X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/d71fa6fb2e2421b9d8deb826d60d5d8636665d91..6e040042bae080a3fe8907e3772c103deab91c13:/src/common/string.cpp diff --git a/src/common/string.cpp b/src/common/string.cpp index efe9a4e186..beb618873d 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -38,6 +38,8 @@ #include "wx/thread.h" #endif +#include "wx/regex.h" // for wxString::Matches() + #include #include #include @@ -64,15 +66,6 @@ // NB: EXTRA_ALLOC must be >= 0! #define EXTRA_ALLOC (19 - nLen % 16) -#if defined(__VISAGECPP__) && __IBMCPP__ >= 400 -// must define this in .cpp for VA or else you get multiply defined symbols everywhere - -// maximum possible length for a string means "take all string" everywhere -// (as sizeof(StringData) is unknown here, we substract 100) -const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100; - -#endif - // --------------------------------------------------------------------------- // static class variables definition // --------------------------------------------------------------------------- @@ -94,6 +87,12 @@ static const struct wxChar dummy; } g_strEmpty = { {-1, 0, 0}, wxT('\0') }; +#if defined(__VISAGECPP__) && __IBMCPP__ >= 400 +// must define this static for VA or else you get multiply defined symbols +// everywhere +const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100; +#endif // Visual Age + // empty C style string: points to 'string data' byte of g_strEmpty extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy; @@ -116,9 +115,11 @@ extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy; // function wxVsnprintfA (A for ANSI), should also find one for Unicode // strings in Unicode build #ifdef __WXMSW__ - #if (defined(__VISUALC__) || defined(wxUSE_NORLANDER_HEADERS)) && !defined(__MINGW32__) + #if defined(__VISUALC__) || (defined(__MINGW32__) && wxUSE_NORLANDER_HEADERS) #define wxVsnprintfA _vsnprintf #endif +#elif defined(__WXMAC__) + #define wxVsnprintfA vsnprintf #else // !Windows #ifdef HAVE_VSNPRINTF #define wxVsnprintfA vsnprintf @@ -133,9 +134,7 @@ extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy; #if defined(__VISUALC__) #pragma message("Using sprintf() because no snprintf()-like function defined") - #elif defined(__GNUG__) && !defined(__UNIX__) - #warning "Using sprintf() because no snprintf()-like function defined" - #elif defined(__MWERKS__) + #elif defined(__GNUG__) #warning "Using sprintf() because no snprintf()-like function defined" #endif //compiler #endif // no vsnprintf @@ -156,7 +155,7 @@ extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy; // // ATTN: you can _not_ use both of these in the same program! -istream& operator>>(istream& is, wxString& WXUNUSED(str)) +wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str)) { #if 0 int w = is.width(0); @@ -187,7 +186,7 @@ istream& operator>>(istream& is, wxString& WXUNUSED(str)) return is; } -ostream& operator<<(ostream& os, const wxString& str) +wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) { os << str.c_str(); return os; @@ -204,7 +203,8 @@ extern int WXDLLEXPORT wxVsnprintf(wxChar *buf, size_t len, int iLen = s.PrintfV(format, argptr); if ( iLen != -1 ) { - wxStrncpy(buf, s.c_str(), iLen); + wxStrncpy(buf, s.c_str(), len); + buf[len-1] = wxT('\0'); } return iLen; @@ -293,10 +293,12 @@ void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength) { Init(); - wxASSERT( nPos <= wxStrlen(psz) ); + // if the length is not given, assume the string to be NUL terminated + if ( nLength == wxSTRING_MAXLEN ) { + wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") ); - if ( nLength == wxSTRING_MAXLEN ) nLength = wxStrlen(psz + nPos); + } STATISTICS_ADD(InitialLength, nLength); @@ -344,15 +346,15 @@ wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength) #if wxUSE_WCHAR_T // from wide string -wxString::wxString(const wchar_t *pwz) +wxString::wxString(const wchar_t *pwz, wxMBConv& conv) { // first get necessary size - size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0; + size_t nLen = pwz ? conv.WC2MB((char *) NULL, pwz, 0) : 0; // empty? if ( (nLen != 0) && (nLen != (size_t)-1) ) { AllocBuffer(nLen); - wxWC2MB(m_pchData, pwz, nLen); + conv.WC2MB(m_pchData, pwz, nLen); } else { Init(); @@ -531,6 +533,12 @@ void wxString::UngetWriteBuf() GetStringData()->Validate(TRUE); } +void wxString::UngetWriteBuf(size_t nLen) +{ + GetStringData()->nDataLength = nLen; + GetStringData()->Validate(TRUE); +} + // --------------------------------------------------------------------------- // data access // --------------------------------------------------------------------------- @@ -771,6 +779,35 @@ wxString wxString::Mid(size_t nFirst, size_t nCount) const return dest; } +// check that the tring starts with prefix and return the rest of the string +// in the provided pointer if it is not NULL, otherwise return FALSE +bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const +{ + wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") ); + + // first check if the beginning of the string matches the prefix: note + // that we don't have to check that we don't run out of this string as + // when we reach the terminating NUL, either prefix string ends too (and + // then it's ok) or we break out of the loop because there is no match + const wxChar *p = c_str(); + while ( *prefix ) + { + if ( *prefix++ != *p++ ) + { + // no match + return FALSE; + } + } + + if ( rest ) + { + // put the rest of the string into provided pointer + *rest = p; + } + + return TRUE; +} + // extract nCount last (rightmost) characters wxString wxString::Right(size_t nCount) const { @@ -907,6 +944,8 @@ bool wxString::IsWord() const bool wxString::IsNumber() const { const wxChar *s = (const wxChar*) *this; + if (wxStrlen(s)) + if ((s[0] == '-') || (s[0] == '+')) s++; while(*s){ if(!wxIsdigit(*s)) return(FALSE); s++; @@ -950,14 +989,22 @@ wxString& wxString::MakeLower() // trimming and padding // --------------------------------------------------------------------------- +// some compilers (VC++ 6.0 not to name them) return TRUE for a call to +// isspace('ê') in the C locale which seems to be broken to me, but we have to +// live with this by checking that the character is a 7 bit one - even if this +// may fail to detect some spaces (I don't know if Unicode doesn't have +// space-like symbols somewhere except in the first 128 chars), it is arguably +// still better than trimming away accented letters +inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); } + // trims spaces (in the sense of isspace) from left or right side wxString& wxString::Trim(bool bFromRight) { // first check if we're going to modify the string at all if ( !IsEmpty() && ( - (bFromRight && wxIsspace(GetChar(Len() - 1))) || - (!bFromRight && wxIsspace(GetChar(0u))) + (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) || + (!bFromRight && wxSafeIsspace(GetChar(0u))) ) ) { @@ -968,7 +1015,7 @@ wxString& wxString::Trim(bool bFromRight) { // find last non-space character wxChar *psz = m_pchData + GetStringData()->nDataLength - 1; - while ( wxIsspace(*psz) && (psz >= m_pchData) ) + while ( wxSafeIsspace(*psz) && (psz >= m_pchData) ) psz--; // truncate at trailing space start @@ -979,7 +1026,7 @@ wxString& wxString::Trim(bool bFromRight) { // find first non-space character const wxChar *psz = m_pchData; - while ( wxIsspace(*psz) ) + while ( wxSafeIsspace(*psz) ) psz++; // fix up data and length @@ -1107,7 +1154,7 @@ wxString wxString::Format(const wxChar *pszFormat, ...) wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr) { wxString s; - s.Printf(pszFormat, argptr); + s.PrintfV(pszFormat, argptr); return s; } @@ -1366,7 +1413,7 @@ int wxString::PrintfV(const wxChar* pszFormat, va_list argptr) // NB: wxVsnprintf() may return either less than the buffer size or -1 if // there is not enough place depending on implementation - int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), pszFormat, argptr); + int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), (char *)pszFormat, argptr); if ( iLen != -1 ) { // the whole string is in szScratch *this = szScratch; @@ -1411,21 +1458,77 @@ int wxString::PrintfV(const wxChar* pszFormat, va_list argptr) // of them) bool wxString::Matches(const wxChar *pszMask) const { - // check char by char - const wxChar *pszTxt; - for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) { +#if wxUSE_REGEX + // first translate the shell-like mask into a regex + wxString pattern; + pattern.reserve(wxStrlen(pszMask)); + + pattern += _T('^'); + while ( *pszMask ) + { + switch ( *pszMask ) + { + case _T('?'): + pattern += _T('.'); + break; + + case _T('*'): + pattern += _T(".