/////////////////////////////////////////////////////////////////////////////
-// Name: string.cpp
+// Name: src/common/string.cpp
// Purpose: wxString class
// Author: Vadim Zeitlin, Ryan Norton
// Modified by:
// Licence: wxWindows licence
/////////////////////////////////////////////////////////////////////////////
-/*
- * 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
// ===========================================================================
#include "wx/wxprec.h"
#ifdef __BORLANDC__
- #pragma hdrstop
+ #pragma hdrstop
#endif
#ifndef WX_PRECOMP
- #include "wx/defs.h"
- #include "wx/string.h"
- #include "wx/intl.h"
- #include "wx/thread.h"
+ #include "wx/string.h"
+ #include "wx/wxcrtvararg.h"
+ #include "wx/intl.h"
+ #include "wx/log.h"
#endif
#include <ctype.h>
+
+#ifndef __WXWINCE__
+ #include <errno.h>
+#endif
+
#include <string.h>
#include <stdlib.h>
-#ifdef __SALFORDC__
- #include <clib.h>
+#include "wx/hashmap.h"
+#include "wx/vector.h"
+#include "wx/xlocale.h"
+
+#ifdef __WXMSW__
+ #include "wx/msw/wrapwin.h"
+#endif // __WXMSW__
+
+#if wxUSE_STD_IOSTREAM
+ #include <sstream>
+#endif
+
+// 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
-// allocating extra space for each string consumes more memory but speeds up
-// the concatenation operations (nLen is the current string's length)
-// NB: EXTRA_ALLOC must be >= 0!
-#define EXTRA_ALLOC (19 - nLen % 16)
+// define a function declared in wx/buffer.h here as we don't have buffer.cpp
+// and don't want to add it just because of this simple function
+namespace wxPrivate
+{
+
+// wxXXXBuffer classes can be (implicitly) used during global statics
+// initialization so wrap the status UntypedBufferData variable in a function
+// to make it safe to access it even before all global statics are initialized
+UntypedBufferData *GetUntypedNullData()
+{
+ static UntypedBufferData s_untypedNullData(NULL, 0);
+
+ return &s_untypedNullData;
+}
+
+} // namespace wxPrivate
// ---------------------------------------------------------------------------
// static class variables definition
// ---------------------------------------------------------------------------
-#if !wxUSE_STL
- //According to STL _must_ be a -1 size_t
- const size_t wxStringBase::npos = (size_t) -1;
-#endif
+//According to STL _must_ be a -1 size_t
+const size_t wxString::npos = (size_t) -1;
-// ----------------------------------------------------------------------------
-// static data
-// ----------------------------------------------------------------------------
+#if wxUSE_STRING_POS_CACHE
-#if wxUSE_STL
+#ifdef wxHAS_COMPILER_TLS
-extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T("");
+wxTLS_TYPE(wxString::Cache) wxString::ms_cache;
-#else
+#else // !wxHAS_COMPILER_TLS
-// for an empty string, GetStringData() will return this address: this
-// structure has the same layout as wxStringData and it's data() method will
-// return the empty string (dummy pointer)
-static const struct
+struct wxStrCacheInitializer
{
- wxStringData data;
- wxChar dummy;
-} g_strEmpty = { {-1, 0, 0}, wxT('\0') };
-
-// empty C style string: points to 'string data' byte of g_strEmpty
-extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy;
+ wxStrCacheInitializer()
+ {
+ // calling this function triggers s_cache initialization in it, and
+ // from now on it becomes safe to call from multiple threads
+ wxString::GetCache();
+ }
+};
-#endif
+/*
+wxString::Cache& wxString::GetCache()
+{
+ static wxTLS_TYPE(Cache) s_cache;
-// ----------------------------------------------------------------------------
-// global functions
-// ----------------------------------------------------------------------------
+ return wxTLS_VALUE(s_cache);
+}
+*/
-#if wxUSE_STD_IOSTREAM
+static wxStrCacheInitializer gs_stringCacheInit;
-// 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!
