// Licence: wxWindows licence
/////////////////////////////////////////////////////////////////////////////
-/*
- * About ref counting:
- * 1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
- * 2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
- * 3) Unlock() decrements nRefs and frees memory if it goes to 0
- */
-
// ===========================================================================
// headers, declarations, constants
// ===========================================================================
#ifndef WX_PRECOMP
#include "wx/string.h"
+ #include "wx/wxcrtvararg.h"
#endif
#include <ctype.h>
#include <clib.h>
#endif
+#include "wx/hashmap.h"
// string handling functions used by wxString:
#if wxUSE_UNICODE_UTF8
{
// FIXME-UTF8: always, not only if wxUSE_UNICODE
#if wxUSE_UNICODE && !defined(__BORLANDC__)
- return os << str.AsWChar();
+ return os << (const wchar_t*)str.AsWCharBuf();
#else
- return os << str.AsChar();
+ return os << (const char*)str.AsCharBuf();
#endif
}
// wxString class core
// ===========================================================================
+#if wxUSE_UNICODE_UTF8
+
+void wxString::PosLenToImpl(size_t pos, size_t len,
+ size_t *implPos, size_t *implLen) const
+{
+ if ( pos == npos )
+ *implPos = npos;
+ else
+ {
+ const_iterator i = begin() + pos;
+ *implPos = wxStringImpl::const_iterator(i.impl()) - m_impl.begin();
+ if ( len == npos )
+ *implLen = npos;
+ else
+ {
+ // too large length is interpreted as "to the end of the string"
+ // FIXME-UTF8: verify this is the case in std::string, assert
+ // otherwise
+ if ( pos + len > length() )
+ len = length() - pos;
+
+ *implLen = (i + len).impl() - i.impl();
+ }
+ }
+}
+
+#endif // wxUSE_UNICODE_UTF8
+
+// ----------------------------------------------------------------------------
+// wxCStrData converted strings caching
+// ----------------------------------------------------------------------------
+
+// FIXME-UTF8: temporarily disabled because it doesn't work with global
+// string objects; re-enable after fixing this bug and benchmarking
+// performance to see if using a hash is a good idea at all
+#if 0
+
+// For backward compatibility reasons, it must be possible to assign the value
+// returned by wxString::c_str() to a char* or wchar_t* variable and work with
+// it. Returning wxCharBuffer from (const char*)c_str() wouldn't do the trick,
+// because the memory would be freed immediately, but it has to be valid as long
+// as the string is not modified, so that code like this still works:
+//
+// const wxChar *s = str.c_str();
+// while ( s ) { ... }
+
+// FIXME-UTF8: not thread safe!
+// FIXME-UTF8: we currently clear the cached conversion only when the string is
+// destroyed, but we should do it when the string is modified, to
+// keep memory usage down
+// FIXME-UTF8: we do the conversion every time As[W]Char() is called, but if we
+// invalidated the cache on every change, we could keep the previous
+// conversion
+// FIXME-UTF8: add tracing of usage of these two methods - new code is supposed
+// to use mb_str() or wc_str() instead of (const [w]char*)c_str()
+
+template<typename T>
+static inline void DeleteStringFromConversionCache(T& hash, const wxString *s)
+{
+ typename T::iterator i = hash.find(wxConstCast(s, wxString));
+ if ( i != hash.end() )
+ {
+ free(i->second);
+ hash.erase(i);
+ }
+}
+
+#if wxUSE_UNICODE
+// NB: non-STL implementation doesn't compile with "const wxString*" key type,
+// so we have to use wxString* here and const-cast when used
+WX_DECLARE_HASH_MAP(wxString*, char*, wxPointerHash, wxPointerEqual,
+ wxStringCharConversionCache);
+static wxStringCharConversionCache gs_stringsCharCache;
+
+const char* wxCStrData::AsChar() const
+{
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsCharCache, m_str);
+
+ // convert the string and keep it:
+ const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
+ m_str->mb_str().release();
+
+ return s + m_offset;
+}
+#endif // wxUSE_UNICODE
+
+#if !wxUSE_UNICODE_WCHAR
+WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
+ wxStringWCharConversionCache);
+static wxStringWCharConversionCache gs_stringsWCharCache;
+
+const wchar_t* wxCStrData::AsWChar() const
+{
