X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/6859e6fcca54ae7a1f09c136b6f3c2214ed05c07..57d3e26697255a5c5154b4a0dc26d6eab23ff871:/src/common/stringimpl.cpp diff --git a/src/common/stringimpl.cpp b/src/common/stringimpl.cpp index e8750dc4f0..ed9a83e476 100644 --- a/src/common/stringimpl.cpp +++ b/src/common/stringimpl.cpp @@ -1,5 +1,5 @@ ///////////////////////////////////////////////////////////////////////////// -// Name: src/common/string.cpp +// Name: src/common/stringimpl.cpp // Purpose: wxString class // Author: Vadim Zeitlin, Ryan Norton // Modified by: @@ -30,6 +30,7 @@ #ifndef WX_PRECOMP #include "wx/stringimpl.h" + #include "wx/wxcrt.h" #endif #include @@ -41,10 +42,6 @@ #include #include -#ifdef __SALFORDC__ - #include -#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! @@ -56,12 +53,10 @@ #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 @@ -80,7 +75,11 @@ const size_t wxStringImpl::npos = (size_t) -1; #if wxUSE_STL_BASED_WXSTRING -extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T(""); +// FIXME-UTF8: get rid of this, have only one wxEmptyString +#if wxUSE_UNICODE_UTF8 +const wxStringCharType WXDLLIMPEXP_BASE *wxEmptyStringImpl = ""; +#endif +const wxChar WXDLLIMPEXP_BASE *wxEmptyString = wxT(""); #else @@ -90,11 +89,17 @@ extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T(""); static const struct { wxStringData data; - wxChar dummy; + wxStringCharType 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; +#if wxUSE_UNICODE_UTF8 +// FIXME-UTF8: get rid of this, have only one wxEmptyString +const wxStringCharType WXDLLIMPEXP_BASE *wxEmptyStringImpl = &g_strEmpty.dummy; +const wxChar WXDLLIMPEXP_BASE *wxEmptyString = wxT(""); +#else +const wxStringCharType WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy; +#endif #endif @@ -106,20 +111,27 @@ extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy; // ---------------------------------------------------------------------------- // this small class is used to gather statistics for performance tuning + +// uncomment this to enable gathering of some statistics about wxString +// efficiency //#define WXSTRING_STATISTICS + #ifdef WXSTRING_STATISTICS class Averager { public: - Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; } + Averager(const wxStringCharType *sz) { m_sz = sz; m_nTotal = m_nCount = 0; } ~Averager() - { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); } + { + wxPrintf("wxString %s: total = %lu, average = %f\n", + m_sz, m_nTotal, ((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; + unsigned long m_nCount, m_nTotal; + const wxStringCharType *m_sz; } g_averageLength("allocation size"), g_averageSummandLength("summand length"), g_averageConcatHit("hit probability in concat"), @@ -147,13 +159,14 @@ void wxStringData::Free() // =========================================================================== // takes nLength elements of psz starting at nPos -void wxStringImpl::InitWith(const wxChar *psz, size_t nPos, size_t nLength) +void wxStringImpl::InitWith(const wxStringCharType *psz, + size_t nPos, size_t nLength) { 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") ); + wxASSERT_MSG( nPos <= wxStrlen(psz), wxT("index out of bounds") ); nLength = wxStrlen(psz + nPos); } @@ -163,7 +176,7 @@ void wxStringImpl::InitWith(const wxChar *psz, size_t nPos, size_t nLength) if ( nLength > 0 ) { // trailing '\0' is written in AllocBuffer() if ( !AllocBuffer(nLength) ) { - wxFAIL_MSG( _T("out of memory in wxStringImpl::InitWith") ); + wxFAIL_MSG( wxT("out of memory in wxStringImpl::InitWith") ); return; } wxStringMemcpy(m_pchData, psz + nPos, nLength); @@ -174,11 +187,11 @@ wxStringImpl::wxStringImpl(const_iterator first, const_iterator last) { if ( last >= first ) { - InitWith(first, 0, last - first); + InitWith(first.GetPtr(), 0, last - first); } else { - wxFAIL_MSG( _T("first must be before last") ); + wxFAIL_MSG( wxT("first must be before last") ); Init(); } } @@ -201,8 +214,8 @@ bool wxStringImpl::AllocBuffer(size_t nLen) wxASSERT( nLen > 0 ); // make sure that we don't overflow - wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) - - (sizeof(wxStringData) + EXTRA_ALLOC + 1) ); + wxCHECK( nLen < (INT_MAX / sizeof(wxStringCharType)) - + (sizeof(wxStringData) + EXTRA_ALLOC + 1), false ); STATISTICS_ADD(Length, nLen); @@ -210,7 +223,7 @@ bool wxStringImpl::AllocBuffer(size_t nLen) // 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)); + malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxStringCharType)); if ( pData == NULL ) { // allocation failures are handled by the caller @@ -269,7 +282,8 @@ bool wxStringImpl::AllocBeforeWrite(size_t nLen) nLen += EXTRA_ALLOC; pData = (wxStringData*) - realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); + realloc(pData, + sizeof(wxStringData) + (nLen + 1)*sizeof(wxStringCharType)); if ( pData == NULL ) { // allocation failures are handled by the caller @@ -297,7 +311,7 @@ wxStringImpl& wxStringImpl::append(size_t n, wxStringCharType ch) size_type len = length(); if ( !Alloc(len + n) || !CopyBeforeWrite() ) { - wxFAIL_MSG( _T("out of memory in wxStringImpl::append") ); + wxFAIL_MSG( wxT("out of memory in wxStringImpl::append") ); return *this; } GetStringData()->nDataLength = len + n; @@ -328,10 +342,12 @@ bool wxStringImpl::Alloc(size_t nLen) wxStringData *pData = GetStringData(); if ( pData->nAllocLength <= nLen ) { if ( pData->IsEmpty() ) { + STATISTICS_ADD(Length, nLen); + nLen += EXTRA_ALLOC; pData = (wxStringData *) - malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); + malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxStringCharType)); if ( pData == NULL ) { // allocation failure handled by caller @@ -352,14 +368,14 @@ bool wxStringImpl::Alloc(size_t nLen) return false; } // +1 to copy the terminator, too - memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxChar)); + memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxStringCharType)); GetStringData()->nDataLength = nOldLen; } else { nLen += EXTRA_ALLOC; pData = (wxStringData *) - realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); + realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxStringCharType)); if ( pData == NULL ) { // allocation failure handled by caller @@ -379,14 +395,14 @@ bool wxStringImpl::Alloc(size_t nLen) wxStringImpl::iterator wxStringImpl::begin() { - if (length() > 0) + if ( !empty() ) CopyBeforeWrite(); return m_pchData; } wxStringImpl::iterator wxStringImpl::end() { - if (length() > 0) + if ( !empty() ) CopyBeforeWrite(); return m_pchData + length(); } @@ -411,7 +427,8 @@ wxStringImpl& wxStringImpl::erase(size_t nStart, size_t nLen) return *this; } -wxStringImpl& wxStringImpl::insert(size_t nPos, const wxChar *sz, size_t n) +wxStringImpl& wxStringImpl::insert(size_t nPos, + const wxStringCharType *sz, size_t n) { wxASSERT( nPos <= length() ); @@ -419,13 +436,13 @@ wxStringImpl& wxStringImpl::insert(size_t nPos, const wxChar *sz, size_t n) if ( n == 0 ) return *this; if ( !Alloc(length() + n) || !CopyBeforeWrite() ) { - wxFAIL_MSG( _T("out of memory in wxStringImpl::insert") ); + wxFAIL_MSG( wxT("out of memory in wxStringImpl::insert") ); return *this; } memmove(m_pchData + nPos + n, m_pchData + nPos, - (length() - nPos) * sizeof(wxChar)); - memcpy(m_pchData + nPos, sz, n * sizeof(wxChar)); + (length() - nPos) * sizeof(wxStringCharType)); + memcpy(m_pchData + nPos, sz, n * sizeof(wxStringCharType)); GetStringData()->nDataLength = length() + n; m_pchData[length()] = '\0'; @@ -487,7 +504,8 @@ size_t wxStringImpl::find(const wxStringImpl& str, size_t nStart) const return p - c_str() + nLenOther <= nLen ? p - c_str() : npos; } -size_t wxStringImpl::find(const wxChar* sz, size_t nStart, size_t n) const +size_t wxStringImpl::find(const wxStringCharType* sz, + size_t nStart, size_t n) const { return find(wxStringImpl(sz, n), nStart); } @@ -510,7 +528,7 @@ size_t wxStringImpl::rfind(const wxStringImpl& str, size_t nStart) const if ( length() >= str.length() ) { // avoids a corner case later - if ( length() == 0 && str.length() == 0 ) + if ( empty() && str.empty() ) return 0; // "top" is the point where search starts from @@ -534,7 +552,8 @@ size_t wxStringImpl::rfind(const wxStringImpl& str, size_t nStart) const return npos; } -size_t wxStringImpl::rfind(const wxChar* sz, size_t nStart, size_t n) const +size_t wxStringImpl::rfind(const wxStringCharType* sz, + size_t nStart, size_t n) const { return rfind(wxStringImpl(sz, n), nStart); } @@ -562,54 +581,45 @@ size_t wxStringImpl::rfind(wxStringCharType ch, size_t nStart) const } wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen, - const wxChar *sz) -{ - wxASSERT_MSG( nStart <= length(), - _T("index out of bounds in wxStringImpl::replace") ); - size_t strLen = length() - nStart; - nLen = strLen < nLen ? strLen : nLen; - - wxStringImpl strTmp; - strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs - - //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]); - - //its safe to do the full version here because - //sz must be a normal c string - strTmp.append(sz); - - for(size_t i2 = nStart + nLen; i2 < length(); ++i2) - strTmp.append(1, this->c_str()[i2]); - - swap(strTmp); - return *this; -} - -wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen, - size_t nCount, wxStringCharType ch) + const wxStringCharType *sz, size_t nCount) { - return replace(nStart, nLen, wxStringImpl(nCount, ch).c_str()); -} + // check and adjust parameters + const size_t lenOld = length(); -wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen, - const wxStringImpl& str, - size_t nStart2, size_t nLen2) -{ - return replace(nStart, nLen, str.substr(nStart2, nLen2)); -} + wxASSERT_MSG( nStart <= lenOld, + wxT("index out of bounds in wxStringImpl::replace") ); + size_t nEnd = nStart + nLen; + if ( nLen > lenOld - nStart ) + { + // nLen may be out of range, as it can be npos, just clump it down + nLen = lenOld - nStart; + nEnd = lenOld; + } -wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen, - const wxChar* sz, size_t nCount) -{ - return replace(nStart, nLen, wxStringImpl(sz, nCount).c_str()); + if ( nCount == npos ) + nCount = wxStrlen(sz); + + // build the new string from 3 pieces: part of this string before nStart, + // the new substring and the part of this string after nStart+nLen + wxStringImpl tmp; + const size_t lenNew = lenOld + nCount - nLen; + if ( lenNew ) + { + tmp.AllocBuffer(lenOld + nCount - nLen); + + wxStringCharType *dst = tmp.m_pchData; + memcpy(dst, m_pchData, nStart*sizeof(wxStringCharType)); + dst += nStart; + + memcpy(dst, sz, nCount*sizeof(wxStringCharType)); + dst += nCount; + + memcpy(dst, m_pchData + nEnd, (lenOld - nEnd)*sizeof(wxStringCharType)); + } + + // and replace this string contents with the new one + swap(tmp); + return *this; } wxStringImpl wxStringImpl::substr(size_t nStart, size_t nLen) const @@ -643,24 +653,25 @@ wxStringImpl& wxStringImpl::operator=(const wxStringImpl& stringSrc) // assigns a single character wxStringImpl& wxStringImpl::operator=(wxStringCharType ch) { - wxChar c(ch); + wxStringCharType c(ch); if ( !AssignCopy(1, &c) ) { - wxFAIL_MSG( _T("out of memory in wxStringImpl::operator=(wxChar)") ); + wxFAIL_MSG( wxT("out of memory in