+#endif // wxHAS_COMPILER_TLS/!wxHAS_COMPILER_TLS
-#include <iostream>
+// 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
+#if wxDEBUG_LEVEL >= 2
-wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str))
+struct wxStrCacheDumper
{
-#if 0
- int w = is.width(0);
- if ( is.ipfx(0) ) {
- streambuf *sb = is.rdbuf();
- str.erase();
- while ( true ) {
- int ch = sb->sbumpc ();
- if ( ch == EOF ) {
- is.setstate(ios::eofbit);
- break;
- }
- else if ( isspace(ch) ) {
- sb->sungetc();
- break;
- }
-
- str += ch;
- if ( --w == 1 )
- break;
+ 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);
+ }
}
- }
+};
- is.isfx();
- if ( str.length() == 0 )
- is.setstate(ios::failbit);
-#endif
- return is;
-}
+void wxDumpStrCache() { wxStrCacheDumper::ShowAll(); }
-wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
-{
-#ifdef __BORLANDC__
- os << str.mb_str();
-#else
- os << str.c_str();
-#endif
- return os;
-}
+#endif // wxDEBUG_LEVEL >= 2
-#endif // wxUSE_STD_IOSTREAM
+#ifdef wxPROFILE_STRING_CACHE
-// ----------------------------------------------------------------------------
-// private classes
-// ----------------------------------------------------------------------------
+wxString::CacheStats wxString::ms_cacheStats;
-// this small class is used to gather statistics for performance tuning
-//#define WXSTRING_STATISTICS
-#ifdef WXSTRING_STATISTICS
- class Averager
- {
- public:
- Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
- ~Averager()
- { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
-
- void Add(size_t n) { m_nTotal += n; m_nCount++; }
-
- private:
- size_t m_nCount, m_nTotal;
- const wxChar *m_sz;
- } g_averageLength("allocation size"),
- g_averageSummandLength("summand length"),
- g_averageConcatHit("hit probability in concat"),
- g_averageInitialLength("initial string length");
-
- #define STATISTICS_ADD(av, val) g_average##av.Add(val)
-#else
- #define STATISTICS_ADD(av, val)
-#endif // WXSTRING_STATISTICS
+struct wxStrCacheStatsDumper
+{
+ ~wxStrCacheStatsDumper()
+ {
+ const wxString::CacheStats& stats = wxString::ms_cacheStats;
-#if !wxUSE_STL
+ 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);
+ }
-// ===========================================================================
-// wxStringData class deallocation
-// ===========================================================================
+ if ( stats.lentot )
+ {
+ printf("\tNumber of calls to length(): %u, hits=%.2f%%\n",
+ stats.lentot, 100.*float(stats.lenhits)/stats.lentot);
+ }
+ }
+};
-#if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
-# pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
-void wxStringData::Free()
-{
- free(this);
-}
-#endif
+static wxStrCacheStatsDumper s_showCacheStats;
-// ===========================================================================
-// wxStringBase
-// ===========================================================================
+#endif // wxPROFILE_STRING_CACHE
-// takes nLength elements of psz starting at nPos
-void wxStringBase::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
-{
- Init();
+#endif // wxUSE_STRING_POS_CACHE
+
+// ----------------------------------------------------------------------------
+// global functions
+// ----------------------------------------------------------------------------
- // if the length is not given, assume the string to be NUL terminated
- if ( nLength == npos ) {
- wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
+#if wxUSE_STD_IOSTREAM
- nLength = wxStrlen(psz + nPos);
- }
+#include <iostream>
- STATISTICS_ADD(InitialLength, nLength);
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
+{
+#if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8
+ const wxScopedCharBuffer buf(str.AsCharBuf());
+ if ( !buf )
+ os.clear(wxSTD ios_base::failbit);
+ else
+ os << buf.data();
- if ( nLength > 0 ) {
- // trailing '\0' is written in AllocBuffer()
- if ( !AllocBuffer(nLength) ) {
- wxFAIL_MSG( _T("out of memory in wxStringBase::InitWith") );
- return;
- }
- wxTmemcpy(m_pchData, psz + nPos, nLength);
- }
+ return os;
+#else
+ return os << str.AsInternal();
+#endif
}
-// poor man's iterators are "void *" pointers
-wxStringBase::wxStringBase(const void *pStart, const void *pEnd)
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
{
- InitWith((const wxChar *)pStart, 0,
- (const wxChar *)pEnd - (const wxChar *)pStart);