*"); + break; + + case _T('^'): + case _T('.'): + case _T('$'): + case _T('('): + case _T(')'): + case _T('|'): + case _T('+'): + case _T('\\'): + // these characters are special in a RE, quote them + // (however note that we don't quote '[' and ']' to allow + // using them for Unix shell like matching) + pattern += _T('\\'); + // fall through + + default: + pattern += *pszMask; + } + + pszMask++; + } + pattern += _T('$'); + + // and now use it + return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str()); +#else // !wxUSE_REGEX + // TODO: this is, of course, awfully inefficient... + + // the char currently being checked + const wxChar *pszTxt = c_str(); + + // the last location where '*' matched + const wxChar *pszLastStarInText = NULL; + const wxChar *pszLastStarInMask = NULL; + +match: + for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) { switch ( *pszMask ) { case wxT('?'): if ( *pszTxt == wxT('\0') ) return FALSE; - // pszText and pszMask will be incremented in the loop statement + // pszTxt and pszMask will be incremented in the loop statement break; case wxT('*'): { + // remember where we started to be able to backtrack later + pszLastStarInText = pszTxt; + pszLastStarInMask = pszMask; + // ignore special chars immediately following this one + // (should this be an error?) while ( *pszMask == wxT('*') || *pszMask == wxT('?') ) pszMask++; @@ -1465,7 +1568,23 @@ bool wxString::Matches(const wxChar *pszMask) const } // match only if nothing left - return *pszTxt == wxT('\0'); + if ( *pszTxt == wxT('\0') ) + return TRUE; + + // if we failed to match, backtrack if we can + if ( pszLastStarInText ) { + pszTxt = pszLastStarInText + 1; + pszMask = pszLastStarInMask; + + pszLastStarInText = NULL; + + // don't bother resetting pszLastStarInMask, it's unnecessary + + goto match; + } + + return FALSE; +#endif // wxUSE_REGEX/!wxUSE_REGEX } // Count the number of chars @@ -1500,8 +1619,34 @@ int wxString::sprintf(const wxChar *pszFormat, ...) // --------------------------------------------------------------------------- // standard C++ library string functions // --------------------------------------------------------------------------- + #ifdef wxSTD_STRING_COMPATIBILITY +void wxString::resize(size_t nSize, wxChar ch) +{ + size_t len = length(); + + if ( nSize < len ) + { + Truncate(nSize); + } + else if ( nSize > len ) + { + *this += wxString(ch, len - nSize); + } + //else: we have exactly the specified length, nothing to do +} + +void wxString::swap(wxString& str) +{ + // this is slightly less efficient than fiddling with m_pchData directly, + // but it is still quite efficient as we don't copy the string here because + // ref count always stays positive + wxString tmp = str; + str = *this; + *this = str; +} + wxString& wxString::insert(size_t nPos, const wxString& str) { wxASSERT( str.GetStringData()->IsValid() ); @@ -1599,7 +1744,7 @@ size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const const wxChar *start = c_str() + nStart; const wxChar *firstOf = wxStrpbrk(start, sz); if ( firstOf ) - return firstOf - start; + return firstOf - c_str(); else return npos; } @@ -1710,13 +1855,15 @@ wxString& wxString::erase(size_t nStart, size_t nLen) wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz) { - wxASSERT( nStart + nLen <= wxStrlen(sz) ); + wxASSERT_MSG( nStart + nLen <= Len(), + _T("index out of bounds in wxString::replace") ); wxString strTmp; + strTmp.Alloc(Len()); // micro optimisation to avoid multiple mem allocs + if ( nStart != 0 ) strTmp.append(c_str(), nStart); - strTmp += sz; - strTmp.append(c_str() + nStart + nLen); + strTmp << sz << c_str() + nStart + nLen; *this = strTmp; return *this; @@ -1781,6 +1928,8 @@ wxArrayString& wxArrayString::operator=(const wxArrayString& src) Copy(src); + m_autoSort = src.m_autoSort; + return *this; } @@ -1797,8 +1946,14 @@ void wxArrayString::Copy(const wxArrayString& src) void wxArrayString::Grow() { // only do it if no more place - if( m_nCount == m_nSize ) { - if( m_nSize == 0 ) { + if ( m_nCount == m_nSize ) { + // 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; m_pItems = new wxChar *[m_nSize]; @@ -1806,13 +1961,6 @@ void wxArrayString::Grow() else { // otherwise when it's called for the first time, nIncrement would be 0 // and the array would never be expanded -#if defined(__VISAGECPP__) && defined(__WXDEBUG__) - int array_size = ARRAY_DEFAULT_INITIAL_SIZE; - wxASSERT( array_size != 0 ); -#else - wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 ); -#endif - // add 50% but not too much size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1; @@ -2090,6 +2238,9 @@ void wxArrayString::Sort(CompareFunction compareFunction) DoSort(); + // reset it to NULL so that Sort(bool) will work the next time + gs_compareFunction = NULL; + END_SORT(); } @@ -2114,3 +2265,17 @@ void wxArrayString::DoSort() qsort(m_pItems, m_nCount, sizeof(wxChar *), 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; +} +