X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/abc79986869dd899ded3f707943d77493f3228d8..4b5d2be39657bb364c9111c4d13b1eecb8f3b69f:/src/common/string.cpp diff --git a/src/common/string.cpp b/src/common/string.cpp index 3031ae6daf..595caf7125 100644 --- a/src/common/string.cpp +++ b/src/common/string.cpp @@ -1,25 +1,15 @@ ///////////////////////////////////////////////////////////////////////////// -// Name: string.cpp +// Name: src/common/string.cpp // Purpose: wxString class -// Author: Vadim Zeitlin +// Author: Vadim Zeitlin, Ryan Norton // Modified by: // Created: 29/01/98 // RCS-ID: $Id$ // Copyright: (c) 1998 Vadim Zeitlin -// Licence: wxWindows license +// (c) 2004 Ryan Norton +// Licence: wxWindows licence ///////////////////////////////////////////////////////////////////////////// -#ifdef __GNUG__ -#pragma implementation "string.h" -#endif - -/* - * About ref counting: - * 1) all empty strings use g_strEmpty, nRefs = -1 (set in Init()) - * 2) AllocBuffer() sets nRefs to 1, Lock() increments it by one - * 3) Unlock() decrements nRefs and frees memory if it goes to 0 - */ - // =========================================================================== // headers, declarations, constants // =========================================================================== @@ -28,293 +18,589 @@ #include "wx/wxprec.h" #ifdef __BORLANDC__ -#pragma hdrstop + #pragma hdrstop #endif #ifndef WX_PRECOMP -#include "wx/defs.h" -#include "wx/string.h" + #include "wx/string.h" + #include "wx/wxcrtvararg.h" #endif #include + +#ifndef __WXWINCE__ + #include +#endif + #include #include -#ifdef WXSTRING_IS_WXOBJECT - IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject) -#endif //WXSTRING_IS_WXOBJECT +#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 + // --------------------------------------------------------------------------- // static class variables definition // --------------------------------------------------------------------------- -#ifdef STD_STRING_COMPATIBILITY - const size_t wxString::npos = STRING_MAXLEN; -#endif +//According to STL _must_ be a -1 size_t +const size_t wxString::npos = (size_t) -1; -// =========================================================================== -// static class data, special inlines -// =========================================================================== +#if wxUSE_STRING_POS_CACHE -// for an empty string, GetStringData() will return this address -static int g_strEmpty[] = { -1, // ref count (locked) - 0, // current length - 0, // allocated memory - 0 }; // string data -// empty C style string: points to 'string data' byte of g_strEmpty -extern const char *g_szNul = (const char *)(&g_strEmpty[3]); +#ifdef wxHAS_COMPILER_TLS -// =========================================================================== -// global functions -// =========================================================================== +wxTLS_TYPE(wxString::Cache) wxString::ms_cache; -#ifdef STD_STRING_COMPATIBILITY +#else // !wxHAS_COMPILER_TLS -// MS Visual C++ version 5.0 provides the new STL headers as well as the old -// iostream ones. -// -// ATTN: you can _not_ use both of these in the same program! -#if 0 // def _MSC_VER - #include - #define NAMESPACE std:: -#else - #include - #define NAMESPACE -#endif //Visual C++ - -NAMESPACE istream& operator>>(NAMESPACE istream& is, wxString& WXUNUSED(str)) +struct wxStrCacheInitializer { -#if 0 - int w = is.width(0); - if ( is.ipfx(0) ) { - NAMESPACE streambuf *sb = is.rdbuf(); - str.erase(); - while ( true ) { - int ch = sb->sbumpc (); - if ( ch == EOF ) { - is.setstate(NAMESPACE ios::eofbit); - break; - } - else if ( isspace(ch) ) { - sb->sungetc(); - break; - } - - str += ch; - if ( --w == 1 ) - break; + wxStrCacheInitializer() + { + // calling this function triggers s_cache initialization in it, and + // from now on it becomes safe to call from multiple threads + wxString::GetCache(); } - } +}; - is.isfx(); - if ( str.length() == 0 ) - is.setstate(NAMESPACE ios::failbit); -#endif - return is; +/* +wxString::Cache& wxString::GetCache() +{ + static wxTLS_TYPE(Cache) s_cache; + + return wxTLS_VALUE(s_cache); } +*/ -#endif //std::string compatibility +static wxStrCacheInitializer gs_stringCacheInit; -// =========================================================================== -// wxString class core -// =========================================================================== +#endif // wxHAS_COMPILER_TLS/!wxHAS_COMPILER_TLS -// --------------------------------------------------------------------------- -// construction -// --------------------------------------------------------------------------- +// gdb seems to be unable to display thread-local variables correctly, at least +// not my 6.4.98 version under amd64, so provide this debugging helper to do it +#ifdef __WXDEBUG__ -// construct an empty string -wxString::wxString() +struct wxStrCacheDumper { - Init(); + static void ShowAll() + { + puts("*** wxString cache dump:"); + for ( unsigned n = 0; n < wxString::Cache::SIZE; n++ ) + { + const wxString::Cache::Element& + c = wxString::GetCacheBegin()[n]; + + printf("\t%u%s\t%p: pos=(%lu, %lu), len=%ld\n", + n, + n == wxString::LastUsedCacheElement() ? " [*]" : "", + c.str, + (unsigned long)c.pos, + (unsigned long)c.impl, + (long)c.len); + } + } +}; + +void wxDumpStrCache() { wxStrCacheDumper::ShowAll(); } + +#endif // __WXDEBUG__ + +#ifdef wxPROFILE_STRING_CACHE + +wxString::CacheStats wxString::ms_cacheStats; + +struct wxStrCacheStatsDumper +{ + ~wxStrCacheStatsDumper() + { + const wxString::CacheStats& stats = wxString::ms_cacheStats; + + if ( stats.postot ) + { + puts("*** wxString cache statistics:"); + printf("\tTotal non-trivial calls to PosToImpl(): %u\n", + stats.postot); + printf("\tHits %u (of which %u not used) or %.2f%%\n", + stats.poshits, + stats.mishits, + 100.*float(stats.poshits - stats.mishits)/stats.postot); + printf("\tAverage position requested: %.2f\n", + float(stats.sumpos) / stats.postot); + printf("\tAverage offset after cached hint: %.2f\n", + float(stats.sumofs) / stats.postot); + } + + if ( stats.lentot ) + { + printf("\tNumber of calls to length(): %u, hits=%.2f%%\n", + stats.lentot, 100.*float(stats.lenhits)/stats.lentot); + } + } +}; + +static wxStrCacheStatsDumper s_showCacheStats; + +#endif // wxPROFILE_STRING_CACHE + +#endif // wxUSE_STRING_POS_CACHE + +// ---------------------------------------------------------------------------- +// global functions +// ---------------------------------------------------------------------------- + +#if wxUSE_STD_IOSTREAM + +#include + +wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str) +{ +#if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8 + return os << (const char *)str.AsCharBuf(); +#else + return os << str.AsInternal(); +#endif } -// copy constructor -wxString::wxString(const wxString& stringSrc) +wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str) { - wxASSERT( stringSrc.GetStringData()->IsValid() ); + return os << str.c_str(); +} - if ( stringSrc.IsEmpty() ) { - // nothing to do for an empty string - Init(); - } - else { - m_pchData = stringSrc.m_pchData; // share same data - GetStringData()->Lock(); // => one more copy - } +wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str) +{ + return os << str.data(); } -// constructs string of copies of character -wxString::wxString(char ch, size_t nLength) +#ifndef __BORLANDC__ +wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str) { - Init(); + return os << str.data(); +} +#endif - if ( nLength > 0 ) { - AllocBuffer(nLength); +#if wxUSE_UNICODE && defined(HAVE_WOSTREAM) - wxASSERT( sizeof(char) == 1 ); // can't use memset if not +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str) +{ + return wos << str.wc_str(); +} - memset(m_pchData, ch, nLength); - } +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str) +{ + return wos << str.AsWChar(); } -// takes nLength elements of psz starting at nPos -void wxString::InitWith(const char *psz, size_t nPos, size_t nLength) +wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str) { - Init(); + return wos << str.data(); +} - wxASSERT( nPos <= Strlen(psz) ); +#endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM) - if ( nLength == STRING_MAXLEN ) - nLength = Strlen(psz + nPos); +#endif // wxUSE_STD_IOSTREAM - if ( nLength > 0 ) { - // trailing '\0' is written in AllocBuffer() - AllocBuffer(nLength); - memcpy(m_pchData, psz + nPos, nLength*sizeof(char)); - } -} - -// take first nLength characters of C string psz -// (default value of STRING_MAXLEN means take all the string) -wxString::wxString(const char *psz, size_t nLength) +// =========================================================================== +// wxString class core +// =========================================================================== + +#if wxUSE_UNICODE_UTF8 + +void wxString::PosLenToImpl(size_t pos, size_t len, + size_t *implPos, size_t *implLen) const { - InitWith(psz, 0, nLength); + if ( pos == npos ) + { + *implPos = npos; + } + else // have valid start position + { + const const_iterator b = GetIterForNthChar(pos); + *implPos = wxStringImpl::const_iterator(b.impl()) - m_impl.begin(); + if ( len == npos ) + { + *implLen = npos; + } + else // have valid length too + { + // we need to handle the case of length specifying a substring + // going beyond the end of the