X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/bea561cec04633a5810f3166107bd2d0288d4091..27c78e4552aaefac9a4db0d4453eff09cdfef2ad:/src/common/intl.cpp diff --git a/src/common/intl.cpp b/src/common/intl.cpp index f0512af3cc..548776f481 100644 --- a/src/common/intl.cpp +++ b/src/common/intl.cpp @@ -18,10 +18,6 @@ // headers // ---------------------------------------------------------------------------- -#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) - #pragma implementation "intl.h" -#endif - #if defined(__BORLAND__) && !defined(__WXDEBUG__) // There's a bug in Borland's compiler that breaks wxLocale with -O2, // so make sure that flag is not used for this file: @@ -72,12 +68,15 @@ #endif #include "wx/file.h" +#include "wx/filename.h" #include "wx/tokenzr.h" #include "wx/module.h" #include "wx/fontmap.h" #include "wx/encconv.h" #include "wx/hashmap.h" #include "wx/ptr_scpd.h" +#include "wx/app.h" +#include "wx/apptrait.h" #if defined(__WXMAC__) #include "wx/mac/private.h" // includes mac headers @@ -99,14 +98,13 @@ typedef wxUint32 size_t32; const size_t32 MSGCATALOG_MAGIC = 0x950412de; const size_t32 MSGCATALOG_MAGIC_SW = 0xde120495; -// extension of ".mo" files -#define MSGCATALOG_EXTENSION _T(".mo") - // the constants describing the format of lang_LANG locale string static const size_t LEN_LANG = 2; static const size_t LEN_SUBLANG = 2; static const size_t LEN_FULL = LEN_LANG + 1 + LEN_SUBLANG; // 1 for '_' +#define TRACE_I18N _T("i18n") + // ---------------------------------------------------------------------------- // global functions // ---------------------------------------------------------------------------- @@ -498,14 +496,13 @@ public: ~wxPluralFormsCalculator() {} void init(wxPluralFormsToken::Number nplurals, wxPluralFormsNode* plural); - wxString getString() const; private: wxPluralFormsToken::Number m_nplurals; wxPluralFormsNodePtr m_plural; }; -wxDEFINE_SCOPED_PTR_TYPE(wxPluralFormsCalculator); +wxDEFINE_SCOPED_PTR_TYPE(wxPluralFormsCalculator) void wxPluralFormsCalculator::init(wxPluralFormsToken::Number nplurals, wxPluralFormsNode* plural) @@ -1074,18 +1071,27 @@ static wxString GetFullSearchPath(const wxChar *lang) } // open disk file and read in it's contents -bool wxMsgCatalogFile::Load(const wxChar *szDirPrefix, const wxChar *szName0, +bool wxMsgCatalogFile::Load(const wxChar *szDirPrefix, const wxChar *szName, wxPluralFormsCalculatorPtr& rPluralFormsCalculator) { - /* We need to handle locales like de_AT.iso-8859-1 - For this we first chop off the .CHARSET specifier and ignore it. - FIXME: UNICODE SUPPORT: must use CHARSET specifier! - */ - wxString szName = szName0; - if(szName.Find(wxT('.')) != wxNOT_FOUND) // contains a dot - szName = szName.Left(szName.Find(wxT('.'))); + wxString searchPath; - wxString searchPath = GetFullSearchPath(szDirPrefix); +#if wxUSE_FONTMAP + // first look for the catalog for this language and the current locale: + // notice that we don't use the system name for the locale as this would + // force us to install catalogs in different locations depending on the + // system but always use the canonical name + wxFontEncoding encSys = wxLocale::GetSystemEncoding(); + if ( encSys != wxFONTENCODING_SYSTEM ) + { + wxString fullname(szDirPrefix); + fullname << _T('.') << wxFontMapperBase::GetEncodingName(encSys); + searchPath << GetFullSearchPath(fullname) << wxPATH_SEP; + } +#endif // wxUSE_FONTMAP + + + searchPath += GetFullSearchPath(szDirPrefix); const wxChar *sublocale = wxStrchr(szDirPrefix, wxT('_')); if ( sublocale ) { @@ -1097,9 +1103,6 @@ bool wxMsgCatalogFile::Load(const wxChar *szDirPrefix, const wxChar *szName0, << wxPATH_SEP; } - wxString strFile = szName; - strFile += MSGCATALOG_EXTENSION; - // don't give translation errors here because the wxstd catalog might // not yet be loaded (and it's normal) // @@ -1107,30 +1110,38 @@ bool wxMsgCatalogFile::Load(const wxChar *szDirPrefix, const wxChar *szName0, NoTransErr noTransErr; wxLogVerbose(_("looking for catalog '%s' in path '%s'."), - szName.c_str(), searchPath.c_str()); + szName, searchPath.c_str()); + wxLogTrace(TRACE_I18N, _T("Looking for \"%s.mo\" in \"%s\""), + szName, searchPath.c_str()); + wxFileName fn(szName); + fn.SetExt(_T("mo")); wxString strFullName; - if ( !wxFindFileInPath(&strFullName, searchPath, strFile) ) { - wxLogVerbose(_("catalog file for domain '%s' not found."), szName.c_str()); + if ( !wxFindFileInPath(&strFullName, searchPath, fn.GetFullPath()) ) { + wxLogVerbose(_("catalog file for domain '%s' not found."), szName); + wxLogTrace(TRACE_I18N, _T("Catalog \"%s.mo\" not found"), szName); return false; } // open file - wxLogVerbose(_("using catalog '%s' from '%s'."), - szName.c_str(), strFullName.c_str()); + wxLogVerbose(_("using catalog '%s' from '%s'."), szName, strFullName.c_str()); + wxLogTrace(TRACE_I18N, _T("Using catalog \"%s\"."), strFullName.c_str()); wxFile fileMsg(strFullName); if ( !fileMsg.IsOpened() ) return false; // get the file size (assume it is less than 4Gb...) - wxFileOffset nSize = fileMsg.Length(); - if ( nSize == wxInvalidOffset ) + wxFileOffset lenFile = fileMsg.Length(); + if ( lenFile == wxInvalidOffset ) return false; + size_t nSize = wx_truncate_cast(size_t, lenFile); + wxASSERT_MSG( nSize == lenFile + size_t(0), _T("message catalog bigger than 4GB?") ); + // read the whole file in memory m_pData = new size_t8[nSize]; - if ( fileMsg.Read(m_pData, (size_t)nSize) != nSize ) { + if ( fileMsg.Read(m_pData, nSize) != lenFile ) { wxDELETEA(m_pData); return false; } @@ -1205,8 +1216,7 @@ bool wxMsgCatalogFile::Load(const wxChar *szDirPrefix, const wxChar *szName0, } else { - wxLogVerbose(_("Cannot parse Plural-Forms:'%s'"), - pfs.c_str()); + wxLogVerbose(_("Cannot parse Plural-Forms:'%s'"), pfs.c_str()); } } } @@ -1224,16 +1234,51 @@ void wxMsgCatalogFile::FillHash(wxMessagesHash& hash, const wxString& msgIdCharset, bool convertEncoding) const { -#if wxUSE_WCHAR_T - wxCSConv *csConv = NULL; - if ( !m_charset.IsEmpty() ) - csConv = new wxCSConv(m_charset); +#if wxUSE_FONTMAP + // determine if we need any conversion at all + if ( convertEncoding ) + { + wxFontEncoding encCat = wxFontMapperBase::GetEncodingFromName(m_charset); + if ( encCat == wxLocale::GetSystemEncoding() ) + { + // no need to convert + convertEncoding = false; + } + } +#endif // wxUSE_FONTMAP - wxMBConv& inputConv = csConv ? *((wxMBConv*)csConv) : *wxConvCurrent; +#if wxUSE_WCHAR_T + // conversion to use to convert catalog strings to the GUI encoding + wxMBConv *inputConv, + *csConv = NULL; // another ptr just to be able to delete it later + if ( convertEncoding ) + { + if ( m_charset.empty() ) + { + inputConv = wxConvCurrent; + } + else + { + inputConv = + csConv = new wxCSConv(m_charset); + } + } + else // no need to convert the encoding + { + // we still need the conversion for Unicode build +#if wxUSE_UNICODE + inputConv = wxConvCurrent; +#else // !wxUSE_UNICODE + inputConv = NULL; +#endif // wxUSE_UNICODE/!wxUSE_UNICODE + } - wxCSConv *sourceConv = NULL; - if ( !msgIdCharset.empty() && (m_charset != msgIdCharset) ) - sourceConv = new wxCSConv(msgIdCharset); + // conversion to apply to msgid strings before looking them up: we only + // need it if the msgids are neither in 7 bit ASCII nor in the same + // encoding as the catalog + wxCSConv *sourceConv = msgIdCharset.empty() || (msgIdCharset == m_charset) + ? NULL + : new wxCSConv(msgIdCharset); #elif wxUSE_FONTMAP wxASSERT_MSG( msgIdCharset == NULL, @@ -1243,7 +1288,7 @@ void wxMsgCatalogFile::FillHash(wxMessagesHash& hash, if ( convertEncoding ) { wxFontEncoding targetEnc = wxFONTENCODING_SYSTEM; - wxFontEncoding enc = wxFontMapper::Get()->CharsetToEncoding(m_charset, false); + wxFontEncoding enc = wxFontMapperBase::Get()->CharsetToEncoding(m_charset, false); if ( enc == wxFONTENCODING_SYSTEM ) { convertEncoding = false; // unknown encoding @@ -1272,18 +1317,18 @@ void wxMsgCatalogFile::FillHash(wxMessagesHash& hash, #endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T (void)convertEncoding; // get rid of warnings about unused parameter - for (size_t i = 0; i < m_numStrings; i++) + for (size_t32 i = 0; i < m_numStrings; i++) { const char *data = StringAtOfs(m_pOrigTable, i); #if wxUSE_UNICODE - wxString msgid(data, inputConv); -#else + wxString msgid(data, *inputConv); +#else // ASCII wxString msgid; -#if wxUSE_WCHAR_T - if ( convertEncoding && sourceConv ) - msgid = wxString(inputConv.cMB2WC(data), *sourceConv); - else -#endif + #if wxUSE_WCHAR_T + if ( inputConv && sourceConv ) + msgid = wxString(inputConv->cMB2WC(data), *sourceConv); + else + #endif msgid = data; #endif // wxUSE_UNICODE @@ -1296,10 +1341,10 @@ void wxMsgCatalogFile::FillHash(wxMessagesHash& hash, wxString msgstr; #if wxUSE_WCHAR_T #if wxUSE_UNICODE - msgstr = wxString(data + offset, inputConv); + msgstr = wxString(data + offset, *inputConv); #else - if ( convertEncoding ) - msgstr = wxString(inputConv.cMB2WC(data + offset), wxConvLocal); + if ( inputConv ) + msgstr = wxString(inputConv->cMB2WC(data + offset), wxConvLocal); else msgstr = wxString(data + offset); #endif @@ -1379,7 +1424,7 @@ const wxChar *wxMsgCatalog::GetString(const wxChar *sz, size_t n) const #include "wx/arrimpl.cpp" WX_DECLARE_EXPORTED_OBJARRAY(wxLanguageInfo, wxLanguageInfoArray); -WX_DEFINE_OBJARRAY(wxLanguageInfoArray); +WX_DEFINE_OBJARRAY(wxLanguageInfoArray) wxLanguageInfoArray *wxLocale::ms_languagesDB = NULL; @@ -1402,7 +1447,9 @@ wxLanguageInfoArray *wxLocale::ms_languagesDB = NULL; void wxLocale::DoCommonInit() { m_pszOldLocale = NULL; - m_pOldLocale = NULL; + + m_pOldLocale = wxSetLocale(this); + m_pMsgCat = NULL; m_language = wxLANGUAGE_UNKNOWN; m_initialized = false; @@ -1460,7 +1507,7 @@ bool wxLocale::Init(const wxChar *szName, // the short name will be used to look for catalog files as well, // so we need something here - if ( m_strShort.IsEmpty() ) { + if ( m_strShort.empty() ) { // FIXME I don't know how these 2 letter abbreviations are formed, // this wild guess is surely wrong if ( szLocale && szLocale[0] ) @@ -1471,15 +1518,23 @@ bool wxLocale::Init(const wxChar *szName, } } - // save the old locale to be able to restore it later - m_pOldLocale = wxSetLocale(this); - // load the default catalog with wxWidgets standard messages m_pMsgCat = NULL; bool bOk = true; if ( bLoadDefault ) + { bOk = AddCatalog(wxT("wxstd")); + // there may be a catalog with toolkit specific overrides, it is not + // an error if this does not exist + if ( bOk && wxTheApp ) + { + wxAppTraits *traits = wxTheApp->GetTraits(); + if (traits) + AddCatalog(traits->GetToolkitInfo().name.BeforeFirst(wxT('/')).MakeLower()); + } + } + return bOk; } @@ -1545,48 +1600,83 @@ bool wxLocale::Init(int language, int flags) wxString locale; // Set the locale: -#if defined(__UNIX__) && !defined(__WXMAC__) - if (language == wxLANGUAGE_DEFAULT) - locale = wxEmptyString; - else +#if defined(__OS2__) + wxMB2WXbuf retloc = wxSetlocale(LC_ALL , wxEmptyString); +#elif defined(__UNIX__) && !defined(__WXMAC__) + if (language != wxLANGUAGE_DEFAULT) locale = info->CanonicalName; wxMB2WXbuf retloc = wxSetlocaleTryUTF(LC_ALL, locale); + const wxString langOnly = locale.Left(2); if ( !retloc ) { // Some C libraries don't like xx_YY form and require xx only - retloc = wxSetlocaleTryUTF(LC_ALL, locale.Mid(0,2)); + retloc = wxSetlocaleTryUTF(LC_ALL, langOnly); } + +#if wxUSE_FONTMAP + // some systems (e.g. FreeBSD and HP-UX) don't have xx_YY aliases but + // require the full xx_YY.encoding form, so try using UTF-8 because this is + // the only thing we can do generically + // + // TODO: add encodings applicable to each language to the lang DB and try + // them all in turn here if ( !retloc ) { - // Some C libraries (namely glibc) still use old ISO 639, - // so will translate the abbrev for them - wxString mid = locale.Mid(0,2); - if (mid == wxT("he")) - locale = wxT("iw") + locale.Mid(3); - else if (mid == wxT("id")) - locale = wxT("in") + locale.Mid(3); - else if (mid == wxT("yi")) - locale = wxT("ji") + locale.Mid(3); - else if (mid == wxT("nb")) - locale = wxT("no_NO"); - else if (mid == wxT("nn")) - locale = wxT("no_NY"); - - retloc = wxSetlocaleTryUTF(LC_ALL, locale); + const wxChar **names = + wxFontMapperBase::GetAllEncodingNames(wxFONTENCODING_UTF8); + while ( *names ) + { + retloc = wxSetlocale(LC_ALL, locale + _T('.') + *names++); + if ( retloc ) + break; + } } +#endif // wxUSE_FONTMAP + if ( !retloc ) { - // (This time, we changed locale in previous if-branch, so try again.) - // Some C libraries don't like xx_YY form and require xx only - retloc = wxSetlocaleTryUTF(LC_ALL, locale.Mid(0,2)); + // Some C libraries (namely glibc) still use old ISO 639, + // so will translate the abbrev for them + wxString localeAlt; + if ( langOnly == wxT("he") ) + localeAlt = wxT("iw") + locale.Mid(3); + else if ( langOnly == wxT("id") ) + localeAlt = wxT("in") + locale.Mid(3); + else if ( langOnly == wxT("yi") ) + localeAlt = wxT("ji") + locale.Mid(3); + else if ( langOnly == wxT("nb") ) + localeAlt = wxT("no_NO"); + else if ( langOnly == wxT("nn") ) + localeAlt = wxT("no_NY"); + + if ( !localeAlt.empty() ) + { + retloc = wxSetlocaleTryUTF(LC_ALL, localeAlt); + if ( !retloc ) + retloc = wxSetlocaleTryUTF(LC_ALL, locale.Left(2)); + } } + if ( !retloc ) { wxLogError(wxT("Cannot set locale to '%s'."), locale.c_str()); return