Copy about.htm

[wxWidgets.git] / src / common / strconv.cpp

diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp
index db0147d6bc7706eaff14e6e4a6c3a1bb92ef2f95..88e49338d756679cd9696d859e6afdd8f4bd3064 100644 (file)
--- a/src/common/strconv.cpp
+++ b/src/common/strconv.cpp
@@ -55,9 +55,6 @@
 #include <ctype.h>
 #include <string.h>
 #include <stdlib.h>
 #include <ctype.h>
 #include <string.h>
 #include <stdlib.h>

-#ifdef HAVE_LANGINFO_H
-  #include <langinfo.h>
-#endif

 
 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
     #define wxHAVE_WIN32_MB2WC
 
 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
     #define wxHAVE_WIN32_MB2WC


@@ -155,7 +152,7 @@ static size_t decode_utf16(const wxUint16* input, wxUint32& output)
         output = *input;
         return 1;
     }
         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;
     {
         output = *input;
         return (size_t)-1;


@@ -360,50 +357,39 @@ size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
     return wxWC2MB(buf, psz, n);
 }
 
     return wxWC2MB(buf, psz, n);
 }
 

+#ifdef __UNIX__
+

 // ----------------------------------------------------------------------------
 // ----------------------------------------------------------------------------

-// wxConvBrokenFileNames is made for GTK2 in Unicode mode when
-// files are accidentally written in an encoding which is not
-// the system encoding. Typically, the system encoding will be
-// UTF8 but there might be files stored in ISO8859-1 on disk.
+// wxConvBrokenFileNames 

 // ----------------------------------------------------------------------------
 
 // ----------------------------------------------------------------------------
 

-class wxConvBrokenFileNames: public wxMBConvLibc
-{
-public:
-    wxConvBrokenFileNames() : m_utf8conv(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL) { }
-    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
-    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
-    inline bool UseUTF8() const;
-private:
-    wxMBConvUTF8 m_utf8conv;
-};
-
-bool wxConvBrokenFileNames::UseUTF8() const
+wxConvBrokenFileNames::wxConvBrokenFileNames(const wxChar *charset)

 {
 {

-#if defined HAVE_LANGINFO_H && defined CODESET
-    char *codeset = nl_langinfo(CODESET);
-    return strcmp(codeset, "UTF-8") == 0;
-#else
-    return false;
-#endif
+    if ( !charset || wxStricmp(charset, _T("UTF-8")) == 0
+                  || wxStricmp(charset, _T("UTF8")) == 0  )
+        m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL);
+    else
+        m_conv = new wxCSConv(charset);

 }
 
 }
 

-size_t wxConvBrokenFileNames::MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const
+size_t
+wxConvBrokenFileNames::MB2WC(wchar_t *outputBuf,
+                             const char *psz,
+                             size_t outputSize) const

 {
 {

-    if (UseUTF8())
-        return m_utf8conv.MB2WC( outputBuf, psz, outputSize );
-    else
-        return wxMBConvLibc::MB2WC( outputBuf, psz, outputSize );
+    return m_conv->MB2WC( outputBuf, psz, outputSize );

 }
 
 }
 

-size_t wxConvBrokenFileNames::WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const
+size_t
+wxConvBrokenFileNames::WC2MB(char *outputBuf,
+                             const wchar_t *psz,
+                             size_t outputSize) const

 {
 {

-    if (UseUTF8())
-        return m_utf8conv.WC2MB( outputBuf, psz, outputSize );
-    else
-        return wxMBConvLibc::WC2MB( outputBuf, psz, outputSize );
+    return m_conv->WC2MB( outputBuf, psz, outputSize );

 }
 
 }
 

+#endif
+

 // ----------------------------------------------------------------------------
 // UTF-7
 // ----------------------------------------------------------------------------
 // ----------------------------------------------------------------------------
 // UTF-7
 // ----------------------------------------------------------------------------


@@ -558,7 +544,7 @@ size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
         }
 #ifndef WC_UTF16
         else if (((wxUint32)cc) > 0xffff)
         }
 #ifndef WC_UTF16
         else if (((wxUint32)cc) > 0xffff)

-           {
+        {

             // no surrogate pair generation (yet?)
             return (size_t)-1;
         }
             // no surrogate pair generation (yet?)
             return (size_t)-1;
         }


@@ -638,6 +624,15 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
             if (buf)
                 *buf++ = cc;
             len++;
             if (buf)
                 *buf++ = cc;
             len++;

+
+            // escape the escape character for octal escapes
+            if ((m_options & MAP_INVALID_UTF8_TO_OCTAL)
+                    && cc == '\\' && (!buf || len < n))
+            {
+                if (buf)
+                    *buf++ = cc;
+                len++;
+            }

         }
         else
         {
         }
         else
         {


@@ -728,9 +723,9 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
                         {
                             unsigned char n = *opsz;
                             *buf++ = L'\\';
                         {
                             unsigned char n = *opsz;
                             *buf++ = L'\\';

-                            *buf++ = L'0' + n / 0100;
-                            *buf++ = L'0' + (n % 0100) / 010;
-                            *buf++ = L'0' + n % 010;
+                            *buf++ = (wchar_t)( L'0' + n / 0100 );
+                            *buf++ = (wchar_t)( L'0' + (n % 0100) / 010 );
+                            *buf++ = (wchar_t)( L'0' + n % 010 );

                         }
                         opsz++;
                         len += 4;
                         }
                         opsz++;
                         len += 4;


@@ -775,15 +770,23 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
                 *buf++ = (char)(cc - wxUnicodePUA);
             len++;
         }
                 *buf++ = (char)(cc - wxUnicodePUA);
             len++;
         }

+        else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL)
+                    && cc == L'\\' && psz[0] == L'\\' )
+        {
+            if (buf)
+                *buf++ = (char)cc;
+            psz++;
+            len++;
+        }

         else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) &&
                     cc == L'\\' &&
                         isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) )
         {
             if (buf)
             {
         else if ( (m_options & MAP_INVALID_UTF8_TO_OCTAL) &&
                     cc == L'\\' &&
                         isoctal(psz[0]) && isoctal(psz[1]) && isoctal(psz[2]) )
         {
             if (buf)
             {

-                *buf++ = (char) (psz[0] - L'0')*0100 +
-                                (psz[1] - L'0')*010 +
-                                (psz[2] - L'0');
+                *buf++ = (char) ((psz[0] - L'0')*0100 +
+                                 (psz[1] - L'0')*010 +
+                                 (psz[2] - L'0'));

             }
 
             psz += 3;
             }
 
             psz += 3;


@@ -2736,7 +2739,6 @@ static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM);
 static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
 static wxMBConvUTF7 wxConvUTF7Obj;
 static wxMBConvUTF8 wxConvUTF8Obj;
 static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
 static wxMBConvUTF7 wxConvUTF7Obj;
 static wxMBConvUTF8 wxConvUTF8Obj;

-static wxConvBrokenFileNames wxConvBrokenFileNamesObj;

 
 WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
 WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;
 
 WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
 WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;


@@ -2747,8 +2749,6 @@ WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj;
 WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = &
 #ifdef __WXOSX__
                                     wxConvUTF8Obj;
 WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName = &
 #ifdef __WXOSX__
                                     wxConvUTF8Obj;

-#elif __WXGTK20__
-                                    wxConvBrokenFileNamesObj;

 #else
                                     wxConvLibcObj;
 #endif
 #else
                                     wxConvLibcObj;
 #endif