X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/76046d7390d8c6f02dbbd851d1d896354bfc07cc..29fbbf8967e4c4b3174a4bc45a9007b92ecb0de7:/src/common/string.cpp?ds=sidebyside diff --git a/src/common/string.cpp b/src/common/string.cpp index f26486112a..6b65824622 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -10,13 +10,6 @@ // 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 // =========================================================================== @@ -30,8 +23,7 @@ #ifndef WX_PRECOMP #include "wx/string.h" - #include "wx/intl.h" - #include "wx/thread.h" + #include "wx/wxcrtvararg.h" #endif #include @@ -43,47 +35,28 @@ #include #include -#ifdef __SALFORDC__ - #include +#include "wx/hashmap.h" + +// string handling functions used by wxString: +#if wxUSE_UNICODE_UTF8 + #define wxStringMemcpy memcpy + #define wxStringMemcmp memcmp + #define wxStringMemchr memchr + #define wxStringStrlen strlen +#else + #define wxStringMemcpy wxTmemcpy + #define wxStringMemcmp wxTmemcmp + #define wxStringMemchr wxTmemchr + #define wxStringStrlen wxStrlen #endif -// allocating extra space for each string consumes more memory but speeds up -// the concatenation operations (nLen is the current string's length) -// NB: EXTRA_ALLOC must be >= 0! -#define EXTRA_ALLOC (19 - nLen % 16) // --------------------------------------------------------------------------- // static class variables definition // --------------------------------------------------------------------------- -#if !wxUSE_STL_BASED_WXSTRING - //According to STL _must_ be a -1 size_t - const size_t wxStringBase::npos = (size_t) -1; -#endif - -// ---------------------------------------------------------------------------- -// static data -// ---------------------------------------------------------------------------- - -#if wxUSE_STL_BASED_WXSTRING - -extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T(""); - -#else - -// for an empty string, GetStringData() will return this address: this -// structure has the same layout as wxStringData and it's data() method will -// return the empty string (dummy pointer) -static const struct -{ - wxStringData data; - wxChar dummy; -} g_strEmpty = { {-1, 0, 0}, wxT('\0') }; - -// empty C style string: points to 'string data' byte of g_strEmpty -extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy; - -#endif +//According to STL _must_ be a -1 size_t +const size_t wxString::npos = (size_t) -1; // ---------------------------------------------------------------------------- // global functions @@ -93,868 +66,667 @@ extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy; #include -wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) -{ - return os << str.c_str(); -} - wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str) { -#if wxUSE_UNICODE && !defined(__BORLANDC__) - return os << str.AsWChar(); +#if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8 + return os << (const char *)str.AsCharBuf(); #else - return os << str.AsChar(); + return os << str.AsInternal(); #endif } -#endif // wxUSE_STD_IOSTREAM - -// ---------------------------------------------------------------------------- -// private classes -// ---------------------------------------------------------------------------- - -// this small class is used to gather statistics for performance tuning -//#define WXSTRING_STATISTICS -#ifdef WXSTRING_STATISTICS - class Averager - { - public: - Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; } - ~Averager() - { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); } - - void Add(size_t n) { m_nTotal += n; m_nCount++; } - - private: - size_t m_nCount, m_nTotal; - const wxChar *m_sz; - } g_averageLength("allocation size"), - g_averageSummandLength("summand length"), - g_averageConcatHit("hit probability in concat"), - g_averageInitialLength("initial string length"); - - #define STATISTICS_ADD(av, val) g_average##av.Add(val) -#else - #define STATISTICS_ADD(av, val) -#endif // WXSTRING_STATISTICS - -#if !wxUSE_STL_BASED_WXSTRING +wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) +{ + return os << str.c_str(); +} -// =========================================================================== -// wxStringData class deallocation -// =========================================================================== +wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str) +{ + return os << str.data(); +} -#if defined(__VISUALC__) && defined(_MT) && !defined(_DLL) -# pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!") -void wxStringData::Free() +#ifndef __BORLANDC__ +wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str) { - free(this); + return os << str.data(); } #endif -// =========================================================================== -// wxStringBase -// =========================================================================== +#if wxUSE_UNICODE && defined(HAVE_WOSTREAM) -// takes nLength elements of psz starting at nPos -void wxStringBase::InitWith(const wxChar *psz, size_t nPos, size_t nLength) +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str) { - Init(); - - // if the length is not given, assume the string to be NUL terminated - if ( nLength == npos ) { - wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") ); - - nLength = wxStrlen(psz + nPos); - } - - STATISTICS_ADD(InitialLength, nLength); - - if ( nLength > 0 ) { - // trailing '\0' is written in AllocBuffer() - if ( !AllocBuffer(nLength) ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::InitWith") ); - return; - } - wxTmemcpy(m_pchData, psz + nPos, nLength); - } + return wos << str.wc_str(); } -// poor man's iterators are "void *" pointers -wxStringBase::wxStringBase(const void *pStart, const void *pEnd) +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str) { - if ( pEnd >= pStart ) - { - InitWith((const wxChar *)pStart, 0, - (const wxChar *)pEnd - (const wxChar *)pStart); - } - else - { - wxFAIL_MSG( _T("pStart is not before pEnd") ); - Init(); - } + return wos << str.AsWChar(); } -wxStringBase::wxStringBase(size_type n, wxUniChar ch) +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str) { - Init(); - append(n, ch); + return wos << str.data(); } -// --------------------------------------------------------------------------- -// memory allocation -// --------------------------------------------------------------------------- - -// allocates memory needed to store a C string of length nLen -bool wxStringBase::AllocBuffer(size_t nLen) -{ - // allocating 0 sized buffer doesn't make sense, all empty strings should - // reuse g_strEmpty - wxASSERT( nLen > 0 ); - - // make sure that we don't overflow - wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) - - (sizeof(wxStringData) + EXTRA_ALLOC + 1) ); +#endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM) - STATISTICS_ADD(Length, nLen); +#endif // wxUSE_STD_IOSTREAM - // allocate memory: - // 1) one extra character for '\0' termination - // 2) sizeof(wxStringData) for housekeeping info - wxStringData* pData = (wxStringData*) - malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar)); +// =========================================================================== +// wxString class core +// =========================================================================== - if ( pData == NULL ) { - // allocation failures are handled by the caller - return false; - } +#if wxUSE_UNICODE_UTF8 - pData->nRefs = 1; - pData->nDataLength = nLen; - pData->nAllocLength = nLen + EXTRA_ALLOC; - m_pchData = pData->data(); // data starts after wxStringData - m_pchData[nLen] = wxT('\0'); - return true; -} - -// must be called before changing this string -bool wxStringBase::CopyBeforeWrite() +void wxString::PosLenToImpl(size_t pos, size_t len, + size_t *implPos, size_t *implLen) const { - wxStringData* pData = GetStringData(); + 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; - if ( pData->IsShared() ) { - pData->Unlock(); // memory not freed because shared - size_t nLen = pData->nDataLength; - if ( !AllocBuffer(nLen) ) { - // allocation failures are handled by the caller - return false; + *implLen = (i + len).impl() - i.impl(); + } } - wxTmemcpy(m_pchData, pData->data(), nLen); - } - - wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner - - return true; } -// must be called before replacing contents of this string -bool wxStringBase::AllocBeforeWrite(size_t nLen) -{ - wxASSERT( nLen != 0 ); // doesn't make any sense +#endif // wxUSE_UNICODE_UTF8 - // must not share string and must have enough space - wxStringData* pData = GetStringData(); - if ( pData->IsShared() || pData->IsEmpty() ) { - // can't work with old buffer, get new one - pData->Unlock(); - if ( !AllocBuffer(nLen) ) { - // allocation failures are handled by the caller - return false; +// ---------------------------------------------------------------------------- +// 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 +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); } - } - else { - if ( nLen > pData->nAllocLength ) { - // realloc the buffer instead of calling malloc() again, this is more - // efficient - STATISTICS_ADD(Length, nLen); +} - nLen += EXTRA_ALLOC; +#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; - pData = (wxStringData*) - realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); +const char* wxCStrData::AsChar() const +{ + // remove previously cache value, if any (see FIXMEs above): + DeleteStringFromConversionCache(gs_stringsCharCache, m_str); - if ( pData == NULL ) { - // allocation failures are handled by the caller - // keep previous data since reallocation failed - return false; - } + // convert the string and keep it: + const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] = + m_str->mb_str().release(); - pData->nAllocLength = nLen; - m_pchData = pData->data(); - } - } + return s + m_offset; +} +#endif // wxUSE_UNICODE - wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner +#if !wxUSE_UNICODE_WCHAR +WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual, + wxStringWCharConversionCache); +static wxStringWCharConversionCache gs_stringsWCharCache; - // it doesn't really matter what the string length is as it's going to be - // overwritten later but, for extra safety, set it to 0 for now as we may - // have some junk in m_pchData - GetStringData()->nDataLength = 0; +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 true; + return s + m_offset; } +#endif // !wxUSE_UNICODE_WCHAR -wxStringBase& wxStringBase::append(size_t n, wxUniChar ch) +wxString::~wxString() { - size_type len = length(); - - if ( !Alloc(len + n) || !CopyBeforeWrite() ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::append") ); - } - GetStringData()->nDataLength = len + n; - m_pchData[len + n] = '\0'; - for ( size_t i = 0; i < n; ++i ) - m_pchData[len + i] = ch; - return *this; +#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 -void wxStringBase::resize(size_t nSize, wxUniChar ch) +#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY +const char* wxCStrData::AsChar() const { - size_t len = length(); - - if ( nSize < len ) +#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) ) { - erase(begin() + nSize, end()); + // 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 if ( nSize > len ) + else { - append(nSize - len, ch); + str->m_convertedToChar = buf.release(); } - //else: we have exactly the specified length, nothing to do + + // and keep it: + return str->m_convertedToChar + m_offset; } +#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY -// allocate enough memory for nLen characters -bool wxStringBase::Alloc(size_t nLen) +#if !wxUSE_UNICODE_WCHAR +const wchar_t* wxCStrData::AsWChar() const { - wxStringData *pData = GetStringData(); - if ( pData->nAllocLength <= nLen ) { - if ( pData->IsEmpty() ) { - nLen += EXTRA_ALLOC; + wxString *str = wxConstCast(m_str, wxString); - pData = (wxStringData *) - malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); + // convert the string: + wxWCharBuffer buf(str->wc_str()); - if ( pData == NULL ) { - // allocation failure handled by caller - return false; - } + // 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 - pData->nRefs = 1; - pData->nDataLength = 0; - pData->nAllocLength = nLen; - m_pchData = pData->data(); // data starts after wxStringData - m_pchData[0u] = wxT('\0'); - } - else if ( pData->IsShared() ) { - pData->Unlock(); // memory not freed because shared - size_t nOldLen = pData->nDataLength; - if ( !AllocBuffer(nLen) ) { - // allocation failure handled by caller - return false; - } - // +1 to copy the terminator, too - memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxChar)); - GetStringData()->nDataLength = nOldLen; + // 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 { - nLen += EXTRA_ALLOC; - - pData = (wxStringData *) - realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); - - if ( pData == NULL ) { - // allocation failure handled by caller - // keep previous data since reallocation failed - return false; - } - - // it's not important if the pointer changed or not (the check for this - // is not faster than assigning to m_pchData in all cases) - pData->nAllocLength = nLen; - m_pchData = pData->data(); + else + { + str->m_convertedToWChar = buf.release(); } - } - //else: we've already got enough - return true; -} -wxStringBase::iterator wxStringBase::begin() -{ - if (length() > 0) - CopyBeforeWrite(); - return m_pchData; + // and keep it: + return str->m_convertedToWChar + m_offset; } +#endif // !wxUSE_UNICODE_WCHAR -wxStringBase::iterator wxStringBase::end() -{ - if (length() > 0) - CopyBeforeWrite(); - return m_pchData + length(); -} - -wxStringBase::iterator wxStringBase::erase(iterator it) -{ - size_type idx = it - begin(); - erase(idx, 1); - return begin() + idx; -} - -wxStringBase& wxStringBase::erase(size_t nStart, size_t nLen) -{ - wxASSERT(nStart <= length()); - size_t strLen = length() - nStart; - // delete nLen or up to the end of the string characters - nLen = strLen < nLen ? strLen : nLen; - wxString strTmp(c_str(), nStart); - strTmp.append(c_str() + nStart + nLen, length() - nStart - nLen); +// =========================================================================== +// wxString class core +// =========================================================================== - swap(strTmp); - return *this; -} +// --------------------------------------------------------------------------- +// construction and conversion +// --------------------------------------------------------------------------- -wxStringBase& wxStringBase::insert(size_t nPos, const wxChar *sz, size_t n) +#if wxUSE_UNICODE_WCHAR +/* static */ +wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv) { - wxASSERT( nPos <= length() ); - - if ( n == npos ) n = wxStrlen(sz); - if ( n == 0 ) return *this; - - if ( !Alloc(length() + n) || !CopyBeforeWrite() ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::insert") ); - } - - memmove(m_pchData + nPos + n, m_pchData + nPos, - (length() - nPos) * sizeof(wxChar)); - memcpy(m_pchData + nPos, sz, n * sizeof(wxChar)); - GetStringData()->nDataLength = length() + n; - m_pchData[length()] = '\0'; + // anything to do? + if ( !psz || nLength == 0 ) + return SubstrBufFromMB(L"", 0); - return *this; -} + if ( nLength == npos ) + nLength = wxNO_LEN; -void wxStringBase::swap(wxStringBase& str) -{ - wxChar* tmp = str.m_pchData; - str.m_pchData = m_pchData; - m_pchData = tmp; + size_t wcLen; + wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen)); + if ( !wcLen ) + return SubstrBufFromMB(_T(""), 0); + else + return SubstrBufFromMB(wcBuf, wcLen); } +#endif // wxUSE_UNICODE_WCHAR -size_t wxStringBase::find(const wxStringBase& str, size_t nStart) const +#if wxUSE_UNICODE_UTF8 +/* static */ +wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv) { - // deal with the special case of empty string first - const size_t nLen = length(); - const size_t nLenOther = str.length(); + // anything to do? + if ( !psz || nLength == 0 ) + return SubstrBufFromMB("", 0); - if ( !nLenOther ) + // if psz is already in UTF-8, we don't have to do the roundtrip to + // wchar_t* and back: + if ( conv.IsUTF8() ) { - // empty string is a substring of anything - return 0; + // 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 ( !nLen ) - { - // the other string is non empty so can't be our substring - return npos; - } + if ( nLength == npos ) + nLength = wxNO_LEN; - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart <= nLen ); + // first convert to wide string: + size_t wcLen; + wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen)); + if ( !wcLen ) + return SubstrBufFromMB("", 0); - const wxChar * const other = str.c_str(); + // and then to UTF-8: + 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") ); - // anchor - const wxChar* p = (const wxChar*)wxTmemchr(c_str() + nStart, - *other, - nLen - nStart); + return buf; +} +#endif // wxUSE_UNICODE_UTF8 - if ( !p ) - return npos; +#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("", 0); - while ( p - c_str() + nLenOther <= nLen && wxTmemcmp(p, other, nLenOther) ) - { - p++; + if ( nLength == npos ) + nLength = wxNO_LEN; - // anchor again - p = (const wxChar*)wxTmemchr(p, *other, nLen - (p - c_str())); + size_t mbLen; + wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen)); + if ( !mbLen ) + return SubstrBufFromWC("", 0); + else + return SubstrBufFromWC(mbBuf, mbLen); +} +#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE - if ( !p ) - return npos; - } - return p - c_str() + nLenOther <= nLen ? p - c_str() : npos; -} +#if wxUSE_UNICODE_WCHAR -size_t wxStringBase::find(const wxChar* sz, size_t nStart, size_t n) const +//Convert wxString in Unicode mode to a multi-byte string +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - return find(wxStringBase(sz, n), nStart); + return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL); } -size_t wxStringBase::find(wxUniChar ch, size_t nStart) const -{ - wxASSERT( nStart <= length() ); - - const wxChar *p = (const wxChar*)wxTmemchr(c_str() + nStart, ch, length() - nStart); +#elif wxUSE_UNICODE_UTF8 - return p == NULL ? npos : p - c_str(); +const wxWCharBuffer wxString::wc_str() const +{ + return wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size, not length + NULL + ); } -size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart == npos || nStart <= length() ); + if ( conv.IsUTF8() ) + return wxCharBuffer::CreateNonOwned(m_impl.c_str()); - if ( length() >= str.length() ) - { - // avoids a corner case later - if ( length() == 0 && str.length() == 0 ) - return 0; + // FIXME-UTF8: use wc_str() here once we have buffers with length - // "top" is the point where search starts from - size_t top = length() - str.length(); + size_t wcLen; + wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size + &wcLen + )); + if ( !wcLen ) + return wxCharBuffer(""); - if ( nStart == npos ) - nStart = length() - 1; - if ( nStart < top ) - top = nStart; + return conv.cWC2MB(wcBuf, wcLen+1, NULL); +} - const wxChar *cursor = c_str() + top; - do - { - if ( wxTmemcmp(cursor, str.c_str(), - str.length()) == 0 ) - { - return cursor - c_str(); - } - } while ( cursor-- > c_str() ); - } +#else // ANSI - return npos; +//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); } -size_t wxStringBase::rfind(const wxChar* sz, size_t nStart, size_t n) const +#endif // Unicode/ANSI + +// shrink to minimal size (releasing extra memory) +bool wxString::Shrink() { - return rfind(wxStringBase(sz, n), nStart); + wxString tmp(begin(), end()); + swap(tmp); + return tmp.length() == length(); } -size_t wxStringBase::rfind(wxUniChar ch, size_t nStart) const +// deprecated compatibility code: +#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +wxStringCharType *wxString::GetWriteBuf(size_t nLen) { - if ( nStart == npos ) - { - nStart = length(); - } - else - { - wxASSERT( nStart <= length() ); - } - - const wxChar *actual; - for ( actual = c_str() + ( nStart == npos ? length() : nStart + 1 ); - actual > c_str(); --actual ) - { - if ( *(actual - 1) == ch ) - return (actual - 1) - c_str(); - } - - return npos; + return DoGetWriteBuf(nLen); } -size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart) const +void wxString::UngetWriteBuf() { - wxASSERT(nStart <= length()); - - size_t len = wxStrlen(sz); - - size_t i; - for(i = nStart; i < this->length(); ++i) - { - if (wxTmemchr(sz, *(c_str() + i), len)) - break; - } - - if(i == this->length()) - return npos; - else - return i; + DoUngetWriteBuf(); } -size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart, - size_t n) const +void wxString::UngetWriteBuf(size_t nLen) { - return find_first_of(wxStringBase(sz, n), nStart); + DoUngetWriteBuf(nLen); } +#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 -size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart) const -{ - if ( nStart == npos ) - { - nStart = length() - 1; - } - else - { - wxASSERT_MSG( nStart <= length(), - _T("invalid index in find_last_of()") ); - } - size_t len = wxStrlen(sz); +// --------------------------------------------------------------------------- +// data access +// --------------------------------------------------------------------------- - for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p ) - { - if ( wxTmemchr(sz, *p, len) ) - return p - c_str(); - } +// all functions are inline in string.h - return npos; -} +// --------------------------------------------------------------------------- +// concatenation operators +// --------------------------------------------------------------------------- -size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart, - size_t n) const -{ - return find_last_of(wxStringBase(sz, n), nStart); -} +/* + * concatenation functions come in 5 flavours: + * string + string + * char + string and string + char + * C str + string and string + C str + */ -size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart) const +wxString operator+(const wxString& str1, const wxString& str2) { - if ( nStart == npos ) - { - nStart = length(); - } - else - { - wxASSERT( nStart <= length() ); - } - - size_t len = wxStrlen(sz); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str1.IsValid() ); + wxASSERT( str2.IsValid() ); +#endif - size_t i; - for(i = nStart; i < this->length(); ++i) - { - if (!wxTmemchr(sz, *(c_str() + i), len)) - break; - } + wxString s = str1; + s += str2; - if(i == this->length()) - return npos; - else - return i; + return s; } -size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart, - size_t n) const +wxString operator+(const wxString& str, wxUniChar ch) { - return find_first_not_of(wxStringBase(sz, n), nStart); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s = str; + s += ch; + + return s; } -size_t wxStringBase::find_first_not_of(wxUniChar ch, size_t nStart) const +wxString operator+(wxUniChar ch, const wxString& str) { - wxASSERT( nStart <= length() ); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - for ( const_iterator p = begin() + nStart; (bool)*p; ++p ) // FIXME-DMARS - { - if ( *p != ch ) - return p - begin(); - } + wxString s = ch; + s += str; - return npos; + return s; } -size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart) const +wxString operator+(const wxString& str, const char *psz) { - if ( nStart == npos ) - { - nStart = length() - 1; - } - else - { - wxASSERT( nStart <= length() ); - } - - size_t len = wxStrlen(sz); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p ) - { - if ( !wxTmemchr(sz, *p,len) ) - return p - c_str(); + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); } + s += str; + s += psz; - return npos; -} - -size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart, - size_t n) const -{ - return find_last_not_of(wxStringBase(sz, n), nStart); + return s; } -size_t wxStringBase::find_last_not_of(wxUniChar ch, size_t nStart) const +wxString operator+(const wxString& str, const wchar_t *pwz) { - if ( nStart == npos ) - { - nStart = length() - 1; - } - else - { - wxASSERT( nStart <= length() ); - } +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - for ( const_iterator p = begin() + nStart; p != begin(); --p ) - { - if ( *p != ch ) - return p - begin(); + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); } + s += str; + s += pwz; - return npos; + return s; } -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxChar *sz) +wxString operator+(const char *psz, const wxString& str) { - wxASSERT_MSG( nStart <= length(), - _T("index out of bounds in wxStringBase::replace") ); - size_t strLen = length() - nStart; - nLen = strLen < nLen ? strLen : nLen; +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - wxStringBase strTmp; - strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = psz; + s += str; - //This is kind of inefficient, but its pretty good considering... - //we don't want to use character access operators here because on STL - //it will freeze the reference count of strTmp, which means a deep copy - //at the end when swap is called - // - //Also, we can't use append with the full character pointer and must - //do it manually because this string can contain null characters - for(size_t i1 = 0; i1 < nStart; ++i1) - strTmp.append(1, this->c_str()[i1]); + return s; +} - //its safe to do the full version here because - //sz must be a normal c string - strTmp.append(sz); +wxString operator+(const wchar_t *pwz, const wxString& str) +{ +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - for(size_t i2 = nStart + nLen; i2 < length(); ++i2) - strTmp.append(1, this->c_str()[i2]); + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = pwz; + s += str; - swap(strTmp); - return *this; + return s; } -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - size_t nCount, wxUniChar ch) -{ - return replace(nStart, nLen, wxStringBase(nCount, ch).c_str()); -} +// --------------------------------------------------------------------------- +// string comparison +// --------------------------------------------------------------------------- -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxStringBase& str, - size_t nStart2, size_t nLen2) +bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const { - return replace(nStart, nLen, str.substr(nStart2, nLen2)); + return (length() == 1) && (compareWithCase ? GetChar(0u) == c + : wxToupper(GetChar(0u)) == wxToupper(c)); } -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxChar* sz, size_t nCount) +#ifdef HAVE_STD_STRING_COMPARE + +// NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with +// UTF-8 encoded strings too, thanks to UTF-8's design which allows us to +// sort strings in characters code point order by sorting the byte sequence +// in byte values order (i.e. what strcmp() and memcmp() do). + +int wxString::compare(const wxString& str) const { - return replace(nStart, nLen, wxStringBase(sz, nCount).c_str()); + return m_impl.compare(str.m_impl); } -wxStringBase wxStringBase::substr(size_t nStart, size_t nLen) const +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str) const { - if ( nLen == npos ) - nLen = length() - nStart; - return wxStringBase(*this, nStart, nLen); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + return m_impl.compare(pos, len, str.m_impl); } -// assigns one string to another -wxStringBase& wxStringBase::operator=(const wxStringBase& stringSrc) +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { - wxASSERT( stringSrc.GetStringData()->IsValid() ); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - // don't copy string over itself - if ( m_pchData != stringSrc.m_pchData ) { - if ( stringSrc.GetStringData()->IsEmpty() ) { - Reinit(); - } - else { - // adjust references - GetStringData()->Unlock(); - m_pchData = stringSrc.m_pchData; - GetStringData()->Lock(); - } - } + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); - return *this; + return m_impl.compare(pos, len, str.m_impl, pos2, len2); } -// assigns a single character -wxStringBase& wxStringBase::operator=(wxUniChar ch) +int wxString::compare(const char* sz) const { - wxChar c(ch); - if ( !AssignCopy(1, &c) ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(wxChar)") ); - } - return *this; + return m_impl.compare(ImplStr(sz)); } -// assigns