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
+
+ // convert the string and keep it:
+ const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
+ m_str->wc_str().release();
+
+ return s + m_offset;
+}
+#endif // !wxUSE_UNICODE_WCHAR
+
+wxString::~wxString()
+{
+#if wxUSE_UNICODE
+ // FIXME-UTF8: do this only if locale is not UTF8 if wxUSE_UNICODE_UTF8
+ DeleteStringFromConversionCache(gs_stringsCharCache, this);
+#endif
+#if !wxUSE_UNICODE_WCHAR
+ DeleteStringFromConversionCache(gs_stringsWCharCache, this);
+#endif
+}
+#endif
+
+#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+const char* wxCStrData::AsChar() const
+{
+#if wxUSE_UNICODE_UTF8
+ if ( wxLocaleIsUtf8 )
+ return AsInternal();
+#endif
+ // under non-UTF8 locales, we have to convert the internal UTF-8
+ // representation using wxConvLibc and cache the result
+
+ wxString *str = wxConstCast(m_str, wxString);
+
+ // convert the string:
+ wxCharBuffer buf(str->mb_str());
+
+ // FIXME-UTF8: do the conversion in-place in the existing buffer
+ if ( str->m_convertedToChar &&
+ strlen(buf) == strlen(str->m_convertedToChar) )
+ {
+ // keep the same buffer for as long as possible, so that several calls
+ // to c_str() in a row still work:
+ strcpy(str->m_convertedToChar, buf);
+ }
+ else
+ {
+ str->m_convertedToChar = buf.release();
+ }
+
+ // and keep it:
+ return str->m_convertedToChar + m_offset;
+}
+#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+
+#if !wxUSE_UNICODE_WCHAR
+const wchar_t* wxCStrData::AsWChar() const
+{
+ wxString *str = wxConstCast(m_str, wxString);
+
+ // convert the string:
+ wxWCharBuffer buf(str->wc_str());
+
+ // FIXME-UTF8: do the conversion in-place in the existing buffer
+ if ( str->m_convertedToWChar &&
+ wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) )
+ {
+ // keep the same buffer for as long as possible, so that several calls
+ // to c_str() in a row still work:
+ memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf));
+ }
+ else
+ {
+ str->m_convertedToWChar = buf.release();
+ }
+
+ // and keep it:
+ return str->m_convertedToWChar + m_offset;
+}
+#endif // !wxUSE_UNICODE_WCHAR
+
+// ===========================================================================
+// wxString class core
+// ===========================================================================
+
// ---------------------------------------------------------------------------
// construction and conversion
// ---------------------------------------------------------------------------
-#if wxUSE_UNICODE
+#if wxUSE_UNICODE_WCHAR
/* static */
wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
const wxMBConv& conv)
{
// anything to do?
if ( !psz || nLength == 0 )
- return SubstrBufFromMB();
+ return SubstrBufFromMB(L"", 0);
if ( nLength == npos )
nLength = wxNO_LEN;
size_t wcLen;
wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
if ( !wcLen )
- return SubstrBufFromMB();
+ return SubstrBufFromMB(_T(""), 0);
else
return SubstrBufFromMB(wcBuf, wcLen);
}
-#else
+#endif // wxUSE_UNICODE_WCHAR
+
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+ const wxMBConv& conv)
+{
+ // anything to do?
+ if ( !psz || nLength == 0 )
+ return SubstrBufFromMB("", 0);
+
+ // if psz is already in UTF-8, we don't have to do the roundtrip to
+ // wchar_t* and back:
+ if ( conv.IsUTF8() )
+ {
+ // we need to validate the input because UTF8 iterators assume valid
+ // UTF-8 sequence and psz may be invalid:
+ if ( wxStringOperations::IsValidUtf8String(psz, nLength) )
+ {
+ return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
+ }
+ // else: do the roundtrip through wchar_t*
+ }
+
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
+
+ // first convert to wide string:
+ size_t wcLen;
+ wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ if ( !wcLen )
+ return SubstrBufFromMB("", 0);
+
+ // and then to UTF-8:
+ SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxConvUTF8));
+ // widechar -> UTF-8 conversion isn't supposed to ever fail:
+ wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") );
+
+ return buf;
+}
+#endif // wxUSE_UNICODE_UTF8
+
+#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
/* static */
wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
const wxMBConv& conv)
{
// anything to do?
