
#define TO_LOOP_MIN_NEEDED_FROM		4
#define TO_LOOP_MAX_NEEDED_FROM		4
#define TO_LOOP_MIN_NEEDED_TO		1
#define TO_LOOP_MAX_NEEDED_TO		3
#define PREPARE_LOOP \
  int save_curcs;							      \
  int *curcsp = &data->__statep->__count;
#define EXTRA_LOOP_ARGS		, curcsp

/* Definitions of initialization and destructor function.  */
#define DEFINE_INIT	1
#define DEFINE_FINI	1


/* Since this is a stateful encoding we have to provide code which resets
   the output state to the initial state.  This has to be done during the
   flushing.  */
#define EMIT_SHIFT_TO_INIT \
  if ((data->__statep->__count & ~7) != sb)				      \
    {									      \
      if (FROM_DIRECTION)						      \
	data->__statep->__count &= 7;					      \
      else								      \
	{								      \
	  /* We are not in the initial state.  To switch back we have	      \
	     to emit `SI'.  */						      \
	  if (__glibc_unlikely (outbuf >= outend))			      \
	    /* We don't have enough room in the output buffer.  */	      \
	    status = __GCONV_FULL_OUTPUT;				      \
	  else								      \
	    {								      \
	      /* Write out the shift sequence.  */			      \
	      *outbuf++ = SI;						      \
	      data->__statep->__count &= 7;				      \
	    }								      \
	}								      \
    }


/* Since we might have to reset input pointer we must be able to save
   and retore the state.  */
#define SAVE_RESET_STATE(Save) \
  if (Save)								      \
    save_curcs = *curcsp;						      \
  else									      \
    *curcsp = save_curcs


/* Current codeset type.  */
enum
{
  sb = 0,
  db = 64
};

/* First, define the conversion function from IBM-939 to UCS4.  */
#define MIN_NEEDED_INPUT	FROM_LOOP_MIN_NEEDED_FROM
#define MAX_NEEDED_INPUT	FROM_LOOP_MAX_NEEDED_FROM
#define MIN_NEEDED_OUTPUT	FROM_LOOP_MIN_NEEDED_TO
#define MAX_NEEDED_OUTPUT	FROM_LOOP_MAX_NEEDED_TO
#define LOOPFCT 		FROM_LOOP
#define BODY \
  {									      \
    uint32_t ch = *inptr;						      \
    uint32_t res;							      \
									      \
    if (__builtin_expect (ch, 0) == SO)					      \
      {									      \
	/* Shift OUT, change to DBCS converter.  */			      \
	if (curcs == db)						      \
	  {								      \
	    result = __GCONV_ILLEGAL_INPUT;				      \
	    break;							      \
	  }								      \
	curcs = db;							      \
	++inptr;							      \
	continue;							      \
      }									      \
    else if (__builtin_expect (ch, 0) == SI)				      \
      {									      \
	/* Shift IN, change to SBCS converter.  */			      \
	if (curcs == sb)						      \
	  {								      \
	    result = __GCONV_ILLEGAL_INPUT;				      \
	    break;							      \
	  }								      \
	curcs = sb;							      \
	++inptr;							      \
	continue;							      \
      }									      \
									      \
    if (curcs == sb)							      \
      {									      \
	/* Use the IBM939 table for single byte.  */			      \
	res = __ibm939sb_to_ucs4[ch];					      \
	if (__builtin_expect (res == L'\0', 0) && ch != '\0')		      \
	  {								      \
	    /* This is an illegal character.  */			      \
	    STANDARD_FROM_LOOP_ERR_HANDLER (1);				      \
	  }								      \
	else								      \
	  {								      \
	    put32 (outptr, res);					      \
	    outptr += 4;						      \
	  }								      \
	++inptr;							      \
      }									      \
    else								      \
      {									      \
	/* Use the IBM939 table for double byte.  */			      \
	const struct gap *rp2 = __ibm939db_to_ucs4_idx;			      \
									      \
	assert (curcs == db);						      \
									      \
	if (__glibc_unlikely (inptr + 1 >= inend))			      \
	  {								      \
	    /* The second character is not available.  Store the	      \
	       intermediate result. */					      \
	    result = __GCONV_INCOMPLETE_INPUT;				      \
	    break;							      \
	  }								      \
									      \
	ch = (ch * 0x100) + inptr[1];					      \
	while (ch > rp2->end)						      \
	  ++rp2;							      \
									      \
	if (__builtin_expect (rp2->start == 0xffff, 0)			      \
	    || __builtin_expect (ch < rp2->start, 0)			      \