C string -wxStringBase& wxStringBase::operator=(const wxChar *psz) +int wxString::compare(const wchar_t* sz) const { - if ( !AssignCopy(wxStrlen(psz), psz) ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(const wxChar *)") ); - } - return *this; + return m_impl.compare(ImplStr(sz)); } -// helper function: does real copy -bool wxStringBase::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData) +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const { - if ( nSrcLen == 0 ) { - Reinit(); - } - else { - if ( !AllocBeforeWrite(nSrcLen) ) { - // allocation failure handled by caller - return false; - } - memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar)); - GetStringData()->nDataLength = nSrcLen; - m_pchData[nSrcLen] = wxT('\0'); - } - return true; -} - -// --------------------------------------------------------------------------- -// string concatenation -// --------------------------------------------------------------------------- - -// add something to this string -bool wxStringBase::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, - size_t nMaxLen) -{ - STATISTICS_ADD(SummandLength, nSrcLen); - - nSrcLen = nSrcLen < nMaxLen ? nSrcLen : nMaxLen; - - // concatenating an empty string is a NOP - if ( nSrcLen > 0 ) { - wxStringData *pData = GetStringData(); - size_t nLen = pData->nDataLength; - size_t nNewLen = nLen + nSrcLen; + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - // alloc new buffer if current is too small - if ( pData->IsShared() ) { - STATISTICS_ADD(ConcatHit, 0); + SubstrBufFromMB str(ImplStr(sz, nCount)); - // we have to allocate another buffer - wxStringData* pOldData = GetStringData(); - if ( !AllocBuffer(nNewLen) ) { - // allocation failure handled by caller - return false; - } - memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar)); - pOldData->Unlock(); - } - else if ( nNewLen > pData->nAllocLength ) { - STATISTICS_ADD(ConcatHit, 0); - - reserve(nNewLen); - // we have to grow the buffer - if ( capacity() < nNewLen ) { - // allocation failure handled by caller - return false; - } - } - else { - STATISTICS_ADD(ConcatHit, 1); - - // the buffer is already big enough - } - - // should be enough space - wxASSERT( nNewLen <= GetStringData()->nAllocLength ); - - // fast concatenation - all is done in our buffer - memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar)); - - m_pchData[nNewLen] = wxT('\0'); // put terminating '\0' - GetStringData()->nDataLength = nNewLen; // and fix the length - } - //else: the string to append was empty - return true; + return m_impl.compare(pos, len, str.data, str.len); } -// --------------------------------------------------------------------------- -// simple sub-string extraction -// --------------------------------------------------------------------------- - -// helper function: clone the data attached to this string -bool wxStringBase::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const { - if ( nCopyLen == 0 ) { - dest.Init(); - } - else { - if ( !dest.AllocBuffer(nCopyLen) ) { - // allocation failure handled by caller - return false; - } - memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar)); - } - return true; -} - -#endif // !wxUSE_STL_BASED_WXSTRING + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); -#if !wxUSE_STL_BASED_WXSTRING || !defined(HAVE_STD_STRING_COMPARE) + SubstrBufFromWC str(ImplStr(sz, nCount)); -#if !wxUSE_STL_BASED_WXSTRING - #define STRINGCLASS wxStringBase -#else - #define STRINGCLASS wxString -#endif + return m_impl.compare(pos, len, str.data, str.len); +} + +#else // !HAVE_STD_STRING_COMPARE -static inline int wxDoCmp(const wxChar* s1, size_t l1, - const wxChar* s2, size_t l2) +static inline int wxDoCmp(const wxStringCharType* s1, size_t l1, + const wxStringCharType* s2, size_t l2) { if( l1 == l2 ) - return wxTmemcmp(s1, s2, l1); + return wxStringMemcmp(s1, s2, l1); else if( l1 < l2 ) { - int ret = wxTmemcmp(s1, s2, l1); + int ret = wxStringMemcmp(s1, s2, l1); return ret == 0 ? -1 : ret; } else { - int ret = wxTmemcmp(s1, s2, l2); + int ret = wxStringMemcmp(s1, s2, l2); return ret == 0 ? +1 : ret; } } -int STRINGCLASS::compare(const wxStringBase& str) const +int wxString::compare(const wxString& str) const { - return ::wxDoCmp(data(), length(), str.data(), str.length()); + return ::wxDoCmp(m_impl.data(), m_impl.length(), + str.m_impl.data(), str.m_impl.length()); } -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxStringBase& str) const +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str) const { wxASSERT(nStart <= length()); size_type strLen = length() - nStart; nLen = strLen < nLen ? strLen : nLen; - return ::wxDoCmp(data() + nStart, nLen, str.data(), str.length()); + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data(), str.m_impl.length()); } -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxStringBase& str, - size_t nStart2, size_t nLen2) const +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { wxASSERT(nStart <= length()); wxASSERT(nStart2 <= str.length()); @@ -962,326 +734,282 @@ int STRINGCLASS::compare(size_t nStart, size_t nLen, strLen2 = str.length() - nStart2; nLen = strLen < nLen ? strLen : nLen; nLen2 = strLen2 < nLen2 ? strLen2 : nLen2; - return ::wxDoCmp(data() + nStart, nLen, str.data() + nStart2, nLen2); + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); + + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data() + pos2, len2); +} + +int wxString::compare(const char* sz) const +{ + SubstrBufFromMB str(ImplStr(sz, npos)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); + return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len); } -int STRINGCLASS::compare(const wxChar* sz) const +int wxString::compare(const wchar_t* sz) const { - size_t nLen = wxStrlen(sz); - return ::wxDoCmp(data(), length(), sz, nLen); + SubstrBufFromWC str(ImplStr(sz, npos)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); + return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len); } -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxChar* sz, size_t nCount) const +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const { wxASSERT(nStart <= length()); size_type strLen = length() - nStart; nLen = strLen < nLen ? strLen : nLen; - if( nCount == npos ) - nCount = wxStrlen(sz); - - return ::wxDoCmp(data() + nStart, nLen, sz, nCount); -} - -#undef STRINGCLASS -#endif // !wxUSE_STL_BASED_WXSTRING || !defined(HAVE_STD_STRING_COMPARE) - -// =========================================================================== -// wxString class core -// =========================================================================== + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); -// --------------------------------------------------------------------------- -// construction and conversion -// --------------------------------------------------------------------------- + SubstrBufFromMB str(ImplStr(sz, nCount)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); -#if wxUSE_UNICODE + return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len); +} -// from multibyte string -wxString::wxString(const char *psz, const wxMBConv& conv, size_t nLength) +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const { - // anything to do? - if ( psz && nLength != 0 ) - { - if ( nLength == npos ) - { - nLength = wxNO_LEN; - } + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; - size_t nLenWide; - wxWCharBuffer wbuf = conv.cMB2WC(psz, nLength, &nLenWide); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - if ( nLenWide ) - assign(wbuf, nLenWide); - } -} + SubstrBufFromWC str(ImplStr(sz, nCount)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); -//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 ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len); } -#else // ANSI +#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE -#if wxUSE_WCHAR_T -// from wide string -wxString::wxString(const wchar_t *pwz, const wxMBConv& conv, size_t nLength) -{ - // anything to do? - if ( pwz && nLength != 0 ) - { - if ( nLength == npos ) - { - nLength = wxNO_LEN; - } +// --------------------------------------------------------------------------- +// find_{first,last}_[not]_of functions +// --------------------------------------------------------------------------- - size_t nLenMB; - wxCharBuffer buf = conv.cWC2MB(pwz, nLength, &nLenMB); +#if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 - if ( nLenMB ) - assign(buf, nLenMB); - } -} +// NB: All these functions are implemented with the argument being wxChar*, +// i.e. widechar string in any Unicode build, even though native string +// representation is char* in the UTF-8 build. This is because we couldn't +// use memchr() to determine if a character is in a set encoded as UTF-8. -//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 +size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const { - return conv.cMB2WC(c_str(), length() + 1 /* size, not length */, NULL); + return find_first_of(sz, nStart, wxStrlen(sz)); } -#endif // wxUSE_WCHAR_T - -#endif // Unicode/ANSI - -// shrink to minimal size (releasing extra memory) -bool wxString::Shrink() +size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const { - wxString tmp(begin(), end()); - swap(tmp); - return tmp.length() == length(); + return find_first_not_of(sz, nStart, wxStrlen(sz)); } -#if !wxUSE_STL_BASED_WXSTRING -// get