false; } + +#ifdef __AIX__ + // at least in AIX 5.2 libc is buggy and the string returned from setlocale(LC_ALL) + // can't be passed back to it because it returns 6 strings (one for each locale + // category), i.e. for C locale we get back "C C C C C C" + // + // this contradicts IBM own docs but this is not of much help, so just work around + // it in the crudest possible manner + wxChar *p = wxStrchr((wxChar *)retloc, _T(' ')); + if ( p ) + *p = _T('\0'); +#endif // __AIX__ + #elif defined(__WIN32__) #if wxUSE_UNICODE && (defined(__VISUALC__) || defined(__MINGW32__)) @@ -1638,7 +1728,7 @@ bool wxLocale::Init(int language, int flags) if (codepage != 0) locale << wxT(".") << buffer; } - if (locale.IsEmpty()) + if (locale.empty()) { wxLogLastError(wxT("SetThreadLocale")); wxLogError(wxT("Cannot set locale to language %s."), name.c_str()); @@ -1687,7 +1777,7 @@ bool wxLocale::Init(int language, int flags) return false; } #elif defined(__WXMAC__) - if (language == wxLANGUAGE_DEFAULT) + if (lang == wxLANGUAGE_DEFAULT) locale = wxEmptyString; else locale = info->CanonicalName; @@ -1704,8 +1794,6 @@ bool wxLocale::Init(int language, int flags) wxLogError(wxT("Cannot set locale to '%s'."), locale.c_str()); return false; } -#elif defined(__WXPM__) - wxMB2WXbuf retloc = wxSetlocale(LC_ALL , wxEmptyString); #else return false; #define WX_NO_LOCALE_SUPPORT @@ -1799,7 +1887,7 @@ void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix) (langFull.Len() == LEN_FULL && langFull[LEN_LANG] == wxT('_')) ) { // 0. Make sure the lang is according to latest ISO 639 - // (this is neccessary because glibc uses iw and in instead + // (this is necessary because glibc uses iw and in instead // of he and id respectively). // the language itself (second part is the dialect/sublang) @@ -2220,35 +2308,9 @@ wxString wxLocale::GetSystemEncodingName() if ( alang ) { - // 7 bit ASCII encoding has several alternative names which we should - // recognize to avoid warnings about unrecognized encoding on each - // program startup - - // nl_langinfo() under Solaris returns 646 by default which stands for - // ISO-646, i.e. 7 bit ASCII - // - // and recent glibc call it ANSI_X3.4-1968... - // - // HP-UX uses HP-Roman8 cset which is not the same as ASCII (see RFC - // 1345 for its definition) but must be recognized as otherwise HP - // users get a warning about it on each program startup, so handle it - // here -- but it would be obviously better to add real supprot to it, - // of course! - if ( strcmp(alang, "646") == 0 - || strcmp(alang, "ANSI_X3.4-1968") == 0 -#ifdef __HPUX__ - || strcmp(alang, "roman8") == 0 -#endif // __HPUX__ - ) - { - encname = _T("US-ASCII"); - } - else - { - encname = wxString::FromAscii( alang ); - } + encname = wxString::FromAscii( alang ); } - else + else // nl_langinfo() failed #endif // HAVE_LANGINFO_H { // if we can't get at the character set directly, try to see if it's in @@ -2285,12 +2347,17 @@ wxFontEncoding wxLocale::GetSystemEncoding() #if defined(__WIN32__) && !defined(__WXMICROWIN__) UINT codepage = ::GetACP(); - // wxWidgets only knows about CP1250-1257, 932, 936, 949, 950 + // wxWidgets only knows about CP1250-1257, 874, 932, 936, 949, 950 if ( codepage >= 1250 && codepage <= 1257 ) { return (wxFontEncoding)(wxFONTENCODING_CP1250 + codepage - 1250); } + if ( codepage == 874 ) + { + return