From c7c3b0e907efac218b329ebbe3fc7432ec4415c5 Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Tue, 29 Aug 2000 21:14:05 +0000 Subject: [PATCH] Update. 2000-08-29 Akira Higuchi * iconv/gconv_db.c (increment_counter): Reset __init_fct, __fct, and __end_fct fields of struct __gconv_step. * iconv/Makefile (tests): Add iconv-bug2. * iconv/iconv-bug2.c: New file. * iconvdata/euc-kr.c (BODY for FROM_LOOP): Pass 'inend - inptr' instead of 'inptr - inend' to ksc5601_to_ucs4. * iconvdata/sjis.c (BODY for FROM_LOOP): Allow 0x7f character. * iconvdata/iso-2022-cn.c (BODY for FROM_LOOP): If an incomplete character or shift sequence is found at the end of the input string, return__GCONV_INCOMPLETE_INPUT instead of __GCONV_EMPTY_INPUT. * iconvdata/iso-2022-jp.c (BODY for FROM_LOOP): Likewise. * iconvdata/iso-2022-kr.c (BODY for FROM_LOOP): Likewise. * iconvdata/iso-2022-jp.c (BODY for FROM_LOOP): Return __GCONV_ILLEGAL_INPUT for 8bit characters. --- ChangeLog | 22 ++++++++++++++++++ iconv/gconv_db.c | 9 ++++++++ iconvdata/Makefile | 2 +- iconvdata/bug-iconv2.c | 47 ++++++++++++++++++++++++++++++++++++++ iconvdata/euc-kr.c | 2 +- iconvdata/iso-2022-cn.c | 4 ++-- iconvdata/iso-2022-jp.c | 16 +++++++++++-- iconvdata/iso-2022-kr.c | 4 ++-- iconvdata/sjis.c | 4 ++-- localedata/ChangeLog | 7 ++++++ localedata/SUPPORTED | 2 ++ localedata/charmaps/GB2312 | 2 +- localedata/charmaps/GBK | 2 +- 13 files changed, 111 insertions(+), 12 deletions(-) create mode 100644 iconvdata/bug-iconv2.c diff --git a/ChangeLog b/ChangeLog index 9caad1f2a6..bef84db3fa 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,25 @@ +2000-08-29 Akira Higuchi + + * iconv/gconv_db.c (increment_counter): Reset __init_fct, __fct, + and __end_fct fields of struct __gconv_step. + * iconv/Makefile (tests): Add iconv-bug2. + * iconv/iconv-bug2.c: New file. + + * iconvdata/euc-kr.c (BODY for FROM_LOOP): Pass 'inend - inptr' + instead of 'inptr - inend' to ksc5601_to_ucs4. + + * iconvdata/sjis.c (BODY for FROM_LOOP): Allow 0x7f character. + + * iconvdata/iso-2022-cn.c (BODY for FROM_LOOP): If an incomplete + character or shift sequence is found at the end of the input + string, return__GCONV_INCOMPLETE_INPUT instead of + __GCONV_EMPTY_INPUT. + * iconvdata/iso-2022-jp.c (BODY for FROM_LOOP): Likewise. + * iconvdata/iso-2022-kr.c (BODY for FROM_LOOP): Likewise. + + * iconvdata/iso-2022-jp.c (BODY for FROM_LOOP): Return + __GCONV_ILLEGAL_INPUT for 8bit characters. + 2000-08-29 Ulrich Drepper * signal/signal.h (sigpause): Move __THROW before __asm__. diff --git a/iconv/gconv_db.c b/iconv/gconv_db.c index 0caf310b91..50cedd6256 100644 --- a/iconv/gconv_db.c +++ b/iconv/gconv_db.c @@ -305,6 +305,15 @@ increment_counter (struct __gconv_step *steps, size_t nsteps) result = __GCONV_NOCONV; break; } + + steps[cnt].__init_fct = steps[cnt].__shlib_handle->init_fct; + steps[cnt].__fct = steps[cnt].__shlib_handle->fct; + steps[cnt].__end_fct = steps[cnt].__shlib_handle->end_fct; + + if (steps[cnt].__end_fct != NULL) + DL_CALL_FCT (steps[cnt].__end_fct, &steps[cnt]); + if (steps[cnt].__init_fct != NULL) + DL_CALL_FCT (steps[cnt].