the pointer to writable buffer of (at least) nLen bytes -wxChar *wxString::DoGetWriteBuf(size_t nLen) +size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const { - if ( !AllocBeforeWrite(nLen) ) { - // allocation failure handled by caller - return NULL; - } + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); - wxASSERT( GetStringData()->nRefs == 1 ); - GetStringData()->Validate(false); - - return m_pchData; -} + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( wxTmemchr(sz, *i, n) ) + return idx; + } -// put string back in a reasonable state after GetWriteBuf -void wxString::DoUngetWriteBuf() -{ - DoUngetWriteBuf(wxStrlen(m_pchData)); + return npos; } -void wxString::DoUngetWriteBuf(size_t nLen) +size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const { - wxStringData * const pData = GetStringData(); + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); - wxASSERT_MSG( nLen < pData->nAllocLength, _T("buffer overrun") ); + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( !wxTmemchr(sz, *i, n) ) + return idx; + } - // the strings we store are always NUL-terminated - pData->data()[nLen] = _T('\0'); - pData->nDataLength = nLen; - pData->Validate(true); + return npos; } -// deprecated compatibility code: -#if WXWIN_COMPATIBILITY_2_8 -wxChar *wxString::GetWriteBuf(size_t nLen) -{ - return DoGetWriteBuf(nLen); -} -void wxString::UngetWriteBuf() +size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const { - DoUngetWriteBuf(); + return find_last_of(sz, nStart, wxStrlen(sz)); } -void wxString::UngetWriteBuf(size_t nLen) +size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const { - DoUngetWriteBuf(nLen); + return find_last_not_of(sz, nStart, wxStrlen(sz)); } -#endif // WXWIN_COMPATIBILITY_2_8 - -#endif // !wxUSE_STL_BASED_WXSTRING - - -// --------------------------------------------------------------------------- -// data access -// --------------------------------------------------------------------------- -// all functions are inline in string.h +size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const +{ + size_t len = length(); -// --------------------------------------------------------------------------- -// assignment operators -// --------------------------------------------------------------------------- + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } -#if !wxUSE_UNICODE + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) + { + if ( wxTmemchr(sz, *i, n) ) + return idx; + } -// same as 'signed char' variant -wxString& wxString::operator=(const unsigned char* psz) -{ - *this = (const char *)psz; - return *this; + return npos; } -#if wxUSE_WCHAR_T -wxString& wxString::operator=(const wchar_t *pwz) +size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const { - wxString str(pwz); - swap(str); - return *this; -} -#endif - -#endif - -/* - * concatenation functions come in 5 flavours: - * string + string - * char + string and string + char - * C str + string and string + C str - */ + size_t len = length(); -wxString operator+(const wxString& str1, const wxString& str2) -{ -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str1.GetStringData()->IsValid() ); - wxASSERT( str2.GetStringData()->IsValid() ); -#endif + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } - wxString s = str1; - s += str2; + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) + { + if ( !wxTmemchr(sz, *i, n) ) + return idx; + } - return s; + return npos; } -wxString operator+(const wxString& str, wxUniChar ch) +size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const { -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); - wxString s = str; - s += ch; + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( *i != ch ) + return idx; + } - return s; + return npos; } -wxString operator+(wxUniChar ch, const wxString& str) +size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const { -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif - - wxString s = ch; - s += str; - - return s; -} + size_t len = length(); -wxString operator+(const wxString& str, const wxChar *psz) -{ -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } - wxString s; - if ( !s.Alloc(wxStrlen(psz) + str.length()) ) { - wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) + { + if ( *i != ch ) + return idx; } - s += str; - s += psz; - return s; + return npos; } -wxString operator+(const wxChar *psz, const wxString& str) -{ -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); +// the functions above were implemented for wchar_t* arguments in Unicode +// build and char* in ANSI build; below are implementations for the other +// version: +#if wxUSE_UNICODE + #define wxOtherCharType char + #define STRCONV (const wxChar*)wxConvLibc.cMB2WC +#else + #define wxOtherCharType wchar_t + #define STRCONV (const wxChar*)wxConvLibc.cWC2MB #endif - wxString s; - if ( !s.Alloc(wxStrlen(psz) + str.length()) ) { - wxFAIL_MSG( _T("out of memory in wxString::operator+") ); - } - s = psz; - s += str; +size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const + { return find_first_of(STRCONV(sz), nStart); } + +size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart, + size_t n) const + { return find_first_of(STRCONV(sz, n, NULL), nStart, n); } +size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const + { return find_last_of(STRCONV(sz), nStart); } +size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart, + size_t n) const + { return find_last_of(STRCONV(sz, n, NULL), nStart, n); } +size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const + { return find_first_not_of(STRCONV(sz), nStart); } +size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart, + size_t n) const + { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); } +size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const + { return find_last_not_of(STRCONV(sz), nStart); } +size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart, + size_t n) const + { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); } - return s; -} +#undef wxOtherCharType +#undef STRCONV + +#endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 // =========================================================================== // other common string functions // =========================================================================== -int wxString::Cmp(const wxString& s) const -{ - return compare(s); -} - -int wxString::Cmp(const wxChar* psz) const +int wxString::CmpNoCase(const wxString& s) const { - return compare(psz); -} + // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added -static inline int wxDoCmpNoCase(const wxChar* s1, size_t l1, - const wxChar* s2, size_t l2) -{ - size_t i; + const_iterator i1 = begin(); + const_iterator end1 = end(); + const_iterator i2 = s.begin(); + const_iterator end2 = s.end(); - if( l1 == l2 ) - { - for(i = 0; i < l1; ++i) - { - if(wxTolower(s1[i]) != wxTolower(s2[i])) - break; - } - return i == l1 ? 0 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1; - } - else if( l1 < l2 ) - { - for(i = 0; i < l1; ++i) - { - if(wxTolower(s1[i]) != wxTolower(s2[i])) - break; - } - return i == l1 ? -1 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1; - } - else + for ( ; i1 != end1 && i2 != end2; ++i1, ++i2 ) { - for(i = 0; i < l2; ++i) - { - if(wxTolower(s1[i]) != wxTolower(s2[i])) - break; - } - return i == l2 ? 1 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1; + wxUniChar lower1 = (wxChar)wxTolower(*i1); + wxUniChar lower2 = (wxChar)wxTolower(*i2); + if ( lower1 != lower2 ) + return lower1 < lower2 ? -1 : 1; } -} - -int wxString::CmpNoCase(const wxString& s) const -{ - return wxDoCmpNoCase(data(), length(), s.data(), s.length()); -} -int wxString::CmpNoCase(const wxChar* psz) const -{ - int nLen = wxStrlen(psz); + size_t len1 = length(); + size_t len2 = s.length(); - return wxDoCmpNoCase(data(), length(), psz, nLen); + if ( len1 < len2 ) + return -1; + else if ( len1 > len2 ) + return 1; + return 0; } @@ -1293,63 +1021,69 @@ int wxString::CmpNoCase(const wxChar* psz) const #endif #endif -wxString wxString::FromAscii(const char *ascii) +wxString wxString::FromAscii(const char *ascii, size_t len) { - if (!ascii) + if (!ascii || len == 0) return wxEmptyString; - size_t len = strlen( ascii ); wxString res; - if ( len ) { - wxStringBuffer buf(res, len); - - wchar_t *dest = buf; + wxStringInternalBuffer buf(res, len); + wxStringCharType *dest = buf; - for ( ;; ) + for ( ; len > 0; --len ) { - 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; } } return res; } -wxString wxString::FromAscii(const char ascii) +wxString wxString::FromAscii(const char *ascii) +{ + return FromAscii(ascii, wxStrlen(ascii)); +} + +wxString wxString::FromAscii(char ascii) { // 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 @@ -1385,28 +1119,15 @@ wxString wxString::Mid(size_t nFirst, size_t nCount) const // check that the string starts with prefix and return the rest of the string // in the provided pointer if it is not NULL, otherwise return false -bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const +bool wxString::StartsWith(const wxString& prefix, wxString *rest) const { - wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") ); - - // first check if the beginning of the string matches the prefix: note - // that we don't have to check that we don't run out of this string as - // when we reach the terminating