X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/e6310bbc5a3bce1033c0e579341e115be9df6fe9..1a8e8d694bdcaec228f59a55f2aca1dfea0679db:/src/common/string.cpp?ds=inline diff --git a/src/common/string.cpp b/src/common/string.cpp index 7b9f77b28b..ff87c095c7 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -233,9 +233,35 @@ const char* wxCStrData::AsChar() const 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()); - // FIXME-UTF8: do the conversion in-place in the existing buffer + // 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) ) { @@ -261,6 +287,10 @@ const wchar_t* wxCStrData::AsWChar() const // 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) ) @@ -340,7 +370,7 @@ wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, return SubstrBufFromMB("", 0); // and then to UTF-8: - SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvUTF8())); + SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8())); // widechar -> UTF-8 conversion isn't supposed to ever fail: wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") ); @@ -382,9 +412,12 @@ const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const const wxWCharBuffer wxString::wc_str() const { - return wxMBConvUTF8().cMB2WC(m_impl.c_str(), - m_impl.length() + 1 /* size, not length */, - NULL); + return wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size, not length + NULL + ); } const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const @@ -395,14 +428,16 @@ const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const // FIXME-UTF8: use wc_str() here once we have buffers with length size_t wcLen; - wxWCharBuffer wcBuf( - wxMBConvUTF8().cMB2WC(m_impl.c_str(), - m_impl.length() + 1 /* size, not length */, - &wcLen)); + wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size + &wcLen + )); if ( !wcLen ) return wxCharBuffer(""); - return conv.cWC2MB(wcBuf, wcLen, NULL); + return conv.cWC2MB(wcBuf, wcLen+1, NULL); } #else // ANSI @@ -980,19 +1015,18 @@ wxString wxString::FromAscii(const char *ascii, size_t len) wxString res; - wxImplStringBuffer buf(res, len); - wxStringCharType *dest = buf; - - for ( ;; ) { - unsigned char c = (unsigned char)*ascii++; - wxASSERT_MSG( c < 0x80, - _T("Non-ASCII value passed to FromAscii().") ); + wxStringInternalBuffer buf(res, len); + wxStringCharType *dest = buf; - *dest++ = (wchar_t)c; + for ( ; len > 0; --len ) + { + unsigned char c = (unsigned char)*ascii++; + wxASSERT_MSG( c < 0x80, + _T("Non-ASCII value passed to FromAscii().") ); - if ( c == '\0' ) - break; + *dest++ = (wchar_t)c; + } } return res; @@ -1000,10 +1034,10 @@ wxString wxString::FromAscii(const char *ascii, size_t len) wxString wxString::FromAscii(const char *ascii) { - return FromAscii(ascii, strlen(ascii)); + return FromAscii(ascii, wxStrlen(ascii)); } -wxString wxString::FromAscii(const char ascii) +wxString wxString::FromAscii(char ascii) { // What do we do with '\0' ? @@ -1387,49 +1421,46 @@ int wxString::Find(wxUniChar ch, bool bFromEnd) const // it out. Note that number extraction works correctly on UTF-8 strings, so // we can use wxStringCharType and wx_str() for maximum efficiency. -template -bool wxStringToIntType(const wxStringCharType *start, - T *val, - int base, - T (*func)(const wxStringCharType*, wxStringCharType**, int)) -{ - wxCHECK_MSG( val, false, _T("NULL output pointer") ); - wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); - #ifndef __WXWINCE__ - errno = 0; + #define DO_IF_NOT_WINCE(x) x +#else + #define DO_IF_NOT_WINCE(x) #endif - wxStringCharType *end; - *val = (*func)(start, &end, 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 - return !*end && (end != start) -#ifndef __WXWINCE__ - && (errno != ERANGE) -#endif - ; -} +#define WX_STRING_TO_INT_TYPE(val, base, func) \ + wxCHECK_MSG( val, false, _T("NULL output pointer") ); \ + wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \ + \ + DO_IF_NOT_WINCE( errno = 0; ) \ + \ + const wxStringCharType *start = wx_str(); \ + wxStringCharType *end; \ + *val = func(start, &end, 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: */ \ + return !*end && (end != start) \ + DO_IF_NOT_WINCE( && (errno != ERANGE) ) bool wxString::ToLong(long *val, int base) const { - return wxStringToIntType(wx_str(), val, base, wxStrtol); + WX_STRING_TO_INT_TYPE(val, base, wxStrtol); } bool wxString::ToULong(unsigned long *val, int base) const { - return wxStringToIntType(wx_str(), val, base, wxStrtoul); + WX_STRING_TO_INT_TYPE(val, base, wxStrtoul); } bool wxString::ToLongLong(wxLongLong_t *val, int base) const { - return wxStringToIntType(wx_str(), val, base, wxStrtoll); + WX_STRING_TO_INT_TYPE(val, base, wxStrtoll); } bool wxString::ToULongLong(wxULongLong_t *val, int base) const { - return wxStringToIntType(wx_str(), val, base, wxStrtoull); + WX_STRING_TO_INT_TYPE(val, base, wxStrtoull); } bool wxString::ToDouble(double *val) const @@ -1640,7 +1671,7 @@ int wxString::PrintfV(const wxString& format, va_list argptr) #if wxUSE_STL_BASED_WXSTRING typedef wxStringTypeBuffer Utf8Buffer; #else - typedef wxImplStringBuffer Utf8Buffer; + typedef wxStringInternalBuffer Utf8Buffer; #endif #endif @@ -1828,3 +1859,32 @@ wxString wxString::Upper() const // convert to lower case, return the copy of the string wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); } + +// ---------------------------------------------------------------------------- +// wxUTF8StringBuffer +// ---------------------------------------------------------------------------- + +#if wxUSE_UNICODE_WCHAR +wxUTF8StringBuffer::~wxUTF8StringBuffer() +{ + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); + + wxStringInternalBuffer wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf); +} + +wxUTF8StringBufferLength::~wxUTF8StringBufferLength() +{ + wxCHECK_RET(m_lenSet, "length not set"); + + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf, m_len); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); + + wxStringInternalBufferLength wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf, m_len); + wbuf.SetLength(wlen); +} +#endif // wxUSE_UNICODE_WCHAR