+ return os << str.c_str();
}
-wxStringBase::wxStringBase(size_type n, wxChar ch)
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxScopedCharBuffer& str)
{
- Init();
- append(n, ch);
+ return os << str.data();
}
-// ---------------------------------------------------------------------------
-// memory allocation
-// ---------------------------------------------------------------------------
-
-// allocates memory needed to store a C string of length nLen
-bool wxStringBase::AllocBuffer(size_t nLen)
+#ifndef __BORLANDC__
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxScopedWCharBuffer& str)
{
- // allocating 0 sized buffer doesn't make sense, all empty strings should
- // reuse g_strEmpty
- wxASSERT( nLen > 0 );
-
- // make sure that we don't overflow
- wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
- (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
-
- STATISTICS_ADD(Length, nLen);
-
- // allocate memory:
- // 1) one extra character for '\0' termination
- // 2) sizeof(wxStringData) for housekeeping info
- wxStringData* pData = (wxStringData*)
- malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
+ return os << str.data();
+}
+#endif
- if ( pData == NULL ) {
- // allocation failures are handled by the caller
- return false;
- }
+#if wxUSE_UNICODE && defined(HAVE_WOSTREAM)
- pData->nRefs = 1;
- pData->nDataLength = nLen;
- pData->nAllocLength = nLen + EXTRA_ALLOC;
- m_pchData = pData->data(); // data starts after wxStringData
- m_pchData[nLen] = wxT('\0');
- return true;
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str)
+{
+ return wos << str.wc_str();
}
-// must be called before changing this string
-bool wxStringBase::CopyBeforeWrite()
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str)
{
- wxStringData* pData = GetStringData();
-
- if ( pData->IsShared() ) {
- pData->Unlock(); // memory not freed because shared
- size_t nLen = pData->nDataLength;
- if ( !AllocBuffer(nLen) ) {
- // allocation failures are handled by the caller
- return false;
- }
- wxTmemcpy(m_pchData, pData->data(), nLen);
- }
-
- wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner
-
- return true;
+ return wos << str.AsWChar();
}
-// must be called before replacing contents of this string
-bool wxStringBase::AllocBeforeWrite(size_t nLen)
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxScopedWCharBuffer& str)
{
- wxASSERT( nLen != 0 ); // doesn't make any sense
+ return wos << str.data();
+}
- // must not share string and must have enough space
- wxStringData* pData = GetStringData();
- if ( pData->IsShared() || pData->IsEmpty() ) {
- // can't work with old buffer, get new one
- pData->Unlock();
- if ( !AllocBuffer(nLen) ) {
- // allocation failures are handled by the caller
- return false;
- }
- }
- else {
- if ( nLen > pData->nAllocLength ) {
- // realloc the buffer instead of calling malloc() again, this is more
- // efficient
- STATISTICS_ADD(Length, nLen);
+#endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM)
- nLen += EXTRA_ALLOC;
+#endif // wxUSE_STD_IOSTREAM
- pData = (wxStringData*)
- realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
+// ===========================================================================
+// wxString class core
+// ===========================================================================
- if ( pData == NULL ) {
- // allocation failures are handled by the caller
- // keep previous data since reallocation failed
- return false;
- }
+#if wxUSE_UNICODE_UTF8
- pData->nAllocLength = nLen;
- m_pchData = pData->data();
+void wxString::PosLenToImpl(size_t pos, size_t len,
+ size_t *implPos, size_t *implLen) const
+{
+ if ( pos == npos )
+ {
+ *implPos = npos;
}
- }
-
- wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner
-
- // it doesn't really matter what the string length is as it's going to be
- // overwritten later but, for extra safety, set it to 0 for now as we may
- // have some junk in m_pchData
- GetStringData()->nDataLength = 0;
+ 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;
+ }
- return true;
+ *implLen = i.impl() - b.impl();
+ }
+ }
}
-wxStringBase& wxStringBase::append(size_t n, wxChar ch)
-{
- size_type len = length();
+#endif // wxUSE_UNICODE_UTF8
- if ( !Alloc(len + n) || !CopyBeforeWrite() ) {
- wxFAIL_MSG( _T("out of memory in wxStringBase::append") );
- }
- GetStringData()->nDataLength = len + n;
