X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/d43088ee78dca3fc64f9d419b184844eb72384da..8aa2cea1d03f39245f54deec0c4fe9ef49b5d910:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 69f9a7a85a..75c6201df8 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 ///////////////////////////////////////////////////////////////////////////// @@ -28,8 +28,13 @@ #pragma hdrstop #endif +#ifndef WX_PRECOMP + #include "wx/intl.h" + #include "wx/log.h" +#endif // WX_PRECOMP + #ifdef __WXMSW__ - #include "wx/msw/private.h" + #include "wx/msw/private.h" #endif #include @@ -37,88 +42,148 @@ #include #include -#ifdef __SALFORDC__ - #include -#endif +#include "wx/module.h" +#include "wx/strconv.h" -#ifdef HAVE_ICONV_H - #include -#endif -#ifdef HAVE_LANGINFO_H - #include -#endif +// ---------------------------------------------------------------------------- +// globals +// ---------------------------------------------------------------------------- -#ifdef __WXMSW__ - #include +#if wxUSE_WCHAR_T + WXDLLEXPORT_DATA(wxMBConv) wxConvLibc; + WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL); + WXDLLEXPORT_DATA(wxCSConv) wxConvISO8859_1(_T("iso-8859-1")); +#else + // stand-ins in absence of wchar_t + WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, + wxConvFile, + wxConvISO8859_1, + wxConvLocal, + wxConvUTF8; +#endif // wxUSE_WCHAR_T + +WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; + +class wxStrConvModule: public wxModule +{ +public: + wxStrConvModule() : wxModule() { } + virtual bool OnInit() { return TRUE; } + virtual void OnExit() + { +#if wxUSE_WCHAR_T + wxConvLocal.Clear(); + wxConvISO8859_1.Clear(); #endif + } -#include "wx/debug.h" -#include "wx/strconv.h" + DECLARE_DYNAMIC_CLASS(wxStrConvModule) +}; -#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; _c #endif -#define BSWAP_UTF32(str, len) BSWAP_UCS4(str, len) -#define BSWAP_UTF16(str, len) BSWAP_UCS2(str, len) -#if SIZEOF_WCHAR_T == 4 -#define WC_NAME "UCS4" -#define WC_BSWAP BSWAP_UCS4 -#elif SIZEOF_WCHAR_T == 2 -#define WC_NAME "UTF16" -#define WC_BSWAP BSWAP_UTF16 -#define WC_UTF16 +#ifdef HAVE_ICONV + #include #endif +#include "wx/encconv.h" +#include "wx/fontmap.h" + // ---------------------------------------------------------------------------- -// globals +// macros // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; +#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c=0x110000) { - return (size_t)-1; - } else { - if (output) { - *output++ = (input >> 10)+0xd7c0; - *output++ = (input&0x3ff)+0xdc00; - } - return 2; - } +static size_t encode_utf16(wxUint32 input, wchar_t *output) +{ + if (input<=0xffff) + { + if (output) *output++ = (wchar_t) input; + return 1; + } + else if (input>=0x110000) + { + return (size_t)-1; + } + else + { + if (output) + { + *output++ = (wchar_t) ((input >> 10)+0xd7c0); + *output++ = (wchar_t) ((input&0x3ff)+0xdc00); + } + return 2; + } } -static size_t decode_utf16(wxUint16*input,wxUint32&output) -{ - if ((*input<0xd800) || (*input>0xdfff)) { - output = *input; - return 1; - } else - if ((input[1]<0xdc00) || (input[1]>=0xdfff)) { - output = *input; - return (size_t)-1; - } else { - output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); - return 2; - } +static size_t decode_utf16(const wchar_t* input, wxUint32& output) +{ + if ((*input<0xd800) || (*input>0xdfff)) + { + output = *input; + return 1; + } + else if ((input[1]<0xdc00) || (input[1]>=0xdfff)) + { + output = *input; + return (size_t)-1; + } + else + { + output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00); + return 2; + } } #endif // WC_UTF16 @@ -127,63 +192,87 @@ static size_t decode_utf16(wxUint16*input,wxUint32&output) // wxMBConv // ---------------------------------------------------------------------------- -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc; +#define IGNORE_LIBC 0 + +wxMBConv::~wxMBConv() +{ + // nothing to do here +} size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { +#if IGNORE_LIBC + if (buf) + { + for (size_t i = 0; i < strlen( psz )+1; i++) + buf[i] = (wchar_t) psz[i]; + return strlen( psz ); + } + else + { + return strlen( psz ); + } +#else return wxMB2WC(buf, psz, n); +#endif } size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { +#if IGNORE_LIBC + if (buf) + { + for (size_t i = 0; i < wxStrlen( psz )+1; i++) + buf[i] = (char) psz[i]; + return wxStrlen( psz ); + } + else + { + return wxStrlen( psz ); + } +#else return