qiurui
/
glibc
mirror of git://sourceware.org/git/glibc.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Update. 

* iconvdata/iso-2022-jp.c: Correct handling of G2 set.
	Patch by Shinya Hanataka <hanataka@abyss.rim.or.jp>.
This commit is contained in:
Ulrich Drepper 1999-01-16 18:38:41 +00:00
parent 482eec0d11
commit 70e57b1382
2 changed files with 159 additions and 120 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
ChangeLog
View File

@ -1,5 +1,8 @@
1999-01-16 Ulrich Drepper <drepper@cygnus.com> 1999-01-16 Ulrich Drepper <drepper@cygnus.com>
* iconvdata/iso-2022-jp.c: Correct handling of G2 set.
Patch by Shinya Hanataka <hanataka@abyss.rim.or.jp>.
* elf/Makefile (tests): Add loadtest. * elf/Makefile (tests): Add loadtest.
Add rules to generate test modules. Add rules to generate test modules.
* Makeconfig (+link): Add $(LDFLAGS-$(@F)) to command line. * Makeconfig (+link): Add $(LDFLAGS-$(@F)) to command line.

276
iconvdata/iso-2022-jp.c
View File

@ -95,7 +95,14 @@ enum
JISX0201_Kana_set, JISX0201_Kana_set,
GB2312_set, GB2312_set,
KSC5601_set, KSC5601_set,
JISX0212_set, JISX0212_set
};
/* The second value stored is the designation of the G2 set. The following
values are possible: */
enum
{
UNSPECIFIED_set = 0,
ISO88591_set, ISO88591_set,
ISO88597_set ISO88597_set
}; };
@ -187,12 +194,13 @@ gconv_end (struct gconv_step *data)
\ \
if (dir == from_iso2022jp) \ if (dir == from_iso2022jp) \
/* It's easy, we don't have to emit anything, we just reset the \ /* It's easy, we don't have to emit anything, we just reset the \
state for the input. */ \ state for the input. Note that this also clears the G2 \
designation. */ \
data->statep->count = ASCII_set; \ data->statep->count = ASCII_set; \
else \ else \
{ \ { \
char *outbuf = data->outbuf; \ char *outbuf = data->outbuf; \
\ \
/* We are not in the initial state. To switch back we have \ /* We are not in the initial state. To switch back we have \
to emit the sequence `Esc ( B'. */ \ to emit the sequence `Esc ( B'. */ \
if (outbuf + 3 > data->outbufend) \ if (outbuf + 3 > data->outbufend) \
@ -205,6 +213,7 @@ gconv_end (struct gconv_step *data)
*outbuf++ = '('; \ *outbuf++ = '('; \
*outbuf++ = 'B'; \ *outbuf++ = 'B'; \
data->outbuf = outbuf; \ data->outbuf = outbuf; \
/* Note that this also clears the G2 designation. */ \
data->statep->count = ASCII_set; \ data->statep->count = ASCII_set; \
} \ } \
} \ } \
@ -319,23 +328,50 @@ gconv_end (struct gconv_step *data)
if (inptr[2] == 'A') \ if (inptr[2] == 'A') \
{ \ { \
/* ISO 8859-1-GR selected. */ \ /* ISO 8859-1-GR selected. */ \
set = ISO88591_set; \ set2 = ISO88591_set; \
inptr += 3; \ inptr += 3; \
continue; \ continue; \
} \ } \
else if (inptr[2] == 'F') \ else if (inptr[2] == 'F') \
{ \ { \
/* ISO 8859-7-GR selected. */ \ /* ISO 8859-7-GR selected. */ \
set = ISO88597_set; \ set2 = ISO88597_set; \
inptr += 3; \ inptr += 3; \
continue; \ continue; \
} \ } \
} \ } \
} \ } \
\ \
if (set == ASCII_set \ if (ch == ESC && var == iso2022jp2 && inptr[1] == 'N') \
|| (var < ISO88591_set && (ch < 0x21 || ch == 0x7f)) \ { \
|| (var >= ISO88591_set && ch < 0x20)) \ if (set2 == ISO88591_set) \
{ \
ch = inptr[2] | 0x80; \
inptr += 3; \
} \
else if (set2 == ISO88597_set) \
{ \
/* We use the table from the ISO 8859-7 module. */ \
if (inptr[2] < 0x20 || inptr[2] > 0x80) \