	    || (res = __ibm939db_to_ucs4[ch + rp2->idx],		      \
		__builtin_expect (res, L'\1') == L'\0' && ch != '\0'))	      \
	  {								      \
	    /* This is an illegal character.  */			      \
	    STANDARD_FROM_LOOP_ERR_HANDLER (2);				      \
	  }								      \
	else								      \
	  {								      \
	    put32 (outptr, res);					      \
	    outptr += 4;						      \
	  }								      \
	inptr += 2;							      \
      }									      \
  }
#define LOOP_NEED_FLAGS
#define EXTRA_LOOP_DECLS	, int *curcsp
#define INIT_PARAMS		int curcs = *curcsp & ~7
#define UPDATE_PARAMS		*curcsp = curcs
#include <iconv/loop.c>

/* Next, define the other direction */
#define MIN_NEEDED_INPUT	TO_LOOP_MIN_NEEDED_FROM
#define MAX_NEEDED_INPUT	TO_LOOP_MAX_NEEDED_FROM
#define MIN_NEEDED_OUTPUT	TO_LOOP_MIN_NEEDED_TO
#define MAX_NEEDED_OUTPUT	TO_LOOP_MAX_NEEDED_TO
#define LOOPFCT			TO_LOOP
#define BODY \
  {									      \
    uint32_t ch = get32 (inptr);					      \
    const struct gap *rp1 = __ucs4_to_ibm939sb_idx;			      \
    const struct gap *rp2 = __ucs4_to_ibm939db_idx;			      \
    const char *cp;							      \
									      \
    if (__glibc_unlikely (ch >= 0xffff))				      \
      {									      \
	UNICODE_TAG_HANDLER (ch, 4);					      \
	goto ibm939_invalid_char;					      \
      }									      \
									      \
    while (ch > rp1->end)						      \
      ++rp1;								      \
									      \
    /* Use the UCS4 table for single byte.  */				      \
    if (__builtin_expect (ch < rp1->start, 0)				      \
	|| (cp = __ucs4_to_ibm939sb[ch + rp1->idx],			      \
	    __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))	      \
      {									      \
	/* Use the UCS4 table for double byte.  */			      \
	while (ch > rp2->end)						      \
	  ++rp2;							      \
									      \
	if (__builtin_expect (ch < rp2->start, 0)			      \
	    || (cp = __ucs4_to_ibm939db[ch + rp2->idx],			      \
		__builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0'))	      \
	  {								      \
	    /* This is an illegal character.  */			      \
	  ibm939_invalid_char:						      \
	    STANDARD_TO_LOOP_ERR_HANDLER (4);				      \
	  }								      \
	else								      \
	  {								      \
	    if (curcs == sb)						      \
	      {								      \
		if (__glibc_unlikely (outptr + 1 > outend))		      \
		  {							      \
		    result = __GCONV_FULL_OUTPUT;			      \
		    break;						      \
		  }							      \
		*outptr++ = SO;						      \
		curcs = db;						      \
	      }								      \
									      \
	    if (__glibc_unlikely (outptr + 2 > outend))			      \
	      {								      \
		result = __GCONV_FULL_OUTPUT;				      \
		break;							      \
	      }								      \
	    *outptr++ = cp[0];						      \
	    *outptr++ = cp[1];						      \
	  }								      \
      }									      \
    else								      \
      {									      \
	if (curcs == db)						      \
	  {								      \
	    if (__glibc_unlikely (outptr + 1 > outend))			      \
	      {								      \
		result = __GCONV_FULL_OUTPUT;				      \
		break;							      \
	      }								      \
	    *outptr++ = SI;						      \
	  }								      \
									      \
	if (__glibc_unlikely (outptr + 1 > outend))			      \
	  {								      \
	    result = __GCONV_FULL_OUTPUT;				      \
	    break;							      \
	  }								      \
	if (ch == 0x7e)							      \
	  *outptr++ = 0xa0;						      \
	else if (ch == 0x5c)						      \
	  *outptr++ = 0xb2;						      \
	else								      \
	  *outptr++ = cp[0];						      \
	curcs = sb;							      \
      }									      \
									      \
    /* Now that we wrote the output increment the input pointer.  */	      \
    inptr += 4;								      \
  }
#define LOOP_NEED_FLAGS
#define EXTRA_LOOP_DECLS	, int *curcsp
#define INIT_PARAMS		int curcs = *curcsp & ~7
#define REINIT_PARAMS		curcs = *curcsp & ~7
#define UPDATE_PARAMS		*curcsp = curcs
#include <iconv/loop.c>

/* Now define the toplevel functions. */
#include <iconv/skeleton.c>
Commit	Line	Data
dbd10df7	1	/* Conversion to and from IBM939.
