mirror of
https://sourceware.org/git/glibc.git
synced 2024-11-18 02:50:09 +00:00
fd1b5c0fb6
2000-04-09 Ulrich Drepper <drepper@redhat.com> Implement handling of restartable conversion functions according to ISO C. * iconv/gconv.h (__gconv_fct): Add additional parameter. * iconv/gconv_int.h (__BUILTIN_TRANS): Likewise. * iconv/gconv.c: Pass additional parameter to conversion function. * iconv/gconv_simple.c (internal_ucs4_loop_single): New function. (internal_ucs4le_loop_single): New function. (__gconv_transform_ascii_internal): Define ONE_DIRECTION. (__gconv_transform_internal_ascii): Likewise. (__gconv_transform_internal_utf8): Likewise. (__gconv_transform_utf8_internal): Likewise. (__gconv_transform_ucs2_internal): Likewise. (__gconv_transform_internal_ucs2): Likewise. (__gconv_transform_ucs2reverse_internal): Likewise. (__gconv_transform_internal_ucs2reverse): Likewise. (internal_ucs4le_loop_unaligned): Before return __GCONV_INCOMPLETE_INPUT check that the remaining bytes really form a valid character. Otherwise return __GCONV_ILLEGAL_INPUT. (__gconv_transform_utf8_internal): Define STORE_REST and UNPACK_BYTES. * iconv/loop.c: Fit in definition of function to convert one character for processing of left-over bytes from the state object. * iconv/skeleton.c (gconv): Rename inbuf to inptrp and inbufend to inend to match names in loop functions. (RESET_INPUT_BUFFER): Change apprpriately. (gconv): If needed, call function to process bytes from the state object. Similar at the end: store left over bytes if input is incomplete. Take extra argument and add new argument to all calls of the conversion function. * iconvdata/iso-2022-cn.c: Adjust numeric values used to store information in the state object to not conflict with length count. * iconvdata/iso-2022-jp.c: Likewise. * iconvdata/iso-2022-kr.c: Likewise. * iconvdata/unicode.c: Adjust for change change in parameters of skeleton function. * iconvdata/utf-16.c: Likewise. * libio/iofwide.c: Add new parameter to all calls of conversion function. * wcsmbs/btowc.c: Likewise. * wcsmbs/mbrtowc.c: Likewise. * wcsmbs/mbsnrtowcs.c: Likewise. * wcsmbs/mbsrtowcs.c: Likewise. * wcsmbs/wcrtomb.c: Likewise. * wcsmbs/wcsnrtombs.c: Likewise. * wcsmbs/wcsrtombs.c: Likewise. * wcsmbs/wctob.c: Likewise. * iconvdata/gbgbk.c: Always define MAX_NEEDED_OUTPUT and MAX_NEEDED_INPUT.
401 lines
14 KiB
C
401 lines
14 KiB
C
/* Conversion module for ISO-2022-CN.
|
|
Copyright (C) 1999, 2000 Free Software Foundation, Inc.
|
|
This file is part of the GNU C Library.
|
|
Contributed by Ulrich Drepper <drepper@cygnus.com>, 1999.
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU Library General Public License as
|
|
published by the Free Software Foundation; either version 2 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
|
|
Library General Public License for more details.