string, just as std::string does + const const_iterator e(end()); + const_iterator i(b); + while ( len && i <= e ) + { + ++i; + --len; + } + + *implLen = i.impl() - b.impl(); + } + } } -// the same as previous constructor, but for compilers using unsigned char -wxString::wxString(const unsigned char* psz, size_t nLength) +#endif // wxUSE_UNICODE_UTF8 + +// ---------------------------------------------------------------------------- +// wxCStrData converted strings caching +// ---------------------------------------------------------------------------- + +// FIXME-UTF8: temporarily disabled because it doesn't work with global +// string objects; re-enable after fixing this bug and benchmarking +// performance to see if using a hash is a good idea at all +#if 0 + +// For backward compatibility reasons, it must be possible to assign the value +// returned by wxString::c_str() to a char* or wchar_t* variable and work with +// it. Returning wxCharBuffer from (const char*)c_str() wouldn't do the trick, +// because the memory would be freed immediately, but it has to be valid as long +// as the string is not modified, so that code like this still works: +// +// const wxChar *s = str.c_str(); +// while ( s ) { ... } + +// FIXME-UTF8: not thread safe! +// FIXME-UTF8: we currently clear the cached conversion only when the string is +// destroyed, but we should do it when the string is modified, to +// keep memory usage down +// FIXME-UTF8: we do the conversion every time As[W]Char() is called, but if we +// invalidated the cache on every change, we could keep the previous +// conversion +// FIXME-UTF8: add tracing of usage of these two methods - new code is supposed +// to use mb_str() or wc_str() instead of (const [w]char*)c_str() + +template +static inline void DeleteStringFromConversionCache(T& hash, const wxString *s) +{ + typename T::iterator i = hash.find(wxConstCast(s, wxString)); + if ( i != hash.end() ) + { + free(i->second); + hash.erase(i); + } +} + +#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; + +const char* wxCStrData::AsChar() const { - InitWith((const char *)psz, 0, nLength); -} - -#ifdef STD_STRING_COMPATIBILITY + // remove previously cache value, if any (see FIXMEs above): + DeleteStringFromConversionCache(gs_stringsCharCache, m_str); + + // convert the string and keep it: + const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] = + m_str->mb_str().release(); + + return s + m_offset; +} +#endif // wxUSE_UNICODE + +#if !wxUSE_UNICODE_WCHAR +WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual, + wxStringWCharConversionCache); +static wxStringWCharConversionCache gs_stringsWCharCache; -// ctor from a substring -wxString::wxString(const wxString& str, size_t nPos, size_t nLen) +const wchar_t* wxCStrData::AsWChar() const { - wxASSERT( str.GetStringData()->IsValid() ); + // remove previously cache value, if any (see FIXMEs above): + DeleteStringFromConversionCache(gs_stringsWCharCache, m_str); - InitWith(str.c_str(), nPos, nLen == npos ? 0 : nLen); + // convert the string and keep it: + const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] = + m_str->wc_str().release(); + + return s + m_offset; } +#endif // !wxUSE_UNICODE_WCHAR -// poor man's iterators are "void *" pointers -wxString::wxString(const void *pStart, const void *pEnd) +wxString::~wxString() { - InitWith((const char *)pStart, 0, - (const char *)pEnd - (const char *)pStart); +#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 -#endif //std::string compatibility +#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY +const char* wxCStrData::AsChar() const +{ +#if wxUSE_UNICODE_UTF8 + if ( wxLocaleIsUtf8 ) + return AsInternal(); +#endif + // under non-UTF8 locales, we have to convert the internal UTF-8 + // representation using wxConvLibc and cache the result + + wxString *str = wxConstCast(m_str, wxString); + + // convert the string: + // + // FIXME-UTF8: we'd like to do the conversion in the existing buffer (if we + // have it) but it's unfortunately not obvious to implement + // because we don't know how big buffer do we need for the + // given string length (in case of multibyte encodings, e.g. + // ISO-2022-JP or UTF-8 when internal representation is wchar_t) + // + // One idea would be to store more than just m_convertedToChar + // in wxString: then we could record the length of the string + // which was converted the last time and try to reuse the same + // buffer if the current length is not greater than it (this + // could still fail because string could have been modified in + // place but it would work most of the time, so we'd do it and + // only allocate the new buffer if in-place conversion returned + // an error). We could also store a bit saying if the string + // was modified since the last conversion (and update it in all + // operation modifying the string, of course) to avoid unneeded + // consequential conversions. But both of these ideas require + // adding more fields to wxString and require profiling results + // to be sure that we really gain enough from them to justify + // doing it. + wxCharBuffer buf(str->mb_str()); + + // if it failed, return empty string and not NULL to avoid crashes in code + // written with either wxWidgets 2 wxString or std::string behaviour in + // mind: neither of them ever returns NULL and so we shouldn't neither + if ( !buf ) + return ""; + + if ( str->m_convertedToChar && + strlen(buf) == strlen(str->m_convertedToChar) ) + { + // 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 + { + str->m_convertedToChar = buf.release(); + } + + // and keep it: + return str->m_convertedToChar + m_offset; +} +#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY -// from wide string -wxString::wxString(const wchar_t *pwz) +#if !wxUSE_UNICODE_WCHAR +const wchar_t* wxCStrData::AsWChar() const { - // first get necessary size - size_t nLen = wcstombs(NULL, pwz, 0); + wxString *str = wxConstCast(m_str, wxString); - // empty? - if ( nLen != 0 ) { - AllocBuffer(nLen); - wcstombs(m_pchData, pwz, nLen); - } - else { - Init(); - } + // convert the string: + wxWCharBuffer buf(str->wc_str()); + + // notice that here, unlike above in AsChar(), conversion can't fail as our + // internal UTF-8 is always well-formed -- or the string was corrupted and + // all bets are off anyhow + + // FIXME-UTF8: do the conversion in-place in the existing buffer + if ( str->m_convertedToWChar && + wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) ) + { + // keep the same buffer for as long as possible, so that several calls + // to c_str() in a row still work: + memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf)); + } + else + { + str->m_convertedToWChar = buf.release(); + } + + // and keep it: + return str->m_convertedToWChar + m_offset; } +#endif // !wxUSE_UNICODE_WCHAR + +// =========================================================================== +// wxString class core +// =========================================================================== // --------------------------------------------------------------------------- -// memory allocation +// construction and conversion // --------------------------------------------------------------------------- -// allocates memory needed to store a C string of length nLen -void wxString::AllocBuffer(size_t nLen) -{ - wxASSERT( nLen > 0 ); // - wxASSERT( nLen <= INT_MAX-1 ); // max size (enough room for 1 extra) +#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 + +#if wxUSE_UNICODE_UTF8 +/* static */ +wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv) +{ + // anything to do? + if ( !psz || nLength == 0 ) + return SubstrBufFromMB("", 0); + + // if psz is already in UTF-8, we don't have to do the roundtrip to + // wchar_t* and back: + if ( conv.IsUTF8() ) + { + // we need to validate the input because UTF8 iterators assume valid + // UTF-8 sequence and psz may be invalid: + if ( wxStringOperations::IsValidUtf8String(psz, nLength) ) + { + // we must pass the real string length to SubstrBufFromMB ctor + if ( nLength == npos ) + nLength = psz ? strlen(psz) : 0; + return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength); + } + // else: do the roundtrip through wchar_t* + } - // allocate memory: - // 1) one extra character for '\0' termination - // 2) sizeof(wxStringData) for housekeeping info - wxStringData* pData = (wxStringData*)new char[sizeof(wxStringData) + - (nLen + 1)*sizeof(char)]; - pData->nRefs = 1; - pData->data()[nLen] = '\0'; - pData->nDataLength = nLen; - pData->nAllocLength = nLen; - m_pchData = pData->data(); // data starts after wxStringData -} + if ( nLength == npos ) + nLength = wxNO_LEN; -// releases the string memory and reinits it -void wxString::Reinit() -{ - GetStringData()->Unlock(); - Init(); + // first convert to wide string: + size_t wcLen; + wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen)); + if ( !wcLen ) + return SubstrBufFromMB("", 0); + + // and then to UTF-8: + SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8())); + // widechar -> UTF-8 conversion isn't supposed to ever fail: + wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") ); + + return buf; } +#endif // wxUSE_UNICODE_UTF8 -// wrapper around wxString::Reinit -void wxString::Empty() +#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE +/* static */ +wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength, + const