if ( !pwz || nLength == 0 )
- return SubstrBufFromWC();
+ return SubstrBufFromWC("", 0);
if ( nLength == npos )
nLength = wxNO_LEN;
size_t mbLen;
wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
if ( !mbLen )
- return SubstrBufFromWC();
+ return SubstrBufFromWC("", 0);
else
return SubstrBufFromWC(mbBuf, mbLen);
}
-#endif
+#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
-#if wxUSE_UNICODE
+#if wxUSE_UNICODE_WCHAR
//Convert wxString in Unicode mode to a multi-byte string
const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
{
- return conv.cWC2MB(c_str(), length() + 1 /* size, not length */, NULL);
+ return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);
}
-#else // ANSI
+#elif wxUSE_UNICODE_UTF8
-#if wxUSE_WCHAR_T
+const wxWCharBuffer wxString::wc_str() const
+{
+ return wxConvUTF8.cMB2WC(m_impl.c_str(),
+ m_impl.length() + 1 /* size, not length */,
+ NULL);
+}
+
+const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+{
+ if ( conv.IsUTF8() )
+ return wxCharBuffer::CreateNonOwned(m_impl.c_str());
+
+ // FIXME-UTF8: use wc_str() here once we have buffers with length
+
+ size_t wcLen;
+ wxWCharBuffer wcBuf(
+ wxConvUTF8.cMB2WC(m_impl.c_str(),
+ m_impl.length() + 1 /* size, not length */,
+ &wcLen));
+ if ( !wcLen )
+ return wxCharBuffer("");
+
+ return conv.cWC2MB(wcBuf, wcLen, NULL);
+}
+
+#else // ANSI
//Converts this string to a wide character string if unicode
//mode is not enabled and wxUSE_WCHAR_T is enabled
const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const
{
- return conv.cMB2WC(c_str(), length() + 1 /* size, not length */, NULL);
+ return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL);
}
-#endif // wxUSE_WCHAR_T
-
#endif // Unicode/ANSI
// shrink to minimal size (releasing extra memory)
// deprecated compatibility code:
#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
-wxChar *wxString::GetWriteBuf(size_t nLen)
+wxStringCharType *wxString::GetWriteBuf(size_t nLen)
{
return DoGetWriteBuf(nLen);
}
if (!ascii)
return wxEmptyString;
- size_t len = strlen( ascii );
+ size_t len = strlen(ascii);
wxString res;
if ( len )
{
- wxStringBuffer buf(res, len);
-
- wchar_t *dest = buf;
+ wxImplStringBuffer buf(res, len);
+ wxStringCharType *dest = buf;
for ( ;; )
{
- if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
- break;
+ unsigned char c = (unsigned char)*ascii++;
+ wxASSERT_MSG( c < 0x80,
+ _T("Non-ASCII value passed to FromAscii().") );
+
+ *dest++ = (wchar_t)c;
+
+ if ( c == '\0' )
+ break;
}
}
{
// What do we do with '\0' ?
- wxString res;
- res += (wchar_t)(unsigned char) ascii;
+ unsigned char c = (unsigned char)ascii;
- return res;
+ wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") );
+
+ // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value
+ return wxString(wxUniChar((wchar_t)c));
}
const wxCharBuffer wxString::ToAscii() const
{
// this will allocate enough space for the terminating NUL too
wxCharBuffer buffer(length());
-
-
char *dest = buffer.data();
- const wchar_t *pwc = c_str();
- for ( ;; )
+ for ( const_iterator i = begin(); i != end(); ++i )
{
- *dest++ = (char)(*pwc > SCHAR_MAX ? wxT('_') : *pwc);
+ wxUniChar c(*i);
+ // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?')
+ *dest++ = c.IsAscii() ? (char)c : '_';
// the output string can't have embedded NULs anyhow, so we can safely
// stop at first of them even if we do have any
- if ( !*pwc++ )
+ if ( !c )
break;
}
return buffer;
}
-#endif // Unicode
+#endif // wxUSE_UNICODE
// extract string of length nCount starting at nFirst
wxString wxString::Mid(size_t nFirst, size_t nCount) const
wxASSERT_MSG( suffix, _T("invalid parameter in wxString::EndssWith") );
int start = length() - wxStrlen(suffix);
- if ( start < 0 || wxStrcmp(wx_str() + start, suffix) != 0 )
+
+ if ( start < 0 || compare(start, npos, suffix) != 0 )
return false;
if ( rest )
bool wxString::IsAscii() const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!isascii(*s)) return(false);
- s++;
- }
- return(true);
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !(*i).IsAscii() )
+ return false;
+ }
+
+ return true;
}
bool wxString::IsWord() const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!wxIsalpha(*s)) return(false);
- s++;
- }
- return(true);
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !wxIsalpha(*i) )
+ return false;
+ }
+
+ return true;
}
bool wxString::IsNumber() const
{
- const wxChar *s = (const wxChar*) *this;
- if (wxStrlen(s))
- if ((s[0] == wxT('-')) || (s[0] == wxT('+'))) s++;
- while(*s){
- if(!wxIsdigit(*s)) return(false);
- s++;
- }
- return(true);
+ if ( empty() )
+ return true;
+
+ const_iterator i = begin();
+
+ if ( *i == _T('-') || *i == _T('+') )
+ ++i;
+
+ for ( ; i != end(); ++i )
+ {
+ if ( !wxIsdigit(*i) )
+ return false;
+ }
+
+ return true;
}
wxString wxString::Strip(stripType w) const
bool wxString::ToLongLong(wxLongLong_t *val, int base) const
{
-#ifdef wxHAS_STRTOLL
return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoll);
-#else
- // TODO: implement this ourselves
- wxUnusedVar(val);
- wxUnusedVar(base);
- return false;
-#endif // wxHAS_STRTOLL
}
bool wxString::ToULongLong(wxULongLong_t *val, int base) const
{
-#ifdef wxHAS_STRTOLL
return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoull);
-#else
- // TODO: implement this ourselves
- wxUnusedVar(val);
- wxUnusedVar(base);
- return false;
-#endif
}
bool wxString::ToDouble(double *val) const
// formatted output
// ---------------------------------------------------------------------------
+#if !wxUSE_UTF8_LOCALE_ONLY
/* static */
#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
-wxString wxStringPrintfMixinBase::DoFormat(const wxChar *format, ...)