- m_pchData[len + n] = '\0';
- for ( size_t i = 0; i < n; ++i )
- m_pchData[len + i] = ch;
- return *this;
-}
+// ----------------------------------------------------------------------------
+// wxCStrData converted strings caching
+// ----------------------------------------------------------------------------
-void wxStringBase::resize(size_t nSize, wxChar ch)
-{
- size_t len = length();
+// 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
- if ( nSize < len )
- {
- erase(begin() + nSize, end());
- }
- else if ( nSize > len )
+// 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<typename T>
+static inline void DeleteStringFromConversionCache(T& hash, const wxString *s)
+{
+ typename T::iterator i = hash.find(wxConstCast(s, wxString));
+ if ( i != hash.end() )
{
- append(nSize - len, ch);
+ free(i->second);
+ hash.erase(i);
}
- //else: we have exactly the specified length, nothing to do
}
-// allocate enough memory for nLen characters
-bool wxStringBase::Alloc(size_t nLen)
+#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
{
- wxStringData *pData = GetStringData();
- if ( pData->nAllocLength <= nLen ) {
- if ( pData->IsEmpty() ) {
- nLen += EXTRA_ALLOC;
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsCharCache, m_str);
- pData = (wxStringData *)
- malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
+ // convert the string and keep it:
+ const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
+ m_str->mb_str().release();
- if ( pData == NULL ) {
- // allocation failure handled by caller
- return false;
- }
+ return s + m_offset;
+}
+#endif // wxUSE_UNICODE
- pData->nRefs = 1;
- pData->nDataLength = 0;
- pData->nAllocLength = nLen;
- m_pchData = pData->data(); // data starts after wxStringData
- m_pchData[0u] = wxT('\0');
- }
- else if ( pData->IsShared() ) {
- pData->Unlock(); // memory not freed because shared
- size_t nOldLen = pData->nDataLength;
- if ( !AllocBuffer(nLen) ) {
- // allocation failure handled by caller
- return false;
- }
- // +1 to copy the terminator, too
- memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxChar));
- GetStringData()->nDataLength = nOldLen;
- }
- else {
- nLen += EXTRA_ALLOC;
+#if !wxUSE_UNICODE_WCHAR
+WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
+ wxStringWCharConversionCache);
+static wxStringWCharConversionCache gs_stringsWCharCache;
- pData = (wxStringData *)
- realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
+const wchar_t* wxCStrData::AsWChar() const
+{
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
- if ( pData == NULL ) {
- // allocation failure handled by caller
- // keep previous data since reallocation failed
- return false;
- }
+ // convert the string and keep it:
+ const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
+ m_str->wc_str().release();
- // it's not important if the pointer changed or not (the check for this
- // is not faster than assigning to m_pchData in all cases)
- pData->nAllocLength = nLen;
- m_pchData = pData->data();
- }
- }
- //else: we've already got enough
- return true;
+ return s + m_offset;
}
+#endif // !wxUSE_UNICODE_WCHAR
-wxStringBase::iterator wxStringBase::begin()
+wxString::~wxString()
{
- if (length() > 0)
- CopyBeforeWrite();
- return m_pchData;
+#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
-wxStringBase::iterator wxStringBase::end()
-{
- if (length() > 0)
- CopyBeforeWrite();
- return m_pchData + length();
-}
+// ===========================================================================
+// wxString class core
+// ===========================================================================
-wxStringBase::iterator wxStringBase::erase(iterator it)
-{
- size_type idx = it - begin();
- erase(idx, 1);
- return begin() + idx;
-}
+// ---------------------------------------------------------------------------
+// construction and conversion
+// ---------------------------------------------------------------------------
-wxStringBase& wxStringBase::erase(size_t nStart, size_t nLen)
+#if wxUSE_UNICODE_WCHAR
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+ const wxMBConv& conv)
{
- wxASSERT(nStart <= length());
- size_t strLen = length() - nStart;
- // delete nLen or up to the end of the string characters
- nLen = strLen < nLen ? strLen : nLen;
- wxString strTmp(c_str(), nStart);
- strTmp.append(c_str() + nStart + nLen, length() - nStart - nLen);
+ // anything to do?
