X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/992527a5738f7fb629d2aaeee64cfc733ddb7e1c..1e510b1e2d0270caf227c3fc0cf34ae2e7dd6794:/src/common/string.cpp?ds=sidebyside diff --git a/src/common/string.cpp b/src/common/string.cpp index 9c8122f952..a398b215e5 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -30,8 +30,6 @@ #ifndef WX_PRECOMP #include "wx/string.h" - #include "wx/intl.h" - #include "wx/thread.h" #endif #include @@ -47,43 +45,28 @@ #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! -#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(""); +#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 -// 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; +// --------------------------------------------------------------------------- +// static class variables definition +// --------------------------------------------------------------------------- -#endif +//According to STL _must_ be a -1 size_t +const size_t wxString::npos = (size_t) -1; // ---------------------------------------------------------------------------- // global functions @@ -93,1195 +76,1263 @@ 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) { +// FIXME-UTF8: always, not only if wxUSE_UNICODE #if wxUSE_UNICODE && !defined(__BORLANDC__) - return os << str.AsWChar(); + return os << (const wchar_t*)str.AsWCharBuf(); #else - return os << str.AsChar(); + return os << (const char*)str.AsCharBuf(); #endif } -#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 - -// =========================================================================== -// wxStringData class deallocation -// =========================================================================== - -#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() +wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) { - free(this); + return os << str.c_str(); } -#endif - -// =========================================================================== -// wxStringBase -// =========================================================================== -// takes nLength elements of psz starting at nPos -void wxStringBase::InitWith(const wxChar *psz, size_t nPos, size_t nLength) +wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& 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 os << str.data(); } -// poor man's iterators are "void *" pointers -wxStringBase::wxStringBase(const void *pStart, const void *pEnd) +#ifndef __BORLANDC__ +wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& 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 os << str.data(); } +#endif -wxStringBase::wxStringBase(size_type n, wxUniChar ch) -{ - Init(); - append(n, ch); -} +#endif // wxUSE_STD_IOSTREAM + +// =========================================================================== +// wxString class core +// =========================================================================== + +#if wxUSE_UNICODE_UTF8 // --------------------------------------------------------------------------- -// memory allocation +// UTF-8 operations // --------------------------------------------------------------------------- -// allocates memory needed to store a C string of length nLen -bool wxStringBase::AllocBuffer(size_t nLen) +// +// Table 3.1B from Unicode spec: Legal UTF-8 Byte Sequences +// +// Code Points | 1st Byte | 2nd Byte | 3rd Byte | 4th Byte | +// -------------------+----------+----------+----------+----------+ +// U+0000..U+007F | 00..7F | | | | +// U+0080..U+07FF | C2..DF | 80..BF | | | +// U+0800..U+0FFF | E0 | A0..BF | 80..BF | | +// U+1000..U+FFFF | E1..EF | 80..BF | 80..BF | | +// U+10000..U+3FFFF | F0 | 90..BF | 80..BF | 80..BF | +// U+40000..U+FFFFF | F1..F3 | 80..BF | 80..BF | 80..BF | +// U+100000..U+10FFFF | F4 | 80..8F | 80..BF | 80..BF | +// -------------------+----------+----------+----------+----------+ + +bool wxString::IsValidUtf8String(const char *str) { - // allocating 0 sized buffer doesn't make sense, all empty strings should - // reuse g_strEmpty - wxASSERT( nLen > 0 ); + if ( !str ) + return true; // empty string is UTF8 string - // make sure that we don't overflow - wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) - - (sizeof(wxStringData) + EXTRA_ALLOC + 1) ); + const unsigned char *c = (const unsigned char*)str; - STATISTICS_ADD(Length, nLen); + for ( ; *c; ++c ) + { + unsigned char b = *c; - // 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)); + if ( b <= 0x7F ) // 00..7F + continue; - if ( pData == NULL ) { - // allocation failures are handled by the caller - return false; - } + else if ( b < 0xC2 ) // invalid lead bytes: 80..C1 + return false; - 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; -} + // two-byte sequences: + else if ( b <= 0xDF ) // C2..DF + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } -// must be called before changing this string -bool wxStringBase::CopyBeforeWrite() -{ - wxStringData* pData = GetStringData(); + // three-byte sequences: + else if ( b == 0xE0 ) + { + b = *(++c); + if ( !(b >= 0xA0 && b <= 0xBF ) ) + return false; + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } + else if ( b <= 0xEF ) // E1..EF + { + for ( int i = 0; i < 2; ++i ) + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } + } - 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; + // four-byte sequences: + else if ( b == 0xF0 ) + { + b = *(++c); + if ( !(b >= 0x90 && b <= 0xBF ) ) + return false; + for ( int i = 0; i < 2; ++i ) + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } + } + else if ( b <= 0xF3 ) // F1..F3 + { + for ( int i = 0; i < 3; ++i ) + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } + } + else if ( b == 0xF4 ) + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0x8F ) ) + return false; + for ( int i = 0; i < 2; ++i ) + { + b = *(++c); + if ( !