From 66175fa825828d3a012c34777fd9e7916b7aa10b Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Sat, 16 May 1998 00:03:15 +0000 Subject: Update. * iconv/loop.c: Let user add code to use and set paramaters. * iconvdata/iso-2022-kr.c: Correct to-loop. * iconvdata/ksc5601.h (ksc5601_to_ucs4): Use correct value to find hanja. * iconvdata/TESTS: Add ISO-2022-KR test. * iconvdata/testdata/ISO-2022-KR: New file. * wcsmbs/btowc.c: Likewise. * wcsmbs/mbrtowcs.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/iso-2022-kr.c | 146 ++++++++++++++++++++---------------------------- 1 file changed, 62 insertions(+), 84 deletions(-) (limited to 'iconvdata/iso-2022-kr.c') diff --git a/iconvdata/iso-2022-kr.c b/iconvdata/iso-2022-kr.c index 166fa15072..6746f9c2ac 100644 --- a/iconvdata/iso-2022-kr.c +++ b/iconvdata/iso-2022-kr.c @@ -44,7 +44,7 @@ #define MAX_NEEDED_TO 4 #define PREPARE_LOOP \ int save_set; \ - int set = data->statep->count; \ + int *setp = &data->statep->count; \ if (!FROM_DIRECTION && !data->internal_use && data->invocation_counter == 0)\ { \ /* Emit the designator sequence. */ \ @@ -56,7 +56,7 @@ *outbuf++ = ')'; \ *outbuf++ = 'C'; \ } -#define EXTRA_LOOP_ARGS , set +#define EXTRA_LOOP_ARGS , setp /* The COUNT element of the state keeps track of the currently selected @@ -72,27 +72,27 @@ enum the output state to the initial state. This has to be done during the flushing. */ #define EMIT_SHIFT_TO_INIT \ - if (data->statep->count != 0) \ + if (data->statep->count != ASCII_set) \ { \ - if (step->data == &from_object) \ + if (FROM_DIRECTION) \ /* It's easy, we don't have to emit anything, we just reset the \ state for the input. */ \ - data->statep->count = 0; \ + data->statep->count = ASCII_set; \ else \ { \ char *outbuf = data->outbuf; \ \ /* We are not in the initial state. To switch back we have \ - to emit `SO'. */ \ + 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++ = SO; \ + *outbuf++ = SI; \ data->outbuf = outbuf; \ - data->statep->count = 0; \ + data->statep->count = ASCII_set; \ } \ } \ } @@ -102,9 +102,9 @@ enum and retore the state. */ #define SAVE_RESET_STATE(Save) \ if (Save) \ - save_set = set; \ + save_set = *setp; \ else \ - set = save_set + *setp = save_set /* First define the conversion function from ISO-2022-JP to UCS4. */ @@ -127,14 +127,14 @@ enum if (ch == ESC) \ { \ /* We don't really have to handle escape sequences since all the \ - switching is done using the SI and SO bytes. Butwe have to \ + switching is done using the SI and SO bytes. But we have to \ recognize `Esc $ ) C' since this is a kind of flag for this \ encoding. We simply ignore it. */ \ if (inptr + 1 > inend \ || (inptr[1] == '$' \ && (inptr + 2 > inend \ || (inptr[2] == ')' && inptr + 3 > inend)))) \ - \ + \ { \ result = GCONV_EMPTY_INPUT; \ break; \ @@ -146,14 +146,14 @@ enum continue; \ } \ } \ - else if (ch == SI) \ + else if (ch == SO) \ { \ /* Switch to use KSC. */ \ ++inptr; \ set = KSC5601_set; \ continue; \ } \ - else if (ch == SO) \ + else if (ch == SI) \ { \ /* Switch to use ASCII. */ \ ++inptr; \ @@ -161,9 +161,16 @@ enum continue; \ } \ \ - if (set == ASCII_set || ch < 0x21 || ch == 0x7f) \ - /* Almost done, just advance the input pointer. */ \ - ++inptr; \ + if (set == ASCII_set) \ + { \ + if (ch >= 