diff options
Diffstat (limited to 'sal/textenc')
| -rw-r--r-- | sal/textenc/tcvteas1.tab | 56 | ||||
| -rw-r--r-- | sal/textenc/tencinfo.c | 9 |
2 files changed, 61 insertions, 4 deletions
diff --git a/sal/textenc/tcvteas1.tab b/sal/textenc/tcvteas1.tab index b939c0fa0..1c5313ded 100644 --- a/sal/textenc/tcvteas1.tab +++ b/sal/textenc/tcvteas1.tab @@ -2,9 +2,9 @@ * * $RCSfile: tcvteas1.tab,v $ * - * $Revision: 1.3 $ + * $Revision: 1.4 $ * - * last change: $Author: sb $ $Date: 2001-10-12 10:44:53 $ + * last change: $Author: sb $ $Date: 2002-03-18 14:43:42 $ * * The Contents of this file are made available subject to the terms of * either of the following licenses @@ -177,6 +177,58 @@ static ImplTextEncodingData const aImplMS874TextEncodingData /* ======================================================================= */ +/* TIS 620-2533 + * + * A good source of information is <http://www.inet.co.th/cyberclub/trin/ + * thairef/index.html> as of 18 Mar 2002. + * + * Single byte encoding, from which MS874 is derived (although it is the other + * way around in this implementation): + * + * 0x00--9F map to U+0000--009F + * 0xA0 is questionable (unassigned or U+00A0 NO BREAK SPACE), to ease + * implementation, it maps to U+00A0 + * 0xA1--DA map to U+0E01--0E3A (TIS 620, same for MS874) + * 0xDB--DE are unassigned (TIS 620, same for MS874) + * 0xDF--FB map to U+0E3F--0E5B (TIS 620, same for MS874) + * 0xFC--FF are unassigned (TIS 620, same for MS874) + */ + +static ImplByteConvertData const aImplTis620ByteCvtData = +{ + aImplMS874ToUniTab + (0xA0 - MS874UNI_START), + aImpl8090SameToUniTab, + 0xA0, MS874UNI_END, + SAME8090UNI_START, SAME8090UNI_END, + aImplMS874ToCharTab, + aImpl8090SameToCharTab, + aImplMS874ToCharTabEx, + MS874CHAR_START, MS874CHAR_END, + SAME8090CHAR_START, SAME8090CHAR_END, + 1 +}; + +static ImplTextEncodingData const aImplTis620TextEncodingData + = { { &aImplTis620ByteCvtData, + ImplCharToUnicode, + ImplUnicodeToChar, + NULL, + NULL, + NULL, + NULL, + NULL, + NULL }, + RTL_TEXTENCODING_TIS_620, + 1, + 1, + 1, + 222, + "iso8859-1", /* TODO! correct? */ + "TIS-620", + RTL_TEXTENCODING_INFO_ASCII | RTL_TEXTENCODING_INFO_MIME }; + +/* ======================================================================= */ + /* MS-1258 */ /* Windows Standard CharSet for Vietnamease */ /* 1-Byte, 0x00-0x7F ASCII ohne Ausnahme */ diff --git a/sal/textenc/tencinfo.c b/sal/textenc/tencinfo.c index cb1f1829f..1478fb4d1 100644 --- a/sal/textenc/tencinfo.c +++ b/sal/textenc/tencinfo.c @@ -2,9 +2,9 @@ * * $RCSfile: tencinfo.c,v $ * - * $Revision: 1.14 $ + * $Revision: 1.15 $ * - * last change: $Author: sb $ $Date: 2002-01-15 17:00:15 $ + * last change: $Author: sb $ $Date: 2002-03-18 14:43:42 $ * * The Contents of this file are made available subject to the terms of * either of the following licenses @@ -445,6 +445,10 @@ rtl_TextEncoding SAL_CALL rtl_getTextEncodingFromMacTextEncoding( sal_uInt32 nMa rtl_TextEncoding SAL_CALL rtl_getTextEncodingFromUnixCharset( const sal_Char* pUnixCharset ) { + /* See <ftp://ftp.x.org/pub/DOCS/registry>, section 14 ("Font Charset + * (Registry and Encoding) Names"). + */ + /* All Identifiers in the tables are lower case The function search */ /* for the first matching string in the tables. */ /* Sort order: unique (first 14, than 1), important */ @@ -912,6 +916,7 @@ rtl_TextEncoding SAL_CALL rtl_getTextEncodingFromMimeCharset( const sal_Char* pM { "gb18030", RTL_TEXTENCODING_GB_18030 }, /* This is no actual MIME character set name, it's only a guess */ { "big5hkscs", RTL_TEXTENCODING_BIG5_HKSCS }, + { "tis620", RTL_TEXTENCODING_TIS_620 }, { NULL, RTL_TEXTENCODING_DONTKNOW } }; |