X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/95c8801cdf222d27d92350a0f67938554fea8a63..803c55373484d1e2bfed85b46cbbfff8bfad1276:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 95fc0d9c07..33b934ef9f 100644 --- a/src/common/strconv.cpp +++ b/src/common/strconv.cpp @@ -54,6 +54,8 @@ #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) @@ -91,34 +93,43 @@ WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc; static size_t encode_utf16(wxUint32 input,wxUint16*output) { - if (input<=0xffff) { - if (output) *output++ = input; - return 1; - } else - if (input>=0x110000) { - return (size_t)-1; - } else { - if (output) { - *output++ = (input >> 10)+0xd7c0; - *output++ = (input&0x3ff)+0xdc00; - } - return 2; - } + if (input<=0xffff) + { + if (output) *output++ = input; + return 1; + } + else if (input>=0x110000) + { + return (size_t)-1; + } + else + { + if (output) + { + *output++ = (input >> 10)+0xd7c0; + *output++ = (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; - } + 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 @@ -198,26 +209,32 @@ 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 @@ -259,86 +276,110 @@ 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; + 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++; + if (buf) + *buf++ = res; + len++; #endif - } + } + } } - } - 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++ = 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 && (lenMB2WC(buf,psz,n) : (size_t)-1; } - - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) + + 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; } public: wxMBConv*work; }; + #ifdef HAVE_ICONV_H + +// 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 + class IC_CharSet : public wxCharacterSet { public: - IC_CharSet(const wxChar*name) - : wxCharacterSet(name), m2w((iconv_t)-1), w2m((iconv_t)-1) {} - ~IC_CharSet() + IC_CharSet(const wxChar *name) + : wxCharacterSet(name) { - if (m2w!=(iconv_t)-1) iconv_close(m2w); - if (w2m!=(iconv_t)-1) iconv_close(w2m); + m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname)); + w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME); } - - 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); + + ~IC_CharSet() + { + if ( m2w != (iconv_t)-1 ) + iconv_close(m2w); + if ( w2m != (iconv_t)-1 ) + iconv_close(w2m); } - - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) { - LoadM2W(); 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)); + // 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 #ifdef WX_ICONV_TAKES_CHAR - cres = iconv( m2w, (char**)&psz, &inbuf, (char**)&buf, &outbuf ); + cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf); #else - cres = iconv( m2w, &psz, &inbuf, (char**)&buf, &outbuf ); + cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf); #endif - res = n-(outbuf/SIZEOF_WCHAR_T); + res = n - (outbuf / SIZEOF_WCHAR_T); // convert to native endianness - WC_BSWAP(buf, res) +#ifdef WC_NEED_BSWAP + WC_BSWAP(buf /* _not_ bufPtr */, res) +#endif } else { @@ -472,39 +531,38 @@ public: wchar_t tbuf[8]; res = 0; do { - buf = tbuf; outbuf = 8*SIZEOF_WCHAR_T; + bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T; #ifdef WX_ICONV_TAKES_CHAR - cres = iconv( m2w, (char**)&psz, &inbuf, (char**)&buf, &outbuf ); + cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); #else - cres = iconv( m2w, &psz, &inbuf, (char**)&buf, &outbuf ); + cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); #endif res += 8-(outbuf/SIZEOF_WCHAR_T); } while ((cres==(size_t)-1) && (errno==E2BIG)); } - - if (cres==(size_t)-1) + + if (ICONV_FAILED(cres, inbuf)) return (size_t)-1; - + return res; } - - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) + + size_t WC2MB(char *buf, const wchar_t *psz, size_t n) { - LoadW2M(); #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) - size_t inbuf = std::wcslen(psz); + size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; #else - size_t inbuf = ::wcslen(psz); + size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T; #endif 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); + 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; @@ -538,15 +596,15 @@ public: #ifdef WC_NEED_BSWAP free(tmpbuf); #endif - if (cres==(size_t)-1) + if (ICONV_FAILED(cres, inbuf)) return (size_t)-1; - + return res; } - + bool usable() - { return TRUE; } - + { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } + public: iconv_t m2w, w2m; }; @@ -556,23 +614,25 @@ public: class CP_CharSet : public wxCharacterSet { public: - CP_CharSet(const wxChar*name) + CP_CharSet(const wxChar*name) : wxCharacterSet(name), CodePage(CharsetToCodepage(name)) {} - - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) + + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) { - size_t len = MultiByteToWideChar(CodePage,0,psz,-1,buf,buf?n:0); + 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 WC2MB(char *buf, const wchar_t *psz, size_t n) { - size_t len = WideCharToMultiByte(CodePage,0,psz,-1,buf,buf?n:0,NULL,NULL); + 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; } + { return CodePage != -1; } public: long CodePage; @@ -584,36 +644,36 @@ 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); } - - size_t MB2WC(wchar_t*buf, const char*psz, size_t n) + + 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); + if (buf) + m2w.Convert(psz,buf); return inbuf; } - - size_t WC2MB(char*buf, const wchar_t*psz, size_t n) + + 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); } @@ -627,7 +687,7 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name) wxCharacterSet *cset = NULL; if (name) { - if (!wxStricmp(name, wxT("UTF8")) || !wxStricmp(name, wxT("UTF-8"))) + if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0) { cset = new ID_CharSet(name, &wxConvUTF8); } @@ -638,32 +698,42 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name) #endif } } - + if (cset && cset->usable()) return cset; - if (cset) delete cset; + if (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 (cset->usable()) + return cset; + + delete cset; +#endif // __WIN32__ + cset = new EC_CharSet(name); - if (cset->usable()) return cset; + if (cset->usable()) + return cset; + delete cset; + wxLogError(_("Unknown encoding '%s'!"), name); return NULL; } 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) @@ -677,33 +747,15 @@ void wxCSConv::SetName(const wxChar *charset) void wxCSConv::LoadNow() { -// wxPrintf(wxT("Conversion request\n")); 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); m_deferred = FALSE; } @@ -712,26 +764,26 @@ void wxCSConv::LoadNow() 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); - + size_t len = strlen(psz); + if (buf) { - for (size_t c=0; c<=len; c++) + 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); @@ -743,24 +795,31 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const #endif if (buf) { - for (size_t c=0; c<=len; c++) - buf[c] = (psz[c]>0xff) ? '?' : psz[c]; + 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(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) + { + 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; @@ -769,28 +828,30 @@ public: #else size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf ); #endif - if (res==(size_t)-1) return (size_t)-1; - return n-outbuf; + if (res == (size_t)-1) + return (size_t)-1; + return (n - outbuf); } public: iconv_t cnv; }; -#endif + +#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; };