b168057a	2	Copyright (C) 2000-2015 Free Software Foundation, Inc.
dbd10df7 UD	3	This file is part of the GNU C Library.
	4	Contributed by Masahide Washizawa <washi@yamato.ibm.co.jp>, 2000.
	5
	6	The GNU C Library is free software; you can redistribute it and/or
41bdb6e2 AJ	7	modify it under the terms of the GNU Lesser General Public
	8	License as published by the Free Software Foundation; either
	9	version 2.1 of the License, or (at your option) any later version.
dbd10df7 UD	10
	11	The GNU C Library is distributed in the hope that it will be useful,
	12	but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
41bdb6e2	14	Lesser General Public License for more details.
dbd10df7	15
41bdb6e2	16	You should have received a copy of the GNU Lesser General Public
59ba27a6 PE	17	License along with the GNU C Library; if not, see
59ba27a6 PE	18	<http://www.gnu.org/licenses/>. */
dbd10df7	19
4de3b51e CD	20	/* IBM939 is designed for the representation of Japanese Latin/Kanji
	21	using a stateful EBCDIC encoding scheme. It is also known as
	22	CCSID 939 or CP939. See:
	23	https://www-01.ibm.com/software/globalization/ccsid/ccsid939.html */
	24
dbd10df7 UD	25	#include <dlfcn.h>
	26	#include <stdint.h>
	27	#include <wchar.h>
	28	#include <byteswap.h>
	29	#include "ibm939.h"
	30
	31	/* The shift sequences for this charset (it does not use ESC). */
	32	#define SI 0x0F /* Shift In, host code to turn DBCS off. */
	33	#define SO 0x0E /* Shift Out, host code to turn DBCS on. */
	34
	35	/* Definitions used in the body of the `gconv' function. */
	36	#define CHARSET_NAME "IBM939//"
	37	#define FROM_LOOP from_ibm939
	38	#define TO_LOOP to_ibm939
13e402e7	39	#define ONE_DIRECTION 0
faaa6f62 UD	40	#define FROM_LOOP_MIN_NEEDED_FROM 1
	41	#define FROM_LOOP_MAX_NEEDED_FROM 2
	42	#define FROM_LOOP_MIN_NEEDED_TO 4
	43	#define FROM_LOOP_MAX_NEEDED_TO 4
	44	#define TO_LOOP_MIN_NEEDED_FROM 4
	45	#define TO_LOOP_MAX_NEEDED_FROM 4
	46	#define TO_LOOP_MIN_NEEDED_TO 1
	47	#define TO_LOOP_MAX_NEEDED_TO 3
f1813b56 UD	48	#define PREPARE_LOOP \
	49	int save_curcs; \
	50	int *curcsp = &data->__statep->__count;
	51	#define EXTRA_LOOP_ARGS , curcsp
dbd10df7 UD	52
	53	/* Definitions of initialization and destructor function. */
	54	#define DEFINE_INIT 1
	55	#define DEFINE_FINI 1
	56
f1813b56 UD	57
	58	/* Since this is a stateful encoding we have to provide code which resets
	59	the output state to the initial state. This has to be done during the
	60	flushing. */
	61	#define EMIT_SHIFT_TO_INIT \
	62	if ((data->__statep->__count & ~7) != sb) \
	63	{ \
	64	if (FROM_DIRECTION) \
	65	data->__statep->__count &= 7; \
	66	else \
	67	{ \
f1813b56 UD	68	/* We are not in the initial state. To switch back we have \
f1813b56 UD	69	to emit `SI'. */ \
a1ffb40e	70	if (__glibc_unlikely (outbuf >= outend)) \
f1813b56 UD	71	/* We don't have enough room in the output buffer. */ \
	72	status = __GCONV_FULL_OUTPUT; \
	73	else \
	74	{ \
	75	/* Write out the shift sequence. */ \
	76	*outbuf++ = SI; \
f1813b56 UD	77	data->__statep->__count &= 7; \
	78	} \
	79	} \
	80	}
	81
	82
	83	/* Since we might have to reset input pointer we must be able to save