{ \
result = GCONV_ILLEGAL_INPUT; \
break; \
} \
ch = iso88597_to_ucs4[inptr[2] - 0x20]; \
if (ch == 0) \
{ \
result = GCONV_ILLEGAL_INPUT; \
break; \
} \
inptr += 3; \
} \
else \
{ \
result = GCONV_ILLEGAL_INPUT; \
break; \
} \
} \
else if (set == ASCII_set || (ch < 0x21 || ch == 0x7f)) \
/* Almost done, just advance the input pointer. */ \ /* Almost done, just advance the input pointer. */ \
++inptr; \ ++inptr; \
else if (set == JISX0201_Roman_set) \ else if (set == JISX0201_Roman_set) \
@ -360,24 +396,6 @@ gconv_end (struct gconv_step *data)
} \ } \
++inptr; \ ++inptr; \
} \ } \
else if (set == ISO88591_set) \
{ \
/* This is quite easy. All characters are defined and the \
ISO 10646 value is computed by adding 0x80. */ \
ch |= 0x80; \
++inptr; \
} \
else if (set == ISO88597_set) \
{ \
/* We use the table from the ISO 8859-7 module. */ \
ch = iso88597_to_ucs4[(ch & 0x7f) - 0x20]; \
if (ch == 0) \
{ \
result = GCONV_ILLEGAL_INPUT; \
break; \
} \
++inptr; \
} \
else \ else \
{ \ { \
if (set == JISX0208_1978_set || set == JISX0208_1983_set) \ if (set == JISX0208_1978_set || set == JISX0208_1983_set) \
@ -419,8 +437,8 @@ gconv_end (struct gconv_step *data)
*((uint32_t *) outptr)++ = ch; \ *((uint32_t *) outptr)++ = ch; \
} }
#define EXTRA_LOOP_DECLS , enum variant var, int *setp #define EXTRA_LOOP_DECLS , enum variant var, int *setp
#define INIT_PARAMS int set = *setp #define INIT_PARAMS int set = *setp % 0x100, set2 = *setp / 0x100
#define UPDATE_PARAMS *setp = set #define UPDATE_PARAMS *setp = (set2 << 8) + set
#include <iconv/loop.c> #include <iconv/loop.c>
@ -448,6 +466,9 @@ gconv_end (struct gconv_step *data)
*outptr++ = ch; \ *outptr++ = ch; \
written = 1; \ written = 1; \
} \ } \
/* At the beginning of a line, G2 designation is cleared. */ \
if (var == iso2022jp2 && ch == 0x0a) \
set2 = UNSPECIFIED_set; \
} \ } \
else if (set == JISX0201_Roman_set) \ else if (set == JISX0201_Roman_set) \
{ \ { \
@ -473,30 +494,6 @@ gconv_end (struct gconv_step *data)
else \ else \
written = UNKNOWN_10646_CHAR; \ written = UNKNOWN_10646_CHAR; \
} \ } \
else if (set == ISO88591_set) \
{ \
if (ch >= 0x80 && ch <= 0xff) \
{ \
*outptr++ = ch; \
written = 1; \
} \
} \
else if (set == ISO88597_set) \
{ \
const struct gap *rp = from_idx; \
\
while (ch > rp->end) \
++rp; \
if (ch >= rp->start) \
{ \
unsigned char res = iso88597_from_ucs4[ch + rp->idx]; \
if (res != '\0') \
{ \
*outptr++ = res | 0x80; \
written = 1; \
} \
} \
} \
else \ else \
{ \ { \
if (set == JISX0208_1978_set || set == JISX0208_1983_set) \ if (set == JISX0208_1978_set || set == JISX0208_1983_set) \
@ -528,6 +525,38 @@ gconv_end (struct gconv_step *data)
outptr += written; \ outptr += written; \
} \ } \
\ \
if (written == UNKNOWN_10646_CHAR || written == 0) \
{ \
if (set2 == ISO88591_set) \
{ \
if (ch >= 0x80 && ch <= 0xff) \
{ \
*outptr++ = ESC; \
*outptr++ = 'N'; \
*outptr++ = ch & 0x7f; \
written = 3; \
} \
} \
else if (set2 == ISO88597_set) \
{ \
const struct gap *rp = from_idx; \
\
while (ch > rp->end) \
++rp; \
if (ch >= rp->start) \
{ \
unsigned char res = iso88597_from_ucs4[ch - 0xa0 + rp->idx]; \
if (res != '\0') \
{ \
*outptr++ = ESC; \
*outptr++ = 'N'; \
*outptr++ = res; \
written = 3; \
} \
} \
} \
} \
\
if (written == UNKNOWN_10646_CHAR || written == 0) \ if (written == UNKNOWN_10646_CHAR || written == 0) \