wxWC2MB(buf, psz, n); +#endif } const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const { - if (psz) + if ( psz ) { - size_t nLen = MB2WC((wchar_t *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxWCharBuffer((wchar_t *) NULL); - wxWCharBuffer buf(nLen); - MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen); - return buf; + // calculate the length of the buffer needed first + size_t nLen = MB2WC(NULL, psz, 0); + if ( nLen != (size_t)-1 ) + { + // now do the actual conversion + wxWCharBuffer buf(nLen); + MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL + + return buf; + } } - else - return wxWCharBuffer((wchar_t *) NULL); + + wxWCharBuffer buf((wchar_t *)NULL); + + return buf; } -const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const +const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const { - if (psz) + if ( pwz ) { - size_t nLen = WC2MB((char *) NULL, psz, 0); - if (nLen == (size_t)-1) - return wxCharBuffer((char *) NULL); - wxCharBuffer buf(nLen); - WC2MB((char *)(const char *) buf, psz, nLen); - return buf; - } - else - return wxCharBuffer((char *) NULL); -} + size_t nLen = WC2MB(NULL, pwz, 0); + if ( nLen != (size_t)-1 ) + { + wxCharBuffer buf(nLen); + WC2MB(buf.data(), pwz, nLen + 1); -// ---------------------------------------------------------------------------- -// standard file conversion -// ---------------------------------------------------------------------------- - -WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile; + return buf; + } + } -// just use the libc conversion for now -size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const -{ - return wxMB2WC(buf, psz, n); -} + wxCharBuffer buf((char *)NULL); -size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const -{ - return wxWC2MB(buf, psz, n); + return buf; } // ---------------------------------------------------------------------------- @@ -198,26 +287,32 @@ WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - if (buf) { - return gdk_mbstowcs((GdkWChar *)buf, psz, n); - } else { - GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; - size_t len = gdk_mbstowcs(nbuf, psz, n); - delete [] nbuf; - return len; - } + if (buf) + { + return gdk_mbstowcs((GdkWChar *)buf, psz, n); + } + else + { + GdkWChar *nbuf = new GdkWChar[n=strlen(psz)]; + size_t len = gdk_mbstowcs(nbuf, psz, n); + delete[] nbuf; + return len; + } } size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const { - char *mbstr = gdk_wcstombs((GdkWChar *)psz); - size_t len = mbstr ? strlen(mbstr) : 0; - if (buf) { - if (len > n) len = n; - memcpy(buf, psz, len); - if (len < n) buf[len] = 0; - } - return len; + char *mbstr = gdk_wcstombs((GdkWChar *)psz); + size_t len = mbstr ? strlen(mbstr) : 0; + if (buf) + { + if (len > n) + len = n; + memcpy(buf, psz, len); + if (len < n) + buf[len] = 0; + } + return len; } #endif // GTK > 1.0 @@ -259,437 +354,722 @@ size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8; -static wxUint32 utf8_max[]={0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff,0xffffffff}; +static wxUint32 utf8_max[]= + { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff }; size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const { - size_t len = 0; - - while (*psz && ((!buf) || (len>cnt); - while (cnt--) { - cc = *psz++; - if ((cc&0xC0)!=0x80) { - // invalid UTF-8 sequence - return (size_t)-1; - } - res=(res<<6)|(cc&0x3f); - } - if (res<=utf8_max[ocnt]) { - // illegal UTF-8 encoding - return (size_t)-1; + size_t len = 0; + + while (*psz && ((!buf) || (len < n))) + { + unsigned char cc = *psz++, fc = cc; + unsigned cnt; + for (cnt = 0; fc & 0x80; cnt++) + fc <<= 1; + if (!cnt) + { + // plain ASCII char + if (buf) + *buf++ = cc; + len++; } + else + { + cnt--; + if (!cnt) + { + // invalid UTF-8 sequence + return (size_t)-1; + } + else + { + unsigned ocnt = cnt - 1; + wxUint32 res = cc & (0x3f >> cnt); + while (cnt--) + { + cc = *psz++; + if ((cc & 0xC0) != 0x80) + { + // invalid UTF-8 sequence + return (size_t)-1; + } + res = (res << 6) | (cc & 0x3f); + } + if (res <= utf8_max[ocnt]) + { + // illegal UTF-8 encoding + return (size_t)-1; + } #ifdef WC_UTF16 - size_t pa = encode_utf16(res, buf); - if (pa == (size_t)-1) - return (size_t)-1; - if (buf) buf+=pa; - len+=pa; -#else - if (buf) *buf++=res; - len++; -#endif - } + size_t pa = encode_utf16(res, buf); + if (pa == (size_t)-1) + return (size_t)-1; + if (buf) + buf += pa; + len += pa; +#else // !WC_UTF16 + if (buf) + *buf++ = res; + len++; +#endif // WC_UTF16/!WC_UTF16 + } + } } - } - if (buf && (lenutf8_max[cnt]; cnt++); - if (!cnt) { - // plain ASCII char - if (buf) *buf++=cc; - len++; - } else { - len+=cnt+1; - if (buf) { - *buf++=(-128>>cnt)|((cc>>(cnt*6))&(0x3f>>cnt)); - while (cnt--) - *buf++=0x80|((cc>>(cnt*6))&0x3f); - } - } - } - if (buf && (len utf8_max[cnt]; cnt++) {} + if (!cnt) + { + // plain ASCII char + if (buf) + *buf++ = (char) cc; + len++; + } -// ---------------------------------------------------------------------------- -// specified character set -// ---------------------------------------------------------------------------- + else + { + len += cnt + 1; + if (buf) + { + *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt))); + while (cnt--) + *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f)); + } + } + } -WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL); + if (buf && (len wxConvUTF8) -// - move wxEncodingConverter meat in here - -#ifdef __WIN32__ -#include "wx/msw/registry.h" -// this should work if M$ Internet Exploiter is installed -static long CharsetToCodepage(const wxChar *name) -{ - if (!name) return GetACP(); - long CP=-1; - wxString cn(name); - do { - wxString path(wxT("MIME\\Database\\Charset\\")); - path += cn; - wxRegKey key(wxRegKey::HKCR,path); - - /* two cases: either there's an AliasForCharset string, - * or there are Codepage and InternetEncoding dwords. - * The InternetEncoding gives us the actual encoding, - * the Codepage just says which Windows character set to - * use when displaying the data. - */ - if (key.QueryValue(wxT("InternetEncoding"),&CP)) break; - // no encoding, see if it's an alias - if (!key.QueryValue(wxT("AliasForCharset"),cn)) break; - } while (1); - return CP; + return len; } -#endif + +// ============================================================================ +// wxCharacterSet and derived classes +// ============================================================================ + +// ---------------------------------------------------------------------------- +// wxCharacterSet is the ABC for the classes below +// ---------------------------------------------------------------------------- class wxCharacterSet { public: - const wxChar*cname; - 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: - wxMBConv*work; - ID_CharSet(const wxChar*name,wxMBConv*cnv) : wxCharacterSet(name), work(cnv) {} - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) - { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; } - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) - { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } - bool usable() { return work!=NULL; } + ID_CharSet(const wxChar *name, wxMBConv *cnv) + : wxCharacterSet(name), work(cnv) {} + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) + { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; } + + 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 +// yet another bug in glibc) the only case when iconv() returns with (size_t)-1 +// (which means error) and says there are 0 bytes left in the input buffer -- +// when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence, +// this alternative test for iconv() failure. +// [This bug does not appear in glibc 2.2.] +#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1 +#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \ + (errno != E2BIG || bufLeft != 0)) +#else +#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: - iconv_t m2w, w2m; - IC_CharSet(const wxChar*name) : wxCharacterSet(name), m2w((iconv_t)-1), w2m((iconv_t)-1) {} - ~IC_CharSet() { - if (m2w!=(iconv_t)-1) iconv_close(m2w); - if (w2m!=(iconv_t)-1) iconv_close(w2m); - } - void LoadM2W() { if (m2w==(iconv_t)-1) m2w=iconv_open(WC_NAME,wxConvLibc.cWX2MB(cname)); } - void LoadW2M() { if (w2m==(iconv_t)-1) w2m=iconv_open(wxConvLibc.cWX2MB(cname),WC_NAME); } - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) { - LoadM2W(); + 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) +{ + // Do it the hard way + char cname[100]; + for (size_t i = 0; i < wxStrlen(name)+1; i++) + cname[i] = (char) name[i]; + + // check for charset that represents wchar_t: + if (ms_wcCharsetName == NULL) + { + ms_wcNeedsSwap = FALSE; + + // try charset with explicit bytesex info (e.g. "UCS-4LE"): + ms_wcCharsetName = WC_NAME_BEST; + m2w = iconv_open(ms_wcCharsetName, cname); + + 