wxFONTENCODING_CP874; + } + if ( codepage == 932 ) { return wxFONTENCODING_CP932; @@ -2319,16 +2386,15 @@ wxFontEncoding wxLocale::GetSystemEncoding() #endif return wxMacGetFontEncFromSystemEnc( encoding ) ; #elif defined(__UNIX_LIKE__) && wxUSE_FONTMAP - wxString encname = GetSystemEncodingName(); + const wxString encname = GetSystemEncodingName(); if ( !encname.empty() ) { - wxFontEncoding enc = wxFontMapper::Get()-> - CharsetToEncoding(encname, false /* not interactive */); + wxFontEncoding enc = wxFontMapperBase::GetEncodingFromName(encname); // on some modern Linux systems (RedHat 8) the default system locale // is UTF8 -- but it isn't supported by wxGTK in ANSI build at all so // don't even try to use it in this case -#if !wxUSE_UNICODE +#if !wxUSE_UNICODE && defined(__WXGTK__) if ( enc == wxFONTENCODING_UTF8 ) { // the most similar supported encoding... @@ -2336,13 +2402,11 @@ wxFontEncoding wxLocale::GetSystemEncoding() } #endif // !wxUSE_UNICODE - // this should probably be considered as a bug in CharsetToEncoding(): - // it shouldn't return wxFONTENCODING_DEFAULT at all - but it does it - // for US-ASCII charset - // - // we, OTOH, definitely shouldn't return it as it doesn't make sense at - // all (which encoding is it?) - if ( enc != wxFONTENCODING_DEFAULT ) + // GetEncodingFromName() returns wxFONTENCODING_DEFAULT for C locale + // (a.k.a. US-ASCII) which is arguably a bug but keep it like this for + // backwards compatibility and just take care to not return + // wxFONTENCODING_DEFAULT from here as this surely doesn't make sense + if ( enc != wxFONTENCODING_MAX && enc != wxFONTENCODING_DEFAULT ) { return enc; } @@ -2449,9 +2513,8 @@ wxLocale::~wxLocale() delete pTmpCat; } - // restore old locale if we had changed it - if ( m_pOldLocale ) - wxSetLocale(m_pOldLocale); + // restore old locale pointer + wxSetLocale(m_pOldLocale); // FIXME #ifndef __WXWINCE__ @@ -2504,20 +2567,12 @@ const wxChar *wxLocale::GetString(const wxChar *szOrigString, { NoTransErr noTransErr; - if ( szDomain != NULL ) - { - wxLogTrace(_T("i18n"), - _T("string '%s'[%lu] not found in domain '%s' for locale '%s'."), - szOrigString, (unsigned long)n, - szDomain, m_strLocale.c_str()); - - } - else - { - wxLogTrace(_T("i18n"), - _T("string '%s'[%lu] not found in locale '%s'."), - szOrigString, (unsigned long)n, m_strLocale.c_str()); - } + wxLogTrace(TRACE_I18N, + _T("string \"%s\"[%ld] not found in %slocale '%s'."), + szOrigString, (long)n, + szDomain ? wxString::Format(_T("domain '%s' "), szDomain).c_str() + : _T(""), + m_strLocale.c_str()); } #endif // __WXDEBUG__ @@ -2547,14 +2602,14 @@ wxString wxLocale::GetHeaderValue( const wxChar* szHeader, if ( pMsgCat == NULL ) return wxEmptyString; - pszTrans = pMsgCat->GetString(wxT(""), (size_t)-1); + pszTrans = pMsgCat->GetString(wxEmptyString, (size_t)-1); } else { // search in all domains for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) { - pszTrans = pMsgCat->GetString(wxT(""), (size_t)-1); + pszTrans = pMsgCat->GetString(wxEmptyString, (size_t)-1); if ( pszTrans != NULL ) // take the first found break; } @@ -3514,10 +3569,9 @@ void wxLocale::InitLanguagesDB() LNG(wxLANGUAGE_ZHUANG, "za" , 0 , 0 , "Zhuang") LNG(wxLANGUAGE_ZULU, "zu" , 0 , 0 , "Zulu") -}; +} #undef LNG // --- --- --- generated code ends here --- --- --- #endif // wxUSE_INTL -