X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/dccce9eae10e099d791cc055dd02d3dda731778e..4aaef122cbbd5bbe0e70b824e320458e2329dd13:/src/common/strconv.cpp?ds=inline

diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp
index 33b934ef9f..bd39f333b0 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 <string.h>
 #include <stdlib.h>
 
+
+#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 <clib.h>
 #endif
@@ -44,54 +63,46 @@
 #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)
+#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
+    #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
+    #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
 
-// ----------------------------------------------------------------------------
-// globals
-// ----------------------------------------------------------------------------
-
-WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
-
-// ============================================================================
-// implementation
-// ============================================================================
-
-#if wxUSE_WCHAR_T
 
 #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)
     {
@@ -113,7 +124,7 @@ 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))
     {
@@ -351,7 +362,7 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
     {
         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;
@@ -396,10 +407,62 @@ 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__)
+
+#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)
         return GetACP();
@@ -412,7 +475,7 @@ static long CharsetToCodepage(const wxChar *name)
         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.
@@ -429,7 +492,10 @@ static long CharsetToCodepage(const wxChar *name)
 
     return CP;
 }
-#endif
+
+#endif // !wxUSE_GUI || !wxUSE_FONTMAP
+
+#endif // Win32
 
 class wxCharacterSet
 {
@@ -469,6 +535,9 @@ 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
@@ -489,11 +558,75 @@ public:
     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;
 
-    ~IC_CharSet()
+            // 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() 
     {
         if ( m2w != (iconv_t)-1 )
             iconv_close(m2w);
@@ -519,10 +652,12 @@ 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
         {
@@ -542,7 +677,11 @@ public:
         }
 
         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;
     }
@@ -556,17 +695,21 @@ public:
 #endif
         size_t outbuf = n;
         size_t res, cres;
+        
+        wchar_t *tmpbuf = 0;
+
+        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
@@ -593,11 +736,18 @@ 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;
     }
@@ -605,47 +755,56 @@ public:
     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)
-        : 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;
+            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; }
+        { return m_CodePage != -1; }
 
 public:
-    long CodePage;
+    long m_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),
-                                    enc(wxFONTENCODING_SYSTEM)
+    EC_CharSet(const wxChar* name) : wxCharacterSet(name),
+                                     enc(wxFONTENCODING_SYSTEM)
     {
         if (name)
             enc = wxTheFontMapper->CharsetToEncoding(name, FALSE);
@@ -663,7 +822,8 @@ public:
 
     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
     {
-#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
+#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
+    || ( defined(__MWERKS__) && defined(__WXMSW__) )
         size_t inbuf = std::wcslen(psz);
 #else
         size_t inbuf = ::wcslen(psz);
@@ -682,6 +842,8 @@ public:
     wxEncodingConverter m2w, w2m;
 };
 
+#endif // wxUSE_FONTMAP
+
 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
 {
     wxCharacterSet *cset = NULL;
@@ -699,14 +861,16 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
         }
     }
 
-    if (cset && cset->usable()) return cset;
+    if (cset && cset->usable())
+        return cset;
+
     if (cset)
     {
         delete cset;
         cset = NULL;
     }
 
-#ifdef __WIN32__
+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
     cset = new CP_CharSet(name); // may take NULL
     if (cset->usable())
         return cset;
@@ -714,9 +878,11 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
     delete cset;
 #endif // __WIN32__
 
+#if wxUSE_FONTMAP
     cset = new EC_CharSet(name);
     if (cset->usable())
         return cset;
+#endif // wxUSE_FONTMAP
 
     delete cset;
     wxLogError(_("Unknown encoding '%s'!"), name);
@@ -727,6 +893,8 @@ wxCSConv::wxCSConv(const wxChar *charset)
 {
     m_name = (wxChar *)NULL;
     m_cset = (wxCharacterSet *) NULL;
+    m_deferred = TRUE;
+
     SetName(charset);
 }
 
@@ -756,7 +924,8 @@ void wxCSConv::LoadNow()
                 SetName(name);
         }
 
-        m_cset = wxGetCharacterSet(m_name);
+        // wxGetCharacterSet() complains about NULL name
+        m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
         m_deferred = FALSE;
     }
 }
@@ -788,7 +957,8 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
         return m_cset->WC2MB(buf, psz, n);
 
     // latin-1 (direct)
-#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
+#if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
+    || ( defined(__MWERKS__) && defined(__WXMSW__) )
     size_t len=std::wcslen(psz);
 #else
     size_t len=::wcslen(psz);
@@ -842,10 +1012,10 @@ public:
 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);