diff options
Diffstat (limited to 'iconvdata')
37 files changed, 657 insertions, 388 deletions
diff --git a/iconvdata/8bit-gap.c b/iconvdata/8bit-gap.c index 7caa5427d5..e349270b8d 100644 --- a/iconvdata/8bit-gap.c +++ b/iconvdata/8bit-gap.c @@ -1,6 +1,6 @@ /* Generic conversion to and from 8bit charsets, converting from UCS using gaps. - Copyright (C) 1997, 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. @@ -83,7 +83,10 @@ struct gap unsigned char res; \ \ if (__builtin_expect (ch, 0) >= 0xffff) \ - rp = NULL; \ + { \ + UNICODE_TAG_HANDLER (ch, 4); \ + rp = NULL; \ + } \ else \ while (ch > rp->end) \ ++rp; \ diff --git a/iconvdata/8bit-generic.c b/iconvdata/8bit-generic.c index bc05ccd8db..3a81080c4c 100644 --- a/iconvdata/8bit-generic.c +++ b/iconvdata/8bit-generic.c @@ -1,5 +1,5 @@ /* Generic conversion to and from 8bit charsets. - Copyright (C) 1997, 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. @@ -67,6 +67,8 @@ if (__builtin_expect (ch >= sizeof (from_ucs4) / sizeof (from_ucs4[0]), 0)\ || (__builtin_expect (from_ucs4[ch], '\1') == '\0' && ch != 0)) \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* This is an illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/ansi_x3.110.c b/iconvdata/ansi_x3.110.c index e1dd0ed0a7..a06a4f0658 100644 --- a/iconvdata/ansi_x3.110.c +++ b/iconvdata/ansi_x3.110.c @@ -1,5 +1,5 @@ /* Generic conversion to and from ANSI_X3.110-1983. - Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. @@ -549,6 +549,8 @@ static const char from_ucs4[][2] = cp = "\xd5"; \ else \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal characters. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/big5.c b/iconvdata/big5.c index e06b44609e..4d1f1acca4 100644 --- a/iconvdata/big5.c +++ b/iconvdata/big5.c @@ -1,5 +1,5 @@ /* Mapping tables for Big5 handling. - Copyright (C) 1997, 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. @@ -8564,6 +8564,7 @@ static const char from_ucs4_tab15[][2] = cp = from_ucs4_tab15[ch - 0xff01]; \ break; \ default: \ + UNICODE_TAG_HANDLER (ch, 4); \ /* Illegal character. */ \ cp = ""; \ break; \ diff --git a/iconvdata/big5hkscs.c b/iconvdata/big5hkscs.c index 7f4a8c3bf2..dff5456868 100644 --- a/iconvdata/big5hkscs.c +++ b/iconvdata/big5hkscs.c @@ -1,5 +1,5 @@ /* Mapping tables for Big5-HKSCS handling. - Copyright (C) 1997, 1998, 2000 Free Software Foundation, Inc. + Copyright (C) 1997, 1998, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. Modified for Big5-HKSCS by Roger So <spacehunt@e-fever.org>, 2000. @@ -11772,6 +11772,8 @@ static struct \ if (cp[0] == '\0' && ch != 0) \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/cp1255.c b/iconvdata/cp1255.c index 127f8be406..d81e7da85c 100644 --- a/iconvdata/cp1255.c +++ b/iconvdata/cp1255.c @@ -451,7 +451,10 @@ static const struct { else if (ch == 0x2122) \ res = 0x99; \ else \ - res = 0; \ + { \ + UNICODE_TAG_HANDLER (ch, 4); \ + res = 0; \ + } \ \ if (__builtin_expect (res != 0, 1)) \ { \ diff --git a/iconvdata/cp1258.c b/iconvdata/cp1258.c index b4168d9781..4868e7559c 100644 --- a/iconvdata/cp1258.c +++ b/iconvdata/cp1258.c @@ -777,7 +777,10 @@ static const struct { else if (ch == 0x2122) \ res = 0x99; \ else \ - res = 0; \ + { \ + UNICODE_TAG_HANDLER (ch, 