+wxString wxStringPrintfMixinBase::DoFormatWchar(const wxChar *format, ...)
#else
-wxString wxString::DoFormat(const wxChar *format, ...)
+wxString wxString::DoFormatWchar(const wxChar *format, ...)
#endif
{
va_list argptr;
return s;
}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
+
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString wxString::DoFormatUtf8(const char *format, ...)
+{
+ va_list argptr;
+ va_start(argptr, format);
+
+ wxString s;
+ s.PrintfV(format, argptr);
+
+ va_end(argptr);
+
+ return s;
+}
+#endif // wxUSE_UNICODE_UTF8
/* static */
wxString wxString::FormatV(const wxString& format, va_list argptr)
return s;
}
+#if !wxUSE_UTF8_LOCALE_ONLY
#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
-int wxStringPrintfMixinBase::DoPrintf(const wxChar *format, ...)
+int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...)
#else
-int wxString::DoPrintf(const wxChar *format, ...)
+int wxString::DoPrintfWchar(const wxChar *format, ...)
#endif
{
va_list argptr;
return iLen;
}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
-int wxString::PrintfV(const wxString& format, va_list argptr)
+#if wxUSE_UNICODE_UTF8
+int wxString::DoPrintfUtf8(const char *format, ...)
+{
+ va_list argptr;
+ va_start(argptr, format);
+
+ int iLen = PrintfV(format, argptr);
+
+ va_end(argptr);
+
+ return iLen;
+}
+#endif // wxUSE_UNICODE_UTF8
+
+#if wxUSE_UNICODE_UTF8
+template<typename BufferType>
+#else
+// we only need one version in non-UTF8 builds and at least two Windows
+// compilers have problems with this function template, so use just one
+// normal function here
+#endif
+static int DoStringPrintfV(wxString& str,
+ const wxString& format, va_list argptr)
{
int size = 1024;
for ( ;; )
{
- wxStringBuffer tmp(*this, size + 1);
+#if wxUSE_UNICODE_UTF8
+ BufferType tmp(str, size + 1);
+ typename BufferType::CharType *buf = tmp;
+#else
+ wxStringBuffer tmp(str, size + 1);
wxChar *buf = tmp;
+#endif
if ( !buf )
{
}
// we could have overshot
- Shrink();
+ str.Shrink();
- return length();
+ return str.length();
+}
+
+int wxString::PrintfV(const wxString& format, va_list argptr)
+{
+ va_list argcopy;
+ wxVaCopy(argcopy, argptr);
+
+#if wxUSE_UNICODE_UTF8
+ #if wxUSE_STL_BASED_WXSTRING
+ typedef wxStringTypeBuffer<char> Utf8Buffer;
+ #else
+ typedef wxImplStringBuffer Utf8Buffer;
+ #endif
+#endif
+
+#if wxUSE_UTF8_LOCALE_ONLY
+ return DoStringPrintfV<Utf8Buffer>(*this, format, argcopy);
+#else
+ #if wxUSE_UNICODE_UTF8
+ if ( wxLocaleIsUtf8 )
+ return DoStringPrintfV<Utf8Buffer>(*this, format, argcopy);
+ else
+ // wxChar* version
+ return DoStringPrintfV<wxStringBuffer>(*this, format, argcopy);
+ #else
+ return DoStringPrintfV(*this, format, argcopy);
+ #endif // UTF8/WCHAR
+#endif
}
// ----------------------------------------------------------------------------
// convert to lower case, return the copy of the string
wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
-
projects / wxWidgets.git / blobdiff