wxMBConv& conv) { - if ( GetStringData()->nDataLength != 0 ) - Reinit(); + // anything to do? + if ( !pwz || nLength == 0 ) + return SubstrBufFromWC("", 0); + + if ( nLength == npos ) + nLength = wxNO_LEN; - wxASSERT( GetStringData()->nDataLength == 0 ); - wxASSERT( GetStringData()->nAllocLength == 0 ); + size_t mbLen; + wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen)); + if ( !mbLen ) + return SubstrBufFromWC("", 0); + else + return SubstrBufFromWC(mbBuf, mbLen); } +#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE + + +#if wxUSE_UNICODE_WCHAR -// must be called before changing this string -void wxString::CopyBeforeWrite() +//Convert wxString in Unicode mode to a multi-byte string +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - wxStringData* pData = GetStringData(); + return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL); +} - if ( pData->IsShared() ) { - pData->Unlock(); // memory not freed because shared - AllocBuffer(pData->nDataLength); - memcpy(m_pchData, pData->data(), (pData->nDataLength + 1)*sizeof(char)); - } +#elif wxUSE_UNICODE_UTF8 - wxASSERT( !pData->IsShared() ); // we must be the only owner +const wxWCharBuffer wxString::wc_str() const +{ + return wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size, not length + NULL + ); } -// must be called before replacing contents of this string -void wxString::AllocBeforeWrite(size_t nLen) +const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const { - wxASSERT( nLen != 0 ); // doesn't make any sense + if ( conv.IsUTF8() ) + return wxCharBuffer::CreateNonOwned(m_impl.c_str()); - // must not share string and must have enough space - register wxStringData* pData = GetStringData(); - if ( pData->IsShared() || (nLen > pData->nAllocLength) ) { - // can't work with old buffer, get new one - pData->Unlock(); - AllocBuffer(nLen); - } + // FIXME-UTF8: use wc_str() here once we have buffers with length + + size_t wcLen; + wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC + ( + m_impl.c_str(), + m_impl.length() + 1, // size + &wcLen + )); + if ( !wcLen ) + return wxCharBuffer(""); - wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner + return conv.cWC2MB(wcBuf, wcLen+1, NULL); } -// get the pointer to writable buffer of (at least) nLen bytes -char *wxString::GetWriteBuf(int nLen) +#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 { - AllocBeforeWrite(nLen); + return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL); +} + +#endif // Unicode/ANSI - wxASSERT( GetStringData()->nRefs == 1 ); - GetStringData()->Validate(FALSE); +// shrink to minimal size (releasing extra memory) +bool wxString::Shrink() +{ + wxString tmp(begin(), end()); + swap(tmp); + return tmp.length() == length(); +} - return m_pchData; +// deprecated compatibility code: +#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +wxStringCharType *wxString::GetWriteBuf(size_t nLen) +{ + return DoGetWriteBuf(nLen); } -// put string back in a reasonable state after GetWriteBuf void wxString::UngetWriteBuf() { - GetStringData()->nDataLength = strlen(m_pchData); - GetStringData()->Validate(TRUE); + DoUngetWriteBuf(); } -// dtor frees memory if no other strings use it -wxString::~wxString() +void wxString::UngetWriteBuf(size_t nLen) { - GetStringData()->Unlock(); + DoUngetWriteBuf(nLen); } +#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 + // --------------------------------------------------------------------------- // data access @@ -323,281 +609,683 @@ wxString::~wxString() // all functions are inline in string.h // --------------------------------------------------------------------------- -// assignment operators +// concatenation operators // --------------------------------------------------------------------------- -// helper function: does real copy -void wxString::AssignCopy(size_t nSrcLen, const char *pszSrcData) +/* + * 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) { - if ( nSrcLen == 0 ) { - Reinit(); - } - else { - AllocBeforeWrite(nSrcLen); - memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(char)); - GetStringData()->nDataLength = nSrcLen; - m_pchData[nSrcLen] = '\0'; - } +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str1.IsValid() ); + wxASSERT( str2.IsValid() ); +#endif + + wxString s = str1; + s += str2; + + return s; } -// assigns one string to another -wxString& wxString::operator=(const wxString& stringSrc) +wxString operator+(const wxString& str, wxUniChar ch) { - wxASSERT( stringSrc.GetStringData()->IsValid() ); +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif - // 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(); - } - } + wxString s = str; + s += ch; - return *this; + return s; } -// assigns a single character -wxString& wxString::operator=(char ch) +wxString operator+(wxUniChar ch, const wxString& str) { - AssignCopy(1, &ch); - return *this; +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s = ch; + s += str; + + return s; } -// assigns C string -wxString& wxString::operator=(const char *psz) +wxString operator+(const wxString& str, const char *psz) { - AssignCopy(Strlen(psz), psz); - return *this; +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s += str; + s += psz; + + return s; } -// same as 'signed char' variant -wxString& wxString::operator=(const unsigned char* psz) +wxString operator+(const wxString& str, const wchar_t *pwz) { - *this = (const char *)psz; - return *this; +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s += str; + s += pwz; + + return s; } -wxString& wxString::operator=(const wchar_t *pwz) +wxString operator+(const char *psz, const wxString& str) { - wxString str(pwz); - *this = str; - return *this; +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s; + if ( !s.Alloc(strlen(psz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = psz; + s += str; + + return s; +} + +wxString operator+(const wchar_t *pwz, const wxString& str) +{ +#if !wxUSE_STL_BASED_WXSTRING + wxASSERT( str.IsValid() ); +#endif + + wxString s; + if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) { + wxFAIL_MSG( _T("out of memory in wxString::operator+") ); + } + s = pwz; + s += str; + + return s; } // --------------------------------------------------------------------------- -// string concatenation +// string comparison // --------------------------------------------------------------------------- -// concatenate two sources -// NB: assume that 'this' is a new wxString object -void wxString::ConcatCopy(int nSrc1Len, const char *pszSrc1Data, - int nSrc2Len, const char *pszSrc2Data) -{ - int nNewLen = nSrc1Len + nSrc2Len; - if ( nNewLen != 0 ) - { - AllocBuffer(nNewLen); - memcpy(m_pchData, pszSrc1Data, nSrc1Len*sizeof(char)); - memcpy(m_pchData + nSrc1Len, pszSrc2Data, nSrc2Len*sizeof(char)); - } +bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const +{ + return (length() == 1) && (compareWithCase ? GetChar(0u) == c + : wxToupper(GetChar(0u)) == wxToupper(c)); +} + +#ifdef HAVE_STD_STRING_COMPARE + +// NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with +// UTF-8 encoded strings too, thanks to UTF-8's design which allows us to +// sort strings in characters code point order by sorting the byte sequence +// in byte values order (i.e. what strcmp() and memcmp() do). + +int wxString::compare(const wxString& str) const +{ + return m_impl.compare(str.m_impl); +} + +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); } -// add something to this string -void wxString::ConcatSelf(int nSrcLen, const char *pszSrcData) +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { - // concatenating an empty string is a NOP - if ( nSrcLen != 0 ) { - register wxStringData *pData = GetStringData(); + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - // alloc new buffer if current is too small - if ( pData->IsShared() || - pData->nDataLength + nSrcLen > pData->nAllocLength ) { - // we have to grow the buffer, use the ConcatCopy routine - // (which will allocate memory) - wxStringData* pOldData = GetStringData(); - ConcatCopy(pOldData->nDataLength, m_pchData, nSrcLen, pszSrcData); - pOldData->Unlock(); - } - else { - // fast concatenation when buffer big enough - memcpy(m_pchData + pData->nDataLength, pszSrcData, nSrcLen*sizeof(char)); - pData->nDataLength += nSrcLen; + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); - // should be enough space - wxASSERT( pData->nDataLength <= pData->nAllocLength ); + return m_impl.compare(pos, len, str.m_impl, pos2, len2); +} + +int wxString::compare(const char* sz) const +{ + return m_impl.compare(ImplStr(sz)); +} + +int wxString::compare(const wchar_t* sz) const +{ + return m_impl.compare(ImplStr(sz)); +} + +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const +{ + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + SubstrBufFromMB str(ImplStr(sz, nCount)); + + return m_impl.compare(pos, len, str.data, str.len); +} + +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const +{ + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + SubstrBufFromWC str(ImplStr(sz, nCount)); + + return m_impl.compare(pos, len, str.data, str.len); +} - m_pchData[pData->nDataLength] = '\0'; // put terminating '\0' +#else // !HAVE_STD_STRING_COMPARE + +static inline int wxDoCmp(const wxStringCharType* s1, size_t l1, + const wxStringCharType* s2, size_t l2) +{ + 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; } - } } -/* - * string