+ if ( !psz || nLength == 0 )
+ return SubstrBufFromMB(wxWCharBuffer(L""), 0);
- swap(strTmp);
- return *this;
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
+
+ size_t wcLen;
+ wxScopedWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ if ( !wcLen )
+ return SubstrBufFromMB(wxWCharBuffer(L""), 0);
+ else
+ return SubstrBufFromMB(wcBuf, wcLen);
}
+#endif // wxUSE_UNICODE_WCHAR
-wxStringBase& wxStringBase::insert(size_t nPos, const wxChar *sz, size_t n)
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+ const wxMBConv& conv)
{
- wxASSERT( nPos <= length() );
-
- if ( n == npos ) n = wxStrlen(sz);
- if ( n == 0 ) return *this;
+ // anything to do?
+ if ( !psz || nLength == 0 )
+ return SubstrBufFromMB(wxCharBuffer(""), 0);
- if ( !Alloc(length() + n) || !CopyBeforeWrite() ) {
- wxFAIL_MSG( _T("out of memory in wxStringBase::insert") );
+ // 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(wxScopedCharBuffer::CreateNonOwned(psz, nLength),
+ nLength);
+ }
+ // else: do the roundtrip through wchar_t*
}
- memmove(m_pchData + nPos + n, m_pchData + nPos,
- (length() - nPos) * sizeof(wxChar));
- memcpy(m_pchData + nPos, sz, n * sizeof(wxChar));
- GetStringData()->nDataLength = length() + n;
- m_pchData[length()] = '\0';
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
- return *this;
-}
+ // first convert to wide string:
+ size_t wcLen;
+ wxScopedWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ if ( !wcLen )
+ return SubstrBufFromMB(wxCharBuffer(""), 0);
-void wxStringBase::swap(wxStringBase& str)
-{
- wxChar* tmp = str.m_pchData;
- str.m_pchData = m_pchData;
- m_pchData = tmp;
+ // 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, wxT("conversion to UTF-8 failed") );
+
+ return buf;
}
+#endif // wxUSE_UNICODE_UTF8
-size_t wxStringBase::find(const wxStringBase& str, size_t nStart) const
+#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
+/* static */
+wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
+ const wxMBConv& conv)
{
- wxASSERT( str.GetStringData()->IsValid() );
- wxASSERT( nStart <= length() );
+ // anything to do?
+ if ( !pwz || nLength == 0 )
+ return SubstrBufFromWC(wxCharBuffer(""), 0);
- //anchor
- const wxChar* p = (const wxChar*)wxTmemchr(c_str() + nStart,
- str.c_str()[0],
- length() - nStart);
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
- if(!p)
- return npos;
+ size_t mbLen;
+ wxScopedCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
+ if ( !mbLen )
+ return SubstrBufFromWC(wxCharBuffer(""), 0);
+ else
+ return SubstrBufFromWC(mbBuf, mbLen);
+}
+#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
- while(p - c_str() + str.length() <= length() &&
- wxTmemcmp(p, str.c_str(), str.length()) )
+// This std::string::c_str()-like method returns a wide char pointer to string
+// contents. In wxUSE_UNICODE_WCHAR case it is trivial as it can simply return
+// a pointer to the internal representation. Otherwise a conversion is required
+// and it returns a temporary buffer.
+//
+// However for compatibility with c_str() and to avoid breaking existing code
+// doing
+//
+// for ( const wchar_t *p = s.wc_str(); *p; p++ )
+// ... use *p...
