/////////////////////////////////////////////////////////////////////////////
// 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
/////////////////////////////////////////////////////////////////////////////
#include <string.h>
#include <stdlib.h>
-#ifdef __SALFORDC__
- #include <clib.h>
-#endif
-
-#ifdef HAVE_ICONV_H
- #include <iconv.h>
-#endif
-#ifdef HAVE_LANGINFO_H
- #include <langinfo.h>
-#endif
-
-#ifdef __WXMSW__
- #include <windows.h>
-#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; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
-#define BSWAP_UCS2(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
-#define WC_NEED_BSWAP
-#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
-#endif
-
// ----------------------------------------------------------------------------
// globals
// ----------------------------------------------------------------------------
WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
+
// ============================================================================
// implementation
// ============================================================================
#if wxUSE_WCHAR_T
+#ifdef __SALFORDC__
+ #include <clib.h>
+#endif
+
+#ifdef HAVE_ICONV
+ #include <iconv.h>
+#endif
+
+#ifdef __WXMSW__
+ #include <windows.h>
+#endif
+
+#define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
+#define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
+
+// under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
+// it might be not defined - assume the most common value
+#ifndef SIZEOF_WCHAR_T
+ #define SIZEOF_WCHAR_T 2
+#endif // !defined(SIZEOF_WCHAR_T)
+
+#if SIZEOF_WCHAR_T == 4
+ #define WC_NAME "UCS4"
+ #define WC_BSWAP BSWAP_UCS4
+ #ifdef WORDS_BIGENDIAN
+ #define WC_NAME_BEST "UCS-4BE"
+ #else
+ #define WC_NAME_BEST "UCS-4LE"
+ #endif
+#elif SIZEOF_WCHAR_T == 2
+ #define WC_NAME "UTF16"
+ #define WC_BSWAP BSWAP_UTF16
+ #define WC_UTF16
+ #ifdef WORDS_BIGENDIAN
+ #define WC_NAME_BEST "UTF-16BE"
+ #else
+ #define WC_NAME_BEST "UTF-16LE"
+ #endif
+#else // sizeof(wchar_t) != 2 nor 4
+ // I don't know what to do about this
+ #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
+#endif
+
+
#ifdef WC_UTF16
-static size_t encode_utf16(wxUint32 input,wxUint16*output)
+static size_t encode_utf16(wxUint32 input, wchar_t *output)
{
- if (input<=0xffff)
+ if (input<=0xffff)
{
if (output) *output++ = input;
return 1;
- }
- else if (input>=0x110000)
+ }
+ else if (input>=0x110000)
{
return (size_t)-1;
- }
- else
+ }
+ else
{
- if (output)
+ if (output)
{
*output++ = (input >> 10)+0xd7c0;
*output++ = (input&0x3ff)+0xdc00;
}
}
-static size_t decode_utf16(wxUint16*input,wxUint32&output)
+static size_t decode_utf16(const wchar_t* input, wxUint32& output)
{
- if ((*input<0xd800) || (*input>0xdfff))
+ if ((*input<0xd800) || (*input>0xdfff))
{
output = *input;
return 1;
- }
- else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
+ }
+ else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
{
output = *input;
return (size_t)-1;
- }
- else
+ }
+ else
{
output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
return 2;
size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
- if (buf)
+ if (buf)
{
return gdk_mbstowcs((GdkWChar *)buf, psz, n);
- }
- else
+ }
+ else
{
GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
size_t len = gdk_mbstowcs(nbuf, psz, n);
{
char *mbstr = gdk_wcstombs((GdkWChar *)psz);
size_t len = mbstr ? strlen(mbstr) : 0;
- if (buf)
+ if (buf)
{
- if (len > n)
+ if (len > n)
len = n;
memcpy(buf, psz, len);
- if (len < n)
+ if (len < n)
buf[len] = 0;
}
return len;
WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
-static wxUint32 utf8_max[]=
+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 < n)))
+ while (*psz && ((!buf) || (len < n)))
{
unsigned char cc = *psz++, fc = cc;
unsigned cnt;
- for (cnt = 0; fc & 0x80; cnt++)
+ for (cnt = 0; fc & 0x80; cnt++)
fc <<= 1;
- if (!cnt)
+ if (!cnt)
{
// plain ASCII char
- if (buf)
+ if (buf)
*buf++ = cc;
len++;
- }
- else
+ }
+ else
{
cnt--;
- if (!cnt)
+ if (!cnt)
{
// invalid UTF-8 sequence
return (size_t)-1;
- }
- else
+ }
+ else
{
unsigned ocnt = cnt - 1;
wxUint32 res = cc & (0x3f >> cnt);
- while (cnt--)
+ while (cnt--)
{
cc = *psz++;
- if ((cc & 0xC0) != 0x80)
+ if ((cc & 0xC0) != 0x80)
{
// invalid UTF-8 sequence
return (size_t)-1;
}
res = (res << 6) | (cc & 0x3f);
}
- if (res <= utf8_max[ocnt])
+ if (res <= utf8_max[ocnt])
{
// illegal UTF-8 encoding
return (size_t)-1;
size_t pa = encode_utf16(res, buf);
if (pa == (size_t)-1)
return (size_t)-1;
- if (buf)
+ if (buf)
buf += pa;
len += pa;
#else
- if (buf)
+ if (buf)
*buf++ = res;
len++;
#endif
}
}
}
- if (buf && (len < n))
+ if (buf && (len < n))
*buf = 0;
return len;
}
{
size_t len = 0;
- while (*psz && ((!buf) || (len < n)))
+ while (*psz && ((!buf) || (len < n)))
{
wxUint32 cc;
#ifdef WC_UTF16
- size_t pa = decode_utf16(psz,cc);
+ size_t pa = decode_utf16(psz, cc);
psz += (pa == (size_t)-1) ? 1 : pa;
#else
cc=(*psz++) & 0x7fffffff;
#endif
unsigned cnt;
for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
- if (!cnt)
+ if (!cnt)
{
// plain ASCII char
- if (buf)
+ if (buf)
*buf++ = cc;
len++;
- }
-
- else
+ }
+
+ else
{
len += cnt + 1;
- if (buf)
+ if (buf)
{
*buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt));
while (cnt--)
// - perhaps common encodings to objects ("UTF8" -> wxConvUTF8)
// - move wxEncodingConverter meat in here
-#ifdef __WIN32__
+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
+
+#if wxUSE_GUI
+
+// VZ: the new version of wxCharsetToCodepage() is more politically correct