may be concatenated with other string, C string or a character - */ +int wxString::compare(const wxString& str) const +{ + return ::wxDoCmp(m_impl.data(), m_impl.length(), + str.m_impl.data(), str.m_impl.length()); +} -void wxString::operator+=(const wxString& string) +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str) const { - wxASSERT( string.GetStringData()->IsValid() ); + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); - ConcatSelf(string.Len(), string); + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data(), str.m_impl.length()); } -void wxString::operator+=(const char *psz) +int wxString::compare(size_t nStart, size_t nLen, + const wxString& str, + size_t nStart2, size_t nLen2) const { - ConcatSelf(Strlen(psz), psz); + 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; + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + size_t pos2, len2; + str.PosLenToImpl(nStart2, nLen2, &pos2, &len2); + + return ::wxDoCmp(m_impl.data() + pos, len, + str.m_impl.data() + pos2, len2); } -void wxString::operator+=(char ch) +int wxString::compare(const char* sz) const { - ConcatSelf(1, &ch); + 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); } -/* - * Same as above but return the result - */ +int wxString::compare(const wchar_t* sz) const +{ + 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); +} -wxString& wxString::operator<<(const wxString& string) +int wxString::compare(size_t nStart, size_t nLen, + const char* sz, size_t nCount) const { - wxASSERT( string.GetStringData()->IsValid() ); + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; - ConcatSelf(string.Len(), string); - return *this; + 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); } -wxString& wxString::operator<<(const char *psz) +int wxString::compare(size_t nStart, size_t nLen, + const wchar_t* sz, size_t nCount) const { - ConcatSelf(Strlen(psz), psz); - return *this; + wxASSERT(nStart <= length()); + size_type strLen = length() - nStart; + nLen = strLen < nLen ? strLen : nLen; + + size_t pos, len; + PosLenToImpl(nStart, nLen, &pos, &len); + + 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); } -wxString& wxString::operator<<(char ch) +#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE + + +// --------------------------------------------------------------------------- +// find_{first,last}_[not]_of functions +// --------------------------------------------------------------------------- + +#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. + +size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const { - ConcatSelf(1, &ch); - return *this; + return find_first_of(sz, nStart, wxStrlen(sz)); } -/* - * concatenation functions come in 5 flavours: - * string + string - * char + string and string + char - * C str + string and string + C str - */ +size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const +{ + return find_first_not_of(sz, nStart, wxStrlen(sz)); +} -wxString operator+(const wxString& string1, const wxString& string2) +size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const { - wxASSERT( string1.GetStringData()->IsValid() ); - wxASSERT( string2.GetStringData()->IsValid() ); + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); + + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( wxTmemchr(sz, *i, n) ) + return idx; + } - wxString s; - s.ConcatCopy(string1.GetStringData()->nDataLength, string1.m_pchData, - string2.GetStringData()->nDataLength, string2.m_pchData); - return s; + return npos; } -wxString operator+(const wxString& string1, char ch) +size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const { - wxASSERT( string1.GetStringData()->IsValid() ); + wxASSERT_MSG( nStart <= length(), _T("invalid index") ); + + size_t idx = nStart; + for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i ) + { + if ( !wxTmemchr(sz, *i, n) ) + return idx; + } - wxString s; - s.ConcatCopy(string1.GetStringData()->nDataLength, string1.m_pchData, 1, &ch); - return s; + return npos; } -wxString operator+(char ch, const wxString& string) + +size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const { - wxASSERT( string.GetStringData()->IsValid() ); + return find_last_of(sz, nStart, wxStrlen(sz)); +} - wxString s; - s.ConcatCopy(1, &ch, string.GetStringData()->nDataLength, string.m_pchData); - 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& string, const char *psz) +size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const { - wxASSERT( string.GetStringData()->IsValid() ); + size_t len = length(); - wxString s; - s.ConcatCopy(string.GetStringData()->nDataLength, string.m_pchData, - Strlen(psz), psz); - return s; + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } + + 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; +} + +size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const +{ + size_t len = length(); + + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } + + 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 char *psz, const wxString& string) +size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const { - wxASSERT( string.GetStringData()->IsValid() ); + 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; +} + +size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const +{ + size_t len = length(); + + if ( nStart == npos ) + { + nStart = len - 1; + } + else + { + wxASSERT_MSG( nStart <= len, _T("invalid index") ); + } + + size_t idx = nStart; + for ( const_reverse_iterator i = rbegin() + (len - nStart - 1); + i != rend(); --idx, ++i ) + { + if ( *i != ch ) + return idx; + } - wxString s; - s.ConcatCopy(Strlen(psz), psz, - string.GetStringData()->nDataLength, string.m_pchData); - return s; + 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 // =========================================================================== -// --------------------------------------------------------------------------- -// simple sub-string extraction -// --------------------------------------------------------------------------- +int wxString::CmpNoCase(const wxString& s) const +{ + // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added + + const_iterator i1 = begin(); + const_iterator end1 = end(); + const_iterator i2 = s.begin(); + const_iterator end2 = s.end(); + + for ( ; i1 != end1 && i2 != end2; ++i1, ++i2 ) + { + wxUniChar lower1 = (wxChar)wxTolower(*i1); + wxUniChar lower2 = (wxChar)wxTolower(*i2); + if ( lower1 != lower2 ) + return lower1 < lower2 ? -1 : 1; + } + + size_t len1 = length(); + size_t len2 = s.length(); -// helper function: clone the data attached to this string -void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const + if ( len1 < len2 ) + return -1; + else if ( len1 > len2 ) + return 1; + return 0; +} + + +#if wxUSE_UNICODE + +#ifdef __MWERKS__ +#ifndef __SCHAR_MAX__ +#define __SCHAR_MAX__ 127 +#endif +#endif + +wxString wxString::FromAscii(const char *ascii, size_t len) { - if ( nCopyLen == 0 ) - { - dest.Init(); - } - else - { - dest.AllocBuffer(nCopyLen); - memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(char)); - } + if (!ascii || len == 0) + return wxEmptyString; + + wxString res; + + { + wxStringInternalBuffer buf(res, len); + wxStringCharType *dest = buf; + + for ( ; len > 0; --len ) + { + unsigned char c = (unsigned char)*ascii++; + wxASSERT_MSG( c < 0x80, + _T("Non-ASCII value passed to FromAscii().") ); + + *dest++ = (wchar_t)c; + } + } + + return res; +} + +wxString wxString::FromAscii(const char *ascii) +{ + return FromAscii(ascii, wxStrlen(ascii)); +} + +wxString wxString::FromAscii(char ascii) +{ + // What do we do with '\0' ? + + unsigned char c = (unsigned char)ascii; + + wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") ); + + // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value + return wxString(wxUniChar((wchar_t)c)); +} + +const wxCharBuffer wxString::ToAscii() const +{ + // this will allocate enough space for the terminating NUL too + wxCharBuffer buffer(length()); + char *dest = buffer.data(); + + for ( const_iterator i = begin(); i != end(); ++i ) + { + wxUniChar c(*i); + // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?') + *dest++ = c.IsAscii() ? (char)c : '_'; + + // the output string can't have embedded NULs anyhow, so we can safely + // stop at first of them even if we do have any + if ( !c ) + break; + } + + return buffer; } -// extract string of length nCount starting at nFirst -// default value of nCount is 0 and means "till the end" -wxString wxString::Mid(size_t nFirst, size_t nCount) const +#endif // wxUSE_UNICODE + +// extract string of length nCount starting at nFirst +wxString wxString::Mid(size_t nFirst, size_t nCount) const +{ + size_t nLen = length(); + + // default value of nCount is npos and means "till the end" + if ( nCount == npos ) + { + nCount = nLen - nFirst; + } + + // out-of-bounds requests return sensible things + if ( nFirst + nCount > nLen ) + { + nCount = nLen - nFirst; + } + + if ( nFirst > nLen ) + { + // AllocCopy() will return empty string + return wxEmptyString; + } + + wxString dest(*this, nFirst, nCount); + if ( dest.length() != nCount ) + { + wxFAIL_MSG( _T("out of memory in wxString::Mid") ); + } + + return dest; +} + +// check that the string starts with prefix and return the rest of the string +// in the provided pointer if it is not NULL, otherwise return false +bool wxString::StartsWith(const