(b >= 0x80 && b <= 0xBF ) ) + return false; + } + } + else // otherwise, it's invalid lead byte + return false; } - wxTmemcpy(m_pchData, pData->data(), nLen); - } - wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner - - return true; + return true; } -// must be called before replacing contents of this string -bool wxStringBase::AllocBeforeWrite(size_t nLen) +#ifdef __WXDEBUG__ +/* static */ +bool wxString::IsValidUtf8LeadByte(unsigned char c) { - wxASSERT( nLen != 0 ); // doesn't make any sense - - // 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; - } - } - else { - if ( nLen > pData->nAllocLength ) { - // realloc the buffer instead of calling malloc() again, this is more - // efficient - STATISTICS_ADD(Length, nLen); - - nLen += EXTRA_ALLOC; - - pData = (wxStringData*) - realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); - - if ( pData == NULL ) { - // allocation failures are handled by the caller - // keep previous data since reallocation failed - return false; - } - - pData->nAllocLength = nLen; - m_pchData = pData->data(); - } - } + return (c <= 0x7F) || (c >= 0xC2 && c <= 0xF4); +} +#endif - wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner +unsigned char wxString::ms_utf8IterTable[256] = { + // single-byte sequences (ASCII): + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 00..0F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 10..1F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 20..2F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 30..3F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 40..4F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 50..5F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 60..6F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 70..7F + + // these are invalid, we use step 1 to skip + // over them (should never happen): + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 80..8F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // 90..9F + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // A0..AF + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, // B0..BF + 1, 1, // C0,C1 + + // two-byte sequences: + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // C2..CF + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // D0..DF + + // three-byte sequences: + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // E0..EF + + // four-byte sequences: + 4, 4, 4, 4, 4, // F0..F4 + + // these are invalid again (5- or 6-byte + // sequences and sequences for code points + // above U+10FFFF, as restricted by RFC 3629): + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 // F5..FF +}; - // 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; +/* static */ +void wxString::DecIter(wxStringImpl::const_iterator& i) +{ + wxASSERT( IsValidUtf8LeadByte(*i) ); - return true; + // Non-lead bytes are all in the 0x80..0xBF range (i.e. 10xxxxxx in + // binary), so we just have to go back until we hit a byte that is either + // < 0x80 (i.e. 0xxxxxxx in binary) or 0xC0..0xFF (11xxxxxx in binary; this + // includes some invalid values, but we can ignore it here, because we + // assume valid UTF-8 input for the purpose of efficient implementation). + --i; + while ( ((*i) & 0xC0) == 0x80 /* 2 highest bits are '10' */ ) + --i; } -wxStringBase& wxStringBase::append(size_t n, wxUniChar ch) +/* static */ +void wxString::DecIter(wxStringImpl::iterator& i) { - 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; + // FIXME-UTF8: use template instead + wxASSERT( IsValidUtf8LeadByte(*i) ); + --i; + while ( ((*i) & 0xC0) == 0x80 /* 2 highest bits are '10' */ ) + --i; } -void wxStringBase::resize(size_t nSize, wxUniChar ch) +/* static */ +wxStringImpl::const_iterator +wxString::AddToIter(wxStringImpl::const_iterator i, int n) { - size_t len = length(); + wxStringImpl::const_iterator out(i); - if ( nSize < len ) + if ( n > 0 ) { - erase(begin() + nSize, end()); + for ( int j = 0; j < n; ++j ) + IncIter(out); } - else if ( nSize > len ) + else if ( n < 0 ) { - append(nSize - len, ch); + for ( int j = 0; j > n; --j ) + DecIter(out); } - //else: we have exactly the specified length, nothing to do + + return out; } -// allocate enough memory for nLen characters -bool wxStringBase::Alloc(size_t nLen) +wxStringImpl::iterator +wxString::AddToIter(wxStringImpl::iterator i, int n) { - wxStringData *pData = GetStringData(); - if ( pData->nAllocLength <= nLen ) { - if ( pData->IsEmpty() ) { - nLen += EXTRA_ALLOC; + // FIXME-UTF8: use template instead + wxStringImpl::iterator out(i); - pData = (wxStringData *) - malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar)); - - if ( pData == NULL ) { - // allocation failure handled by caller - return false; - } - - 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; + if ( n > 0 ) + { + for ( int j = 0; j < n; ++j ) + IncIter(out); } - 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 if ( n < 0 ) + { + for ( int j = 0; j > n; --j ) + DecIter(out); } - } - //else: we've already got enough - return true; -} -wxStringBase::iterator wxStringBase::begin() -{ - if (length() > 0) - CopyBeforeWrite(); - return m_pchData; + return out; } -wxStringBase::iterator wxStringBase::end() -{ - if (length() > 0) - CopyBeforeWrite(); - return m_pchData + length(); -} -wxStringBase::iterator wxStringBase::erase(iterator it) +/* static */ +int wxString::DiffIters(wxStringImpl::const_iterator i1, + wxStringImpl::const_iterator i2) { - size_type idx = it - begin(); - erase(idx, 1); - return begin() + idx; -} + int dist = 0; -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); + if ( i1 < i2 ) + { + while ( i1 != i2 ) + { + IncIter(i1); + dist--; + } + } + else if ( i2 < i1 ) + { + while ( i2 != i1 ) + { + IncIter(i2); + dist++; + } + } - swap(strTmp); - return *this; + return dist; } -wxStringBase& wxStringBase::insert(size_t nPos, const wxChar *sz, size_t n) +int wxString::DiffIters(wxStringImpl::iterator i1, wxStringImpl::iterator i2) { - wxASSERT( nPos <= length() ); + // FIXME-UTF8: use template instead + int dist = 0; - 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") ); + if ( i1 < i2 ) + { + while ( i1 != i2 ) + { + IncIter(i1); + dist--; + } + } + else if ( i2 < i1 ) + { + while ( i2 != i1 ) + { + IncIter(i2); + dist++; + } } - 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'; - - return *this; -} - -void wxStringBase::swap(wxStringBase& str) -{ - wxChar* tmp = str.m_pchData; - str.m_pchData = m_pchData; - m_pchData = tmp; + return dist; } -size_t wxStringBase::find(const wxStringBase& str, size_t nStart) const -{ - // deal with the special case of empty string first - const size_t nLen = length(); - const size_t nLenOther = str.length(); - - if ( !nLenOther ) +/* static */ +wxString::Utf8CharBuffer wxString::EncodeChar(wxUniChar ch) +{ + Utf8CharBuffer buf; + char *out = buf.data; + + wxUniChar::value_type code = ch.GetValue(); + + // Char. number range | UTF-8 octet sequence + // (hexadecimal) | (binary) + // ----------------------+--------------------------------------------- + // 0000 0000 - 0000 007F | 0xxxxxxx + // 0000 0080 - 0000 07FF | 110xxxxx 10xxxxxx + // 0000 0800 - 0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx + // 0001 0000 - 0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx + // + // Code point value is stored in bits marked with 'x', lowest-order bit + // of the value on the right side in the diagram above. + // (from RFC 3629) + + if ( code <= 0x7F ) { - // empty string is a substring of anything - return 0; + out[1] = 0; + out[0] = (char)code; } - - if ( !nLen ) + else if ( code <= 0x07FF ) { - // the other string is non empty so can't be our substring - return npos; + out[2] = 0; + // NB: this line takes 6 least significant bits, encodes them as + // 10xxxxxx and discards them so that the next byte can be encoded: + out[1] = 0x80 | (code & 0x3F); code >>= 6; + out[0] = 0xC0 | code; } - - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart <= nLen ); - - const wxChar * const other = str.c_str(); - - // anchor - const wxChar* p = (const wxChar*)wxTmemchr(c_str() + nStart, - *other, - nLen - nStart); - - if ( !p ) - return npos; - - while ( p - c_str() + nLenOther <= nLen && wxTmemcmp(p, other, nLenOther) ) + else if ( code < 0xFFFF ) { - p++; - - // anchor again - p = (const wxChar*)wxTmemchr(p, *other, nLen - (p - c_str())); - - if ( !p ) - return npos; + out[3] = 0; + out[2] = 0x80 | (code & 0x3F); code >>= 6; + out[1] = 0x80 | (code & 0x3F); code >>= 6; + out[0] = 0xE0 | code; + } + else if ( code <= 0x10FFFF ) + { + out[4] = 0; + out[3] = 0x80 | (code & 0x3F); code >>= 6; + out[2] = 0x80 | (code & 0x3F); code >>= 6; + out[1] = 0x80 | (code & 0x3F); code >>= 6; + out[0] = 0xF0 | code; + } + else + { + wxFAIL_MSG( _T("trying to encode undefined Unicode character") ); + out[0] = 0; } - return p - c_str() + nLenOther <= nLen ? p - c_str() : npos; + return buf; } -size_t wxStringBase::find(const wxChar* sz, size_t nStart, size_t n) const -{ - return find(wxStringBase(sz, n), nStart); -} +/* static */ +wxUniChar wxUniCharRef::DecodeChar(wxStringImpl::const_iterator i) +{ + wxASSERT( wxString::IsValidUtf8LeadByte(*i) ); // FIXME-UTF8: no "wxString::" + + wxUniChar::value_type code = 0; + size_t len = wxString::GetUtf8CharLength(*i); + wxASSERT_MSG( len <= 4, _T("invalid UTF-8 sequence length") ); + + // Char. number range | UTF-8 octet sequence + // (hexadecimal) | (binary) + // ----------------------+--------------------------------------------- + // 0000 0000 - 0000 007F | 0xxxxxxx + // 0000 0080 - 0000 07FF | 110xxxxx 10xxxxxx + // 0000 0800 - 0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx + // 0001 0000 - 0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx + // + // Code point value is stored in bits marked with 'x', lowest-order bit + // of the value on the right side in the diagram above. + // (from RFC 3629) + + // mask to extract lead byte's value ('x' bits above), by sequence's length: + static const unsigned char s_leadValueMask[4] = { 0x7F, 0x1F, 0x0F, 0x07 }; +#ifdef __WXDEBUG__ + // mask and value of lead byte's most significant bits, by length: + static const unsigned char s_leadMarkerMask[4] = { 0x80, 0xE0, 0xF0, 0xF8 }; + static const unsigned char s_leadMarkerVal[4] = { 0x00, 0xC0, 0xE0, 0xF0 }; +#endif -size_t wxStringBase::find(wxUniChar ch, size_t nStart) const -{ - wxASSERT( nStart <= length() ); + // extract the lead byte's value bits: + wxASSERT_MSG( ((unsigned char)*i & s_leadMarkerMask[len-1]) == + s_leadMarkerVal[len-1], + _T("invalid UTF-8 lead byte") ); + code = (unsigned char)*i & s_leadValueMask[len-1]; - const wxChar *p = (const wxChar*)wxTmemchr(c_str() + nStart, ch, length() - nStart); + // all remaining bytes, if any, are handled in the same way regardless of + // sequence's length: + for ( ++i ; len > 1; --len, ++i ) + { + wxASSERT_MSG( ((unsigned char)*i & 0xC0) == 0x80, + _T("invalid UTF-8 byte") ); + + code <<= 6; + code |= (unsigned char)*i & 0x3F; + } - return p == NULL ? npos : p - c_str(); + return wxUniChar(code); } -size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const +/* static */ +wxCharBuffer wxString::EncodeNChars(size_t n, wxUniChar ch) { - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart == npos || nStart <= length() ); + Utf8CharBuffer once(EncodeChar(ch)); + // the IncIter() table can be used to determine the length of ch's encoding: + size_t len = ms_utf8IterTable[(unsigned char)once.data[0]]; - if ( length() >= str.length() ) + wxCharBuffer buf(n * len); + char *ptr = buf.data(); + for ( size_t i = 0; i < n; i++, ptr += len ) { - // avoids a corner case later - if ( length() == 0 && str.length() == 0 ) - return 0; - - // "top" is the point where search starts from - size_t top = length() - str.length(); - - if ( nStart == npos ) - nStart = length() - 1; - if ( nStart < top ) - top = nStart; - - const wxChar *cursor = c_str() + top; - do - { - if ( wxTmemcmp(cursor, str.c_str(), - str.length()) == 0 ) - { - return cursor - c_str(); - } - } while ( cursor-- > c_str() ); + memcpy(ptr, once.data, len); } - return npos; + return buf; } -size_t wxStringBase::rfind(const wxChar* sz, size_t nStart, size_t n) const -{ - return rfind(wxStringBase(sz, n), nStart); -} -size_t