	84	and retore the state. */
	85	#define SAVE_RESET_STATE(Save) \
	86	if (Save) \
	87	save_curcs = *curcsp; \
	88	else \
	89	*curcsp = save_curcs
	90
dbd10df7 UD	91
	92	/* Current codeset type. */
	93	enum
	94	{
f1813b56 UD	95	sb = 0,
f1813b56 UD	96	db = 64
dbd10df7 UD	97	};
	98
	99	/* First, define the conversion function from IBM-939 to UCS4. */
faaa6f62 UD	100	#define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM
	101	#define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM
	102	#define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO
	103	#define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO
dbd10df7 UD	104	#define LOOPFCT FROM_LOOP
	105	#define BODY \
	106	{ \
	107	uint32_t ch = *inptr; \
	108	uint32_t res; \
dbd10df7 UD	109	\
	110	if (__builtin_expect (ch, 0) == SO) \
	111	{ \
dbd10df7 UD	112	/* Shift OUT, change to DBCS converter. */ \
	113	if (curcs == db) \
	114	{ \
	115	result = __GCONV_ILLEGAL_INPUT; \
	116	break; \
	117	} \
	118	curcs = db; \
	119	++inptr; \
f1813b56	120	continue; \
dbd10df7 UD	121	} \
	122	else if (__builtin_expect (ch, 0) == SI) \
	123	{ \
dbd10df7 UD	124	/* Shift IN, change to SBCS converter. */ \
	125	if (curcs == sb) \
	126	{ \
	127	result = __GCONV_ILLEGAL_INPUT; \
	128	break; \
	129	} \
	130	curcs = sb; \
	131	++inptr; \
f1813b56	132	continue; \
dbd10df7 UD	133	} \
dbd10df7 UD	134	\
f1813b56	135	if (curcs == sb) \
dbd10df7 UD	136	{ \
dbd10df7 UD	137	/* Use the IBM939 table for single byte. */ \
f1813b56	138	res = __ibm939sb_to_ucs4[ch]; \
e438a468	139	if (__builtin_expect (res == L'\0', 0) && ch != '\0') \
dbd10df7 UD	140	{ \
dbd10df7 UD	141	/* This is an illegal character. */ \
e438a468	142	STANDARD_FROM_LOOP_ERR_HANDLER (1); \
dbd10df7 UD	143	} \
	144	else \
	145	{ \
dbd10df7 UD	146	put32 (outptr, res); \
dbd10df7 UD	147	outptr += 4; \
dbd10df7	148	} \
f1813b56	149	++inptr; \
dbd10df7	150	} \
f1813b56	151	else \
dbd10df7 UD	152	{ \
dbd10df7 UD	153	/* Use the IBM939 table for double byte. */ \
f1813b56 UD	154	const struct gap *rp2 = __ibm939db_to_ucs4_idx; \
	155	\
	156	assert (curcs == db); \
	157	\
a1ffb40e	158	if (__glibc_unlikely (inptr + 1 >= inend)) \
dbd10df7 UD	159	{ \
	160	/* The second character is not available. Store the \
	161	intermediate result. */ \
	162	result = __GCONV_INCOMPLETE_INPUT; \
	163	break; \
	164	} \
	165	\
	166	ch = (ch * 0x100) + inptr[1]; \
	167	while (ch > rp2->end) \
	168	++rp2; \
	169	\
41488498	170	if (__builtin_expect (rp2->start == 0xffff, 0) \
dbd10df7 UD	171	\|\| __builtin_expect (ch < rp2->start, 0) \
	172	\|\| (res = __ibm939db_to_ucs4[ch + rp2->idx], \
	173	__builtin_expect (res, L'\1') == L'\0' && ch != '\0')) \
	174	{ \
	175	/* This is an illegal character. */ \
e438a468	176	STANDARD_FROM_LOOP_ERR_HANDLER (2); \
dbd10df7 UD	177	} \
	178	else \
	179	{ \
	180	put32 (outptr, res); \
	181	outptr += 4; \
dbd10df7	182	} \
f1813b56	183	inptr += 2; \
dbd10df7 UD	184	} \
	185	}
	186	#define LOOP_NEED_FLAGS
f1813b56 UD	187	#define EXTRA_LOOP_DECLS , int *curcsp
	188	#define INIT_PARAMS int curcs = *curcsp & ~7