wxString& prefix, wxString *rest) const +{ + if ( compare(0, prefix.length(), prefix) != 0 ) + return false; + + if ( rest ) + { + // put the rest of the string into provided pointer + rest->assign(*this, prefix.length(), npos); + } + + return true; +} + + +// check that the string ends with suffix and return the rest of it in the +// provided pointer if it is not NULL, otherwise return false +bool wxString::EndsWith(const wxString& suffix, wxString *rest) const { - // out-of-bounds requests return sensible things - if ( nCount == 0 ) - nCount = GetStringData()->nDataLength - nFirst; + int start = length() - suffix.length(); - if ( nFirst + nCount > (size_t)GetStringData()->nDataLength ) - nCount = GetStringData()->nDataLength - nFirst; - if ( nFirst > (size_t)GetStringData()->nDataLength ) - nCount = 0; + if ( start < 0 || compare(start, npos, suffix) != 0 ) + return false; - wxString dest; - AllocCopy(dest, nCount, nFirst); - return dest; + if ( rest ) + { + // put the rest of the string into provided pointer + rest->assign(*this, 0, start); + } + + return true; } + // extract nCount last (rightmost) characters wxString wxString::Right(size_t nCount) const { - if ( nCount > (size_t)GetStringData()->nDataLength ) - nCount = GetStringData()->nDataLength; + if ( nCount > length() ) + nCount = length(); - wxString dest; - AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount); + wxString dest(*this, length() - nCount, nCount); + if ( dest.length() != nCount ) { + wxFAIL_MSG( _T("out of memory in wxString::Right") ); + } return dest; } // get all characters after the last occurence of ch // (returns the whole string if ch not found) -wxString wxString::Right(char ch) const +wxString wxString::AfterLast(wxUniChar ch) const { wxString str; - int iPos = Find(ch, TRUE); - if ( iPos == NOT_FOUND ) + int iPos = Find(ch, true); + if ( iPos == wxNOT_FOUND ) str = *this; else - str = c_str() + iPos + 1; + str = wx_str() + iPos + 1; return str; } @@ -605,32 +1293,32 @@ wxString wxString::Right(char ch) const // extract nCount first (leftmost) characters wxString wxString::Left(size_t nCount) const { - if ( nCount > (size_t)GetStringData()->nDataLength ) - nCount = GetStringData()->nDataLength; + if ( nCount > length() ) + nCount = length(); - wxString dest; - AllocCopy(dest, nCount, 0); + wxString dest(*this, 0, nCount); + if ( dest.length() != nCount ) { + wxFAIL_MSG( _T("out of memory in wxString::Left") ); + } return dest; } // get all characters before the first occurence of ch // (returns the whole string if ch not found) -wxString wxString::Left(char ch) const +wxString wxString::BeforeFirst(wxUniChar ch) const { - wxString str; - for ( const char *pc = m_pchData; *pc != '\0' && *pc != ch; pc++ ) - str += *pc; - - return str; + int iPos = Find(ch); + if ( iPos == wxNOT_FOUND ) iPos = length(); + return wxString(*this, 0, iPos); } /// get all characters before the last occurence of ch /// (returns empty string if ch not found) -wxString wxString::Before(char ch) const +wxString wxString::BeforeLast(wxUniChar ch) const { wxString str; - int iPos = Find(ch, TRUE); - if ( iPos != NOT_FOUND && iPos != 0 ) + int iPos = Find(ch, true); + if ( iPos != wxNOT_FOUND && iPos != 0 ) str = wxString(c_str(), iPos); return str; @@ -638,700 +1326,811 @@ wxString wxString::Before(char ch) const /// get all characters after the first occurence of ch /// (returns empty string if ch not found) -wxString wxString::After(char ch) const +wxString wxString::AfterFirst(wxUniChar ch) const { wxString str; int iPos = Find(ch); - if ( iPos != NOT_FOUND ) - str = c_str() + iPos + 1; + if ( iPos != wxNOT_FOUND ) + str = wx_str() + iPos + 1; return str; } // replace first (or all) occurences of some substring with another one -uint wxString::Replace(const char *szOld, const char *szNew, bool bReplaceAll) -{ - uint uiCount = 0; // count of replacements made +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( !strOld.empty(), 0, + _T("wxString::Replace(): invalid parameter") ); + + wxSTRING_INVALIDATE_CACHE(); + + size_t uiCount = 0; // count of replacements made + + // optimize the special common case: replacement of one character by + // another one (in UTF-8 case we can only do this for ASCII characters) + // + // benchmarks show that this special version is around 3 times faster + // (depending on the proportion of matching characters and UTF-8/wchar_t + // build) + if ( strOld.m_impl.length() == 1 && strNew.m_impl.length() == 1 ) + { + const wxStringCharType chOld = strOld.m_impl[0], + chNew = strNew.m_impl[0]; + + // this loop is the simplified version of the one below + for ( size_t pos = 0; ; ) + { + pos = m_impl.find(chOld, pos); + if ( pos == npos ) + break; - uint uiOldLen = Strlen(szOld); + m_impl[pos++] = chNew; - wxString strTemp; - const char *pCurrent = m_pchData; - const char *pSubstr; - while ( *pCurrent != '\0' ) { - pSubstr = strstr(pCurrent, szOld); - if ( pSubstr == NULL ) { - // strTemp is unused if no replacements were made, so avoid the copy - if ( uiCount == 0 ) - return 0; + uiCount++; - strTemp += pCurrent; // copy the rest - break; // exit the loop - } - else { - // take chars before match - strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent); - strTemp += szNew; - pCurrent = pSubstr + uiOldLen; // restart after match - - uiCount++; - - // stop now? - if ( !bReplaceAll ) { - strTemp += pCurrent; // copy the rest - break; // exit the loop - } + if ( !bReplaceAll ) + break; + } } - } + else // general case + { + const size_t uiOldLen = strOld.m_impl.length(); + const size_t uiNewLen = strNew.m_impl.length(); + + for ( size_t pos = 0; ; ) + { + pos = m_impl.find(strOld.m_impl, pos); + if ( pos == npos ) + break; + + // replace this occurrence of the old string with the new one + m_impl.replace(pos, uiOldLen, strNew.m_impl); - // only done if there were replacements, otherwise would have returned above - *this = strTemp; + // move up pos past the string that was replaced + pos += uiNewLen; + + // increase replace count + uiCount++; + + // stop after the first one? + if ( !bReplaceAll ) + break; + } + } - return uiCount; + return uiCount; } bool wxString::IsAscii() const { - const char *s = (const char*) *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 char *s = (const char*) *this; - while(*s){ - if(!isalpha(*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 char *s = (const char*) *this; - while(*s){ - if(!isdigit(*s)) return(FALSE); - s++; - } - return(TRUE); -} + if ( empty() ) + return true; + + const_iterator i = begin(); + + if ( *i == _T('-') || *i == _T('+') ) + ++i; -// kludge: we don't have declaraton of wxStringData here, so we add offsets -// manually to get to the "length" field of wxStringData structure -bool wxString::IsEmpty() const { return Len() == 0; } + for ( ; i != end(); ++i ) + { + if ( !wxIsdigit(*i) ) + return false; + } + + return true; +} wxString wxString::Strip(stripType w) const { wxString s = *this; - if ( w & leading ) s.Trim(FALSE); - if ( w & trailing ) s.Trim(TRUE); + if ( w & leading ) s.Trim(false); + if ( w & trailing ) s.Trim(true); return s; } -/// case-insensitive strcmp() (platform independent) -int Stricmp(const char *psz1, const char *psz2) -{ -#if defined(_MSC_VER) - return _stricmp(psz1, psz2); -#elif defined(__BORLANDC__) - return stricmp(psz1, psz2); -#elif defined(__UNIX__) || defined(__GNUWIN32__) - return strcasecmp(psz1, psz2); -#else - // almost all compilers/libraries provide this function (unfortunately under - // different names), that's why we don't implement our own which will surely - // be more efficient than this code (uncomment to use): - /* - register char c1, c2; - do { - c1 = tolower(*psz1++); - c2 = tolower(*psz2++); - } while ( c1 && (c1 == c2) ); - - return c1 - c2; - */ - - #error "Please define string case-insensitive compare for your OS/compiler" -#endif // OS/compiler -} - // --------------------------------------------------------------------------- // case conversion // --------------------------------------------------------------------------- wxString& wxString::MakeUpper() { - CopyBeforeWrite(); - - for ( char *p = m_pchData; *p; p++ ) - *p = (char)toupper(*p); + for ( iterator it = begin(), en = end(); it != en; ++it ) + *it = (wxChar)wxToupper(*it); return *this; } wxString& wxString::MakeLower() { - CopyBeforeWrite(); - - for ( char *p = m_pchData; *p; p++ ) - *p = (char)tolower(*p); + for ( iterator it = begin(), en = end(); it != en; ++it ) + *it = (wxChar)wxTolower(*it); return *this; } +wxString& wxString::MakeCapitalized() +{ + const iterator en = end(); + iterator it = begin(); + if ( it != en ) + { + *it = (wxChar)wxToupper(*it); + for ( ++it; it != en; ++it ) + *it = (wxChar)wxTolower(*it); + } + + return *this; +} + // --------------------------------------------------------------------------- // trimming and padding // --------------------------------------------------------------------------- +// some compilers (VC++ 6.0 not to name them) return true for a call to +// isspace('\xEA') in the C locale which seems to be broken to me, but we have +// to live with this by checking that the character is a 7 bit one - even if +// this may fail to detect some spaces (I don't know if Unicode doesn't have +// space-like symbols somewhere except in the first 128 chars), it is arguably +// still better than trimming away accented letters +inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); } + // trims spaces (in the sense of isspace) from left or right side wxString& wxString::Trim(bool bFromRight) { - CopyBeforeWrite(); + // first check if we're going to modify the string at all + if ( !empty() && + ( + (bFromRight && wxSafeIsspace(GetChar(length() - 1))) || + (!bFromRight && wxSafeIsspace(GetChar(0u))) + ) + ) + { + if ( bFromRight ) + { + // find last non-space character + reverse_iterator psz = rbegin(); + while ( (psz != rend()) && wxSafeIsspace(*psz) ) + ++psz; - if ( bFromRight ) - { - // find last non-space character - char *psz = m_pchData + GetStringData()->nDataLength - 1; - while ( isspace(*psz) && (psz >= m_pchData) ) - psz--; + // truncate at trailing space start + erase(psz.base(), end()); + } + else + { + // find first non-space character + iterator psz = begin(); + while ( (psz != end()) && wxSafeIsspace(*psz) ) + ++psz; - // truncate at trailing space start - *++psz = '\0'; - GetStringData()->nDataLength = psz - m_pchData; - } - else - { - // find first non-space character - const char *psz = m_pchData; - while ( isspace(*psz) ) - psz++; - - // fix up data and length - int nDataLength = GetStringData()->nDataLength - (psz - m_pchData); - memmove(m_pchData, psz, (nDataLength + 1)*sizeof(char)); - GetStringData()->nDataLength = nDataLength; - } + // fix up data and length + erase(begin(), psz); + } + } - return *this; + return *this; } // adds nCount characters chPad to the string from either side -wxString& wxString::Pad(size_t nCount, char chPad, bool bFromRight) +wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight) { - wxString s(chPad, nCount); + wxString s(chPad, nCount); - if ( bFromRight ) - *this += s; - else - { - s += *this; - *this = s; - } + if ( bFromRight ) + *this += s; + else + { + s += *this; + swap(s); + } - return *this; + return *this; } // truncate the string wxString& wxString::Truncate(size_t uiLen) { - *(m_pchData + uiLen) = '\0'; - GetStringData()->nDataLength = uiLen; + if ( uiLen < length() ) + { + erase(begin() + uiLen, end()); + } + //else: nothing to do, string is already short enough - return *this; + return *this; } // --------------------------------------------------------------------------- -// finding (return NOT_FOUND if not found and index otherwise) +// finding (return wxNOT_FOUND if not found and index otherwise) // --------------------------------------------------------------------------- // find a character -int wxString::Find(char ch, bool bFromEnd) const +int wxString::Find(wxUniChar ch, bool bFromEnd) const { - const char *psz = bFromEnd ? strrchr(m_pchData, ch) : strchr(m_pchData, ch); + size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch); - return (psz == NULL) ? NOT_FOUND : psz - m_pchData; + return (idx == npos) ? wxNOT_FOUND : (int)idx; } -// find a sub-string (like strstr) -int wxString::Find(const char *pszSub) const -{ - const char *psz = strstr(m_pchData, pszSub); - - return (psz == NULL) ? NOT_FOUND : psz - m_pchData; -} +// ---------------------------------------------------------------------------- +// conversion to numbers +// ---------------------------------------------------------------------------- -// --------------------------------------------------------------------------- -// formatted output -// --------------------------------------------------------------------------- -int wxString::Printf(const char *pszFormat, ...) -{ - va_list argptr; - va_start(argptr, pszFormat); +// The implementation of all the functions below is exactly the same so factor +// it out. Note that number extraction works correctly on UTF-8 strings, so +// we can use wxStringCharType and wx_str() for maximum efficiency. - int iLen = PrintfV(pszFormat, argptr); +#ifndef __WXWINCE__ + #define DO_IF_NOT_WINCE(x) x +#else + #define DO_IF_NOT_WINCE(x) +#endif - va_end(argptr); +#define WX_STRING_TO_INT_TYPE(out, base, func, T) \ + wxCHECK_MSG( out, false, _T("NULL output pointer") ); \ + wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \ + \ + DO_IF_NOT_WINCE( errno = 0; ) \ + \ + const wxStringCharType *start = wx_str(); \ + wxStringCharType *end; \ + T val = func(start, &end, base); \ + \ + /* return true only if scan was stopped by the terminating NUL and */ \ + /* if the string was not empty to start with and no under/overflow */ \ + /* occurred: */ \ + if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \ + return false; \ + *out = val; \ + return true - return iLen; +bool wxString::ToLong(long *pVal, int base) const +{ + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long); } -int wxString::PrintfV(const char* pszFormat, va_list argptr) +bool wxString::ToULong(unsigned long *pVal, int base) const { - static char s_szScratch[1024]; - - int iLen = vsprintf(s_szScratch, pszFormat, argptr); - AllocBeforeWrite(iLen); - strcpy(m_pchData, s_szScratch); - - return iLen; + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long); } -#if 0 -int wxString::Scanf(const char *pszFormat, ...) const +bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const { - va_list argptr; - va_start(argptr, pszFormat); - - int iLen = ScanfV(pszFormat, argptr); - - va_end(argptr); - - return iLen; + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t); } -int wxString::ScanfV(const char *pszFormat, va_list argptr) const +bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const { -#ifdef __WXMSW__ - wxMessageBox("ScanfV not implemented"); - return 0; -#else - return vsscanf(c_str(), pszFormat, argptr); -#endif + WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t); } -#endif -// ---------------------------------------------------------------------------- -// misc other operations -// ---------------------------------------------------------------------------- -bool wxString::Matches(const char *pszMask) const +bool wxString::ToDouble(double *pVal) const { - // check char by char - const char *pszTxt; - for ( pszTxt = c_str(); *pszMask != '\0'; pszMask++, pszTxt++ ) { - switch ( *pszMask ) { - case '?': - if ( *pszTxt == '\0' ) - return FALSE; - - pszTxt++; - pszMask++; - break; - - case '*': - { - // ignore special chars immediately following this one - while ( *pszMask == '*' || *pszMask == '?' ) - pszMask++; - - // if there is nothing more, match - if ( *pszMask == '\0' ) - return TRUE; - - // are there any other metacharacters in the mask? - uint uiLenMask; - const char *pEndMask = strpbrk(pszMask, "*?"); + wxCHECK_MSG( pVal, false, _T("NULL output pointer") ); - if ( pEndMask != NULL ) { - // we have to match the string between two metachars - uiLenMask = pEndMask - pszMask; - } - else { - // we have to match the remainder of the string - uiLenMask = strlen(pszMask); - } + DO_IF_NOT_WINCE( errno = 0; ) - wxString strToMatch(pszMask, uiLenMask); - const char* pMatch = strstr(pszTxt, strToMatch); - if ( pMatch == NULL ) - return FALSE; + const wxChar *start = c_str(); + wxChar *end; + double val = wxStrtod(start, &end); - // -1 to compensate "++" in the loop - pszTxt = pMatch + uiLenMask - 1; - pszMask += uiLenMask - 1; - } - break; + // return true only if scan was stopped by the terminating NUL and if the + // string was not empty to start with and no under/overflow occurred + if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) + return false; - default: - if ( *pszMask != *pszTxt ) - return FALSE; - break; - } - } + *pVal = val; - // match only if nothing left - return *pszTxt == '\0'; + return true; } // --------------------------------------------------------------------------- -// standard C++ library string functions +// formatted output // --------------------------------------------------------------------------- -#ifdef STD_STRING_COMPATIBILITY -wxString& wxString::insert(size_t nPos, const wxString& str) +#if !wxUSE_UTF8_LOCALE_ONLY +/* static */ +#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN +wxString wxStringPrintfMixinBase::DoFormatWchar(const wxChar *format, ...) +#else +wxString wxString::DoFormatWchar(const wxChar *format, ...) +#endif { - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nPos <= Len() ); + va_list argptr; + va_start(argptr, format); - wxString strTmp; - char *pc = strTmp.GetWriteBuf(Len() + str.Len()); - strncpy(pc, c_str(), nPos); - strcpy(pc + nPos, str); - strcpy(pc + nPos + str.Len(), c_str() + nPos); - strTmp.UngetWriteBuf(); - *this = strTmp; - - return *this; -} + wxString s; + s.PrintfV(format, argptr); -size_t wxString::find(const wxString& str, size_t nStart) const -{ - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart <= Len() ); + va_end(argptr); - const char *p = strstr(c_str() + nStart, str); - - return p == NULL ? npos : p - c_str(); + return s; } +#endif // !wxUSE_UTF8_LOCALE_ONLY -// VC++ 1.5 can't cope with the default argument in the header. -#if ! (defined(_MSC_VER) && !defined(__WIN32__)) -size_t wxString::find(const char* sz, size_t nStart, size_t n) const -{ - return find(wxString(sz, n == npos ? 