wxStringBase::rfind(wxUniChar ch, size_t nStart) const +void wxString::PosLenToImpl(size_t pos, size_t len, + size_t *implPos, size_t *implLen) const { - if ( nStart == npos ) - { - nStart = length(); - } + if ( pos == npos ) + *implPos = npos; else { - wxASSERT( nStart <= length() ); - } + 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; - 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(); + *implLen = (i + len).impl() - i.impl(); + } } - - return npos; } -size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart) const -{ - wxASSERT(nStart <= length()); +#endif // wxUSE_UNICODE_UTF8 - size_t len = wxStrlen(sz); +// ---------------------------------------------------------------------------- +// wxCStrData converted strings caching +// ---------------------------------------------------------------------------- - size_t i; - for(i = nStart; i < this->length(); ++i) +// 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() ) { - if (wxTmemchr(sz, *(c_str() + i), len)) - break; + free(i->second); + hash.erase(i); } - - if(i == this->length()) - return npos; - else - return i; } -size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart, - size_t n) const -{ - return find_first_of(wxStringBase(sz, n), nStart); -} +#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; -size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart) const +const char* wxCStrData::AsChar() const { - if ( nStart == npos ) - { - nStart = length() - 1; - } - else - { - wxASSERT_MSG( nStart <= length(), - _T("invalid index in find_last_of()") ); - } + // remove previously cache value, if any (see FIXMEs above): + DeleteStringFromConversionCache(gs_stringsCharCache, m_str); - size_t len = wxStrlen(sz); + // convert the string and keep it: + const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] = + m_str->mb_str().release(); - for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p ) - { - if ( wxTmemchr(sz, *p, len) ) - return p - c_str(); - } - - return npos; + return s + m_offset; } +#endif // wxUSE_UNICODE -size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart, - size_t n) const -{ - return find_last_of(wxStringBase(sz, n), nStart); -} +#if !wxUSE_UNICODE_WCHAR +WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual, + wxStringWCharConversionCache); +static wxStringWCharConversionCache gs_stringsWCharCache; -size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart) const +const wchar_t* wxCStrData::AsWChar() const { - if ( nStart == npos ) - { - nStart = length(); - } - else - { - wxASSERT( nStart <= length() ); - } - - size_t len = wxStrlen(sz); + // remove previously cache value, if any (see FIXMEs above): + DeleteStringFromConversionCache(gs_stringsWCharCache, m_str); - size_t i; - for(i = nStart; i < this->length(); ++i) - { - if (!wxTmemchr(sz, *(c_str() + i), len)) - break; - } + // convert the string and keep it: + const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] = + m_str->wc_str().release(); - if(i == this->length()) - return npos; - else - return i; + return s + m_offset; } +#endif // !wxUSE_UNICODE_WCHAR -size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart, - size_t n) const +wxString::~wxString() { - return find_first_not_of(wxStringBase(sz, n), nStart); +#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 -size_t wxStringBase::find_first_not_of(wxUniChar ch, size_t nStart) const +#if wxUSE_UNICODE +const char* wxCStrData::AsChar() const { - wxASSERT( nStart <= length() ); - - for ( const_iterator p = begin() + nStart; (bool)*p; ++p ) // FIXME-DMARS - { - if ( *p != ch ) - return p - begin(); - } + wxString *str = wxConstCast(m_str, wxString); - return npos; -} + // convert the string: + wxCharBuffer buf(str->mb_str()); -size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart) const -{ - if ( nStart == npos ) + // FIXME-UTF8: do the conversion in-place in the existing buffer + if ( str->m_convertedToChar && + strlen(buf) == strlen(str->m_convertedToChar) ) { - nStart = length() - 1; + // 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 { - wxASSERT( nStart <= length() ); - } - - size_t len = wxStrlen(sz); - - for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p ) - { - if ( !wxTmemchr(sz, *p,len) ) - return p - c_str(); + str->m_convertedToChar = buf.release(); } - return npos; + // and keep it: + return str->m_convertedToChar + m_offset; } +#endif // wxUSE_UNICODE -size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart, - size_t n) const +#if !wxUSE_UNICODE_WCHAR +const wchar_t* wxCStrData::AsWChar() const { - return find_last_not_of(wxStringBase(sz, n), nStart); -} + wxString *str = wxConstCast(m_str, wxString); -size_t wxStringBase::find_last_not_of(wxUniChar ch, size_t nStart) const -{ - if ( nStart == npos ) + // convert the string: + wxWCharBuffer buf(str->wc_str()); + + // FIXME-UTF8: do the conversion in-place in the existing buffer + if ( str->m_convertedToWChar && + wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) ) { - nStart = length() - 1; + // 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 { - wxASSERT( nStart <= length() ); + str->m_convertedToWChar = buf.release(); } - for ( const_iterator p = begin() + nStart; p != begin(); --p ) - { - if ( *p != ch ) - return p - begin(); - } + // and keep it: + return str->m_convertedToWChar + m_offset; +} +#endif // !wxUSE_UNICODE_WCHAR - return npos; +// =========================================================================== +// wxString class core +// =========================================================================== + +// --------------------------------------------------------------------------- +// construction and conversion +// --------------------------------------------------------------------------- + +#if wxUSE_UNICODE_WCHAR +/* static */ +wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv) +{ + // anything to do? + if ( !psz || nLength == 0 ) + return SubstrBufFromMB(L"", 0); + + if ( nLength == npos ) + nLength = wxNO_LEN; + + 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 -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxChar *sz) +#if wxUSE_UNICODE_UTF8 +/* static */ +wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv) { - wxASSERT_MSG( nStart <= length(), - _T("index out of bounds in wxStringBase::replace") ); - size_t strLen = length() - nStart; - nLen = strLen < nLen ? strLen : nLen; + // FIXME-UTF8: return as-is without copying under UTF8 locale, return + // converted string under other locales - needs wxCharBuffer + // changes - wxStringBase strTmp; - strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs + // anything to do? + if ( !psz || nLength == 0 ) + return SubstrBufFromMB("", 0); - //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]); + if ( nLength == npos ) + nLength = wxNO_LEN; - //its safe to do the full version here because - //sz must be a normal c string - strTmp.append(sz); + // first convert to wide string: + size_t wcLen; + wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen)); + if ( !wcLen ) + return SubstrBufFromMB("", 0); - for(size_t i2 = nStart + nLen; i2 < length(); ++i2) - strTmp.append(1, this->c_str()[i2]); + // and then to UTF-8: + SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxConvUTF8)); + // widechar -> UTF-8 conversion isn't supposed to ever fail: + wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") ); - swap(strTmp); - return *this; + return buf; } +#endif // wxUSE_UNICODE_UTF8 -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - size_t nCount, wxUniChar ch) +#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE +/* static */ +wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength, + const wxMBConv& conv) { - return replace(nStart, nLen, wxStringBase(nCount, ch).c_str()); + // anything to do? + if ( !pwz || nLength == 0 ) + return SubstrBufFromWC("", 0); + + if ( nLength == npos ) + nLength = wxNO_LEN; + + 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 -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxStringBase& str, - size_t nStart2, size_t nLen2) + +#if wxUSE_UNICODE_WCHAR + +//Convert wxString in Unicode mode to a multi-byte string +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - return replace(nStart, nLen, str.substr(nStart2, nLen2)); + return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL); } -wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen, - const wxChar* sz, size_t nCount) +#elif wxUSE_UNICODE_UTF8 + +const wxWCharBuffer wxString::wc_str() const { - return replace(nStart, nLen, wxStringBase(sz, nCount).c_str()); + return wxConvUTF8.cMB2WC(m_impl.c_str(), + m_impl.length() + 1 /* size, not length */, + NULL); } -wxStringBase wxStringBase::substr(size_t nStart, size_t nLen) const +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - if ( nLen == npos ) - nLen = length() - nStart; - return wxStringBase(*this, nStart, nLen); + // FIXME-UTF8: optimize the case when conv==wxConvUTF8 or wxConvLibc + // under UTF8 locale + // FIXME-UTF8: use wc_str() here once we have buffers with length + + size_t wcLen; + wxWCharBuffer wcBuf( + wxConvUTF8.cMB2WC(m_impl.c_str(), + m_impl.length() + 1 /* size, not length */, + &wcLen)); + if ( !wcLen ) + return wxCharBuffer(""); + + return conv.cWC2MB(wcBuf, wcLen, NULL); } -// assigns one string to another -wxStringBase& wxStringBase::operator=(const wxStringBase& stringSrc) +#else // ANSI + +//Converts this string to a wide character string if unicode +//mode is not enabled and wxUSE_WCHAR_T is enabled +const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const { - wxASSERT( stringSrc.GetStringData()->IsValid() ); + return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL); +} - // 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(); - } - } +#endif // Unicode/ANSI - return *this; +// shrink to minimal size (releasing extra memory) +bool wxString::Shrink() +{ + wxString tmp(begin(), end()); + swap(tmp); + return tmp.length() == length(); } -// assigns a single character -wxStringBase& wxStringBase::operator=(wxUniChar ch) +// deprecated compatibility code: +#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +wxChar *wxString::GetWriteBuf(size_t nLen) { - wxChar c(ch); - if ( !AssignCopy(1, &c) ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(wxChar)") ); - } - return *this; + return DoGetWriteBuf(nLen); } -// assigns C string -wxStringBase& wxStringBase::operator=(const wxChar *psz) +void wxString::UngetWriteBuf() { - if ( !AssignCopy(wxStrlen(psz), psz) ) { - wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(const wxChar *)") ); - } - return *this; + DoUngetWriteBuf(); } -// helper function: does real copy -bool wxStringBase::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData) +void wxString::UngetWriteBuf(size_t nLen) { - 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; + DoUngetWriteBuf(nLen); } +#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 + // --------------------------------------------------------------------------- -// string concatenation +// data access // --------------------------------------------------------------------------- -// 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; +// all functions are inline in string.h - // concatenating an empty string is a NOP - if ( nSrcLen > 0 ) { - wxStringData *pData = GetStringData(); - size_t nLen = pData->nDataLength; - size_t nNewLen = nLen + nSrcLen; +// --------------------------------------------------------------------------- +// concatenation operators +// --------------------------------------------------------------------------- - // alloc new buffer if current is too small - if ( pData->IsShared() ) { - STATISTICS_ADD(ConcatHit, 0); +/* + * concatenation functions come in 5 flavours: + * string + string + * char + string and string + char + * C str + string and string + C str + */ - // 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); +wxString operator+(const wxString& str1, const