0x80) \ + { \ + result = GCONV_ILLEGAL_INPUT; \ + break; \ + } \ + /* Almost done, just advance the input pointer. */ \ + ++inptr; \ + } \ else \ { \ assert (set == KSC5601_set); \ @@ -186,7 +193,9 @@ enum \ *((uint32_t *) outptr)++ = ch; \ } -#define EXTRA_LOOP_DECLS , int set +#define EXTRA_LOOP_DECLS , int *setp +#define INIT_PARAMS int set = *setp +#define UPDATE_PARAMS *setp = set #include @@ -197,97 +206,66 @@ enum #define LOOPFCT TO_LOOP #define BODY \ { \ - unsigned char ch; \ + uint32_t ch; \ size_t written = 0; \ \ ch = *((uint32_t *) inptr); \ \ /* First see whether we can write the character using the currently \ selected character set. */ \ - if (set == ASCII_set || (ch >= 0x01 && (ch < 0x21 || ch == 0x7f))) \ + if (ch < 0x80) \ { \ - /* Please note that the NUL byte is *not* matched if we are not \ - currently using the ASCII charset. This is because we must \ - switch to the initial state whenever a NUL byte is written. */ \ - if (ch <= 0x7f) \ + if (set != ASCII_set) \ { \ - *outptr++ = ch; \ - written = 1; \ + *outptr++ = SI; \ + set = ASCII_set; \ + if (NEED_LENGTH_TEST && outptr == outend) \ + { \ + result = GCONV_FULL_OUTPUT; \ + break; \ + } \ } \ + \ + *outptr++ = ch; \ + written = 1; \ } \ else \ { \ - assert (set == KSC5601_set); \ + char buf[2]; \ \ - written = ucs4_to_ksc5601 (ch, outptr, \ - (NEED_LENGTH_TEST ? outend - outptr : 2)); \ + written = ucs4_to_ksc5601 (ch, buf, 2); \ \ - if (NEED_LENGTH_TEST && written == 0) \ + if (written == UNKNOWN_10646_CHAR) \ { \ - result = GCONV_FULL_OUTPUT; \ + /* Illegal character. */ \ + result = GCONV_ILLEGAL_INPUT; \ break; \ } \ - if (written == UNKNOWN_10646_CHAR) \ - { \ - /* Either this is an unknown character or we have to switch \ - the currently selected character set. The character sets \ - do not code entirely separate parts of ISO 10646 and \ - therefore there is no single correct result. If we choose \ - the character set to use wrong we might be end up with \ - using yet another character set for the next character \ - though the current and the next could be encoded with one \ - character set. We leave this kind of optimization for \ - later and now simply use a fixed order in which we test for \ - availability */ \ - \ - if (ch <= 0x7f) \ - { \ - /* We must encode using ASCII. First write out the \ - escape sequence. */ \ - *outptr++ = SO; \ - set = ASCII_set; \ - \ - if (NEED_LENGTH_TEST && outptr == outend) \ - { \ - result = GCONV_FULL_OUTPUT; \ - break; \ - } \ - \ - *outptr++ = ch; \ - } \ - else \ - { \ - char buf[2]; \ + assert (written == 2); \ \ - written = ucs4_to_ksc5601 (ch, buf, 2); \ - if (written != UNKNOWN_10646_CHAR) \ - { \ - /* We use KSC 5601. */ \ - *outptr++ = SI; \ - set = KSC5601_set; \ - \ - if (NEED_LENGTH_TEST && outptr + 2 > outend) \ - { \ - result = GCONV_FULL_OUTPUT; \ - break; \ - } \ + /* We use KSC 5601. */ \ + if (set != KSC5601_set) \ + { \ + *outptr++ = SO; \ + set = KSC5601_set; \ + } \ \ - *outptr++ = buf[0]; \ - *outptr++ = buf[1]; \ - } \ - else \ - { \ - result = GCONV_ILLEGAL_INPUT; \ - break; \ - } \ - } \ + 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 set +#define EXTRA_LOOP_DECLS , int *setp +#define INIT_PARAMS int set = *setp +#define UPDATE_PARAMS *setp = set #include -- cgit v1.2.3