X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/04ef50df3a0fa3c343800c554e609f98fc7575cc..9b69526274b023fa1460b29a92bea8bf82e4703f:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 45d3599040..0631cb5b68 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -1,11 +1,11 @@ ///////////////////////////////////////////////////////////////////////////// // Name: strconv.cpp // Purpose: Unicode conversion classes -// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin +// Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ -// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin +// Copyright: (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik // Licence: wxWindows license ///////////////////////////////////////////////////////////////////////////// @@ -37,11 +37,30 @@ #include #include + +#include "wx/debug.h" +#include "wx/strconv.h" +#include "wx/intl.h" +#include "wx/log.h" + +// ---------------------------------------------------------------------------- +// globals +// ---------------------------------------------------------------------------- + +WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; + + +// ============================================================================ +// implementation +// ============================================================================ + +#if wxUSE_WCHAR_T + #ifdef __SALFORDC__ #include #endif -#ifdef HAVE_ICONV_H +#ifdef HAVE_ICONV #include #endif @@ -49,21 +68,8 @@ #include #endif -#include "wx/debug.h" -#include "wx/strconv.h" -#include "wx/intl.h" -#include "wx/log.h" - -#if defined(WORDS_BIGENDIAN) || defined(__STDC_ISO_10646__) -#define BSWAP_UCS4(str, len) -#define BSWAP_UCS2(str, len) -#else #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _cCharsetToEncoding(name, FALSE); + if ( enc == wxFONTENCODING_SYSTEM ) + return -1; + + // the use the helper function + return wxEncodingToCodepage(enc); +} + +#endif // wxUSE_FONTMAP + +#endif // wxUSE_GUI + +// include old wxCharsetToCodepage() by OK if needed +#if !wxUSE_GUI || !wxUSE_FONTMAP + #include "wx/msw/registry.h" -// this should work if M$ Internet Exploiter is installed -static long CharsetToCodepage(const wxChar *name) + +// this should work if Internet Exploiter is installed +extern long wxCharsetToCodepage(const wxChar *name) { if (!name) return GetACP(); @@ -435,29 +492,39 @@ static long CharsetToCodepage(const wxChar *name) return CP; } -#endif + +#endif // !wxUSE_GUI || !wxUSE_FONTMAP + +#endif // Win32 + +// ============================================================================ +// wxCharacterSet and derived classes +// ============================================================================ + +// ---------------------------------------------------------------------------- +// wxCharacterSet is the ABC for the classes below +// ---------------------------------------------------------------------------- class wxCharacterSet { public: - wxCharacterSet(const wxChar*name) - : cname(name) {} - virtual ~wxCharacterSet() - {} - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) - { return (size_t)-1; } - virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) - { return (size_t)-1; } - virtual bool usable() - { return FALSE; } + wxCharacterSet(const wxChar*name) : cname(name) {} + virtual ~wxCharacterSet() {} + virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0; + virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0; + virtual bool usable() const = 0; public: const wxChar*cname; }; +// ---------------------------------------------------------------------------- +// ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv +// ---------------------------------------------------------------------------- + class ID_CharSet : public wxCharacterSet { public: - ID_CharSet(const wxChar *name,wxMBConv *cnv) + ID_CharSet(const wxChar *name, wxMBConv *cnv) : wxCharacterSet(name), work(cnv) {} size_t MB2WC(wchar_t *buf, const char *psz, size_t n) @@ -466,14 +533,18 @@ public: size_t WC2MB(char *buf, const wchar_t *psz, size_t n) { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } - bool usable() + bool usable() const { return work!=NULL; } public: wxMBConv*work; }; -#ifdef HAVE_ICONV_H +// ============================================================================ +// The classes doing conversion using the iconv_xxx() functions +// ============================================================================ + +#ifdef HAVE_ICONV // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG // if output buffer is _exactly_ as big as needed. Such case is (unless there's @@ -489,144 +560,266 @@ public: #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) #endif +#define ICONV_CHAR_CAST(x) ((ICONV_CONST char **)(x)) + +// ---------------------------------------------------------------------------- +// IC_CharSet: encapsulates an iconv character set +// ---------------------------------------------------------------------------- + class