#ifndef WX_PRECOMP
#include "wx/string.h"
#include "wx/wxcrtvararg.h"
+ #include "wx/log.h"
#endif
#include <ctype.h>
#include "wx/hashmap.h"
#include "wx/vector.h"
+#include "wx/xlocale.h"
+
+#ifdef __WXMSW__
+ #include "wx/msw/wrapwin.h"
+#endif // __WXMSW__
// string handling functions used by wxString:
#if wxUSE_UNICODE_UTF8
namespace wxPrivate
{
-static UntypedBufferData s_untypedNullData(NULL);
+static UntypedBufferData s_untypedNullData(NULL, 0);
UntypedBufferData * const untypedNullDataPtr = &s_untypedNullData;
// gdb seems to be unable to display thread-local variables correctly, at least
// not my 6.4.98 version under amd64, so provide this debugging helper to do it
-#ifdef __WXDEBUG__
+#if wxDEBUG_LEVEL >= 2
struct wxStrCacheDumper
{
void wxDumpStrCache() { wxStrCacheDumper::ShowAll(); }
-#endif // __WXDEBUG__
+#endif // wxDEBUG_LEVEL >= 2
#ifdef wxPROFILE_STRING_CACHE
wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
{
#if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8
- const wxCharBuffer buf(str.AsCharBuf());
+ const wxScopedCharBuffer buf(str.AsCharBuf());
if ( !buf )
os.clear(wxSTD ios_base::failbit);
else
return os << str.c_str();
}
-wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str)
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxScopedCharBuffer& str)
{
return os << str.data();
}
#ifndef __BORLANDC__
-wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str)
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxScopedWCharBuffer& str)
{
return os << str.data();
}
return wos << str.AsWChar();
}
-wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str)
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxScopedWCharBuffer& str)
{
return wos << str.data();
}
}
#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:
- //
- // FIXME-UTF8: we'd like to do the conversion in the existing buffer (if we
- // have it) but it's unfortunately not obvious to implement
- // because we don't know how big buffer do we need for the
- // given string length (in case of multibyte encodings, e.g.
- // ISO-2022-JP or UTF-8 when internal representation is wchar_t)
- //
- // One idea would be to store more than just m_convertedToChar
- // in wxString: then we could record the length of the string
- // which was converted the last time and try to reuse the same
- // buffer if the current length is not greater than it (this
- // could still fail because string could have been modified in
- // place but it would work most of the time, so we'd do it and
- // only allocate the new buffer if in-place conversion returned
- // an error). We could also store a bit saying if the string
- // was modified since the last conversion (and update it in all
- // operation modifying the string, of course) to avoid unneeded
- // consequential conversions. But both of these ideas require
- // adding more fields to wxString and require profiling results
- // to be sure that we really gain enough from them to justify
- // doing it.
- wxCharBuffer buf(str->mb_str());
-
- // if it failed, return empty string and not NULL to avoid crashes in code
- // written with either wxWidgets 2 wxString or std::string behaviour in
- // mind: neither of them ever returns NULL and so we shouldn't neither
- if ( !buf )
- return "";
-
- 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());
-
- // notice that here, unlike above in AsChar(), conversion can't fail as our
- // internal UTF-8 is always well-formed -- or the string was corrupted and
- // all bets are off anyhow
-
- // 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
// ===========================================================================
{
// anything to do?
if ( !psz || nLength == 0 )
- return SubstrBufFromMB(L"", 0);
+ return SubstrBufFromMB(wxWCharBuffer(L""), 0);
if ( nLength == npos )
nLength = wxNO_LEN;
size_t wcLen;
- wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ wxScopedWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
if ( !wcLen )
- return SubstrBufFromMB(_T(""), 0);
+ return SubstrBufFromMB(wxWCharBuffer(L""), 0);
else
return SubstrBufFromMB(wcBuf, wcLen);
}
{
// anything to do?
if ( !psz || nLength == 0 )
- return SubstrBufFromMB("", 0);
+ return SubstrBufFromMB(wxCharBuffer(""), 0);
// if psz is already in UTF-8, we don't have to do the roundtrip to
// wchar_t* and back:
// we must pass the real string length to SubstrBufFromMB ctor
if ( nLength == npos )
nLength = psz ? strlen(psz) : 0;
- return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
+ return SubstrBufFromMB(wxScopedCharBuffer::CreateNonOwned(psz, nLength),
+ nLength);
}
// else: do the roundtrip through wchar_t*
}
// first convert to wide string:
size_t wcLen;
- wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ wxScopedWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
if ( !wcLen )
- return SubstrBufFromMB("", 0);
+ return SubstrBufFromMB(wxCharBuffer(""), 0);
// and then to UTF-8:
SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8()));
{
// anything to do?
if ( !pwz || nLength == 0 )
- return SubstrBufFromWC("", 0);
+ return SubstrBufFromWC(wxCharBuffer(""), 0);
if ( nLength == npos )
nLength = wxNO_LEN;
size_t mbLen;
- wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
+ wxScopedCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
if ( !mbLen )
- return SubstrBufFromWC("", 0);
+ return SubstrBufFromWC(wxCharBuffer(""), 0);
else
return SubstrBufFromWC(mbBuf, mbLen);
}
#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
+// This std::string::c_str()-like method returns a wide char pointer to string
+// contents. In wxUSE_UNICODE_WCHAR case it is trivial as it can simply return
+// a pointer to the internal representation. Otherwise a conversion is required
+// and it returns a temporary buffer.
