X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/b1ac3b56e6b5b7342d69aa3b33744c345edb3d1e..878711c01c1b9ad5b97d35f379a048b8ce1bfb49:/src/common/string.cpp diff --git a/src/common/string.cpp b/src/common/string.cpp index e85bcf2b6b..791c5ce049 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -6,7 +6,7 @@ // Created: 29/01/98 // RCS-ID: $Id$ // Copyright: (c) 1998 Vadim Zeitlin -// Licence: wxWindows license +// Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// #ifdef __GNUG__ @@ -46,13 +46,6 @@ #include #endif -#if wxUSE_UNICODE - #undef wxUSE_EXPERIMENTAL_PRINTF - #ifndef wvsnprintf - #define wxUSE_EXPERIMENTAL_PRINTF 1 - #endif -#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! @@ -88,54 +81,6 @@ static const struct // empty C style string: points to 'string data' byte of g_strEmpty extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy; -// ---------------------------------------------------------------------------- -// conditional compilation -// ---------------------------------------------------------------------------- - -#if !defined(__WXSW__) && wxUSE_UNICODE - #ifdef wxUSE_EXPERIMENTAL_PRINTF - #undef wxUSE_EXPERIMENTAL_PRINTF - #endif - #define wxUSE_EXPERIMENTAL_PRINTF 1 -#endif - -// we want to find out if the current platform supports vsnprintf()-like -// function: for Unix this is done with configure, for Windows we test the -// compiler explicitly. -// -// FIXME currently, this is only for ANSI (!Unicode) strings, so we call this -// function wxVsnprintfA (A for ANSI), should also find one for Unicode -// strings in Unicode build -#ifdef __WXMSW__ - #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 - #endif -#endif // Windows/!Windows - -#ifndef wxVsnprintfA - // in this case we'll use vsprintf() (which is ANSI and thus should be - // always available), but it's unsafe because it doesn't check for buffer - // size - so give a warning - #define wxVsnprintfA(buf, len, format, arg) vsprintf(buf, format, arg) - - #if defined(__VISUALC__) - #pragma message("Using sprintf() because no snprintf()-like function defined") - #elif defined(__GNUG__) - #warning "Using sprintf() because no snprintf()-like function defined" - #endif //compiler -#endif // no vsnprintf - -#if defined(_AIX) - // AIX has vsnprintf, but there's no prototype in the system headers. - extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap); -#endif - // ---------------------------------------------------------------------------- // global functions // ---------------------------------------------------------------------------- @@ -186,74 +131,6 @@ wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) #endif //std::string compatibility -#ifndef wxVsnprintf -int WXDLLEXPORT wxVsnprintf(wxChar *buf, size_t len, - const wxChar *format, va_list argptr) -{ -#if wxUSE_UNICODE - wxString s; - int iLen = s.PrintfV(format, argptr); - if ( iLen != -1 ) - { - wxStrncpy(buf, s.c_str(), len); - buf[len-1] = wxT('\0'); - } - - return iLen; -#else // ANSI - // vsnprintf() will not terminate the string with '\0' if there is not - // enough place, but we want the string to always be NUL terminated - int rc = wxVsnprintfA(buf, len - 1, format, argptr); - if ( rc == -1 ) - { - buf[len] = 0; - } - - return rc; -#endif // Unicode/ANSI -} -#else -// GNU libc 2.2 only has for wxVsnprintf for Unicode called vswprintf -// so we imitate wxVsprintf using it. -int WXDLLEXPORT wxVsprintf(wxChar *buf, - const wxChar *format, - va_list argptr) -{ - return vswprintf( buf, 10000, format, argptr ); -} -#endif - -#ifndef wxSnprintf -int WXDLLEXPORT wxSnprintf(wxChar *buf, size_t len, - const wxChar *format, ...) -{ - va_list argptr; - va_start(argptr, format); - - int iLen = wxVsnprintf(buf, len, format, argptr); - - va_end(argptr); - - return iLen; -} -#else -// GNU libc 2.2 only has for wxSnprintf for Unicode called swprintf -// so we imitate wxSprintf using it. -int WXDLLEXPORT wxSprintf(wxChar *buf, - const wxChar *format, - ...) ATTRIBUTE_PRINTF_2 -{ - va_list argptr; - va_start(argptr, format); - - int iLen = swprintf(buf, 10000, format, argptr); - - va_end(argptr); - - return iLen; -} -#endif - // ---------------------------------------------------------------------------- // private classes // ---------------------------------------------------------------------------- @@ -264,15 +141,15 @@ int WXDLLEXPORT wxSprintf(wxChar *buf, class Averager { public: - Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; } + Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; } ~Averager() - { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); } + { 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 char *m_sz; + const wxChar *m_sz; } g_averageLength("allocation size"), g_averageSummandLength("summand length"), g_averageConcatHit("hit probability in concat"), @@ -303,8 +180,9 @@ wxString::wxString(wxChar ch, size_t nLength) } #if wxUSE_UNICODE - // memset only works on char - for (size_t n=0; n nLength)) - nLen = nLength; + // anything to do? + if ( (nLen != 0) && (nLen != (size_t)-1) ) + { + if ( !AllocBuffer(nLen) ) + { + wxFAIL_MSG( _T("out of memory in wxString::wxString") ); + return; + } - // empty? - if ( (nLen != 0) && (nLen != (size_t)-1) ) { - if ( !AllocBuffer(nLen) ) { - wxFAIL_MSG( _T("out of memory in wxString::wxString") ); - return; + // MB2WC wants the buffer size, not the string length + if ( conv.MB2WC(m_pchData, psz, nLen + 1) != (size_t)-1 ) + { + // initialized ok + return; + } + //else: the conversion failed -- leave the string empty (what else?) } - conv.MB2WC(m_pchData, psz, nLen); - } - else { + Init(); - } } #else // ANSI @@ -377,27 +269,38 @@ wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength) // from wide string wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength) { - // first get necessary size - size_t nLen = 0; - if (pwz) - { - if (nLength == wxSTRING_MAXLEN) - nLen = conv.WC2MB((char *) NULL, pwz, 0); + // first get the size of the buffer we need + size_t nLen; + if ( pwz ) + { + // calculate the needed size ourselves or use a provide one + nLen = nLength == wxSTRING_MAXLEN ? conv.WC2MB(NULL, pwz, 0) : nLength; + } else - nLen = nLength; - } + { + // nothing to convert + nLen = 0; + } - // empty? - if ( (nLen != 0) && (nLen != (size_t)-1) ) { - if ( !AllocBuffer(nLen) ) { - wxFAIL_MSG( _T("out of memory in wxString::wxString") ); - return; + // anything to do? + if ( (nLen != 0) && (nLen != (size_t)-1) ) + { + if ( !AllocBuffer(nLen) ) + { + wxFAIL_MSG( _T("out of memory in wxString::wxString") ); + return; + } + + // WC2MB wants the buffer size, not the string length + if ( conv.WC2MB(m_pchData, pwz, nLen + 1) != (size_t)-1 ) + { + // initialized ok + return; + } + //else: the conversion failed -- leave the string empty (what else?) } - conv.WC2MB(m_pchData, pwz, nLen); - } - else { + Init(); - } } #endif // wxUSE_WCHAR_T @@ -425,12 +328,12 @@ bool wxString::AllocBuffer(size_t nLen) // 2) sizeof(wxStringData) for housekeeping info wxStringData* pData = (wxStringData*) malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar)); - + if ( pData == NULL ) { // allocation failures are handled by the caller return FALSE; } - + pData->nRefs = 1; pData->nDataLength = nLen; pData->nAllocLength = nLen + EXTRA_ALLOC; @@ -484,7 +387,7 @@ bool wxString::AllocBeforeWrite(size_t nLen) 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 @@ -519,7 +422,7 @@ bool wxString::Alloc(size_t nLen) // allocation failure handled by caller return FALSE; } - + pData->nRefs = 1; pData->nDataLength = 0; pData->nAllocLength = nLen; @@ -705,7 +608,7 @@ wxString& wxString::operator=(const wchar_t *pwz) // --------------------------------------------------------------------------- // add something to this string -bool wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData) +bool wxString::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData) { STATISTICS_ADD(SummandLength, nSrcLen); @@ -827,43 +730,63 @@ wxString operator+(const wxChar *psz, const wxString& str) // =========================================================================== #if wxUSE_UNICODE -wxString wxString::FromAscii( char *ascii ) + +wxString wxString::FromAscii(const char *ascii) { if (!ascii) return wxEmptyString; - + size_t len = strlen( ascii ); wxString res; - res.AllocBuffer( len ); - wchar_t *dest = (wchar_t*)(const wchar_t*) res.c_str(); + + if ( len ) + { + wxStringBuffer buf(res, len); + + wchar_t *dest = buf; + + for ( ;; ) + { + if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' ) + break; + } + } + + return