|
|
|
|
You should have received a copy of the GNU Library General Public
|
|
License along with the GNU C Library; see the file COPYING.LIB. If not,
|
|
write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
Boston, MA 02111-1307, USA. */
|
|
|
|
#include <gconv.h>
|
|
#include <stdint.h>
|
|
#include <string.h>
|
|
#include "gb2312.h"
|
|
#include "cns11643l1.h"
|
|
#include "cns11643l2.h"
|
|
|
|
#include <assert.h>
|
|
|
|
/* This makes obvious what everybody knows: 0x1b is the Esc character. */
|
|
#define ESC 0x1b
|
|
|
|
/* We have single-byte shift-in and shift-out sequences, and the single
|
|
shift sequence SS2 which replaces the SS2 designation for the next
|
|
two bytes. */
|
|
#define SI 0x0f
|
|
#define SO 0x0e
|
|
#define SS2_0 ESC
|
|
#define SS2_1 0x4e
|
|
|
|
/* Definitions used in the body of the `gconv' function. */
|
|
#define CHARSET_NAME "ISO-2022-CN//"
|
|
#define DEFINE_INIT 1
|
|
#define DEFINE_FINI 1
|
|
#define FROM_LOOP from_iso2022cn_loop
|
|
#define TO_LOOP to_iso2022cn_loop
|
|
#define MIN_NEEDED_FROM 1
|
|
#define MAX_NEEDED_FROM 4
|
|
#define MIN_NEEDED_TO 4
|
|
#define MAX_NEEDED_TO 4
|
|
#define PREPARE_LOOP \
|
|
int save_set; \
|
|
int *setp = &data->__statep->__count;
|
|
#define EXTRA_LOOP_ARGS , setp
|
|
|
|
|
|
/* The COUNT element of the state keeps track of the currently selected
|
|
character set. The possible values are: */
|
|
enum
|
|
{
|
|
ASCII_set = 0,
|
|
GB2312_set = 8,
|
|
CNS11643_1_set = 16,
|
|
CNS11643_2_set = 24,
|
|
CURRENT_SEL_MASK = 24,
|
|
GB2312_ann = 32,
|
|
CNS11643_1_ann = 64,
|
|
CNS11643_2_ann = 128,
|
|
CURRENT_ANN_MASK = 224
|
|
};
|
|
|
|
|
|
/* 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 != ASCII_set) \
|
|
{ \
|
|
if (FROM_DIRECTION) \
|
|
/* It's easy, we don't have to emit anything, we just reset the \
|
|
state for the input. */ \
|
|
data->__statep->__count = ASCII_set; \
|
|
else \
|
|
{ \
|
|
unsigned char *outbuf = data->__outbuf; \
|
|
\
|
|
/* We are not in the initial state. To switch back we have \
|
|
to emit `SI'. */ \
|
|
if (outbuf == data->__outbufend) \
|
|
/* We don't have enough room in the output buffer. */ \
|
|
status = __GCONV_FULL_OUTPUT; \
|
|
else \
|
|
{ \
|
|
/* Write out the shift sequence. */ \
|
|
*outbuf++ = SI; \
|
|
if (data->__is_last) \
|
|
*written += 1; \
|
|
data->__outbuf = outbuf; \
|
|
data->__statep->__count = ASCII_set; \
|
|
} \
|
|
} \
|
|
}
|
|
|
|
|
|
/* 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_set = *setp; \
|
|
else \
|
|
*setp = save_set
|
|
|
|
|
|
/* First define the conversion function from ISO-2022-CN to UCS4. */
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
|
#define MAX_NEEDED_INPUT MAX_NEEDED_FROM
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
|
#define LOOPFCT FROM_LOOP
|
|
#define BODY \
|
|
{ \
|
|
uint32_t ch = *inptr; \
|
|
\
|
|
/* This is a 7bit character set, disallow all 8bit characters. */ \
|
|
if (ch > 0x7f) \
|
|
{ \
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
|
break; \
|
|
} \
|
|
\
|
|
/* Recognize escape sequences. */ \
|
|
if (ch == ESC) \
|
|
{ \
|
|
/* There are two kinds of escape sequences we have to handle: \
|
|
- those announcing the use of GB and CNS characters on the \
|
|
line; we can simply ignore them \
|
|
- the initial byte of the SS2 sequence. \
|
|
*/ \
|
|
if (NEED_LENGTH_TEST \
|
|