	189	#define UPDATE_PARAMS *curcsp = curcs
dbd10df7 UD	190	#include <iconv/loop.c>
	191
	192	/* Next, define the other direction */
faaa6f62 UD	193	#define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM
	194	#define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM
	195	#define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO
	196	#define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO
dbd10df7 UD	197	#define LOOPFCT TO_LOOP
	198	#define BODY \
	199	{ \
	200	uint32_t ch = get32 (inptr); \
	201	const struct gap *rp1 = __ucs4_to_ibm939sb_idx; \
	202	const struct gap *rp2 = __ucs4_to_ibm939db_idx; \
	203	const char *cp; \
	204	\
a1ffb40e	205	if (__glibc_unlikely (ch >= 0xffff)) \
dbd10df7	206	{ \
601d2942	207	UNICODE_TAG_HANDLER (ch, 4); \
363dd976	208	goto ibm939_invalid_char; \
dbd10df7	209	} \
363dd976 UD	210	\
	211	while (ch > rp1->end) \
	212	++rp1; \
dbd10df7 UD	213	\
dbd10df7 UD	214	/* Use the UCS4 table for single byte. */ \
363dd976	215	if (__builtin_expect (ch < rp1->start, 0) \
dbd10df7 UD	216	\|\| (cp = __ucs4_to_ibm939sb[ch + rp1->idx], \
	217	__builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0')) \
	218	{ \
	219	/* Use the UCS4 table for double byte. */ \
	220	while (ch > rp2->end) \
	221	++rp2; \
	222	\
363dd976	223	if (__builtin_expect (ch < rp2->start, 0) \
dbd10df7 UD	224	\|\| (cp = __ucs4_to_ibm939db[ch + rp2->idx], \
	225	__builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0')) \
	226	{ \
	227	/* This is an illegal character. */ \
363dd976	228	ibm939_invalid_char: \
e438a468	229	STANDARD_TO_LOOP_ERR_HANDLER (4); \
dbd10df7 UD	230	} \
	231	else \
	232	{ \
f1813b56	233	if (curcs == sb) \
dbd10df7	234	{ \
a1ffb40e	235	if (__glibc_unlikely (outptr + 1 > outend)) \
dbd10df7 UD	236	{ \
	237	result = __GCONV_FULL_OUTPUT; \
	238	break; \
	239	} \
	240	*outptr++ = SO; \
	241	curcs = db; \
	242	} \
	243	\
a1ffb40e	244	if (__glibc_unlikely (outptr + 2 > outend)) \
dbd10df7 UD	245	{ \
	246	result = __GCONV_FULL_OUTPUT; \
	247	break; \
	248	} \
	249	*outptr++ = cp[0]; \
	250	*outptr++ = cp[1]; \
	251	} \
	252	} \
	253	else \
	254	{ \
	255	if (curcs == db) \
	256	{ \
a1ffb40e	257	if (__glibc_unlikely (outptr + 1 > outend)) \
dbd10df7 UD	258	{ \
	259	result = __GCONV_FULL_OUTPUT; \
	260	break; \
	261	} \
	262	*outptr++ = SI; \
	263	} \
	264	\
a1ffb40e	265	if (__glibc_unlikely (outptr + 1 > outend)) \
dbd10df7 UD	266	{ \
	267	result = __GCONV_FULL_OUTPUT; \
	268	break; \
	269	} \
	270	if (ch == 0x7e) \
	271	*outptr++ = 0xa0; \
	272	else if (ch == 0x5c) \
	273	*outptr++ = 0xb2; \
	274	else \
	275	*outptr++ = cp[0]; \
	276	curcs = sb; \
	277	} \
	278	\
	279	/* Now that we wrote the output increment the input pointer. */ \
	280	inptr += 4; \
	281	}
	282	#define LOOP_NEED_FLAGS
f1813b56 UD	283	#define EXTRA_LOOP_DECLS , int *curcsp
f1813b56 UD	284	#define INIT_PARAMS int curcs = *curcsp & ~7
4b1b449d	285	#define REINIT_PARAMS curcs = *curcsp & ~7
f1813b56	286	#define UPDATE_PARAMS *curcsp = curcs
dbd10df7 UD	287	#include <iconv/loop.c>
	288
	289	/* Now define the toplevel functions. */
	290	#include <iconv/skeleton.c>