res; +} + +wxString wxString::FromAscii(const char ascii) +{ + // What do we do with '\0' ? + + wxString res; + res += (wchar_t)(unsigned char) ascii; - for (size_t i = 0; i < len+1; i++) - dest[i] = (wchar_t) ascii[i]; - return res; } const wxCharBuffer wxString::ToAscii() const { - if (IsNull()) - return wxCharBuffer( (const char*)NULL ); + // this will allocate enough space for the terminating NUL too + wxCharBuffer buffer(length()); - size_t len = Len(); - wxCharBuffer buffer( len ); // allocates len+1 - - char *dest = (char*)(const char*) buffer; - - for (size_t i = 0; i < len+1; i++) + signed char *dest = (signed char *)buffer.data(); + + const wchar_t *pwc = c_str(); + for ( ;; ) { - if (m_pchData[i] > 127) - dest[i] = '_'; - else - dest[i] = (char) m_pchData[i]; + *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc; + + // the output string can't have embedded NULs anyhow, so we can safely + // stop at first of them even if we do have any + if ( !*pwc++ ) + break; } - + return buffer; } -#endif + +#endif // Unicode // --------------------------------------------------------------------------- // simple sub-string extraction @@ -1116,7 +1039,7 @@ wxString& wxString::MakeUpper() wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") ); return *this; } - + for ( wxChar *p = m_pchData; *p; p++ ) *p = (wxChar)wxToupper(*p); @@ -1331,281 +1254,38 @@ int wxString::Printf(const wxChar *pszFormat, ...) int wxString::PrintfV(const wxChar* pszFormat, va_list argptr) { -#if wxUSE_EXPERIMENTAL_PRINTF - // the new implementation - - // buffer to avoid dynamic memory allocation each time for small strings - char szScratch[1024]; - - Reinit(); - for (size_t n = 0; pszFormat[n]; n++) - if (pszFormat[n] == wxT('%')) { - static char s_szFlags[256] = "%"; - size_t flagofs = 1; - bool adj_left = FALSE, in_prec = FALSE, - prec_dot = FALSE, done = FALSE; - int ilen = 0; - size_t min_width = 0, max_width = wxSTRING_MAXLEN; - do { -#define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; } - switch (pszFormat[++n]) { - case wxT('\0'): - done = TRUE; - break; - case wxT('%'): - *this += wxT('%'); - done = TRUE; - break; - case wxT('#'): - case wxT('0'): - case wxT(' '): - case wxT('+'): - case wxT('\''): - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('-'): - CHECK_PREC - adj_left = TRUE; - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('.'): - CHECK_PREC - in_prec = TRUE; - prec_dot = FALSE; - max_width = 0; - // dot will be auto-added to s_szFlags if non-negative number follows - break; - case wxT('h'): - ilen = -1; - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('l'): - ilen = 1; - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('q'): - case wxT('L'): - ilen = 2; - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('Z'): - ilen = 3; - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - break; - case wxT('*'): - { - int len = va_arg(argptr, int); - if (in_prec) { - if (len<0) break; - CHECK_PREC - max_width = len; - } else { - if (len<0) { - adj_left = !adj_left; - s_szFlags[flagofs++] = '-'; - len = -len; - } - min_width = len; - } - flagofs += ::sprintf(s_szFlags+flagofs,"%d",len); - } - break; - case wxT('1'): case wxT('2'): case wxT('3'): - case wxT('4'): case wxT('5'): case wxT('6'): - case wxT('7'): case wxT('8'): case wxT('9'): - { - int len = 0; - CHECK_PREC - while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) { - s_szFlags[flagofs++] = pszFormat[n]; - len = len*10 + (pszFormat[n] - wxT('0')); - n++; - } - if (in_prec) max_width = len; - else min_width = len; - n--; // the main loop pre-increments n again - } - break; - case wxT('d'): - case wxT('i'): - case wxT('o'): - case wxT('u'): - case wxT('x'): - case wxT('X'): - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - s_szFlags[flagofs] = '\0'; - if (ilen == 0 ) { - int val = va_arg(argptr, int); - ::sprintf(szScratch, s_szFlags, val); - } - else if (ilen == -1) { - short int val = va_arg(argptr, short int); - ::sprintf(szScratch, s_szFlags, val); - } - else if (ilen == 1) { - long int val = va_arg(argptr, long int); - ::sprintf(szScratch, s_szFlags, val); - } - else if (ilen == 2) { -#if SIZEOF_LONG_LONG - long long int val = va_arg(argptr, long