IC_CharSet : public wxCharacterSet { public: - IC_CharSet(const wxChar *name) - : wxCharacterSet(name) + IC_CharSet(const wxChar *name); + virtual ~IC_CharSet(); + + virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n); + virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n); + + bool usable() const + { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } + +protected: + // the iconv handlers used to translate from multibyte to wide char and in + // the other direction + iconv_t m2w, + w2m; + +private: + // the name (for iconv_open()) of a wide char charset - if none is + // available on this machine, it will remain NULL + static const char *ms_wcCharsetName; + + // true if the wide char encoding we use (i.e. ms_wcCharsetName) has + // different endian-ness than the native one + static bool ms_wcNeedsSwap; +}; + +const char *IC_CharSet::ms_wcCharsetName = NULL; +bool IC_CharSet::ms_wcNeedsSwap = FALSE; + +IC_CharSet::IC_CharSet(const wxChar *name) + : wxCharacterSet(name) +{ + // check for charset that represents wchar_t: + if (ms_wcCharsetName == NULL) { - m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname)); - w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME); + ms_wcNeedsSwap = FALSE; + + // try charset with explicit bytesex info (e.g. "UCS-4LE"): + ms_wcCharsetName = WC_NAME_BEST; + m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + + if (m2w == (iconv_t)-1) + { + // try charset w/o bytesex info (e.g. "UCS4") + // and check for bytesex ourselves: + ms_wcCharsetName = WC_NAME; + m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + + // last bet, try if it knows WCHAR_T pseudo-charset + if (m2w == (iconv_t)-1) + { + ms_wcCharsetName = "WCHAR_T"; + m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); + } + + if (m2w != (iconv_t)-1) + { + char buf[2], *bufPtr; + wchar_t wbuf[2], *wbufPtr; + size_t insz, outsz; + size_t res; + + buf[0] = 'A'; + buf[1] = 0; + wbuf[0] = 0; + insz = 2; + outsz = SIZEOF_WCHAR_T * 2; + wbufPtr = wbuf; + bufPtr = buf; + + res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz, + (char**)&wbufPtr, &outsz); + + if (ICONV_FAILED(res, insz)) + { + ms_wcCharsetName = NULL; + wxLogLastError(wxT("iconv")); + wxLogError(_("Convertion to charset '%s' doesn't work."), name); + } + else + { + ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0]; + } + } + else + { + ms_wcCharsetName = NULL; + + // VS: we must not output an error here, since wxWindows will safely + // fall back to using wxEncodingConverter. + wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name); + //wxLogError( + } + } + wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap); + } + else // we already have ms_wcCharsetName + { + m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name)); } - ~IC_CharSet() + // NB: don't ever pass NULL to iconv_open(), it may crash! + if ( ms_wcCharsetName ) { - if ( m2w != (iconv_t)-1 ) - iconv_close(m2w); - if ( w2m != (iconv_t)-1 ) - iconv_close(w2m); + w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName); } + else + { + w2m = (iconv_t)-1; + } +} - size_t MB2WC(wchar_t *buf, const char *psz, size_t n) +IC_CharSet::~IC_CharSet() +{ + if ( m2w != (iconv_t)-1 ) + iconv_close(m2w); + if ( w2m != (iconv_t)-1 ) + iconv_close(w2m); +} + +size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n) +{ + size_t inbuf = strlen(psz); + size_t outbuf = n * SIZEOF_WCHAR_T; + size_t res, cres; + // VS: Use these instead of psz, buf because iconv() modifies its arguments: + wchar_t *bufPtr = buf; + const char *pszPtr = psz; + + if (buf) { - size_t inbuf = strlen(psz); - size_t outbuf = n * SIZEOF_WCHAR_T; - size_t res, cres; - // VS: Use these instead of psz, buf because iconv() modifies its arguments: - wchar_t *bufPtr = buf; - const char *pszPtr = psz; + // have destination buffer, convert there + cres = iconv(m2w, + ICONV_CHAR_CAST(&pszPtr), &inbuf, + (char**)&bufPtr, &outbuf); + res = n - (outbuf / SIZEOF_WCHAR_T); - if (buf) + if (ms_wcNeedsSwap) { - // have destination buffer, convert there -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf); -#else - cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf); -#endif - res = n - (outbuf / SIZEOF_WCHAR_T); // convert to native endianness -#ifdef WC_NEED_BSWAP WC_BSWAP(buf /* _not_ bufPtr */, res) -#endif - } - else - { - // no destination buffer... convert using temp buffer - // to calculate destination buffer requirement - wchar_t tbuf[8]; - res = 0; - do { - bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T; -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); -#else - cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); -#endif - res += 8-(outbuf/SIZEOF_WCHAR_T); - } while ((cres==(size_t)-1) && (errno==E2BIG)); } - - if (ICONV_FAILED(cres, inbuf)) - return (size_t)-1; - - return res; + } + else + { + // no destination buffer... convert using temp buffer + // to calculate destination buffer