&& (inptr + 1 > inend \
|
|
|| (inptr[1] == '$' \
|
|
&& (inptr + 2 > inend \
|
|
|| (inptr[2] == ')' && inptr + 3 > inend) \
|
|
|| (inptr[2] == '*' && inptr + 3 > inend))) \
|
|
|| (inptr[1] == SS2_1 && inptr + 3 > inend))) \
|
|
{ \
|
|
result = __GCONV_EMPTY_INPUT; \
|
|
break; \
|
|
} \
|
|
if (inptr[1] == '$' \
|
|
&& ((inptr[2] == ')' && (inptr[3] == 'A' || inptr[3] == 'G')) \
|
|
|| (inptr[2] == '*' && inptr[3] == 'H'))) \
|
|
{ \
|
|
/* OK, we accept those character sets. */ \
|
|
if (inptr[3] == 'A') \
|
|
ann = GB2312_ann; \
|
|
else if (inptr[3] == 'G') \
|
|
ann = CNS11643_1_ann; \
|
|
inptr += 4; \
|
|
continue; \
|
|
} \
|
|
} \
|
|
else if (ch == SO) \
|
|
{ \
|
|
/* Switch to use GB2312 or CNS 11643 plane 1, depending on which \
|
|
S0 designation came last. The only problem is what to do with \
|
|
faulty input files where no designator came. \
|
|
XXX For now I'll default to use GB2312. If this is not the \
|
|
best behaviour (e.g., we should flag an error) let me know. */ \
|
|
++inptr; \
|
|
set = ann == CNS11643_1_ann ? CNS11643_1_set : GB2312_set; \
|
|
continue; \
|
|
} \
|
|
else if (ch == SI) \
|
|
{ \
|
|
/* Switch to use ASCII. */ \
|
|
++inptr; \
|
|
set = ASCII_set; \
|
|
continue; \
|
|
} \
|
|
\
|
|
if (ch == ESC && inptr[1] == SS2_1) \
|
|
{ \
|
|
/* This is a character from CNS 11643 plane 2. \
|
|
XXX We could test here whether the use of this character \
|
|
set was announced. */ \
|
|
inptr += 2; \
|
|
ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \
|
|
if (ch == __UNKNOWN_10646_CHAR) \
|
|
{ \
|
|
inptr -= 2; \
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
|
break; \
|
|
} \
|
|
} \
|
|
else if (set == ASCII_set) \
|
|
{ \
|
|
/* Almost done, just advance the input pointer. */ \
|
|
++inptr; \
|
|
} \
|
|
else \
|
|
{ \
|
|
/* That's pretty easy, we have a dedicated functions for this. */ \
|
|
if (set == GB2312_set) \
|
|
ch = gb2312_to_ucs4 (&inptr, \
|
|
NEED_LENGTH_TEST ? inend - inptr : 2, 0); \
|
|
else \
|
|
{ \
|
|
assert (set == CNS11643_1_set); \
|
|
ch = cns11643l1_to_ucs4 (&inptr, \
|
|
NEED_LENGTH_TEST ? inend - inptr : 2, 0);\
|
|
} \
|
|
\
|
|
if (NEED_LENGTH_TEST && ch == 0) \
|
|
{ \
|
|
result = __GCONV_EMPTY_INPUT; \
|
|
break; \
|
|
} \
|
|
else if (ch == __UNKNOWN_10646_CHAR) \
|
|
{ \
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
|
break; \
|
|
} \
|
|
} \
|
|
\
|
|
put32 (outptr, ch); \
|
|
outptr += 4; \
|
|
}
|
|
#define EXTRA_LOOP_DECLS , int *setp
|
|
#define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \
|
|
int ann = *setp & CURRENT_ANN_MASK
|
|
#define UPDATE_PARAMS *setp = set | ann
|
|
#include <iconv/loop.c>
|
|
|
|
|
|
/* Next, define the other direction. */
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_TO
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_FROM
|
|
#define MAX_NEEDED_OUTPUT MAX_NEEDED_FROM
|
|
#define LOOPFCT TO_LOOP
|
|
#define BODY \
|
|
{ \
|
|
uint32_t ch; \
|
|
size_t written = 0; \
|
|
\
|
|
ch = get32 (inptr); \
|
|
\
|
|
/* First see whether we can write the character using the currently \
|
|
selected character set. */ \
|
|
if (ch < 0x80) \
|
|
{ \
|
|
if (set != ASCII_set) \
|
|
{ \
|
|
*outptr++ = SI; \
|
|
set = ASCII_set; \
|
|
if (NEED_LENGTH_TEST && outptr == outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
} \
|
|
\
|
|
*outptr++ = ch; \
|
|
written = 1; \
|
|
\
|