wxString& str2) +{ +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str1.IsValid() ); + wxASSERT( str2.IsValid() ); +#endif - reserve(nNewLen); - // we have to grow the buffer - if ( capacity() < nNewLen ) { - // allocation failure handled by caller - return false; - } - } - else { - STATISTICS_ADD(ConcatHit, 1); + wxString s = str1; + s += str2; - // the buffer is already big enough - } + return s; +} - // should be enough space - wxASSERT( nNewLen <= GetStringData()->nAllocLength ); +wxString operator+(const wxString& str, wxUniChar ch) +{ +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - // fast concatenation - all is done in our buffer - memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar)); + wxString s = str; + s += ch; - 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 s; } -// --------------------------------------------------------------------------- -// simple sub-string extraction -// --------------------------------------------------------------------------- - -// helper function: clone the data attached to this string -bool wxStringBase::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const +wxString operator+(wxUniChar ch, const wxString& str) { - 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; -} +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif -#endif // !wxUSE_STL_BASED_WXSTRING + wxString s = ch; + s += str; -#if !wxUSE_STL_BASED_WXSTRING || !defined(HAVE_STD_STRING_COMPARE) + return s; +} +wxString operator+(const wxString& str, const char *psz) +{ #if !wxUSE_STL_BASED_WXSTRING - #define STRINGCLASS wxStringBase -#else - #define STRINGCLASS wxString + wxASSERT( str.IsValid() ); #endif -static inline int wxDoCmp(const wxChar* s1, size_t l1, - const wxChar* s2, size_t l2) -{ - if( l1 == l2 ) - return wxTmemcmp(s1, s2, l1); - else if( l1 < l2 ) - { - int ret = wxTmemcmp(s1, s2, l1); - return ret == 0 ? -1 : ret; - } - else - { - int ret = wxTmemcmp(s1, s2, l2); - return ret == 0 ? +1 : ret; + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); } -} + s += str; + s += psz; -int STRINGCLASS::compare(const wxStringBase& str) const -{ - return ::wxDoCmp(data(), length(), str.data(), str.length()); + return s; } -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxStringBase& str) const +wxString operator+(const wxString& str, const wchar_t *pwz) { - wxASSERT(nStart <= length()); - size_type strLen = length() - nStart; - nLen = strLen < nLen ? strLen : nLen; - return ::wxDoCmp(data() + nStart, nLen, str.data(), str.length()); -} +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxStringBase& str, - size_t nStart2, size_t nLen2) const -{ - wxASSERT(nStart <= length()); - wxASSERT(nStart2 <= str.length()); - size_type strLen = length() - nStart, - strLen2 = str.length() - nStart2; - nLen = strLen < nLen ? strLen : nLen; - nLen2 = strLen2 < nLen2 ? strLen2 : nLen2; - return ::wxDoCmp(data() + nStart, nLen, str.data() + nStart2, nLen2); -} + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s += str; + s += pwz; -int STRINGCLASS::compare(const wxChar* sz) const -{ - size_t nLen = wxStrlen(sz); - return ::wxDoCmp(data(), length(), sz, nLen); + return s; } -int STRINGCLASS::compare(size_t nStart, size_t nLen, - const wxChar* sz, size_t nCount) const +wxString operator+(const char *psz, const wxString& str) { - wxASSERT(nStart <= length()); - size_type strLen = length() - nStart; - nLen = strLen < nLen ? strLen : nLen; - if( nCount == npos ) - nCount = wxStrlen(sz); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - return ::wxDoCmp(data() + nStart, nLen, sz, nCount); + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = psz; + s += str; + + return s; } -#undef STRINGCLASS +wxString operator+(const wchar_t *pwz, const wxString& str) +{ +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif -#endif // !wxUSE_STL_BASED_WXSTRING || !defined(HAVE_STD_STRING_COMPARE) + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = pwz; + s += str; -// =========================================================================== -// wxString class core -// =========================================================================== + return s; +} // --------------------------------------------------------------------------- -// construction and conversion +// string comparison // --------------------------------------------------------------------------- -#if wxUSE_UNICODE +#ifdef HAVE_STD_STRING_COMPARE -// from multibyte string -wxString::wxString(const char *psz, const wxMBConv& conv, size_t nLength) -{ - // anything to do? - if ( psz && nLength != 0 ) - { - if ( nLength == npos ) - { - nLength = wxNO_LEN; - } +// 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). - size_t nLenWide; - wxWCharBuffer wbuf = conv.cMB2WC(psz, nLength, &nLenWide); +int wxString::compare(const wxString& str) const +{ + return m_impl.compare(str.m_impl); +} - if ( nLenWide ) - assign(wbuf, nLenWide); - } +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str) const +{ + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + return m_impl.compare(pos, len, str.m_impl); } -//Convert wxString in Unicode mode to a multi-byte string -const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { - return conv.cWC2MB(c_str(), length() + 1 /* size, not length */, NULL); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); + + return m_impl.compare(pos, len, str.m_impl, pos2, len2); } -#else // ANSI +int wxString::compare(const char* sz) const +{ + return m_impl.compare(ImplStr(sz)); +} -#if wxUSE_WCHAR_T +int wxString::compare(const wchar_t* sz) const +{ + return m_impl.compare(ImplStr(sz)); +} -// from wide string -wxString::wxString(const wchar_t *pwz, const wxMBConv& conv, size_t nLength) +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const { - // anything to do? - if ( pwz && nLength != 0 ) - { - if ( nLength == npos ) - { - nLength = wxNO_LEN; - } + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - size_t nLenMB; - wxCharBuffer buf = conv.cWC2MB(pwz, nLength, &nLenMB); + SubstrBufFromMB str(ImplStr(sz, nCount)); - if ( nLenMB ) - assign(buf, nLenMB); - } + return m_impl.compare(pos, len, str.data, str.len); } -//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 +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const { - return conv.cMB2WC(c_str(), length() + 1 /* size, not length */, NULL); -} + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); -#endif // wxUSE_WCHAR_T + SubstrBufFromWC str(ImplStr(sz, nCount)); -#endif // Unicode/ANSI + return m_impl.compare(pos, len, str.data, str.len); +} -// shrink to minimal size (releasing extra memory) -bool wxString::Shrink() +#else // !HAVE_STD_STRING_COMPARE + +static inline int wxDoCmp(const wxStringCharType* s1, size_t l1, + const wxStringCharType* s2, size_t l2) { - wxString tmp(begin(), end()); - swap(tmp); - return tmp.length() == length(); + if( l1 == l2 ) + return wxStringMemcmp(s1, s2, l1); + else if( l1 < l2 ) + { + int ret = wxStringMemcmp(s1, s2, l1); + return ret == 0 ? -1 : ret; + } + else + { + int ret = wxStringMemcmp(s1, s2, l2); + return ret == 0 ? +1 : ret; + } } -#if !wxUSE_STL_BASED_WXSTRING -// get the pointer to writable buffer of (at least) nLen bytes -wxChar *wxString::DoGetWriteBuf(size_t nLen) +int wxString::compare(const wxString& str) const { - if ( !AllocBeforeWrite(nLen) ) { - // allocation failure handled by caller - return NULL; - } - - wxASSERT( GetStringData()->nRefs == 1 ); - GetStringData()->Validate(false); - - return m_pchData; + return ::wxDoCmp(m_impl.data(), m_impl.length(), + str.m_impl.data(), str.m_impl.length()); } -// put string back in a reasonable state after GetWriteBuf -void wxString::DoUngetWriteBuf() +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str) const { - DoUngetWriteBuf(wxStrlen(m_pchData)); + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data(), str.m_impl.length()); } -void wxString::DoUngetWriteBuf(size_t nLen) +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { - wxStringData * const pData = GetStringData(); + wxASSERT(nStart <= length()); + wxASSERT(nStart2 <= str.length()); + size_type strLen = length() - nStart, + strLen2 = str.length() - nStart2; + nLen = strLen < nLen ? strLen : nLen; + nLen2 = strLen2 < nLen2 ? strLen2 : nLen2; - wxASSERT_MSG( nLen < pData->nAllocLength, _T("buffer overrun") ); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); - // the strings we store are always NUL-terminated - pData->data()[nLen] = _T('\0'); - pData->nDataLength = nLen; - pData->Validate(true); + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data() + pos2, len2); } -// deprecated compatibility code: -#if WXWIN_COMPATIBILITY_2_8 -wxChar *wxString::GetWriteBuf(size_t nLen) +int wxString::compare(const char* sz) const { - return DoGetWriteBuf(nLen); + 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); } -void wxString::UngetWriteBuf() +int wxString::compare(const wchar_t* sz) const { - DoUngetWriteBuf(); + 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); } -void wxString::UngetWriteBuf(size_t nLen) +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const { - DoUngetWriteBuf(nLen); + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + SubstrBufFromMB str(ImplStr(sz, nCount)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); + + return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len); } -#endif // WXWIN_COMPATIBILITY_2_8 -#endif // !wxUSE_STL +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const +{ + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); -// --------------------------------------------------------------------------- -// data access -// --------------------------------------------------------------------------- + SubstrBufFromWC str(ImplStr(sz, nCount)); + if ( str.len == npos ) + str.len = wxStringStrlen(str.data); + + return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len); +} + +#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE -// all functions are inline in string.h // --------------------------------------------------------------------------- -// assignment operators +// find_{first,last}_[not]_of functions // --------------------------------------------------------------------------- -#if !wxUSE_UNICODE +#if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 + +// 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. -// same as 'signed char' variant -wxString& wxString::operator=(const unsigned char* psz) +size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const { - *this = (const char *)psz; - return *this; + return find_first_of(sz, nStart, wxStrlen(sz)); } -#if wxUSE_WCHAR_T -wxString& wxString::operator=(const wchar_t *pwz) +size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const { - wxString str(pwz); - swap(str); - return *this; + return find_first_not_of(sz, nStart, wxStrlen(sz)); } -#endif -#endif - -/* - * concatenation functions come in 5 flavours: - * string + string - * char + string and string + char - * C str + string and string + C str - */ - -wxString operator+(const wxString& str1, const wxString& str2) +size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const { -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str1.GetStringData()->IsValid() ); - wxASSERT( str2.GetStringData()->IsValid() ); -#endif + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); - wxString s = str1; - s += str2; + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++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(const wxChar* sz, size_t nStart, size_t n) 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 ( !wxTmemchr(sz, *i, n) ) + return idx; + } - return s; + return npos; } -wxString operator+(wxUniChar ch, const wxString& str) -{ -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif - wxString s = ch; - s += str; +size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const +{ + return find_last_of(sz, nStart, wxStrlen(sz)); +} - return s; +size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const +{ + return find_last_not_of(sz, nStart, wxStrlen(sz)); } -wxString operator+(const wxString& str, const wxChar *psz) +size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const { -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif + size_t len = length(); - wxString s; - if ( !s.Alloc(wxStrlen(psz) + str.length()) ) { - wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); } - s += str; - s += psz; - return s; + 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 npos; } -wxString operator+(const wxChar *psz, const wxString& str) +size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const { -#if !wxUSE_STL_BASED_WXSTRING - wxASSERT( str.GetStringData()->IsValid() ); -#endif + size_t len = length(); - wxString s; - if ( !s.Alloc(wxStrlen(psz) + str.length()) ) { - wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); } - s = psz; - s += str; - - return s; -} -// =========================================================================== -// other common string functions -// =========================================================================== + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) + { + if ( !wxTmemchr(sz, *i, n) ) + return idx; + } -int wxString::Cmp(const wxString& s) const -{ - return compare(s); + return npos; } -int wxString::Cmp(const wxChar* psz) const +size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const { - return compare(psz); + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); + + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( *i != ch ) + return idx; + } + + return npos; } -static inline int wxDoCmpNoCase(const wxChar* s1, size_t l1, - const wxChar* s2, size_t l2) +size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const { - size_t i; + size_t len = length(); - if( l1 == l2 ) + if ( nStart == npos ) { - 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; + nStart = len - 1; } - else if( l1 < l2 ) + else { - 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; + wxASSERT_MSG( nStart <= len, _T("invalid index") ); } - else + + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) { - 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; + if ( *i != ch ) + return idx; } + + return npos; } +// 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 + +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); } + +#undef wxOtherCharType +#undef STRCONV + +#endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 + +// =========================================================================== +// other common string functions +// =========================================================================== + int wxString::CmpNoCase(const wxString& s) const { - return wxDoCmpNoCase(data(), length(), s.data(), s.length()); -} + // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added -int wxString::CmpNoCase(const wxChar* psz) const -{ - int nLen = wxStrlen(psz); + size_t idx = 0; + const_iterator i1 = begin(); + const_iterator end1 = end(); + const_iterator i2 = s.begin(); + const_iterator end2 = s.end(); + + for ( ; i1 != end1 && i2 != end2; ++idx, ++i1, ++i2 ) + { + wxUniChar lower1 = (wxChar)wxTolower(*i1); + wxUniChar lower2 = (wxChar)wxTolower(*i2); + if ( lower1 != lower2 ) + return lower1 < lower2 ? -1 : 1; + } - return wxDoCmpNoCase(data(), length(), psz, nLen); + size_t len1 = length(); + size_t len2 = s.length(); + + if ( len1 < len2 ) + return -1; + else if ( len1 > len2 ) + return 1; + return 0; } @@ -1420,7 +1471,8 @@ bool wxString::EndsWith(const wxChar *suffix, wxString *rest) const wxASSERT_MSG( suffix, _T("invalid parameter in wxString::EndssWith") ); int start = length() - wxStrlen(suffix); - if ( start < 0 || wxStrcmp(wx_str() + start, suffix) != 0 ) + + if ( start < 0 || compare(start, npos, suffix) != 0 ) return false; if ( rest ) @@ -1507,32 +1559,32 @@ 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 +1603,43 @@ 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 @@ -1697,14 +1758,6 @@ 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 // ---------------------------------------------------------------------------- @@ -1749,26 +1802,12 @@ bool wxString::ToULong(unsigned long *val, int base) const bool wxString::ToLongLong(wxLongLong_t *val, int base) const { -#ifdef wxHAS_STRTOLL return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoll); -#else - // TODO: implement this ourselves - wxUnusedVar(val); - wxUnusedVar(base); - return false; -#endif // wxHAS_STRTOLL } bool wxString::ToULongLong(wxULongLong_t *val, int base) const { -#ifdef wxHAS_STRTOLL return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoull); -#else - // TODO: implement this ourselves - wxUnusedVar(val); - wxUnusedVar(base); - return false; -#endif } bool wxString::ToDouble(double *val) const @@ -1866,7 +1905,7 @@ int wxString::PrintfV(const wxString& format, va_list argptr) // only a copy va_list argptrcopy; wxVaCopy(argptrcopy, argptr); - int len = wxVsnprintf(buf, size, format, argptrcopy); + int len = wxVsnprintf(buf, size, (const wxChar*)/*FIXME-UTF8*/format, argptrcopy); va_end(argptrcopy); // some implementations of vsnprintf() don't NUL terminate @@ -1922,7 +1961,7 @@ 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 +2012,17 @@ 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 +2109,9 @@ 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; @@ -2076,557 +2123,3 @@ 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 -// ---------------------------------------------------------------------------- - -// 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) -{ - 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; - } - } - } - - str.Shrink(); // release extra memory if we allocated too much - return str; -} - -wxArrayString wxSplit(const wxString& str, const wxChar sep, const wxChar escape) -{ - 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; - } - - // add the last token - if ( !curr.empty() || prev == sep ) - ret.Add(curr); - - return ret; -}