{ \ { \
/* Either this is an unknown character or we have to switch \ /* Either this is an unknown character or we have to switch \
@ -541,31 +570,25 @@ gconv_end (struct gconv_step *data)
later and now simply use a fixed order in which we test for \ later and now simply use a fixed order in which we test for \
availability */ \ availability */ \
\ \
/* First test whether we have at least three more bytes for \
the escape sequence. The two charsets which require four \
bytes will be handled later. */ \
if (NEED_LENGTH_TEST && outptr + 3 > outend) \
{ \
result = GCONV_FULL_OUTPUT; \
break; \
} \
\
if (ch <= 0x7f) \ if (ch <= 0x7f) \
{ \ { \
/* We must encode using ASCII. First write out the \ /* We must encode using ASCII. First write out the \
escape sequence. */ \ escape sequence. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 4 > outend) \
*outptr++ = '('; \
*outptr++ = 'B'; \
set = ASCII_set; \
\
if (NEED_LENGTH_TEST && outptr == outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '('; \
*outptr++ = 'B'; \
set = ASCII_set; \
*outptr++ = ch; \ *outptr++ = ch; \
\
/* At the beginning of a line, G2 designation is cleared. */ \
if (var == iso2022jp2 && ch == 0x0a) \
set2 = UNSPECIFIED_set; \
} \ } \
else \ else \
{ \ { \
@ -580,17 +603,16 @@ gconv_end (struct gconv_step *data)
if (written != UNKNOWN_10646_CHAR && buf[0] < 0x80) \ if (written != UNKNOWN_10646_CHAR && buf[0] < 0x80) \
{ \ { \
/* We use JIS X 0201. */ \ /* We use JIS X 0201. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 4 > outend) \
*outptr++ = '('; \
*outptr++ = 'J'; \
set = JISX0201_Roman_set; \
\
if (NEED_LENGTH_TEST && outptr == outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '('; \
*outptr++ = 'J'; \
set = JISX0201_Roman_set; \
*outptr++ = buf[0]; \ *outptr++ = buf[0]; \
} \ } \
else \ else \
@ -599,17 +621,16 @@ gconv_end (struct gconv_step *data)
if (written != UNKNOWN_10646_CHAR) \ if (written != UNKNOWN_10646_CHAR) \
{ \ { \
/* We use JIS X 0208. */ \ /* We use JIS X 0208. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 5 > outend) \
*outptr++ = '$'; \
*outptr++ = 'B'; \
set = JISX0208_1983_set; \
\
if (NEED_LENGTH_TEST && outptr + 2 > outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '$'; \
*outptr++ = 'B'; \
set = JISX0208_1983_set; \
*outptr++ = buf[0]; \ *outptr++ = buf[0]; \
*outptr++ = buf[1]; \ *outptr++ = buf[1]; \
} \ } \
@ -625,7 +646,7 @@ gconv_end (struct gconv_step *data)
if (written != UNKNOWN_10646_CHAR) \ if (written != UNKNOWN_10646_CHAR) \
{ \ { \
/* We use JIS X 0212. */ \ /* We use JIS X 0212. */ \
if (NEED_LENGTH_TEST && outptr + 4 > outend) \ if (NEED_LENGTH_TEST && outptr + 6 > outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
@ -635,13 +656,6 @@ gconv_end (struct gconv_step *data)
*outptr++ = '('; \ *outptr++ = '('; \
*outptr++ = 'D'; \ *outptr++ = 'D'; \
set = JISX0212_set; \ set = JISX0212_set; \
\
if (NEED_LENGTH_TEST && outptr + 2 > outend) \
{ \
result = GCONV_FULL_OUTPUT; \
break; \
} \
\
*outptr++ = buf[0]; \ *outptr++ = buf[0]; \
*outptr++ = buf[1]; \ *outptr++ = buf[1]; \
} \ } \
@ -651,33 +665,33 @@ gconv_end (struct gconv_step *data)
if (written != UNKNOWN_10646_CHAR && buf[0] >= 0x80) \ if (written != UNKNOWN_10646_CHAR && buf[0] >= 0x80) \
{ \ { \
