X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/4def3b359546ee186b3db723db719f85b761f36a..b838cfc9151aea38402ad2b1ba5d2f97cf94e973:/src/common/strconv.cpp diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp index 54064c7064..82656dd1d2 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 ///////////////////////////////////////////////////////////////////////////// @@ -37,6 +37,25 @@ #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 @@ -44,67 +63,59 @@ #ifdef HAVE_ICONV_H #include #endif -#ifdef HAVE_LANGINFO_H - #include -#endif #ifdef __WXMSW__ #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; _c=0x110000) + } + else if (input>=0x110000) { return (size_t)-1; - } - else + } + else { - if (output) + if (output) { *output++ = (input >> 10)+0xd7c0; *output++ = (input&0x3ff)+0xdc00; @@ -113,19 +124,19 @@ static size_t encode_utf16(wxUint32 input,wxUint16*output) } } -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; @@ -209,11 +220,11 @@ WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk; 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); @@ -226,12 +237,12 @@ 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 (buf) { - if (len > n) + if (len > n) len = n; memcpy(buf, psz, len); - if (len < n) + if (len < n) buf[len] = 0; } return len; @@ -276,49 +287,49 @@ size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf), 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; @@ -327,18 +338,18 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const 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; } @@ -347,29 +358,29 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const { 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--) @@ -396,24 +407,24 @@ WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL); // - perhaps common encodings to objects ("UTF8" -> wxConvUTF8) // - move wxEncodingConverter meat in here -#ifdef __WIN32__ +#if defined(__WIN32__) && !defined(__WXMICROWIN__) #include "wx/msw/registry.h" // this should work if M$ Internet Exploiter is installed static long CharsetToCodepage(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, @@ -421,12 +432,12 @@ static long CharsetToCodepage(const wxChar *name) // 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 @@ -438,7 +449,7 @@ public: : cname(name) {} virtual ~wxCharacterSet() {} - virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) + 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; } @@ -453,13 +464,13 @@ class ID_CharSet : public wxCharacterSet public: 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; } public: @@ -469,6 +480,9 @@ public: #ifdef HAVE_ICONV_H +bool g_wcNeedsSwap = FALSE; +static const char *g_wcCharset = NULL; + // 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 @@ -486,21 +500,85 @@ public: class IC_CharSet : public wxCharacterSet { public: - IC_CharSet(const wxChar *name) + IC_CharSet(const wxChar *name) : wxCharacterSet(name) { - m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname)); - w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME); + // check for charset that represents wchar_t: + if (g_wcCharset == NULL) + { + g_wcNeedsSwap = FALSE; + + // try charset with explicit bytesex info (e.g. "UCS-4LE"): + g_wcCharset = WC_NAME_BEST; + m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); + + if (m2w == (iconv_t)-1) + { + // try charset w/o bytesex info (e.g. "UCS4") + // and check for bytesex ourselves: + g_wcCharset = WC_NAME; + m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); + + // last bet, try if it knows WCHAR_T pseudo-charset + if (m2w == (iconv_t)-1) + { + g_wcCharset = "WCHAR_T"; + m2w = iconv_open(g_wcCharset, 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; + + #ifdef WX_ICONV_TAKES_CHAR + res = iconv(m2w, (char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz); + #else + res = iconv(m2w, (const char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz); + #endif + if (ICONV_FAILED(res, insz)) + { + g_wcCharset = NULL; + wxLogLastError(wxT("iconv")); + wxLogError(_("Convertion to charset '%s' doesn't work."), name); + } + else + { + g_wcNeedsSwap = (wbuf[0] != (wchar_t)buf[0]); + } + } + else + { + g_wcCharset = NULL; + wxLogError(_("Don't know how to convert to/from charset '%s'."), name); + } + } + wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), g_wcCharset, g_wcNeedsSwap); + } + else + m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); + + w2m = iconv_open(wxConvLibc.cWX2MB(name), g_wcCharset); } - ~IC_CharSet() + ~IC_CharSet() { - if ( m2w != (iconv_t)-1 ) + if ( m2w != (iconv_t)-1 ) iconv_close(m2w); - if ( w2m != (iconv_t)-1 ) + if ( w2m != (iconv_t)-1 ) iconv_close(w2m); } - + size_t MB2WC(wchar_t *buf, const char *psz, size_t n) { size_t inbuf = strlen(psz); @@ -519,10 +597,12 @@ public: 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 + + if (g_wcNeedsSwap) + { + // convert to native endianness + WC_BSWAP(buf /* _not_ bufPtr */, res) + } } else { @@ -540,13 +620,17 @@ public: 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) { #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) @@ -556,17 +640,21 @@ public: #endif size_t outbuf = n; size_t res, cres; + + wchar_t *tmpbuf; + + if (g_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; + } -#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 @@ -593,58 +681,71 @@ public: res += 16 - outbuf; } while ((cres==(size_t)-1) && (errno==E2BIG)); } -#ifdef WC_NEED_BSWAP - free(tmpbuf); -#endif + + if (g_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 (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); } - -public: + +protected: iconv_t m2w, w2m; }; #endif -#ifdef __WIN32__ +#if defined(__WIN32__) && !defined(__WXMICROWIN__) 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 len = + size_t len = MultiByteToWideChar(CodePage, 0, psz, -1, buf, buf ? n : 0); - 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; } - + 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; + //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; } public: long CodePage; }; -#endif +#endif // __WIN32__ + +#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), + EC_CharSet(const wxChar*name) : wxCharacterSet(name), enc(wxFONTENCODING_SYSTEM) { if (name) @@ -652,15 +753,15 @@ public: 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) @@ -670,10 +771,10 @@ public: #endif if (buf) w2m.Convert(psz,buf); - + return inbuf; } - + bool usable() { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); } @@ -682,6 +783,8 @@ public: wxEncodingConverter m2w, w2m; }; +#endif // wxUSE_FONTMAP + static wxCharacterSet *wxGetCharacterSet(const wxChar *name) { wxCharacterSet *cset = NULL; @@ -698,17 +801,30 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name) #endif } } - - if (cset && cset->usable()) return cset; - if (cset) delete cset; - cset = NULL; -#ifdef __WIN32__ + + if (cset && cset->usable()) + return cset; + + if (cset) + { + delete cset; + cset = NULL; + } + +#if defined(__WIN32__) && !defined(__WXMICROWIN__) 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__ + +#if wxUSE_FONTMAP cset = new EC_CharSet(name); - if (cset->usable()) return cset; + if (cset->usable()) + return cset; +#endif // wxUSE_FONTMAP + delete cset; wxLogError(_("Unknown encoding '%s'!"), name); return NULL; @@ -716,16 +832,17 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name) 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) @@ -741,34 +858,15 @@ void wxCSConv::LoadNow() { 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; } } @@ -776,26 +874,26 @@ 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); - + 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); @@ -810,26 +908,27 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const 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)); + cnv = iconv_open(wxConvLibc.cWX2MB(to->cname), + wxConvLibc.cWX2MB(from->cname)); } - + ~IC_CharSetConverter() - { - if (cnv != (iconv_t)-1) - iconv_close(cnv); + { + if (cnv != (iconv_t)-1) + iconv_close(cnv); } - + size_t Convert(char *buf, const char *psz, size_t n) { size_t inbuf = strlen(psz); @@ -839,7 +938,7 @@ public: #else size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf ); #endif - if (res == (size_t)-1) + if (res == (size_t)-1) return (size_t)-1; return (n - outbuf); } @@ -847,21 +946,22 @@ public: public: iconv_t cnv; }; -#endif + +#endif // HAVE_ICONV_H class EC_CharSetConverter { public: - EC_CharSetConverter(EC_CharSet*from,EC_CharSet*to) + 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 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; };