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, cname); + + // last bet, try if it knows WCHAR_T pseudo-charset + if (m2w == (iconv_t)-1) + { + ms_wcCharsetName = "WCHAR_T"; + m2w = iconv_open(ms_wcCharsetName, cname); + } + + 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(_("Conversion 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, cname); + } + + // NB: don't ever pass NULL to iconv_open(), it may crash! + if ( ms_wcCharsetName ) + { + w2m = iconv_open( cname, ms_wcCharsetName); + } + else + { + w2m = (iconv_t)-1; + } +} + +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 outbuf = n * SIZEOF_WCHAR_T; size_t res, cres; - fprintf(stderr,"IC Convert to WC using %s\n",(const char*)wxConvLibc.cWX2MB(cname)); - if (buf) { - // have destination buffer, convert there - cres = iconv(m2w,&psz,&inbuf,(char**)&buf,&outbuf); - res = n-(outbuf/SIZEOF_WCHAR_T); - // convert to native endianness - WC_BSWAP(buf, res) - } else { - // no destination buffer... convert using temp buffer - // to calculate destination buffer requirement - wchar_t tbuf[8]; - res = 0; - do { - buf = tbuf; outbuf = 8*SIZEOF_WCHAR_T; - cres = iconv(m2w,&psz,&inbuf,(char**)&buf,&outbuf); - res += 8-(outbuf/SIZEOF_WCHAR_T); - } while ((cres==(size_t)-1) && (errno==E2BIG)); - } - if (cres==(size_t)-1) return (size_t)-1; + // VS: Use these instead of psz, buf because iconv() modifies its arguments: + wchar_t *bufPtr = buf; + const char *pszPtr = psz; + + if (buf) + { + // have destination buffer, convert there + cres = iconv(m2w, + ICONV_CHAR_CAST(&pszPtr), &inbuf, + (char**)&bufPtr, &outbuf); + res = n - (outbuf / SIZEOF_WCHAR_T); + + if (ms_wcNeedsSwap) + { + // convert to native endianness + WC_BSWAP(buf /* _not_ bufPtr */, res) + } + + // NB: iconv was given only strlen(psz) characters on input, and so + // it couldn't convert the trailing zero. Let's do it ourselves + // if there's some room left for it in the output buffer. + if (res < n) + buf[res] = 0; + } + 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)); + } + + 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 WC2MB(char*buf, const wchar_t*psz, size_t n) { - LoadW2M(); -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz); -#else - size_t inbuf = ::wcslen(psz); -#endif +} + +size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n) +{ + size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T; size_t outbuf = n; size_t res, cres; - fprintf(stderr,"IC Convert from WC using %s\n",(const char*)wxConvLibc.cWX2MB(cname)); -#ifdef WC_NEED_BSWAP - // 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); - memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); - WC_BSWAP(tmpbuf, inbuf) - psz=tmpbuf; -#endif - if (buf) { - // have destination buffer, convert there - cres = iconv(w2m,(const char**)&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,(const char**)&psz,&inbuf,&buf,&outbuf); - res += 16 - outbuf; - } while ((cres==(size_t)-1) && (errno==E2BIG)); - } -#ifdef WC_NEED_BSWAP - free(tmpbuf); -#endif - if (cres==(size_t)-1) return (size_t)-1; + + 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) + tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T); + memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T); + WC_BSWAP(tmpbuf, inbuf) + psz=tmpbuf; + } + + if (buf) + { + // have destination buffer, convert there + cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf ); + + res = n-outbuf; + + // NB: iconv was given only wcslen(psz) characters on input, and so + // it couldn't convert the trailing zero. Let's do it ourselves + // if there's some room left for it in the output buffer. + if (res < n) + buf[0] = 0; + } + 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); + } + + 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; - } - bool usable() { return TRUE; } -}; -#endif +} + +#endif // HAVE_ICONV + +// ============================================================================ +// Win32 conversion classes +// ============================================================================ + +#if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) + +extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp -#ifdef __WIN32__ class CP_CharSet : public wxCharacterSet { public: - long CodePage; - CP_CharSet(const wxChar*name) : wxCharacterSet(name), CodePage(CharsetToCodepage(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); - return len?len:(size_t)-1; - } - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) { - size_t len = WideCharToMultiByte(CodePage,0,psz,-1,buf,buf?n:0,NULL,NULL); - return len?len:(size_t)-1; - } - bool usable() { return CodePage!