|
/* At the end of the line we have to clear the `ann' flags since \
|
|
every line must contain this information again. */ \
|
|
if (ch == L'\n') \
|
|
ann = 0; \
|
|
} \
|
|
else \
|
|
{ \
|
|
char buf[2]; \
|
|
int used; \
|
|
\
|
|
if (set == GB2312_set || (ann & CNS11643_1_ann) == 0) \
|
|
{ \
|
|
written = ucs4_to_gb2312 (ch, buf, 2); \
|
|
used = GB2312_set; \
|
|
} \
|
|
else \
|
|
{ \
|
|
written = ucs4_to_cns11643l1 (ch, buf, 2); \
|
|
used = CNS11643_1_set; \
|
|
} \
|
|
\
|
|
if (written == __UNKNOWN_10646_CHAR) \
|
|
{ \
|
|
/* Cannot convert it using the currently selected SO set. \
|
|
Next try the SS2 set. */ \
|
|
written = ucs4_to_cns11643l2 (ch, buf, 2); \
|
|
if (written != __UNKNOWN_10646_CHAR) \
|
|
/* Yep, that worked. */ \
|
|
used = CNS11643_2_set; \
|
|
else \
|
|
{ \
|
|
/* Well, see whether we have to change the SO set. */ \
|
|
if (set == GB2312_set) \
|
|
written = ucs4_to_cns11643l1 (ch, buf, 2); \
|
|
else \
|
|
written = ucs4_to_gb2312 (ch, buf, 2); \
|
|
\
|
|
if (written != __UNKNOWN_10646_CHAR) \
|
|
/* Oh well, then switch SO. */ \
|
|
used = GB2312_set + CNS11643_1_set - set; \
|
|
else \
|
|
{ \
|
|
/* Even this does not work. Error. */ \
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
|
break; \
|
|
} \
|
|
} \
|
|
} \
|
|
assert (written == 2); \
|
|
\
|
|
/* See whether we have to emit an escape sequence. */ \
|
|
if (set != used) \
|
|
{ \
|
|
/* First see whether we announced that we use this \
|
|
character set. */ \
|
|
if ((ann & (2 << used)) == 0) \
|
|
{ \
|
|
const char *escseq; \
|
|
\
|
|
if (NEED_LENGTH_TEST && outptr + 4 > outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
\
|
|
assert (used >= 1 && used <= 3); \
|
|
escseq = "\e$)A\e$)G\e$*H" + (used - 1) * 4; \
|
|
*outptr++ = *escseq++; \
|
|
*outptr++ = *escseq++; \
|
|
*outptr++ = *escseq++; \
|
|
*outptr++ = *escseq++; \
|
|
\
|
|
if (used == GB2312_set) \
|
|
ann = (ann & CNS11643_2_ann) | GB2312_ann; \
|
|
else if (used == CNS11643_1_set) \
|
|
ann = (ann & CNS11643_2_ann) | CNS11643_1_ann; \
|
|
else \
|
|
ann |= CNS11643_2_ann; \
|
|
} \
|
|
\
|
|
if (used == CNS11643_2_set) \
|
|
{ \
|
|
if (outptr + 2 > outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
*outptr++ = SS2_0; \
|
|
*outptr++ = SS2_1; \
|
|
} \
|
|
else \
|
|
{ \
|
|
/* We only have to emit something is currently ASCII is \
|
|
selected. Otherwise we are switching within the \
|
|
SO charset. */ \
|
|
if (set == ASCII_set) \
|
|
{ \
|
|
if (outptr + 1 > outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
*outptr++ = SO; \
|
|
} \
|
|
} \
|
|
\
|
|
/* Always test the length here since we have used up all the \
|
|
guaranteed output buffer slots. */ \
|
|
if (outptr + 2 > outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
} \
|
|
else if (NEED_LENGTH_TEST && outptr + 2 > outend) \
|
|
{ \
|
|
result = __GCONV_FULL_OUTPUT; \
|
|
break; \
|
|
} \
|
|
\
|
|
*outptr++ = buf[0]; \
|
|
*outptr++ = buf[1]; \
|
|
} \
|
|
\
|
|
/* Now that we wrote the output increment the input pointer. */ \
|
|
inptr += 4; \
|
|
}
|
|
#define EXTRA_LOOP_DECLS , int *setp
|
|
#define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \
|
|
int ann = *setp & CURRENT_ANN_MASK
|
|
#define UPDATE_PARAMS *setp = set | ann
|
|
#include <iconv/loop.c>
|
|
|
|
|
|
/* Now define the toplevel functions. */
|
|
#include <iconv/skeleton.c>
|