long int); - ::sprintf(szScratch, s_szFlags, val); -#else - long int val = va_arg(argptr, long int); - ::sprintf(szScratch, s_szFlags, val); -#endif - } - else if (ilen == 3) { - size_t val = va_arg(argptr, size_t); - ::sprintf(szScratch, s_szFlags, val); - } - *this += wxString(szScratch); - done = TRUE; - break; - case wxT('e'): - case wxT('E'): - case wxT('f'): - case wxT('g'): - case wxT('G'): - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - s_szFlags[flagofs] = '\0'; - if (ilen == 2) { - long double val = va_arg(argptr, long double); - ::sprintf(szScratch, s_szFlags, val); - } else { - double val = va_arg(argptr, double); - ::sprintf(szScratch, s_szFlags, val); - } - *this += wxString(szScratch); - done = TRUE; - break; - case wxT('p'): - { - void *val = va_arg(argptr, void *); - CHECK_PREC - s_szFlags[flagofs++] = pszFormat[n]; - s_szFlags[flagofs] = '\0'; - ::sprintf(szScratch, s_szFlags, val); - *this += wxString(szScratch); - done = TRUE; - } - break; - case wxT('c'): - { - wxChar val = va_arg(argptr, int); - // we don't need to honor padding here, do we? - *this += val; - done = TRUE; - } - break; - case wxT('s'): - if (ilen == -1) { - // wx extension: we'll let %hs mean non-Unicode strings - char *val = va_arg(argptr, char *); -#if wxUSE_UNICODE - // ASCII->Unicode constructor handles max_width right - wxString s(val, wxConvLibc, max_width); -#else - size_t len = wxSTRING_MAXLEN; - if (val) { - for (len = 0; val[len] && (len= 1) { - long int *val = va_arg(argptr, long int *); - *val = Len(); - } - done = TRUE; - break; - default: - if (wxIsalpha(pszFormat[n])) - // probably some flag not taken care of here yet - s_szFlags[flagofs++] = pszFormat[n]; - else { - // bad format - *this += wxT('%'); // just to pass the glibc tst-printf.c - n--; - done = TRUE; - } - break; + int size = 1024; + for ( ;; ) + { + wxChar *buf = GetWriteBuf(size + 1); + if ( !buf ) + { + // out of memory + return -1; } -#undef CHECK_PREC - } while (!done); - } else *this += pszFormat[n]; -#else - // buffer to avoid dynamic memory allocation each time for small strings - char szScratch[1024]; - - // 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), (char *)pszFormat, argptr); - if ( iLen != -1 ) { - // the whole string is in szScratch - *this = szScratch; - } - else { - bool outOfMemory = FALSE; - int size = 2*WXSIZEOF(szScratch); - while ( !outOfMemory ) { - char *buf = GetWriteBuf(size); - if ( buf ) - iLen = wxVsnprintfA(buf, size, pszFormat, argptr); - else - outOfMemory = TRUE; - - UngetWriteBuf(); - - if ( iLen != -1 ) { - // ok, there was enough space - break; - } + int len = wxVsnprintf(buf, size, pszFormat, argptr); - // still not enough, double it again - size *= 2; - } + // some implementations of vsnprintf() don't NUL terminate the string + // if there is not enough space for it so always do it manually + buf[size] = _T('\0'); - if ( outOfMemory ) { - // out of memory - return -1; - } - } -#endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF + UngetWriteBuf(); + + if ( len >= 0 ) + { + // ok, there was enough space + break; + } - return Len(); + // still not enough, double it again + size *= 2; + } + + // we could have overshot + Shrink(); + + return Len(); } // ---------------------------------------------------------------------------- @@ -2107,7 +1787,7 @@ void wxArrayString::Copy(const wxArrayString& src) void wxArrayString::Grow(size_t nIncrement) { // only do it if no more place - if ( m_nCount == m_nSize ) { + 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 @@ -2117,6 +1797,8 @@ void wxArrayString::Grow(size_t nIncrement) 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 wxChar *[m_nSize]; } else { @@ -2305,7 +1987,7 @@ size_t wxArrayString::Add(const wxString& str, size_t nInsert) Grow(nInsert); for (size_t i = 0; i < nInsert; i++) - { + { // the string data must not be deleted! str.GetStringData()->Lock(); @@ -2409,7 +2091,7 @@ static wxArrayString::CompareFunction gs_compareFunction = NULL; static bool gs_sortAscending = TRUE; // function which is called by quick sort -extern "C" int LINKAGEMODE +extern "C" int wxC_CALLING_CONV // LINKAGEMODE wxStringCompareFunction(const void *first, const void *second) { wxString *strFirst = (wxString *)first;