__init_fct, &steps[cnt]); } return result; } diff --git a/iconvdata/Makefile b/iconvdata/Makefile index 195a81ec81..b309d751f3 100644 --- a/iconvdata/Makefile +++ b/iconvdata/Makefile @@ -49,7 +49,7 @@ modules := ISO8859-1 ISO8859-2 ISO8859-3 ISO8859-4 ISO8859-5 \ modules.so := $(addsuffix .so, $(modules)) -tests = bug-iconv1 +tests = bug-iconv1 bug-iconv2 include ../Makeconfig diff --git a/iconvdata/bug-iconv2.c b/iconvdata/bug-iconv2.c new file mode 100644 index 0000000000..a2bf44f419 --- /dev/null +++ b/iconvdata/bug-iconv2.c @@ -0,0 +1,47 @@ +/* Test case by Akira Higuchi . */ + +#include +#include +#include + +int +main (void) +{ + const char *dummy_codesets[] = + { + "ISO_8859-1", "ISO_8859-2", "ISO_8859-3", "ISO_8859-4", + "ISO_8859-5", "ISO_8859-6", "ISO_8859-7", "ISO_8859-8" + }; + iconv_t dummy_cd[8], cd_a; + int i; + char buffer[1024], *to = buffer; + char *from = (char *) "foobar"; + size_t to_left = 1024, from_left = 6; + + /* load dummy modules */ + for (i = 0; i < 8; i++) + if ((dummy_cd[i] = iconv_open (dummy_codesets[i], "UTF8")) == (iconv_t) -1) + exit (1); + + /* load a module... */ + if ((cd_a = iconv_open ("EUC-JP", "UTF8")) == (iconv_t) -1) + exit (1); + /* and close it once. we'll reload this later */ + iconv_close (cd_a); + + /* unload dummy modules */ + for (i = 0; i < 8; i++) + iconv_close (dummy_cd[i]); + + /* load the module again */ + if ((cd_a = iconv_open ("EUC-JP", "UTF8")) == (iconv_t) -1) + exit (1); + + puts ("This used to crash"); + printf ("%d\n", iconv (cd_a, &from, &from_left, &to, &to_left)); + iconv_close (cd_a); + + puts ("works now"); + + return 0; +} diff --git a/iconvdata/euc-kr.c b/iconvdata/euc-kr.c index 786695801c..9ea937c005 100644 --- a/iconvdata/euc-kr.c +++ b/iconvdata/euc-kr.c @@ -99,7 +99,7 @@ euckr_from_ucs4 (uint32_t ch, unsigned char *cp) { \ /* Two-byte character. First test whether the next character \ is also available. */ \ - ch = ksc5601_to_ucs4 (&inptr, inptr - inend, 0x80); \ + ch = ksc5601_to_ucs4 (&inptr, inend - inptr, 0x80); \ if (__builtin_expect (ch, 1) == 0) \ { \ /* The second character is not available. */ \ diff --git a/iconvdata/iso-2022-cn.c b/iconvdata/iso-2022-cn.c index 6ffa18dc78..d45ed6b30a 100644 --- a/iconvdata/iso-2022-cn.c +++ b/iconvdata/iso-2022-cn.c @@ -151,7 +151,7 @@ enum || (inptr[1] == SS2_1 \ && __builtin_expect (inptr + 3 > inend, 0))) \ { \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ if (inptr[1] == '$' \ @@ -225,7 +225,7 @@ enum \ if (__builtin_expect (ch, 1) == 0) \ { \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ else if (__builtin_expect (ch, 1) == __UNKNOWN_10646_CHAR) \ diff --git a/iconvdata/iso-2022-jp.c b/iconvdata/iso-2022-jp.c index fc34aaba96..bd2e033746 100644 --- a/iconvdata/iso-2022-jp.c +++ b/iconvdata/iso-2022-jp.c @@ -258,7 +258,7 @@ gconv_end (struct __gconv_step *data) && __builtin_expect (inptr + 3 >= inend, 0))) \ { \ /* Not enough input available. */ \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ \ @@ -399,6 +399,18 @@ gconv_end (struct __gconv_step *data) continue; \ } \ } \ + else if (ch >= 0x80) \ + { \ + if (! ignore_errors_p ()) \ + { \ + result = __GCONV_ILLEGAL_INPUT; \ + break; \ + } \ + \ + ++inptr; \ + ++*irreversible; \ + continue; \ + } \ else if (set == ASCII_set || (ch < 0x21 || ch == 0x7f)) \ /* Almost done, just advance the input pointer. */ \ ++inptr; \ @@ -462,7 +474,7 @@ gconv_end (struct __gconv_step *data) \ if (__builtin_expect (ch, 1) == 0) \ { \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ else if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \ diff --git a/iconvdata/iso-2022-kr.c b/iconvdata/iso-2022-kr.c index 56e17a2f96..c0280fccea 100644 --- a/iconvdata/iso-2022-kr.c +++ b/iconvdata/iso-2022-kr.c @@ -149,7 +149,7 @@ enum && __builtin_expect (inptr + 3 > inend, 0))))) \ \ { \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ if (inptr[1] == '$' && inptr[2] == ')' && inptr[3] == 'C') \ @@ -188,7 +188,7 @@ enum \ if (__builtin_expect (ch, 1) == 0) \ { \ - result = __GCONV_EMPTY_INPUT; \ + result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ else if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \ diff --git a/iconvdata/sjis.c b/iconvdata/sjis.c index 244cec6db4..7bbfef3e5d 100644 --- a/iconvdata/sjis.c +++ b/iconvdata/sjis.c @@ -1941,7 +1941,7 @@ static const char from_ucs4_lat1[0xf8][2] = [0x0075] = "\x75\x00", [0x0076] = "\x76\x00", [0x0077] = "\x77\x00", [0x0078] = "\x78\x00", [0x0079] = "\x79\x00", [0x007a] = "\x7a\x00", [0x007b] = "\x7b\x00", [0x007c] = "\x7c\x00", [0x007d] = "\x7d\x00", - [0x007e] = "\x7e\x00", + [0x007e] = "\x7e\x00", [0x007f] = "\x7f\x00", [0x00a2] = "\x81\x91", [0x00a3] = "\x81\x92", [0x00a5] = "\x5c\x00", [0x00a7] = "\x81\x98", [0x00a8] = "\x81\x4e", [0x00ac] = "\x81\xca", [0x00b0] = "\x81\x8b", [0x00b1] = "\x81\x7d", [0x00b4] = "\x81\x4c", @@ -4353,7 +4353,7 @@ static const char from_ucs4_extra[0x100][2] = ch = 0x203e; \ ++inptr; \ } \ - else if (ch < 0x7e) \ + else if (ch < 0x80) \ ++inptr; \ else if (ch >= 0xa1 && ch <= 0xdf) \ { \ diff --git a/localedata/ChangeLog b/localedata/ChangeLog index 559d001528..4987943264 100644 --- a/localedata/ChangeLog +++ b/localedata/ChangeLog @@ -1,3 +1,10 @@ +2000-08-29 Akira Higuchi + + * charmaps/GB2312: Fix ".." -> "..." in width definitions. + * charmaps/GBK: Likewise. + + * SUPPORTED: Add zh_CN.GB2312 and zh_TW.BIG5. + 2000-08-27 Ulrich Drepper * locales/sv_SE: Remove old LC_COLLATE definition. Use iso14651_t1 diff --git a/localedata/SUPPORTED b/localedata/SUPPORTED index 8522572e92..ae2a0d4db9 100644 --- a/localedata/SUPPORTED +++ b/localedata/SUPPORTED @@ -127,3 +127,5 @@ th_TH TIS-620 tr_TR ISO-8859-9 uk_UA KOI8-U vi_VN UTF-8 +zh_CN GB2312 +zh_TW BIG5 diff --git a/localedata/charmaps/GB2312 b/localedata/charmaps/GB2312 index 3dd7b481b9..30d94effee 100644 --- a/localedata/charmaps/GB2312 +++ b/localedata/charmaps/GB2312 @@ -7634,5 +7634,5 @@ CHARMAP END CHARMAP WIDTH -.. 2 +... 2 END WIDTH diff --git a/localedata/charmaps/GBK b/localedata/charmaps/GBK index 7c8b3bfab2..de48a44183 100644 --- a/localedata/charmaps/GBK +++ b/localedata/charmaps/GBK @@ -21929,5 +21929,5 @@ CHARMAP END CHARMAP WIDTH -.. 2 +... 2 END WIDTH -- 2.43.5