NUL, either prefix string ends too (and - // then it's ok) or we break out of the loop because there is no match - const wxChar *p = c_str(); - while ( *prefix ) - { - if ( *prefix++ != *p++ ) - { - // no match - return false; - } - } + if ( compare(0, prefix.length(), prefix) != 0 ) + return false; if ( rest ) { // put the rest of the string into provided pointer - *rest = p; + rest->assign(*this, prefix.length(), npos); } return true; @@ -1415,12 +1136,11 @@ bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const // check that the string ends with suffix and return the rest of it in the // provided pointer if it is not NULL, otherwise return false -bool wxString::EndsWith(const wxChar *suffix, wxString *rest) const +bool wxString::EndsWith(const wxString& suffix, wxString *rest) const { - wxASSERT_MSG( suffix, _T("invalid parameter in wxString::EndssWith") ); + int start = length() - suffix.length(); - 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 ) @@ -1507,32 +1227,32 @@ wxString wxString::AfterFirst(wxUniChar ch) const } // replace first (or all) occurences of some substring with another one -size_t wxString::Replace(const wxChar *szOld, - const wxChar *szNew, bool bReplaceAll) +size_t wxString::Replace(const wxString& strOld, + const wxString& strNew, bool bReplaceAll) { // if we tried to replace an empty string we'd enter an infinite loop below - wxCHECK_MSG( szOld && *szOld && szNew, 0, + wxCHECK_MSG( !strOld.empty(), 0, _T("wxString::Replace(): invalid parameter") ); size_t uiCount = 0; // count of replacements made - size_t uiOldLen = wxStrlen(szOld); - size_t uiNewLen = wxStrlen(szNew); + size_t uiOldLen = strOld.length(); + size_t uiNewLen = strNew.length(); size_t dwPos = 0; - while ( this->c_str()[dwPos] != wxT('\0') ) + while ( (*this)[dwPos] != wxT('\0') ) { //DO NOT USE STRSTR HERE //this string can contain embedded null characters, //so strstr will function incorrectly - dwPos = find(szOld, dwPos); + dwPos = find(strOld, dwPos); if ( dwPos == npos ) break; // exit the loop else { //replace this occurance of the old string with the new one - replace(dwPos, uiOldLen, szNew, uiNewLen); + replace(dwPos, uiOldLen, strNew, uiNewLen); //move up pos past the string that was replaced dwPos += uiNewLen; @@ -1551,34 +1271,43 @@ size_t wxString::Replace(const wxChar *szOld, 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 @@ -1614,9 +1343,9 @@ wxString& wxString::MakeLower() // --------------------------------------------------------------------------- // some compilers (VC++ 6.0 not to name them) return true for a call to -// isspace('ê') in the C locale which seems to be broken to me, but we have to -// live with this by checking that the character is a 7 bit one - even if this -// may fail to detect some spaces (I don't know if Unicode doesn't have +// isspace('\xEA') in the C locale which seems to be broken to me, but we have +// to live with this by checking that the character is a 7 bit one - even if +// this may fail to detect some spaces (I don't know if Unicode doesn't have // space-like symbols somewhere except in the first 128 chars), it is arguably // still better than trimming away accented letters inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); } @@ -1637,7 +1366,7 @@ wxString& wxString::Trim(bool bFromRight) // find last non-space character reverse_iterator psz = rbegin(); while ( (psz != rend()) && wxSafeIsspace(*psz) ) - psz++; + ++psz; // truncate at trailing space start erase(psz.base(), end()); @@ -1647,7 +1376,7 @@ wxString& wxString::Trim(bool bFromRight) // find first non-space character iterator psz = begin(); while ( (psz != end()) && wxSafeIsspace(*psz) ) - psz++; + ++psz; // fix up data and length erase(begin(), psz); @@ -1697,110 +1426,88 @@ int wxString::Find(wxUniChar ch, bool bFromEnd) const return (idx == npos) ? wxNOT_FOUND : (int)idx; } -// find a sub-string (like strstr) -int wxString::Find(const wxChar *pszSub) const -{ - size_type idx = find(pszSub); - - return (idx == npos) ? wxNOT_FOUND : (int)idx; -} - // ---------------------------------------------------------------------------- // conversion to numbers // ---------------------------------------------------------------------------- -// the implementation of all the functions below is exactly the same so factor -// it out - -template -bool wxStringToIntType(const wxChar *start, - T *val, - int base, - F func) -{ - wxCHECK_MSG( val, false, _T("NULL output pointer") ); - wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); +// The implementation of all the functions below is exactly the same so factor +// it out. Note that number extraction works correctly on UTF-8 strings, so +// we can use wxStringCharType and wx_str() for maximum efficiency. #ifndef __WXWINCE__ - errno = 0; + #define DO_IF_NOT_WINCE(x) x +#else + #define DO_IF_NOT_WINCE(x) #endif - wxChar *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(out, base, func, T) \ + wxCHECK_MSG( out, 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; \ + T 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: */ \ + if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \ + return false; \ + *out = val; \ + return true -bool wxString::ToLong(long *val, int base) const +bool wxString::ToLong(long *pVal, int base) const { - return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtol); + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long); } -bool wxString::ToULong(unsigned long *val, int base) const +bool wxString::ToULong(unsigned long *pVal, int base) const { - return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoul); + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long); } -bool wxString::ToLongLong(wxLongLong_t *val, int base) const +bool wxString::ToLongLong(wxLongLong_t *pVal, 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 + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t); } -bool wxString::ToULongLong(wxULongLong_t *val, int base) const +bool wxString::ToULongLong(wxULongLong_t *pVal, 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 + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t); } -bool wxString::ToDouble(double *val) const +bool wxString::ToDouble(double *pVal) const { - wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToDouble") ); + wxCHECK_MSG( pVal, false, _T("NULL output pointer") ); -#ifndef __WXWINCE__ - errno = 0; -#endif + DO_IF_NOT_WINCE( errno = 0; ) const wxChar *start = c_str(); wxChar *end; - *val = wxStrtod(start, &end); + double val = wxStrtod(start, &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 - return !*end && (end != start) -#ifndef __WXWINCE__ - && (errno != ERANGE) -#endif - ; + if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) + return false; + + *pVal = val; + + return true; } // --------------------------------------------------------------------------- // 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; @@ -1813,6 +1520,23 @@ wxString wxString::DoFormat(const wxChar *format, ...) 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) @@ -1822,10 +1546,11 @@ 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; @@ -1846,19 +1571,106 @@ int wxString::DoPrintf(const wxChar *format, ...) 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 + +/* + Uses wxVsnprintf and places the result into the this string. + + In ANSI build, wxVsnprintf is effectively vsnprintf but in Unicode build + it is vswprintf. Due to a discrepancy between vsnprintf and vswprintf in + the ISO C99 (and thus SUSv3) standard the return value for the case of + an undersized buffer is inconsistent. For conforming vsnprintf + implementations the function must return the number of characters that + would have been printed had the buffer been large enough. For conforming + vswprintf implementations the function must return a negative number + and set errno. + + What vswprintf sets errno to is undefined but Darwin seems to set it to + EOVERFLOW. The only expected errno are EILSEQ and EINVAL. Both of + those are defined in the standard and backed up by several conformance + statements. Note that ENOMEM mentioned in the manual page does not + apply to swprintf, only wprintf and fwprintf. + + Official manual page: + http://www.opengroup.org/onlinepubs/009695399/functions/swprintf.html + + Some conformance statements (AIX, Solaris): + http://www.opengroup.org/csq/view.mhtml?RID=ibm%2FSD1%2F3 + http://www.theopengroup.org/csq/view.mhtml?norationale=1&noreferences=1&RID=Fujitsu%2FSE2%2F10 + + Since EILSEQ and EINVAL are rather common but EOVERFLOW is not and since + EILSEQ and EINVAL are specifically defined to mean the error is other than + an undersized buffer and no other errno are defined we treat those two + as meaning hard errors and everything else gets the old behavior which + is to keep looping and increasing buffer size until the function succeeds. + + In practice it's impossible to determine before compilation which behavior + may be used. The vswprintf function may have vsnprintf-like behavior or + vice-versa. Behavior detected on one release can theoretically change + with an updated release. Not to mention that configure testing for it + would require the test to be run on the host system, not the build system + which makes cross compilation difficult. Therefore, we make no assumptions + about behavior and try our best to handle every known case, including the + case where wxVsnprintf returns a negative number and fails to set errno. + + There is yet one more non-standard implementation and that is our own. + Fortunately, that can be detected at compile-time. + + On top of all that, ISO C99 explicitly defines snprintf to write a null + character to the last position of the specified buffer. That would be at + at the given buffer size minus 1. It is supposed to do this even if it + turns out that the buffer is sized too small. + + Darwin (tested on 10.5) follows the C99 behavior exactly. + + Glibc 2.6 almost follows the C99 behavior except vswprintf never sets + errno even when it fails. However, it only seems to ever fail due + to an undersized buffer. +*/ +#if wxUSE_UNICODE_UTF8 +template +#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 ) { // out of memory + + // in UTF-8 build, leaving uninitialized junk in the buffer + // could result in invalid non-empty UTF-8 string, so just + // reset the string to empty on failure: + buf[0] = '\0'; return -1; } @@ -1866,12 +1678,19 @@ int wxString::PrintfV(const wxString& format, va_list argptr) // only a copy va_list argptrcopy; wxVaCopy(argptrcopy, argptr); + +#ifndef __WXWINCE__ + // Set errno to 0 to make it determinate if wxVsnprintf fails to set it. + errno = 0; +#endif int len = wxVsnprintf(buf, size, format, argptrcopy); va_end(argptrcopy); // some implementations of vsnprintf() don't NUL terminate // the string if there is not enough space for it so // always do it manually + // FIXME: This really seems to be the wrong and would be an off-by-one + // bug except the code above allocates an extra character. buf[size] = _T('\0'); // vsnprintf() may return either -1 (traditional Unix behaviour) or the @@ -1879,27 +1698,47 @@ int wxString::PrintfV(const wxString& format, va_list argptr) // buffer were large enough (newer standards such as Unix98) if ( len < 0 ) { + // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or + // wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF + // is true if *both* of them use our own implementation, + // otherwise we can't be sure #if wxUSE_WXVSNPRINTF // we know that our own implementation of wxVsnprintf() returns -1 // only for a format error - thus there's something wrong with // the user's format string + buf[0] = '\0'; return -1; -#else // assume that system version only returns error if not enough space +#else // possibly using system version + // assume it only returns error if there is not enough space, but + // as we don't know how much we need, double the current size of + // the buffer +#ifndef __WXWINCE__ + if( (errno == EILSEQ) || (errno == EINVAL) ) + // If errno was set to one of the two well-known hard errors + // then fail immediately to avoid an infinite loop. + return -1; + else +#endif // __WXWINCE__ // still not enough, as we don't know how much we need, double the // current size of the buffer - size *= 2; + size *= 2; #endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF } else if ( len >= size ) { #if wxUSE_WXVSNPRINTF - // we know that our own implementation of wxVsnprintf() returns + // we know that our own implementation of wxVsnprintf() returns // size+1 when there's not enough space but that's not the size // of the required buffer! size *= 2; // so we just double the current size of the buffer #else // some vsnprintf() implementations NUL-terminate the buffer and // some don't in len == size case, to be safe always add 1 + // FIXME: I don't quite understand this comment. The vsnprintf + // function is specifically defined to return the number of + // characters printed not including the null terminator. + // So OF COURSE you need to add 1 to get the right buffer size. + // The following line is definitely correct, no question. size = len + 1; #endif } @@ -1910,9 +1749,34 @@ int wxString::PrintfV(const wxString& format, va_list argptr) } // we could have overshot - Shrink(); + str.Shrink(); + + return str.length(); +} + +int wxString::PrintfV(const wxString& format, va_list argptr) +{ +#if wxUSE_UNICODE_UTF8 + #if wxUSE_STL_BASED_WXSTRING + typedef wxStringTypeBuffer Utf8Buffer; + #else + typedef wxStringInternalBuffer Utf8Buffer; + #endif +#endif - return length(); +#if wxUSE_UTF8_LOCALE_ONLY + return DoStringPrintfV(*this, format, argptr); +#else + #if wxUSE_UNICODE_UTF8 + if ( wxLocaleIsUtf8 ) + return DoStringPrintfV(*this, format, argptr); + else + // wxChar* version + return DoStringPrintfV(*this, format, argptr); + #else + return DoStringPrintfV(*this, format, argptr); + #endif // UTF8/WCHAR +#endif } // ---------------------------------------------------------------------------- @@ -1922,7 +1786,7 @@ int wxString::PrintfV(const wxString& format, va_list argptr) // returns true if the string matches the pattern which may contain '*' and // '?' metacharacters (as usual, '?' matches any character and '*' any number // of them) -bool wxString::Matches(const wxChar *pszMask) const +bool wxString::Matches(const wxString& mask) const { // I disable this code as it doesn't seem to be faster (in fact, it seems // to be much slower) than the old, hand-written code below and using it @@ -1973,8 +1837,17 @@ bool wxString::Matches(const wxChar *pszMask) const #else // !wxUSE_REGEX // TODO: this is, of course, awfully inefficient... + // FIXME-UTF8: implement using iterators, remove #if +#if wxUSE_UNICODE_UTF8 + wxWCharBuffer maskBuf = mask.wc_str(); + wxWCharBuffer txtBuf = wc_str(); + const wxChar *pszMask = maskBuf.data(); + const wxChar *pszTxt = txtBuf.data(); +#else + const wxChar *pszMask = mask.wx_str(); // the char currently being checked - const wxChar *pszTxt = c_str(); + const wxChar *pszTxt = wx_str(); +#endif // the last location where '*' matched const wxChar *pszLastStarInText = NULL; @@ -2061,10 +1934,9 @@ match: int wxString::Freq(wxUniChar ch) const { int count = 0; - int len = length(); - for (int i = 0; i < len; i++) + for ( const_iterator i = begin(); i != end(); ++i ) { - if (GetChar(i) == ch) + if ( *i == ch ) count ++; } return count; @@ -2077,556 +1949,31 @@ wxString wxString::Upper() const // convert to lower case, return the copy of the string wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); } -// ============================================================================ -// ArrayString -// ============================================================================ - -#include "wx/arrstr.h" - -wxArrayString::wxArrayString(size_t sz, const wxChar** a) -{ -#if !wxUSE_STL - Init(false); -#endif - for (size_t i=0; i < sz; i++) - Add(a[i]); -} - -wxArrayString::wxArrayString(size_t sz, const wxString* a) -{ -#if !wxUSE_STL - Init(false); -#endif - for (size_t i=0; i < sz; i++) - Add(a[i]); -} - -#if !wxUSE_STL - -// size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT) -#define ARRAY_MAXSIZE_INCREMENT 4096 - -#ifndef ARRAY_DEFAULT_INITIAL_SIZE // also defined in dynarray.h -#define ARRAY_DEFAULT_INITIAL_SIZE (16) -#endif - -// ctor -void wxArrayString::Init(bool autoSort) -{ - m_nSize = - m_nCount = 0; - m_pItems = NULL; - m_autoSort = autoSort; -} - -// copy ctor -wxArrayString::wxArrayString(const wxArrayString& src) -{ - Init(src.m_autoSort); - - *this = src; -} - -// assignment operator -wxArrayString& wxArrayString::operator=(const wxArrayString& src) -{ - if ( m_nSize > 0 ) - Clear(); - - Copy(src); - - m_autoSort = src.m_autoSort; - - return *this; -} - -void wxArrayString::Copy(const wxArrayString& src) -{ - if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE ) - Alloc(src.m_nCount); - - for ( size_t n = 0; n < src.m_nCount; n++ ) - Add(src[n]); -} - -// grow the array -void wxArrayString::Grow(size_t nIncrement) -{ - // only do it if no more place - if ( (m_nSize - m_nCount) < nIncrement ) { - // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would - // be never resized! - #if ARRAY_DEFAULT_INITIAL_SIZE == 0 - #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!" - #endif - - if ( m_nSize == 0 ) { - // was empty, alloc some memory - m_nSize = ARRAY_DEFAULT_INITIAL_SIZE; - if (m_nSize < nIncrement) - m_nSize = nIncrement; - m_pItems = new wxString[m_nSize]; - } - else { - // otherwise when it's called for the first time, nIncrement would be 0 - // and the array would never be expanded - // add 50% but not too much - size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE - ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1; - if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT ) - ndefIncrement = ARRAY_MAXSIZE_INCREMENT; - if ( nIncrement < ndefIncrement ) - nIncrement = ndefIncrement; - m_nSize += nIncrement; - wxString *pNew = new wxString[m_nSize]; - - // copy data to new location - for ( size_t j = 0; j < m_nCount; j++ ) - pNew[j] = m_pItems[j]; - - // delete old memory (but do not release the strings!) - wxDELETEA(m_pItems); - - m_pItems = pNew; - } - } -} - -// deletes all the strings from the list -void wxArrayString::Empty() -{ - m_nCount = 0; -} - -// as Empty, but also frees memory -void wxArrayString::Clear() -{ - m_nSize = - m_nCount = 0; - - wxDELETEA(m_pItems); -} - -// dtor -wxArrayString::~wxArrayString() -{ - wxDELETEA(m_pItems); -} - -void wxArrayString::reserve(size_t nSize) -{ - Alloc(nSize); -} - -// pre-allocates memory (frees the previous data!) -void wxArrayString::Alloc(size_t nSize) -{ - // only if old buffer was not big enough - if ( nSize > m_nSize ) { - wxString *pNew = new wxString[nSize]; - if ( !pNew ) - return; - - for ( size_t j = 0; j < m_nCount; j++ ) - pNew[j] = m_pItems[j]; - delete [] m_pItems; - - m_pItems = pNew; - m_nSize = nSize; - } -} - -// minimizes the memory usage by freeing unused memory -void wxArrayString::Shrink() -{ - // only do it if we have some memory to free - if( m_nCount < m_nSize ) { - // allocates exactly as much memory as we need - wxString *pNew = new wxString[m_nCount]; - - // copy data to new location - for ( size_t j = 0; j < m_nCount; j++ ) - pNew[j] = m_pItems[j]; - delete [] m_pItems; - m_pItems = pNew; - } -} - -// searches the array for an item (forward or backwards) -int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const -{ - if ( m_autoSort ) { - // use binary search in the sorted array - wxASSERT_MSG( bCase && !bFromEnd, - wxT("search parameters ignored for auto sorted array") ); - - size_t i, - lo = 0, - hi = m_nCount; - int res; - while ( lo < hi ) { - i = (lo + hi)/2; - - res = wxStrcmp(sz, m_pItems[i]); - if ( res < 0 ) - hi = i; - else if ( res > 0 ) - lo = i + 1; - else - return i; - } - - return wxNOT_FOUND; - } - else { - // use linear search in unsorted array - if ( bFromEnd ) { - if ( m_nCount > 0 ) { - size_t ui = m_nCount; - do { - if ( m_pItems[--ui].IsSameAs(sz, bCase) ) - return ui; - } - while ( ui != 0 ); - } - } - else { - for( size_t ui = 0; ui < m_nCount; ui++ ) { - if( m_pItems[ui].IsSameAs(sz, bCase) ) - return ui; - } - } - } - - return wxNOT_FOUND; -} - -// add item at the end -size_t wxArrayString::Add(const wxString& str, size_t nInsert) -{ - if ( m_autoSort ) { - // insert the string at the correct position to keep the array sorted - size_t i, - lo = 0, - hi = m_nCount; - int res; - while ( lo < hi ) { - i = (lo + hi)/2; - - res = str.Cmp(m_pItems[i]); - if ( res < 0 ) - hi = i; - else if ( res > 0 ) - lo = i + 1; - else { - lo = hi = i; - break; - } - } - - wxASSERT_MSG( lo == hi, wxT("binary search broken") ); - - Insert(str, lo, nInsert); - - return (size_t)lo; - } - else { - Grow(nInsert); - - for (size_t i = 0; i < nInsert; i++) - { - // just append - m_pItems[m_nCount + i] = str; - } - size_t ret = m_nCount; - m_nCount += nInsert; - return ret; - } -} - -// add item at the given position -void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert) -{ - wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") ); - wxCHECK_RET( m_nCount <= m_nCount + nInsert, - wxT("array size overflow in wxArrayString::Insert") ); - - Grow(nInsert); - - for (int j = m_nCount - nIndex - 1; j >= 0; j--) - m_pItems[nIndex + nInsert + j] = m_pItems[nIndex + j]; - - for (size_t i = 0; i < nInsert; i++) - { - m_pItems[nIndex + i] = str; - } - m_nCount += nInsert; -} - -// range insert (STL 23.2.4.3) -void -wxArrayString::insert(iterator it, const_iterator first, const_iterator last) -{ - const int idx = it - begin(); - - // grow it once - Grow(last - first); - - // reset "it" since it can change inside Grow() - it = begin() + idx; - - while ( first != last ) - { - it = insert(it, *first); - - // insert returns an iterator to the last element inserted but we need - // insert the next after this one, that is before the next one - ++it; - - ++first; - } -} - -// expand the array -void wxArrayString::SetCount(size_t count) -{ - Alloc(count); - - wxString s; - while ( m_nCount < count ) - m_pItems[m_nCount++] = s; -} - -// removes item from array (by index) -void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove) -{ - wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") ); - wxCHECK_RET( nIndex + nRemove <= m_nCount, - wxT("removing too many elements in wxArrayString::Remove") ); - - for ( size_t j = 0; j < m_nCount - nIndex -nRemove; j++) - m_pItems[nIndex + j] = m_pItems[nIndex + nRemove + j]; - - m_nCount -= nRemove; -} - -// removes item from array (by value) -void wxArrayString::Remove(const wxChar *sz) -{ - int iIndex = Index(sz); - - wxCHECK_RET( iIndex != wxNOT_FOUND, - wxT("removing inexistent element in wxArrayString::Remove") ); - - RemoveAt(iIndex); -} - -void wxArrayString::assign(const_iterator first, const_iterator last) -{ - reserve(last - first); - for(; first != last; ++first) - push_back(*first); -} - // ---------------------------------------------------------------------------- -// sorting +// wxUTF8StringBuffer // ---------------------------------------------------------------------------- -// we can only sort one array at a time with the quick-sort based -// implementation -#if wxUSE_THREADS - // need a critical section to protect access to gs_compareFunction and - // gs_sortAscending variables - static wxCriticalSection gs_critsectStringSort; -#endif // wxUSE_THREADS - -// function to use for string comparaison -static wxArrayString::CompareFunction gs_compareFunction = NULL; - -// if we don't use the compare function, this flag tells us if we sort the -// array in ascending or descending order -static bool gs_sortAscending = true; - -// function which is called by quick sort -extern "C" int wxC_CALLING_CONV // LINKAGEMODE -wxStringCompareFunction(const void *first, const void *second) -{ - wxString *strFirst = (wxString *)first; - wxString *strSecond = (wxString *)second; - - if ( gs_compareFunction ) { - return gs_compareFunction(*strFirst, *strSecond); - } - else { - // maybe we should use wxStrcoll - int result = strFirst->Cmp(*strSecond); - - return gs_sortAscending ? result : -result; - } -} - -// sort array elements using passed comparaison function -void wxArrayString::Sort(CompareFunction compareFunction) -{ - wxCRIT_SECT_LOCKER(lockCmpFunc, gs_critsectStringSort); - - wxASSERT( !gs_compareFunction ); // must have been reset to NULL - gs_compareFunction = compareFunction; - - DoSort(); - - // reset it to NULL so that Sort(bool) will work the next time - gs_compareFunction = NULL; -} - -extern "C" -{ - typedef int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first, - const void *second); -} - -void wxArrayString::Sort(CompareFunction2 compareFunction) -{ - qsort(m_pItems, m_nCount, sizeof(wxString), (wxStringCompareFn)compareFunction); -} - -void wxArrayString::Sort(bool reverseOrder) -{ - Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending); -} - -void wxArrayString::DoSort() -{ - wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") ); - - qsort(m_pItems, m_nCount, sizeof(wxString), wxStringCompareFunction); -} - -bool wxArrayString::operator==(const wxArrayString& a) const -{ - if ( m_nCount != a.m_nCount ) - return false; - - for ( size_t n = 0; n < m_nCount; n++ ) - { - if ( Item(n) != a[n] ) - return false; - } - - return true; -} - -#endif // !wxUSE_STL - -int wxCMPFUNC_CONV wxStringSortAscending(wxString* s1, wxString* s2) -{ - return s1->Cmp(*s2); -} - -int wxCMPFUNC_CONV wxStringSortDescending(wxString* s1, wxString* s2) -{ - return -s1->Cmp(*s2); -} - - - -// =========================================================================== -// wxJoin and wxSplit -// =========================================================================== - -#include "wx/tokenzr.h" - -wxString wxJoin(const wxArrayString& arr, const wxChar sep, const wxChar escape) +#if wxUSE_UNICODE_WCHAR +wxUTF8StringBuffer::~wxUTF8StringBuffer() { - size_t count = arr.size(); - if ( count == 0 ) - return wxEmptyString; - - wxString str; - - // pre-allocate memory using the estimation of the average length of the - // strings in the given array: this is very imprecise, of course, but - // better than nothing - str.reserve(count*(arr[0].length() + arr[count-1].length()) / 2); - - if ( escape == wxT('\0') ) - { - // escaping is disabled: - for ( size_t i = 0; i < count; i++ ) - { - if ( i ) - str += sep; - str += arr[i]; - } - } - else // use escape character - { - for ( size_t n = 0; n < count; n++ ) - { - if ( n ) - str += sep; - - for ( wxString::const_iterator i = arr[n].begin(), - end = arr[n].end(); - i != end; - ++i ) - { - const wxChar ch = *i; - if ( ch == sep ) - str += escape; // escape this separator - str += ch; - } - } - } + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); - str.Shrink(); // release extra memory if we allocated too much - return str; + wxStringInternalBuffer wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf); } -wxArrayString wxSplit(const wxString& str, const wxChar sep, const wxChar escape) +wxUTF8StringBufferLength::~wxUTF8StringBufferLength() { - if ( escape == wxT('\0') ) - { - // simple case: we don't need to honour the escape character - return wxStringTokenize(str, sep, wxTOKEN_RET_EMPTY_ALL); - } - - wxArrayString ret; - wxString curr; - wxChar prev = wxT('\0'); - - for ( wxString::const_iterator i = str.begin(), - end = str.end(); - i != end; - ++i ) - { - const wxChar ch = *i; - - if ( ch == sep ) - { - if ( prev == escape ) - { - // remove the escape character and don't consider this - // occurrence of 'sep' as a real separator - *curr.rbegin() = sep; - } - else // real separator - { - ret.push_back(curr); - curr.clear(); - } - } - else // normal character - { - curr += ch; - } - - prev = ch; - } + wxCHECK_RET(m_lenSet, "length not set"); - // add the last token - if ( !curr.empty() || prev == sep ) - ret.Add(curr); + 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?" ); - return ret; + wxStringInternalBufferLength wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf, m_len); + wbuf.SetLength(wlen); } +#endif // wxUSE_UNICODE_WCHAR