/* We use JIS X 0201. */ \ /* We use JIS X 0201. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 4 > outend) \
*outptr++ = '('; \
*outptr++ = 'I'; \
set = JISX0201_Kana_set; \
\
if (NEED_LENGTH_TEST && outptr == outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '('; \
*outptr++ = 'I'; \
set = JISX0201_Kana_set; \
*outptr++ = buf[0] - 0x80; \ *outptr++ = buf[0] - 0x80; \
} \ } \
else if (ch != 0xa5 && ch >= 0x80 && ch <= 0xff) \ else if (ch != 0xa5 && ch >= 0x80 && ch <= 0xff) \
{ \ { \
/* ISO 8859-1 upper half. */ \ /* ISO 8859-1 upper half. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 6 > outend) \
*outptr++ = '.'; \
*outptr++ = 'A'; \
set = ISO88591_set; \
\
if (NEED_LENGTH_TEST && outptr == outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '.'; \
*outptr++ = 'A'; \
set2 = ISO88591_set; \
*outptr++ = ESC; \
*outptr++ = 'N'; \
*outptr++ = ch; \ *outptr++ = ch; \
} \ } \
else \ else \
@ -686,28 +700,27 @@ gconv_end (struct gconv_step *data)
if (written != UNKNOWN_10646_CHAR) \ if (written != UNKNOWN_10646_CHAR) \
{ \ { \
/* We use GB 2312. */ \ /* We use GB 2312. */ \
*outptr++ = ESC; \ if (NEED_LENGTH_TEST && outptr + 5 > outend) \
*outptr++ = '$'; \
*outptr++ = 'A'; \
set = GB2312_set; \
\
if (NEED_LENGTH_TEST && outptr + 2 > outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
} \ } \
\ \
*outptr++ = ESC; \
*outptr++ = '$'; \
*outptr++ = 'A'; \
set = GB2312_set; \
*outptr++ = buf[0]; \ *outptr++ = buf[0]; \
*outptr++ = buf[1]; \ *outptr++ = buf[1]; \
} \ } \
else \ else \
{ \ { \
written = ucs4_to_ksc5601 (ch, buf, 2); \ written = ucs4_to_ksc5601 (ch, buf, 2); \
if (written != UNKNOWN_10646_CHAR) \ if (written != UNKNOWN_10646_CHAR) \
{ \ { \
/* We use KSC 5601. */ \ /* We use KSC 5601. */ \
if (NEED_LENGTH_TEST \ if (NEED_LENGTH_TEST \
&& outptr + 4 > outend) \ && outptr + 6 > outend) \
{ \ { \
result = GCONV_FULL_OUTPUT; \ result = GCONV_FULL_OUTPUT; \
break; \ break; \
@ -717,21 +730,44 @@ gconv_end (struct gconv_step *data)
*outptr++ = '('; \ *outptr++ = '('; \
*outptr++ = 'C'; \ *outptr++ = 'C'; \
set = KSC5601_set; \ set = KSC5601_set; \
\
if (NEED_LENGTH_TEST \
&& outptr + 2 > outend) \
{ \
result = GCONV_FULL_OUTPUT; \
break; \
} \
\
*outptr++ = buf[0]; \ *outptr++ = buf[0]; \
*outptr++ = buf[1]; \ *outptr++ = buf[1]; \
} \ } \
else \ else \
{ \ { \
result = GCONV_ILLEGAL_INPUT; \ const struct gap *rp = from_idx; \
break; \ unsigned char gch = 0; \
\
while (ch > rp->end) \
++rp; \
if (ch >= rp->start) \
{ \
ch = ch - 0xa0 + rp->idx; \
gch = iso88597_from_ucs4[ch]; \
} \
\
if (gch != 0) \
{ \
/* We use ISO 8859-7 greek. */ \
if (NEED_LENGTH_TEST \
&& outptr + 6 > outend) \
{ \
result = GCONV_FULL_OUTPUT; \
break; \
} \
*outptr++ = ESC; \
*outptr++ = '.'; \
*outptr++ = 'F'; \
set2 = ISO88597_set; \
*outptr++ = ESC; \
*outptr++ = 'N'; \
*outptr++ = gch; \
} \
else \
{ \
result = GCONV_ILLEGAL_INPUT; \
break; \
} \
} \ } \
} \ } \
} \ } \
@ -745,8 +781,8 @@ gconv_end (struct gconv_step *data)
inptr += 4; \ inptr += 4; \
} }
#define EXTRA_LOOP_DECLS , enum variant var, int *setp #define EXTRA_LOOP_DECLS , enum variant var, int *setp
#define INIT_PARAMS int set = *setp #define INIT_PARAMS int set = *setp % 0x100, set2 = *setp / 0x100
#define UPDATE_PARAMS *setp = set #define UPDATE_PARAMS *setp = (set2 << 8) + set
#include <iconv/loop.c> #include <iconv/loop.c>