requirement + wchar_t tbuf[8]; + res = 0; + do { + bufPtr = tbuf; + outbuf = 8*SIZEOF_WCHAR_T; + + cres = iconv(m2w, + ICONV_CHAR_CAST(&pszPtr), &inbuf, + (char**)&bufPtr, &outbuf ); + + res += 8-(outbuf/SIZEOF_WCHAR_T); + } while ((cres==(size_t)-1) && (errno==E2BIG)); } - size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + if (ICONV_FAILED(cres, inbuf)) { + //VS: it is ok if iconv fails, hence trace only + wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); + return (size_t)-1; + } + + return res; +} + +size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) +{ #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; + size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; #else - size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; + size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; #endif - size_t outbuf = n; - size_t res, cres; + size_t outbuf = n; + size_t res, cres; -#ifdef WC_NEED_BSWAP + wchar_t *tmpbuf = 0; + + if (ms_wcNeedsSwap) + { // need to copy to temp buffer to switch endianness // this absolutely doesn't rock! // (no, doing WC_BSWAP twice on the original buffer won't help, as it // could be in read-only memory, or be accessed in some other thread) - wchar_t *tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); + tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); WC_BSWAP(tmpbuf, inbuf) psz=tmpbuf; -#endif - if (buf) - { - // have destination buffer, convert there -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf ); -#else - cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf ); -#endif - res = n-outbuf; - } - else - { - // no destination buffer... convert using temp buffer - // to calculate destination buffer requirement - char tbuf[16]; - res = 0; - do { - buf = tbuf; outbuf = 16; -#ifdef WX_ICONV_TAKES_CHAR - cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf ); -#else - cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf ); -#endif - res += 16 - outbuf; - } while ((cres==(size_t)-1) && (errno==E2BIG)); - } -#ifdef WC_NEED_BSWAP + } + + if (buf) + { + // have destination buffer, convert there + cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); + + res = n-outbuf; + } + else + { + // no destination buffer... convert using temp buffer + // to calculate destination buffer requirement + char tbuf[16]; + res = 0; + do { + buf = tbuf; outbuf = 16; + + cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); + + res += 16 - outbuf; + } while ((cres==(size_t)-1) && (errno==E2BIG)); + } + + if (ms_wcNeedsSwap) + { free(tmpbuf); -#endif - if (ICONV_FAILED(cres, inbuf)) - return (size_t)-1; + } - return res; + if (ICONV_FAILED(cres, inbuf)) + { + //VS: it is ok if iconv fails, hence trace only + wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode())); + return (size_t)-1; } - bool usable() - { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } + return res; +} -public: - iconv_t m2w, w2m; -}; -#endif +#endif // HAVE_ICONV + +// ============================================================================ +// Win32 conversion classes +// ============================================================================ #if defined(__WIN32__) && !defined(__WXMICROWIN__) class CP_CharSet : public wxCharacterSet { public: - CP_CharSet(const wxChar*name) - : wxCharacterSet(name), CodePage(CharsetToCodepage(name)) {} + CP_CharSet(const wxChar* name) + : wxCharacterSet(name) + { + m_CodePage = wxCharsetToCodepage(name); + } size_t MB2WC(wchar_t *buf, const char *psz, size_t n) { size_t len = - MultiByteToWideChar(CodePage, 0, psz, -1, buf, buf ? n : 0); + MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0); //VS: returns # of written chars for buf!=NULL and *size* // needed buffer for buf==NULL return len ? (buf ? len : len-1) : (size_t)-1; @@ -634,21 +827,25 @@ public: size_t WC2MB(char *buf, const wchar_t *psz, size_t n) { - size_t len = WideCharToMultiByte(CodePage, 0, psz, -1, buf, + size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf, buf ? n : 0, NULL, NULL); //VS: returns # of written chars for buf!=NULL and *size* // needed buffer for buf==NULL return len ? (buf ? len : len-1) : (size_t)-1; } - bool usable() - { return CodePage != -1; } + bool usable() const + { return m_CodePage != -1; } public: - long CodePage; + long m_CodePage; }; #endif // __WIN32__ +// ============================================================================ +// wxEncodingConverter based conversion classes +// ============================================================================ + #if wxUSE_FONTMAP class EC_CharSet : public wxCharacterSet @@ -656,13 +853,14 @@ class EC_CharSet : public wxCharacterSet public: // temporarily just use wxEncodingConverter stuff, // so that it works while a better implementation is built - EC_CharSet(const wxChar*name) : wxCharacterSet(name), - enc(wxFONTENCODING_SYSTEM) + EC_CharSet(const wxChar* name) : wxCharacterSet(name), + enc(wxFONTENCODING_SYSTEM) { if (name) enc = wxTheFontMapper->CharsetToEncoding(name, FALSE); - m2w.Init(enc, wxFONTENCODING_UNICODE); - w2m.Init(wxFONTENCODING_UNICODE, enc); + + m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) && + w2m.Init(wxFONTENCODING_UNICODE, enc); } size_t MB2WC(wchar_t *buf, const char *psz, size_t n) @@ -675,7 +873,8 @@ public: size_t WC2MB(char *buf, const wchar_t *psz, size_t n) { -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) +#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ + || ( defined(__MWERKS__) && defined(__WXMSW__) ) size_t inbuf = std::wcslen(psz); #else size_t inbuf = ::wcslen(psz); @@ -686,61 +885,87 @@ public: return inbuf; } - bool usable() - { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } + bool usable() const { return m_ok; } public: wxFontEncoding enc; wxEncodingConverter m2w, w2m; + + // were we initialized successfully? + bool m_ok; }; #endif // wxUSE_FONTMAP +// ---------------------------------------------------------------------------- +// the function creating the wxCharacterSet for the specified charset on the +// current system, trying all possibilities +// ---------------------------------------------------------------------------- + static wxCharacterSet *wxGetCharacterSet(const wxChar *name) { - wxCharacterSet *cset = NULL; - if (name) + // check for the special case of ASCII charset +#if wxUSE_FONTMAP + if ( wxTheFontMapper->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT ) +#else // wxUSE_FONTMAP + if ( !name ) +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP { - if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) - { - cset = new ID_CharSet(name, &wxConvUTF8); - } - else - { -#ifdef HAVE_ICONV_H - cset = new IC_CharSet(name); // may not take NULL -#endif - } + // don't convert at all + return NULL; } - if (cset && cset->usable()) - return cset; + // the test above must have taken care of this case + wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") ); + + wxCharacterSet *cset; - if (cset) + if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) { - delete cset; + cset = new ID_CharSet(name, &wxConvUTF8); + } + else + { +#ifdef HAVE_ICONV + cset = new IC_CharSet(name); +#else // !HAVE_ICONV cset = NULL; +#endif // HAVE_ICONV/!HAVE_ICONV } + if ( cset->usable() ) + return cset; + + delete cset; + cset = NULL; + #if defined(__WIN32__) && !defined(__WXMICROWIN__) - cset = new CP_CharSet(name); // may take NULL - if (cset->usable()) + cset = new CP_CharSet(name); + if ( cset->usable() ) return cset; delete cset; + cset = NULL; #endif // __WIN32__ #if wxUSE_FONTMAP cset = new EC_CharSet(name); - if (cset->usable()) + if ( cset->usable() ) return cset; -#endif // wxUSE_FONTMAP delete cset; - wxLogError(_("Unknown encoding '%s'!"), name); + cset = NULL; +#endif // wxUSE_FONTMAP + + wxLogError(_("Cannot convert from encoding '%s'!"), name); + return NULL; } +// ============================================================================ +// wxCSConv implementation +// ============================================================================ + wxCSConv::wxCSConv(const wxChar *charset) { m_name = (wxChar *)NULL; @@ -776,7 +1001,8 @@ void wxCSConv::LoadNow() SetName(name); } - m_cset = wxGetCharacterSet(m_name); + // wxGetCharacterSet() complains about NULL name + m_cset = m_name ? wxGetCharacterSet(m_name) : NULL; m_deferred = FALSE; } } @@ -808,7 +1034,8 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const return m_cset->WC2MB(buf, psz, n); // latin-1 (direct) -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) +#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \ + || ( defined(__MWERKS__) && defined(__WXMSW__) ) size_t len=std::wcslen(psz); #else size_t len=::wcslen(psz); @@ -822,60 +1049,6 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const return len; } -#ifdef HAVE_ICONV_H - -class IC_CharSetConverter -{ -public: - IC_CharSetConverter(IC_CharSet *from, IC_CharSet *to) - { - cnv = iconv_open(wxConvLibc.cWX2MB(to->cname), - wxConvLibc.cWX2MB(from->cname)); - } - - ~IC_CharSetConverter() - { - if (cnv != (iconv_t)-1) - iconv_close(cnv); - } - - size_t Convert(char *buf, const char *psz, size_t n) - { - size_t inbuf = strlen(psz); - size_t outbuf = n; -#ifdef WX_ICONV_TAKES_CHAR - size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf ); -#else - size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf ); -#endif - if (res == (size_t)-1) - return (size_t)-1; - return (n - outbuf); - } - -public: - iconv_t cnv; -}; - -#endif // HAVE_ICONV_H - -class EC_CharSetConverter -{ -public: - EC_CharSetConverter(EC_CharSet* from,EC_CharSet* to) - { cnv.Init(from->enc,to->enc); } - - size_t Convert(char* buf, const char* psz, size_t n) - { - size_t inbuf = strlen(psz); - if (buf) cnv.Convert(psz,buf); - return inbuf; - } - -public: - wxEncodingConverter cnv; -}; - #else // !wxUSE_WCHAR_T // ----------------------------------------------------------------------------