=-1; } + CP_CharSet(const wxChar* name) + : wxCharacterSet(name) + { + m_CodePage = wxCharsetToCodepage(name); + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) + { + const size_t len = ::MultiByteToWideChar + ( + m_CodePage, // code page + 0, // flags (none) + psz, // input string + -1, // its length (NUL-terminated) + buf, // output string + buf ? n : 0 // size of output buffer + ); + + // note that it returns # of written chars for buf != NULL and *size* + // of the needed buffer for buf == NULL + return len ? (buf ? len : len - 1) : (size_t)-1; + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) + { + const size_t len = ::WideCharToMultiByte + ( + m_CodePage, // code page + 0, // flags (none) + psz, // input string + -1, // it is (wide) NUL-terminated + buf, // output buffer + buf ? n : 0, // and its size + NULL, // default "replacement" char + NULL // [out] was it used? + ); + + // see the comment above! + return len ? (buf ? len : len - 1) : (size_t)-1; + } + + bool usable() const + { return m_CodePage != -1; } + +public: + long m_CodePage; }; -#endif +#endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) + +// ============================================================================ +// wxEncodingConverter based conversion classes +// ============================================================================ + +#if wxUSE_FONTMAP class EC_CharSet : public wxCharacterSet { public: - // temporarily just use wxEncodingConverter stuff, - // so that it works while a better implementation is built - wxFontEncoding enc; - wxEncodingConverter m2w, w2m; - 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); - } - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) { - size_t inbuf = strlen(psz); - fprintf(stderr,"EC Convert to WC using %d\n",enc); - if (buf) m2w.Convert(psz,buf); - return inbuf; - } - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) { -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz); -#else - size_t inbuf = ::wcslen(psz); -#endif - fprintf(stderr,"EC Convert from WC using %d\n",enc); - if (buf) w2m.Convert(psz,buf); - return inbuf; - } - bool usable() { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } + // 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) + { + if (name) + enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE); + + m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) && + w2m.Init(wxFONTENCODING_UNICODE, enc); + } + + size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) + { + size_t inbuf = strlen(psz); + if (buf) + m2w.Convert(psz,buf); + return inbuf; + } + + size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) + { + const size_t inbuf = wxWcslen(psz); + if (buf) + w2m.Convert(psz,buf); + + return inbuf; + } + + 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) { - if (!wxStricmp(name, wxT("UTF8")) || !wxStricmp(name, wxT("UTF-8"))) { - cset = new ID_CharSet(name, &wxConvUTF8); - } else { -#ifdef HAVE_ICONV_H - cset = new IC_CharSet(name); // may not take NULL -#endif + // check for the special case of ASCII charset +#if wxUSE_FONTMAP + if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT ) +#else // wxUSE_FONTMAP + if ( !name ) +#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP + { + // don't convert at all + return NULL; } - } - if (cset && cset->usable()) return cset; - if (cset) delete cset; -#ifdef __WIN32__ - cset = new CP_CharSet(name); // may take NULL - if (cset->usable()) return cset; -#endif - if (cset) delete cset; - cset = new EC_CharSet(name); - if (cset->usable()) return cset; - delete cset; - return NULL; + + // the test above must have taken care of this case + wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") ); + + wxCharacterSet *cset; + + if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) + { + 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 + } + + // it can only be NULL in this case +#ifndef HAVE_ICONV + if ( cset ) +#endif // !HAVE_ICONV + { + if ( cset->usable() ) + return cset; + + delete cset; + cset = NULL; + } + +#if defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) + cset = new CP_CharSet(name); + if ( cset->usable() ) + return cset; + + delete cset; + cset = NULL; +#endif // defined(__WIN32__) && !defined(__WXMICROWIN__) && !defined(__WXUNIVERSAL__) + +#if wxUSE_FONTMAP + cset = new EC_CharSet(name); + if ( cset->usable() ) + return cset; + + delete cset; + 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; - m_cset = (wxCharacterSet *) NULL; - m_deferred = TRUE; - SetName(charset); + m_name = (wxChar *)NULL; + m_cset = (wxCharacterSet *) NULL; + m_deferred = TRUE; + + SetName(charset); } wxCSConv::~wxCSConv() { - if (m_name) free(m_name); - if (m_cset) delete m_cset; + Clear(); } -void wxCSConv::SetName(const wxChar *charset) +wxCSConv::wxCSConv(const wxCSConv& conv) + : wxMBConv() { - if (charset) { - m_name = wxStrdup(charset); - m_deferred = TRUE; - } + Clear(); + SetName(conv.m_name); } -void wxCSConv::LoadNow() +wxCSConv& wxCSConv::operator=(const wxCSConv& conv) { -// wxPrintf(wxT("Conversion request\n")); - if (m_deferred) { - if (!m_name) { -#ifdef __UNIX__ -#if defined(HAVE_LANGINFO_H) && defined(CODESET) - // GNU libc provides current character set this way - char*alang = nl_langinfo(CODESET); - if (alang) SetName(wxConvLibc.cMB2WX(alang)); - else -#endif - // if we can't get at the character set directly, - // try to see if it's in the environment variables - // (in most cases this won't work, but I was out of ideas) - { - wxChar *lang = wxGetenv(wxT("LC_ALL")); - if (!lang) lang = wxGetenv(wxT("LC_CTYPE")); - if (!lang) lang = wxGetenv(wxT("LANG")); - wxChar *dot = lang ? wxStrchr(lang, wxT('.')) : (wxChar *)NULL; - if (dot) SetName(dot+1); - } -#endif - } - m_cset = wxGetCharacterSet(m_name); - m_deferred = FALSE; - } + Clear(); + SetName(conv.m_name); + return *this; } -size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const +void wxCSConv::Clear() { - ((wxCSConv *)this)->LoadNow(); // discard constness - if (m_cset) - return m_cset->MB2WC(buf, psz, n); - - // latin-1 (direct) - size_t len=strlen(psz); - if (buf) { - for (size_t c=0; c<=len; c++) - buf[c] = (unsigned char)(psz[c]); - } - return len; + if (m_name) + free(m_name); + if (m_cset) + delete m_cset; + m_name = NULL; + m_cset = NULL; } -size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const +void wxCSConv::SetName(const wxChar *charset) { - ((wxCSConv *)this)->LoadNow(); // discard constness - if (m_cset) - return m_cset->WC2MB(buf, psz, n); + if (charset) + { + m_name = wxStrdup(charset); + m_deferred = TRUE; + } +} - // latin-1 (direct) -#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t len=std::wcslen(psz); -#else - size_t len=::wcslen(psz); -#endif - if (buf) { - for (size_t c=0; c<=len; c++) - buf[c] = (psz[c]>0xff) ? '?' : psz[c]; - } - return len; +void wxCSConv::LoadNow() +{ + if ( m_deferred ) + { + // it would probably be better to make GetSystemEncodingName() always + // available (i.e. even when wxUSE_INTL == 0)? +#if wxUSE_INTL + if ( !m_name ) + { + wxString name = wxLocale::GetSystemEncodingName(); + if ( !name.empty() ) + { + SetName(name); + } + } +#endif // wxUSE_INTL + + // wxGetCharacterSet() complains about NULL name + m_cset = m_name ? wxGetCharacterSet(m_name) : NULL; + m_deferred = FALSE; + } } -#ifdef HAVE_ICONV_H -class IC_CharSetConverter +size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const { -public: - iconv_t cnv; - 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; - size_t res = iconv(cnv,&psz,&inbuf,&buf,&outbuf); - if (res==(size_t)-1) return (size_t)-1; - return n-outbuf; - } -}; -#endif + ((wxCSConv *)this)->LoadNow(); // discard constness + + if (m_cset) + return m_cset->MB2WC(buf, psz, n); + + // latin-1 (direct) + size_t len = strlen(psz); + + if (buf) + { + for (size_t c = 0; c <= len; c++) + buf[c] = (unsigned char)(psz[c]); + } + + return len; +} -class EC_CharSetConverter +size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const { -public: - wxEncodingConverter cnv; - 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; - } -}; + ((wxCSConv *)this)->LoadNow(); // discard constness -#else // !wxUSE_WCHAR_T + if (m_cset) + return m_cset->WC2MB(buf, psz, n); -// ---------------------------------------------------------------------------- -// stand-ins in absence of wchar_t -// ---------------------------------------------------------------------------- + // latin-1 (direct) + const size_t len = wxWcslen(psz); + if (buf) + { + for (size_t c = 0; c <= len; c++) + buf[c] = (psz[c] > 0xff) ? '?' : psz[c]; + } -WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile; + return len; +} #endif // wxUSE_WCHAR_T