+//
+// However for compatibility with c_str() and to avoid breaking existing code
+// doing
+//
+// for ( const wchar_t *p = s.wc_str(); *p; p++ )
+// ... use *p...
+//
+// we actually need to ensure that the returned buffer is _not_ temporary and
+// so we use wxString::m_convertedToWChar to store the returned data
+#if !wxUSE_UNICODE_WCHAR
-#if wxUSE_UNICODE_WCHAR
-
-//Convert wxString in Unicode mode to a multi-byte string
-const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+const wchar_t *wxString::AsWChar(const wxMBConv& conv) const
{
- return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);
-}
+ const char * const strMB = m_impl.c_str();
+ const size_t lenMB = m_impl.length();
-#elif wxUSE_UNICODE_UTF8
+ // find out the size of the buffer needed
+ const size_t lenWC = conv.ToWChar(NULL, 0, strMB, lenMB);
+ if ( lenWC == wxCONV_FAILED )
+ return NULL;
-const wxWCharBuffer wxString::wc_str() const
-{
- return wxMBConvStrictUTF8().cMB2WC
- (
- m_impl.c_str(),
- m_impl.length() + 1, // size, not length
- NULL
- );
+ // keep the same buffer if the string size didn't change: this is not only
+ // an optimization but also ensure that code which modifies string
+ // character by character (without changing its length) can continue to use
+ // the pointer returned by a previous wc_str() call even after changing the
+ // string
+
+ // TODO-UTF8: we could check for ">" instead of "!=" here as this would
+ // allow to save on buffer reallocations but at the cost of
+ // consuming (even) more memory, we should benchmark this to
+ // determine if it's worth doing
+ if ( !m_convertedToWChar.m_str || lenWC != m_convertedToWChar.m_len )
+ {
+ if ( !const_cast<wxString *>(this)->m_convertedToWChar.Extend(lenWC) )
+ return NULL;
+ }
+
+ // finally do convert
+ m_convertedToWChar.m_str[lenWC] = L'\0';
+ if ( conv.ToWChar(m_convertedToWChar.m_str, lenWC,
+ strMB, lenMB) == wxCONV_FAILED )
+ return NULL;
+
+ return m_convertedToWChar.m_str;
}
-const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+#endif // !wxUSE_UNICODE_WCHAR
+
+
+// Same thing for mb_str() which returns a normal char pointer to string
+// contents: this always requires converting it to the specified encoding in
+// non-ANSI build except if we need to convert to UTF-8 and this is what we
+// already use internally.
+#if wxUSE_UNICODE
+
+const char *wxString::AsChar(const wxMBConv& conv) const
{
+#if wxUSE_UNICODE_UTF8
if ( conv.IsUTF8() )
- return wxCharBuffer::CreateNonOwned(m_impl.c_str());
+ return m_impl.c_str();
- // FIXME-UTF8: use wc_str() here once we have buffers with length
+ const wchar_t * const strWC = AsWChar(wxMBConvStrictUTF8());
+ const size_t lenWC = m_convertedToWChar.m_len;
+#else // wxUSE_UNICODE_WCHAR
+ const wchar_t * const strWC = m_impl.c_str();
+ const size_t lenWC = m_impl.length();
+#endif // wxUSE_UNICODE_UTF8/wxUSE_UNICODE_WCHAR
- size_t wcLen;
- wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC
- (
- m_impl.c_str(),
- m_impl.length() + 1, // size
- &wcLen
- ));
- if ( !wcLen )
- return wxCharBuffer("");
+ const size_t lenMB = conv.FromWChar(NULL, 0, strWC, lenWC);
+ if ( lenMB == wxCONV_FAILED )
+ return NULL;
- return conv.cWC2MB(wcBuf, wcLen+1, NULL);
-}
+ if ( !m_convertedToChar.m_str || lenMB != m_convertedToChar.m_len )
+ {
+ if ( !const_cast<wxString *>(this)->m_convertedToChar.Extend(lenMB) )
+ return NULL;
+ }
-#else // ANSI
+ m_convertedToChar.m_str[lenMB] = '\0';
+ if ( conv.FromWChar(m_convertedToChar.m_str, lenMB,
+ strWC, lenWC) == wxCONV_FAILED )
+ return NULL;
-//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(wx_str(), length() + 1 /* size, not length */, NULL);
+ return m_convertedToChar.m_str;
}
-#endif // Unicode/ANSI
+#endif // wxUSE_UNICODE
// shrink to minimal size (releasing extra memory)
bool wxString::Shrink()
int wxString::CmpNoCase(const wxString& s) const
{
-#if wxUSE_UNICODE_UTF8
- // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
+#if defined(__WXMSW__) && !wxUSE_UNICODE_UTF8
+ // prefer to use CompareString() if available as it's more efficient than
+ // doing it manual or even using wxStricmp() (see #10375)
+ switch ( ::CompareString(LOCALE_USER_DEFAULT, NORM_IGNORECASE,
+ m_impl.c_str(), m_impl.length(),
+ s.m_impl.c_str(), s.m_impl.length()) )
+ {
+ case CSTR_LESS_THAN:
+ return -1;
+
+ case CSTR_EQUAL:
+ return 0;
+
+ case CSTR_GREATER_THAN:
+ return 1;
+ default:
+ wxFAIL_MSG( "unexpected CompareString() return value" );
+ // fall through
+
+ case 0:
+ wxLogLastError("CompareString");
+ // use generic code below
+ }