+// and should work on other Windows versions as well but the old version is
+// still needed for !wxUSE_FONTMAP || !wxUSE_GUI case
+
+extern long wxEncodingToCodepage(wxFontEncoding encoding)
+{
+ // translate encoding into the Windows CHARSET
+ wxNativeEncodingInfo natveEncInfo;
+ if ( !wxGetNativeFontEncoding(encoding, &natveEncInfo) )
+ return -1;
+
+ // translate CHARSET to code page
+ CHARSETINFO csetInfo;
+ if ( !::TranslateCharsetInfo((DWORD *)(DWORD)natveEncInfo.charset,
+ &csetInfo,
+ TCI_SRCCHARSET) )
+ {
+ wxLogLastError(_T("TranslateCharsetInfo(TCI_SRCCHARSET)"));
+
+ return -1;
+ }
+
+ return csetInfo.ciACP;
+}
+
+#if wxUSE_FONTMAP
+
+extern long wxCharsetToCodepage(const wxChar *name)
+{
+ // first get the font encoding for this charset
+ if ( !name )
+ return -1;
+
+ wxFontEncoding enc = wxTheFontMapper->CharsetToEncoding(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)
+ if (!name)
return GetACP();
-
+
long CP=-1;
-
+
wxString cn(name);
do {
wxString path(wxT("MIME\\Database\\Charset\\"));
path += cn;
wxRegKey key(wxRegKey::HKCR, path);
- if (!key.Exists()) continue;
-
+ if (!key.Exists()) break;
+
// two cases: either there's an AliasForCharset string,
// or there are Codepage and InternetEncoding dwords.
// The InternetEncoding gives us the actual encoding,
// use when displaying the data.
if (key.HasValue(wxT("InternetEncoding")) &&
key.QueryValue(wxT("InternetEncoding"), &CP)) break;
-
+
// no encoding, see if it's an alias
if (!key.HasValue(wxT("AliasForCharset")) ||
!key.QueryValue(wxT("AliasForCharset"), cn)) break;
} while (1);
-
+
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)
{ 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()
+
+ 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
#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);
}
-
- ~IC_CharSet()
+ else // we already have ms_wcCharsetName
{
- if ( m2w != (iconv_t)-1 )
- iconv_close(m2w);
- if ( w2m != (iconv_t)-1 )
- iconv_close(w2m);
+ m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
}
-
- size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
+
+ // NB: don't ever pass NULL to iconv_open(), it may crash!
+ if ( ms_wcCharsetName )
{
- 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;
+ w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName);
+ }
+ else
+ {
+ w2m = (iconv_t)-1;
+ }
+}
- if (buf)
+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)
+ {
+ // 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)
{
- // 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;
}
-
- size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
+ 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 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;
+
+ wchar_t *tmpbuf = 0;
-#ifdef WC_NEED_BSWAP
+ 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;
}
-
- bool usable()
- { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
-
-public:
- iconv_t m2w, w2m;
-};
-#endif
-#ifdef __WIN32__
+ 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;
+}
+
+#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);
- return len ? len : (size_t)-1;
+ size_t len =
+ 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;
}
-
+
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);
- return len ? len : (size_t)-1;
+ //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
+#endif // __WIN32__
+
+// ============================================================================
+// 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
- 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)
{
size_t inbuf = strlen(psz);
- if (buf)
+ 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)
+#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
+ || ( defined(__MWERKS__) && defined(__WXMSW__) )
size_t inbuf = std::wcslen(psz);
#else
size_t inbuf = ::wcslen(psz);
#endif
if (buf)
w2m.Convert(psz,buf);
-
+
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;
+ }
+
+ // 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
}
-
- if (cset && cset->usable()) return cset;
- if (cset) delete cset;
+
+ if ( cset->usable() )
+ return cset;
+
+ delete cset;
cset = NULL;
-#ifdef __WIN32__
- cset = new CP_CharSet(name); // may take NULL
- if (cset->usable()) return cset;
-#endif
- if (cset) delete cset;
+
+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
+ 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()) return cset;
+ if ( cset->usable() )
+ return cset;
+
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;
+ 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;
+ free(m_name);
+ delete m_cset;
}
void wxCSConv::SetName(const wxChar *charset)
{
if (m_deferred)
{
- if (!m_name)
+ 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
+ wxString name = wxLocale::GetSystemEncodingName();
+ if ( !name.empty() )
+ SetName(name);
}
- m_cset = wxGetCharacterSet(m_name);
+
+ // wxGetCharacterSet() complains about NULL name
+ m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
m_deferred = FALSE;
}
}
size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
{
((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;
}
size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
{
((wxCSConv *)this)->LoadNow(); // discard constness
-
+
if (m_cset)
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);
for (size_t c = 0; c <= len; c++)
buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
}
-
+
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
-
-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
// ----------------------------------------------------------------------------
projects / wxWidgets.git / blobdiff