wxStringImpl::operator=(wxStringCharType)") ); } return *this; } // assigns C string -wxStringImpl& wxStringImpl::operator=(const wxChar *psz) +wxStringImpl& wxStringImpl::operator=(const wxStringCharType *psz) { if ( !AssignCopy(wxStrlen(psz), psz) ) { - wxFAIL_MSG( _T("out of memory in wxStringImpl::operator=(const wxChar *)") ); + wxFAIL_MSG( wxT("out of memory in wxStringImpl::operator=(const wxStringCharType *)") ); } return *this; } // helper function: does real copy -bool wxStringImpl::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData) +bool wxStringImpl::AssignCopy(size_t nSrcLen, + const wxStringCharType *pszSrcData) { if ( nSrcLen == 0 ) { Reinit(); @@ -670,7 +681,7 @@ bool wxStringImpl::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData) // allocation failure handled by caller return false; } - memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar)); + memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxStringCharType)); GetStringData()->nDataLength = nSrcLen; m_pchData[nSrcLen] = wxT('\0'); } @@ -682,7 +693,8 @@ bool wxStringImpl::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData) // --------------------------------------------------------------------------- // add something to this string -bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, +bool wxStringImpl::ConcatSelf(size_t nSrcLen, + const wxStringCharType *pszSrcData, size_t nMaxLen) { STATISTICS_ADD(SummandLength, nSrcLen); @@ -693,6 +705,17 @@ bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, if ( nSrcLen > 0 ) { wxStringData *pData = GetStringData(); size_t nLen = pData->nDataLength; + + // take special care when appending part of this string to itself: the code + // below reallocates our buffer and this invalidates pszSrcData pointer so + // we have to copy it in another temporary string in this case (but avoid + // doing this unnecessarily) + if ( pszSrcData >= m_pchData && pszSrcData < m_pchData + nLen ) + { + wxStringImpl tmp(pszSrcData, nSrcLen); + return ConcatSelf(nSrcLen, tmp.m_pchData, nSrcLen); + } + size_t nNewLen = nLen + nSrcLen; // alloc new buffer if current is too small @@ -705,7 +728,7 @@ bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, // allocation failure handled by caller return false; } - memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar)); + memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxStringCharType)); pOldData->Unlock(); } else if ( nNewLen > pData->nAllocLength ) { @@ -728,7 +751,7 @@ bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, wxASSERT( nNewLen <= GetStringData()->nAllocLength ); // fast concatenation - all is done in our buffer - memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar)); + memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxStringCharType)); m_pchData[nNewLen] = wxT('\0'); // put terminating '\0' GetStringData()->nDataLength = nNewLen; // and fix the length @@ -737,9 +760,8 @@ bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData, return true; } -#if !wxUSE_UNICODE_UTF8 // get the pointer to writable buffer of (at least) nLen bytes -wxChar *wxStringImpl::DoGetWriteBuf(size_t nLen) +wxStringCharType *wxStringImpl::DoGetWriteBuf(size_t nLen) { if ( !AllocBeforeWrite(nLen) ) { // allocation failure handled by caller @@ -762,13 +784,12 @@ void wxStringImpl::DoUngetWriteBuf(size_t nLen) { wxStringData * const pData = GetStringData(); - wxASSERT_MSG( nLen < pData->nAllocLength, _T("buffer overrun") ); + wxASSERT_MSG( nLen < pData->nAllocLength, wxT("buffer overrun") ); // the strings we store are always NUL-terminated - pData->data()[nLen] = _T('\0'); + pData->data()[nLen] = wxT('\0'); pData->nDataLength = nLen; pData->Validate(true); } -#endif // !wxUSE_UNICODE_UTF8 #endif // !wxUSE_STL_BASED_WXSTRING