+#endif // __WXMSW__ && !wxUSE_UNICODE_UTF8
+
+ // do the comparison manually: notice that we can't use wxStricmp() as it
+ // doesn't handle embedded NULs
+
+ // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
const_iterator i1 = begin();
const_iterator end1 = end();
const_iterator i2 = s.begin();
else if ( len1 > len2 )
return 1;
return 0;
-#else // wxUSE_UNICODE_WCHAR or ANSI
- return wxStricmp(m_impl.c_str(), s.m_impl.c_str());
-#endif
}
return wxString(wxUniChar((wchar_t)c));
}
-const wxCharBuffer wxString::ToAscii() const
+const wxScopedCharBuffer wxString::ToAscii() const
{
// this will allocate enough space for the terminating NUL too
wxCharBuffer buffer(length());
#define DO_IF_NOT_WINCE(x)
#endif
-#define WX_STRING_TO_INT_TYPE(out, base, func, T) \
- wxCHECK_MSG( out, false, _T("NULL output pointer") ); \
- wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \
- \
+#define WX_STRING_TO_X_TYPE_START \
+ wxCHECK_MSG( pVal, false, _T("NULL output pointer") ); \
DO_IF_NOT_WINCE( errno = 0; ) \
- \
const wxStringCharType *start = wx_str(); \
- wxStringCharType *end; \
- T val = func(start, &end, base); \
- \
+ wxStringCharType *end;
+
+#define WX_STRING_TO_X_TYPE_END \
/* return true only if scan was stopped by the terminating NUL and */ \
/* if the string was not empty to start with and no under/overflow */ \
/* occurred: */ \
if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \
return false; \
- *out = val; \
- return true
+ *pVal = val; \
+ return true;
bool wxString::ToLong(long *pVal, int base) const
{
- WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long);
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ long val = wxStrtol(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
}
bool wxString::ToULong(unsigned long *pVal, int base) const
{
- WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long);
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ unsigned long val = wxStrtoul(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
}
bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const
{
- WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t);
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ wxLongLong_t val = wxStrtoll(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
}
bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const
{
- WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t);
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ wxULongLong_t val = wxStrtoull(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
}
bool wxString::ToDouble(double *pVal) const
{
- wxCHECK_MSG( pVal, false, _T("NULL output pointer") );
+ WX_STRING_TO_X_TYPE_START
+ double val = wxStrtod(start, &end);
+ WX_STRING_TO_X_TYPE_END
+}
- DO_IF_NOT_WINCE( errno = 0; )
+#if wxUSE_XLOCALE
- const wxChar *start = c_str();
- wxChar *end;
- double val = wxStrtod(start, &end);
+bool wxString::ToCLong(long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
- // return true only if scan was stopped by the terminating NUL and if the
- // string was not empty to start with and no under/overflow occurred
- if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) )
- return false;
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ long val = wxStrtol_lA(start, &end, base, wxCLocale);
+#else
+ long val = wxStrtol_l(start, &end, base, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
+}
- *pVal = val;
+bool wxString::ToCULong(unsigned long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
- return true;
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ unsigned long val = wxStrtoul_lA(start, &end, base, wxCLocale);
+#else
+ unsigned long val = wxStrtoul_l(start, &end, base, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
}
+bool wxString::ToCDouble(double *pVal) const
+{
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ double val = wxStrtod_lA(start, &end, wxCLocale);
+#else
+ double val = wxStrtod_l(start, &end, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
+}
+
+#endif // wxUSE_XLOCALE
+
// ---------------------------------------------------------------------------
// formatted output
// ---------------------------------------------------------------------------
// FIXME-UTF8: implement using iterators, remove #if
#if wxUSE_UNICODE_UTF8
- wxWCharBuffer maskBuf = mask.wc_str();
- wxWCharBuffer txtBuf = wc_str();
+ const wxScopedWCharBuffer maskBuf = mask.wc_str();
+ const wxScopedWCharBuffer txtBuf = wc_str();
const wxChar *pszMask = maskBuf.data();
const wxChar *pszTxt = txtBuf.data();
#else
projects / wxWidgets.git / blobdiff