4); \ + res = 0; \ + } \ \ if (__builtin_expect (res != 0, 1)) \ { \ diff --git a/iconvdata/euc-cn.c b/iconvdata/euc-cn.c index 58f47eda3f..29ed465979 100644 --- a/iconvdata/euc-cn.c +++ b/iconvdata/euc-cn.c @@ -1,5 +1,5 @@ /* Mapping tables for EUC-CN handling. - Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. @@ -140,6 +140,8 @@ { \ if (__builtin_expect (found, 0) == __UNKNOWN_10646_CHAR) \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/euc-jp.c b/iconvdata/euc-jp.c index 8d8429614d..e0512cf12e 100644 --- a/iconvdata/euc-jp.c +++ b/iconvdata/euc-jp.c @@ -1,5 +1,5 @@ /* Mapping tables for EUC-JP handling. - Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. @@ -227,6 +227,8 @@ } \ else \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/euc-kr.c b/iconvdata/euc-kr.c index 5ecdbb8182..7ec613b3f9 100644 --- a/iconvdata/euc-kr.c +++ b/iconvdata/euc-kr.c @@ -1,5 +1,5 @@ /* Mapping tables for EUC-KR handling. - Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Jungshik Shin <jshin@pantheon.yale.edu> and Ulrich Drepper <drepper@cygnus.com>, 1998. @@ -142,10 +142,12 @@ euckr_from_ucs4 (uint32_t ch, unsigned char *cp) \ /* Decomposing Hangul syllables not available in KS C 5601 into \ Jamos should be considered either here or in euckr_from_ucs4() */ \ - euckr_from_ucs4 (ch, cp) ; \ + euckr_from_ucs4 (ch, cp); \ \ if (__builtin_expect (cp[0], '\1') == '\0' && ch != 0) \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/euc-tw.c b/iconvdata/euc-tw.c index 39c2f1f7da..68a80f7c6d 100644 --- a/iconvdata/euc-tw.c +++ b/iconvdata/euc-tw.c @@ -1,5 +1,5 @@ /* Mapping tables for EUC-TW handling. - Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc. + Copyright (C) 1998, 1999, 2000, 2001 Free Software Foundation, Inc. This file is part of the GNU C Library. Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. @@ -61,14 +61,14 @@ } \ else \ { \ - /* Two or more byte character. First test whether the next \ - character is also available. */ \ + /* Two or more byte character. First test whether the next byte \ + is also available. */ \ uint32_t ch2; \ \ if (inptr + 1 >= inend) \ { \ - /* The second character is not available. Store the \ - intermediate result. */ \ + /* The second byte is not available. Store the intermediate \ + result. */ \ result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ @@ -99,7 +99,7 @@ \ if (ch == 0) \ { \ - /* The third or fourth character is not available. Store \ + /* The third or fourth byte is not available. Store \ the intermediate result. */ \ result = __GCONV_INCOMPLETE_INPUT; \ break; \ @@ -170,7 +170,7 @@ *outptr++ = ch; \ else \ { \ - /* Try the JIS character sets. */ \ + /* Try the CNS 11643 planes. */ \ size_t found; \ \ found = ucs4_to_cns11643l1 (ch, outptr, outend - outptr); \ @@ -199,6 +199,8 @@ } \ if (__builtin_expect (found, 0) == __UNKNOWN_10646_CHAR) \ { \ + UNICODE_TAG_HANDLER (ch, 4); \ + \ /* Illegal character. */ \ STANDARD_ERR_HANDLER (4); \ } \ diff --git a/iconvdata/gb18030.c b/iconvdata/gb18030.c index 8b8c472074..e3e9e8a66d 100644 --- a/iconvdata/gb18030.c +++ b/iconvdata/gb18030.c @@ -5683,7 +5683,8 @@ static const uint16_t __fourbyte_to_ucs1[] = [0x0309] = 0x03e5, [0x030a] = 0x03e6, [0x030b] = 0x03e7, [0x030c] = 