0 : n), nStart); -} -#endif - -size_t wxString::find(char ch, size_t nStart) const +#if wxUSE_UNICODE_UTF8 +/* static */ +wxString wxString::DoFormatUtf8(const char *format, ...) { - wxASSERT( nStart <= Len() ); + va_list argptr; + va_start(argptr, format); - const char *p = strchr(c_str() + nStart, ch); - - return p == NULL ? npos : p - c_str(); -} + wxString s; + s.PrintfV(format, argptr); -size_t wxString::rfind(const wxString& str, size_t nStart) const -{ - wxASSERT( str.GetStringData()->IsValid() ); - wxASSERT( nStart <= Len() ); + va_end(argptr); - // # could be quicker than that - const char *p = c_str() + (nStart == npos ? Len() : nStart); - while ( p >= c_str() + str.Len() ) { - if ( strncmp(p - str.Len(), str, str.Len()) == 0 ) - return p - str.Len() - c_str(); - p--; - } - - return npos; + return s; } - -// VC++ 1.5 can't cope with the default argument in the header. -#if ! (defined(_MSC_VER) && !defined(__WIN32__)) -size_t wxString::rfind(const char* sz, size_t nStart, size_t n) const +#endif // wxUSE_UNICODE_UTF8 + +/* static */ +wxString wxString::FormatV(const wxString& format, va_list argptr) { - return rfind(wxString(sz, n == npos ? 0 : n), nStart); + wxString s; + s.PrintfV(format, argptr); + return s; } -size_t wxString::rfind(char ch, size_t nStart) const +#if !wxUSE_UTF8_LOCALE_ONLY +#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN +int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...) +#else +int wxString::DoPrintfWchar(const wxChar *format, ...) +#endif { - wxASSERT( nStart <= Len() ); + va_list argptr; + va_start(argptr, format); - const char *p = strrchr(c_str() + nStart, ch); - - return p == NULL ? npos : p - c_str(); -} +#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN + // get a pointer to the wxString instance; we have to use dynamic_cast<> + // because it's the only cast that works safely for downcasting when + // multiple inheritance is used: + wxString *str = static_cast(this); +#else + wxString *str = this; #endif -wxString wxString::substr(size_t nStart, size_t nLen) const -{ - // npos means 'take all' - if ( nLen == npos ) - nLen = 0; + int iLen = str->PrintfV(format, argptr); - wxASSERT( nStart + nLen <= Len() ); + va_end(argptr); - return wxString(c_str() + nStart, nLen == npos ? 0 : nLen); + return iLen; } +#endif // !wxUSE_UTF8_LOCALE_ONLY -wxString& wxString::erase(size_t nStart, size_t nLen) +#if wxUSE_UNICODE_UTF8 +int wxString::DoPrintfUtf8(const char *format, ...) { - wxString strTmp(c_str(), nStart); - if ( nLen != npos ) { - wxASSERT( nStart + nLen <= Len() ); - - strTmp.append(c_str() + nStart + nLen); - } + va_list argptr; + va_start(argptr, format); - *this = strTmp; - return *this; -} + int iLen = PrintfV(format, argptr); -wxString& wxString::replace(size_t nStart, size_t nLen, const char *sz) -{ - wxASSERT( nStart + nLen <= Strlen(sz) ); + va_end(argptr); - wxString strTmp; - if ( nStart != 0 ) - strTmp.append(c_str(), nStart); - strTmp += sz; - strTmp.append(c_str() + nStart + nLen); - - *this = strTmp; - return *this; + return iLen; } +#endif // wxUSE_UNICODE_UTF8 -wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, char ch) +/* + Uses wxVsnprintf and places the result into the this string. + + In ANSI build, wxVsnprintf is effectively vsnprintf but in Unicode build + it is vswprintf. Due to a discrepancy between vsnprintf and vswprintf in + the ISO C99 (and thus SUSv3) standard the return value for the case of + an undersized buffer is inconsistent. For conforming vsnprintf + implementations the function must return the number of characters that + would have been printed had the buffer been large enough. For conforming + vswprintf implementations the function must return a negative number + and set errno. + + What vswprintf sets errno to is undefined but Darwin seems to set it to + EOVERFLOW. The only expected errno are EILSEQ and EINVAL. Both of + those are defined in the standard and backed up by several conformance + statements. Note that ENOMEM mentioned in the manual page does not + apply to swprintf, only wprintf and fwprintf. + + Official manual page: + http://www.opengroup.org/onlinepubs/009695399/functions/swprintf.html + + Some conformance statements (AIX, Solaris): + http://www.opengroup.org/csq/view.mhtml?RID=ibm%2FSD1%2F3 + http://www.theopengroup.org/csq/view.mhtml?norationale=1&noreferences=1&RID=Fujitsu%2FSE2%2F10 + + Since EILSEQ and EINVAL are rather common but EOVERFLOW is not and since + EILSEQ and EINVAL are specifically defined to mean the error is other than + an undersized buffer and no other errno are defined we treat those two + as meaning hard errors and everything else gets the old behavior which + is to keep looping and increasing buffer size until the function succeeds. + + In practice it's impossible to determine before compilation which behavior + may be used. The vswprintf function may have vsnprintf-like behavior or + vice-versa. Behavior detected on one release can theoretically change + with an updated release. Not to mention that configure testing for it + would require the test to be run on the host system, not the build system + which makes cross compilation difficult. Therefore, we make no assumptions + about behavior and try our best to handle every known case, including the + case where wxVsnprintf returns a negative number and fails to set errno. + + There is yet one more non-standard implementation and that is our own. + Fortunately, that can be detected at compile-time. + + On top of all that, ISO C99 explicitly defines snprintf to write a null + character to the last position of the specified buffer. That would be at + at the given buffer size minus 1. It is supposed to do this even if it + turns out that the buffer is sized too small. + + Darwin (tested on 10.5) follows the C99 behavior exactly. + + Glibc 2.6 almost follows the C99 behavior except vswprintf never sets + errno even when it fails. However, it only seems to ever fail due + to an undersized buffer. +*/ +#if wxUSE_UNICODE_UTF8 +template +#else +// we only need one version in non-UTF8 builds and at least two Windows +// compilers have problems with this function template, so use just one +// normal function here +#endif +static int DoStringPrintfV(wxString& str, + const wxString& format, va_list argptr) { - return replace(nStart, nLen, wxString(ch, nCount)); -} + int size = 1024; -wxString& wxString::replace(size_t nStart, size_t nLen, - const wxString& str, size_t nStart2, size_t nLen2) -{ - return replace(nStart, nLen, str.substr(nStart2, nLen2)); -} + for ( ;; ) + { +#if wxUSE_UNICODE_UTF8 + BufferType tmp(str, size + 1); + typename BufferType::CharType *buf = tmp; +#else + wxStringBuffer tmp(str, size + 1); + wxChar *buf = tmp; +#endif -wxString& wxString::replace(size_t nStart, size_t nLen, - const char* sz, size_t nCount) -{ - return replace(nStart, nLen, wxString(sz, nCount)); -} + if ( !buf ) + { + // out of memory -#endif //std::string compatibility + // in UTF-8 build, leaving uninitialized junk in the buffer + // could result in invalid non-empty UTF-8 string, so just + // reset the string to empty on failure: + buf[0] = '\0'; + return -1; + } -// ============================================================================ -// ArrayString -// ============================================================================ + // wxVsnprintf() may modify the original arg pointer, so pass it + // only a copy + va_list argptrcopy; + wxVaCopy(argptrcopy, argptr); -// size increment = max(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) +#ifndef __WXWINCE__ + // Set errno to 0 to make it determinate if wxVsnprintf fails to set it. + errno = 0; +#endif + int len = wxVsnprintf(buf, size, format, argptrcopy); + va_end(argptrcopy); + + // some implementations of vsnprintf() don't NUL terminate + // the string if there is not enough space for it so + // always do it manually + // FIXME: This really seems to be the wrong and would be an off-by-one + // bug except the code above allocates an extra character. + buf[size] = _T('\0'); + + // vsnprintf() may return either -1 (traditional Unix behaviour) or the + // total number of characters which would have been written if the + // buffer were large enough (newer standards such as Unix98) + if ( len < 0 ) + { + // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or + // wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF + // is true if *both* of them use our own implementation, + // otherwise we can't be sure +#if wxUSE_WXVSNPRINTF + // we know that our own implementation of wxVsnprintf() returns -1 + // only for a format error - thus there's something wrong with + // the user's format string + buf[0] = '\0'; + return -1; +#else // possibly using system version + // assume it only returns error if there is not enough space, but + // as we don't know how much we need, double the current size of + // the buffer +#ifndef __WXWINCE__ + if( (errno == EILSEQ) || (errno == EINVAL) ) + // If errno was set to one of the two well-known hard errors + // then fail immediately to avoid an infinite loop. + return -1; + else +#endif // __WXWINCE__ + // still not enough, as we don't know how much we need, double the + // current size of the buffer + size *= 2; +#endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF + } + else if ( len >= size ) + { +#if wxUSE_WXVSNPRINTF + // we know that our own implementation of wxVsnprintf() returns + // size+1 when there's not enough space but that's not the size + // of the required buffer! + size *= 2; // so we just double the current size of the buffer +#else + // some vsnprintf() implementations NUL-terminate the buffer and + // some don't in len == size case, to be safe always add 1 + // FIXME: I don't quite understand this comment. The vsnprintf + // function is specifically defined to return the number of + // characters printed not including the null terminator. + // So OF COURSE you need to add 1 to get the right buffer size. + // The following line is definitely correct, no question. + size = len + 1; #endif + } + else // ok, there was enough space + { + break; + } + } -#define STRING(p) ((wxString *)(&(p))) + // we could have overshot + str.Shrink(); -// ctor -wxArrayString::wxArrayString() -{ - m_nSize = - m_nCount = 0; - m_pItems = NULL; + return str.length(); } -// copy ctor -wxArrayString::wxArrayString(const wxArrayString& src) +int wxString::PrintfV(const wxString& format, va_list argptr) { - m_nSize = src.m_nSize; - m_nCount = src.m_nCount; - - if ( m_nSize != 0 ) - m_pItems = new char *[m_nSize]; - else - m_pItems = NULL; +#if wxUSE_UNICODE_UTF8 + #if wxUSE_STL_BASED_WXSTRING + typedef wxStringTypeBuffer Utf8Buffer; + #else + typedef wxStringInternalBuffer Utf8Buffer; + #endif +#endif - if ( m_nCount != 0 ) - memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(char *)); +#if wxUSE_UTF8_LOCALE_ONLY + return DoStringPrintfV(*this, format, argptr); +#else + #if wxUSE_UNICODE_UTF8 + if ( wxLocaleIsUtf8 ) + return DoStringPrintfV(*this, format, argptr); + else + // wxChar* version + return DoStringPrintfV(*this, format, argptr); + #else + return DoStringPrintfV(*this, format, argptr); + #endif // UTF8/WCHAR +#endif } -// copy operator -wxArrayString& wxArrayString::operator=(const wxArrayString& src) -{ - DELETEA(m_pItems); - - m_nSize = src.m_nSize; - m_nCount = src.m_nCount; - - if ( m_nSize != 0 ) - m_pItems = new char *[m_nSize]; - else - m_pItems = NULL; - - if ( m_nCount != 0 ) - memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(char *)); +// ---------------------------------------------------------------------------- +// misc other operations +// ---------------------------------------------------------------------------- - return *this; -} +// 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 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 + // here requires always linking with libregex even if the user code doesn't + // use it +#if 0 // 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('$'): + case _T('('): + case _T(')'): + case _T('|'): + case _T('+'): + case _T('\\'): + // these characters are special in a RE, quote them + // (however note that we don't quote '[' and ']' to allow + // using them for Unix shell like matching) + pattern += _T('\\'); + // fall through + + default: + pattern += *pszMask; + } -// grow the array -void wxArrayString::Grow() -{ - // only do it if no more place - if( m_nCount == m_nSize ) { - if( m_nSize == 0 ) { - // was empty, alloc some memory - m_nSize = ARRAY_DEFAULT_INITIAL_SIZE; - m_pItems = new char *[m_nSize]; + pszMask++; } - else { - // add 50% but not too much - size_t nIncrement = m_nSize >> 1; - if ( nIncrement > ARRAY_MAXSIZE_INCREMENT ) - nIncrement = ARRAY_MAXSIZE_INCREMENT; - m_nSize += nIncrement; - char **pNew = new char *[m_nSize]; - - // copy data to new location - memcpy(pNew, m_pItems, m_nCount*sizeof(char *)); + pattern += _T('$'); + + // and now use it + return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str()); +#else // !wxUSE_REGEX + // TODO: this is, of course, awfully inefficient... + + // 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 = wx_str(); +#endif - // delete old memory (but do not release the strings!) - DELETEA(m_pItems); + // the last location where '*' matched + const wxChar *pszLastStarInText = NULL; + const wxChar *pszLastStarInMask = NULL; - m_pItems = pNew; - } - } -} +match: + for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) { + switch ( *pszMask ) { + case wxT('?'): + if ( *pszTxt == wxT('\0') ) + return false; -void wxArrayString::Free() -{ - for ( size_t n = 0; n < m_nCount; n++ ) { - STRING(m_pItems[n])->GetStringData()->Unlock(); - } -} + // pszTxt and pszMask will be incremented in the loop statement -// deletes all the strings from the list -void wxArrayString::Empty() -{ - Free(); + break; - m_nCount = 0; -} + case wxT('*'): + { + // remember where we started to be able to backtrack later + pszLastStarInText = pszTxt; + pszLastStarInMask = pszMask; -// as Empty, but also frees memory -void wxArrayString::Clear() -{ - Free(); + // ignore special chars immediately following this one + // (should this be an error?) + while ( *pszMask == wxT('*') || *pszMask == wxT('?') ) + pszMask++; - m_nSize = - m_nCount = 0; + // if there is nothing more, match + if ( *pszMask == wxT('\0') ) + return true; - DELETEA(m_pItems); - m_pItems = NULL; -} + // are there any other metacharacters in the mask? + size_t uiLenMask; + const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?")); -// dtor -wxArrayString::~wxArrayString() -{ - Free(); + if ( pEndMask != NULL ) { + // we have to match the string between two metachars + uiLenMask = pEndMask - pszMask; + } + else { + // we have to match the remainder of the string + uiLenMask = wxStrlen(pszMask); + } - DELETEA(m_pItems); -} + wxString strToMatch(pszMask, uiLenMask); + const wxChar* pMatch = wxStrstr(pszTxt, strToMatch); + if ( pMatch == NULL ) + return false; -// pre-allocates memory (frees the previous data!) -void wxArrayString::Alloc(size_t nSize) -{ - wxASSERT( nSize > 0 ); + // -1 to compensate "++" in the loop + pszTxt = pMatch + uiLenMask - 1; + pszMask += uiLenMask - 1; + } + break; - // only if old buffer was not big enough - if ( nSize > m_nSize ) { - Free(); - DELETEA(m_pItems); - m_pItems = new char *[nSize]; - m_nSize = nSize; + default: + if ( *pszMask != *pszTxt ) + return false; + break; + } } - m_nCount = 0; -} + // match only if nothing left + if ( *pszTxt == wxT('\0') ) + return true; -// searches the array for an item (forward or backwards) + // if we failed to match, backtrack if we can + if ( pszLastStarInText ) { + pszTxt = pszLastStarInText + 1; + pszMask = pszLastStarInMask; -// Robert Roebling (changed to bool from bool) + pszLastStarInText = NULL; -int wxArrayString::Index(const char *sz, bool bCase, bool bFromEnd) const -{ - if ( bFromEnd ) { - if ( m_nCount > 0 ) { - uint ui = m_nCount; - do { - if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) ) - return ui; - } - while ( ui != 0 ); - } - } - else { - for( uint ui = 0; ui < m_nCount; ui++ ) { - if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) ) - return ui; - } + // don't bother resetting pszLastStarInMask, it's unnecessary + + goto match; } - return NOT_FOUND; + return false; +#endif // wxUSE_REGEX/!wxUSE_REGEX } -// add item at the end -void wxArrayString::Add(const wxString& str) +// Count the number of chars +int wxString::Freq(wxUniChar ch) const { - wxASSERT( str.GetStringData()->IsValid() ); - - Grow(); - - // the string data must not be deleted! - str.GetStringData()->Lock(); - m_pItems[m_nCount++] = (char *)str.c_str(); + int count = 0; + for ( const_iterator i = begin(); i != end(); ++i ) + { + if ( *i == ch ) + count ++; + } + return count; } -// add item at the given position -void wxArrayString::Insert(const wxString& str, size_t nIndex) -{ - wxASSERT( str.GetStringData()->IsValid() ); - - wxCHECK_RET( nIndex <= m_nCount, "bad index in wxArrayString::Insert" ); - - Grow(); - - memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex], - (m_nCount - nIndex)*sizeof(char *)); - - str.GetStringData()->Lock(); - m_pItems[nIndex] = (char *)str.c_str(); - - m_nCount++; -} +// ---------------------------------------------------------------------------- +// wxUTF8StringBuffer +// ---------------------------------------------------------------------------- -// removes item from array (by index) -void wxArrayString::Remove(size_t nIndex) +#if wxUSE_UNICODE_WCHAR +wxUTF8StringBuffer::~wxUTF8StringBuffer() { - wxCHECK_RET( nIndex <= m_nCount, "bad index in wxArrayString::Remove" ); + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); - // release our lock - Item(nIndex).GetStringData()->Unlock(); - - memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1], - (m_nCount - nIndex - 1)*sizeof(char *)); - m_nCount--; + wxStringInternalBuffer wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf); } -// removes item from array (by value) -void wxArrayString::Remove(const char *sz) +wxUTF8StringBufferLength::~wxUTF8StringBufferLength() { - int iIndex = Index(sz); + wxCHECK_RET(m_lenSet, "length not set"); - wxCHECK_RET( iIndex != NOT_FOUND, - "removing inexistent element in wxArrayString::Remove" ); + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf, m_len); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); - Remove((size_t)iIndex); + wxStringInternalBufferLength wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf, m_len); + wbuf.SetLength(wlen); } +#endif // wxUSE_UNICODE_WCHAR -// sort array elements using passed comparaison function +// ---------------------------------------------------------------------------- +// wxCharBufferType +// ---------------------------------------------------------------------------- -void wxArrayString::Sort(bool bCase, bool bReverse) -{ - //@@@@ TO DO - //qsort(m_pItems, m_nCount, sizeof(char *), fCmp); -} +#ifndef __VMS_BROKEN_TEMPLATES +template<> +#endif +wxCharTypeBuffer::Data +wxCharTypeBuffer::NullData(NULL); + +#ifndef __VMS_BROKEN_TEMPLATES +template<> +#endif +wxCharTypeBuffer::Data +wxCharTypeBuffer::NullData(NULL);