0x03e8, [0x030d] = 0x03e9, [0x030e] = 0x03ea, [0x030f] = 0x03eb, [0x0310] = 0x03ec, [0x0311] = 0x03ed, [0x0312] = 0x03ee, [0x0313] = 0x03ef, [0x0314] = 0x03f0, - [0x0315] = 0x03f1, [0x0316] = 0x03f2, [0x0317] = 0x03f3, [0x0324] = 0x0400, + [0x0315] = 0x03f1, [0x0316] = 0x03f2, [0x0317] = 0x03f3, [0x0318] = 0x03f4, + [0x0319] = 0x03f5, [0x0324] = 0x0400, [0x0325] = 0x0402, [0x0326] = 0x0403, [0x0327] = 0x0404, [0x0328] = 0x0405, [0x0329] = 0x0406, [0x032a] = 0x0407, [0x032b] = 0x0408, [0x032c] = 0x0409, [0x032d] = 0x040a, [0x032e] = 0x040b, [0x032f] = 0x040c, [0x0330] = 0x040d, @@ -9517,6 +9518,7 @@ static const unsigned char __ucs_to_gb18030_tab1[][2] = [0x036b] = "\x23\x0f", [0x036c] = "\x23\x10", [0x036d] = "\x23\x11", [0x036e] = "\x23\x12", [0x036f] = "\x23\x13", [0x0370] = "\x23\x14", [0x0371] = "\x23\x15", [0x0372] = "\x23\x16", [0x0373] = "\x23\x17", + [0x0374] = "\x23\x18", [0x0375] = "\x23\x19", [0x0380] = "\x23\x24", [0x0381] = "\xa7\xa7", [0x0382] = "\x23\x25", [0x0383] = "\x23\x26", [0x0384] = "\x23\x27", [0x0385] = "\x23\x28", [0x0386] = "\x23\x29", [0x0387] = "\x23\x2a", [0x0388] = "\x23\x2b", @@ -21954,15 +21956,15 @@ static const unsigned char __ucs_to_gb18030_tab2[][2] = } \ else \ { \ - /* Two or four byte character. First test whether the \ - next character is also available. */ \ + /* Two or four byte character. First test whether the next \ + byte is also available. */ \ uint32_t ch2; \ unsigned long int idx; \ \ if (inptr + 1 >= inend) \ { \ - /* The second character is not available. Store \ - the intermediate result. */ \ + /* The second byte is not available. Store the \ + intermediate result. */ \ result = __GCONV_INCOMPLETE_INPUT; \ break; \ } \ @@ -21992,7 +21994,7 @@ static const unsigned char __ucs_to_gb18030_tab2[][2] = \ if (inptr + 3 >= inend) \ { \ - /* Not all characters are available. Store \ + /* Not all bytes are available. Store \ the intermediate result. */ \ result = __GCONV_INCOMPLETE_INPUT; \ break; \ @@ -22035,17 +22037,19 @@ static const unsigned char __ucs_to_gb18030_tab2[][2] = + ch3 - 0x81) * 10 + ch4 - 0x30; \ \ if (idx <= 0x4A18) \ - ch = __fourbyte_to_ucs1[idx]; \ + ch = __fourbyte_to_ucs1[idx]; \ else if (idx >= 0x4ABD && idx <= 0x4F49) \ - ch = idx + 0x5543; \ + ch = idx + 0x5543; \ else if (idx >= 0x4F4D && idx <= 0x4F83) \ - ch = __fourbyte_to_ucs2[idx-0x4F4D]; \ + ch = __fourbyte_to_ucs2[idx-0x4F4D]; \ else if (idx >= 0x56BD && idx <= 0x8260) \ - ch = idx + 0x5543; \ + ch = idx + 0x5543; \ else if (idx >= 0x93A9 && idx <= 0x99F9) \ - ch = __fourbyte_to_ucs3[idx-0x93A9]; \ + ch = __fourbyte_to_ucs3[idx-0x93A9]; \ + else if (idx >= 0x2E248 && idx <= 0x12E247) \ + ch = idx - 0x1E248; \ else \ - ch = 0; \ + ch = 0; \ \ if (ch == 0 && *inptr != '\0') \ { \ @@ -22148,6 +22152,11 @@ static const unsigned char __ucs_to_gb18030_tab2[][2] = len = 4; \ } \ } \ + else if (ch >= 0x10000 && ch <= 0x10FFFF) \ + { \ + idx = ch + 0x1E248; \ + len = 4; \ + } \ else \ len = 0; \ \ @@ -22175,7 +22184,7 @@ static const unsigned char __ucs_to_gb18030_tab2[][2] = } \ else /* len == 4 */ \ { \ - /* See whether there is enough room for the second byte we \ + /* See whether there is enough room for all four bytes we \ write. */ \ if (__builtin_expect (outptr + 3 >= o |