+//
+// we actually need to ensure that the returned buffer is _not_ temporary and
+// so we use wxString::m_convertedToWChar to store the returned data
+#if !wxUSE_UNICODE_WCHAR
+
+const wchar_t *wxString::AsWChar(const wxMBConv& conv) const
+{
+ const char * const strMB = m_impl.c_str();
+ const size_t lenMB = m_impl.length();
+
+ // find out the size of the buffer needed
+ const size_t lenWC = conv.ToWChar(NULL, 0, strMB, lenMB);
+ if ( lenWC == wxCONV_FAILED )
+ return NULL;
+
+ // keep the same buffer if the string size didn't change: this is not only
+ // an optimization but also ensure that code which modifies string
+ // character by character (without changing its length) can continue to use
+ // the pointer returned by a previous wc_str() call even after changing the
+ // string
+
+ // TODO-UTF8: we could check for ">" instead of "!=" here as this would
+ // allow to save on buffer reallocations but at the cost of
+ // consuming (even) more memory, we should benchmark this to
+ // determine if it's worth doing
+ if ( !m_convertedToWChar.m_str || lenWC != m_convertedToWChar.m_len )
{
- //Previosly passed as the first argument to wxTmemchr,
- //but C/C++ standard does not specify evaluation order
- //of arguments to functions -
- //http://embedded.com/showArticle.jhtml?articleID=9900607
- ++p;
-
- //anchor again
- p = (const wxChar*)wxTmemchr(p,
- str.c_str()[0],
- length() - (p - c_str()));
-
- if(!p)
- return npos;
+ if ( !const_cast<wxString *>(this)->m_convertedToWChar.Extend(lenWC) )
+ return NULL;
}
- return (p - c_str() + str.length() <= length()) ? p - c_str() : npos;
-}
+ // finally do convert
+ m_convertedToWChar.m_str[lenWC] = L'\0';
+ if ( conv.ToWChar(m_convertedToWChar.m_str, lenWC,
+ strMB, lenMB) == wxCONV_FAILED )
+ return NULL;
-size_t wxStringBase::find(const wxChar* sz, size_t nStart, size_t n) const
-{
- return find(wxStringBase(sz, n), nStart);
+ return m_convertedToWChar.m_str;
}
-size_t wxStringBase::find(wxChar ch, size_t nStart) const
-{
- wxASSERT( nStart <= length() );
+#endif // !wxUSE_UNICODE_WCHAR
- const wxChar *p = (const wxChar*)wxTmemchr(c_str() + nStart, ch, length() - nStart);
- return p == NULL ? npos : p - c_str();
-}
+// Same thing for mb_str() which returns a normal char pointer to string
+// contents: this always requires converting it to the specified encoding in
+// non-ANSI build except if we need to convert to UTF-8 and this is what we
+// already use internally.
+#if wxUSE_UNICODE
-size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const
+const char *wxString::AsChar(const wxMBConv& conv) const
{
- wxASSERT( str.GetStringData()->IsValid() );
- wxASSERT( nStart == npos || nStart <= length() );
-
- if ( length() >= str.length() )
- {
- // avoids a corner case later
- if ( length() == 0 && str.length() == 0 )
- return 0;
+#if wxUSE_UNICODE_UTF8
+ if ( conv.IsUTF8() )
+ return m_impl.c_str();
- // "top" is the point where search starts from
- size_t top = length() - str.length();
+ const wchar_t * const strWC = AsWChar(wxMBConvStrictUTF8());
+ const size_t lenWC = m_convertedToWChar.m_len;
+#else // wxUSE_UNICODE_WCHAR
+ const wchar_t * const strWC = m_impl.c_str();
+ const size_t lenWC = m_impl.length();
+#endif // wxUSE_UNICODE_UTF8/wxUSE_UNICODE_WCHAR
- if ( nStart == npos )
- nStart = length() - 1;
- if ( nStart < top )
- top = nStart;
+ const size_t lenMB = conv.FromWChar(NULL, 0, strWC, lenWC);
+ if ( lenMB == wxCONV_FAILED )
+ return NULL;
- const wxChar *cursor = c_str() + top;
- do
- {
- if ( wxTmemcmp(cursor, str.c_str(),
- str.length()) == 0 )
- {
- return cursor - c_str();
- }
- } while ( cursor-- > c_str() );
+ if ( !m_convertedToChar.m_str || lenMB != m_convertedToChar.m_len )
+ {
+ if ( !const_cast<wxString *>(this)->m_convertedToChar.Extend(lenMB) )
+ return NULL;
}
- return npos;
+ m_convertedToChar.m_str[lenMB] = '\0';
+ if ( conv.FromWChar(m_convertedToChar.m_str, lenMB,
+ strWC, lenWC) == wxCONV_FAILED )
+ return NULL;
+
+ return m_convertedToChar.m_str;
}
-size_t wxStringBase::rfind(const wxChar* sz, size_t nStart, size_t n) const
+#endif // wxUSE_UNICODE
+
+// shrink to minimal size (releasing extra memory)
+bool wxString::Shrink()
{
- return rfind(wxStringBase(sz, n), nStart);
+ wxString tmp(begin(), end());
+ swap(tmp);
+ return tmp.length() == length();
}
-size_t wxStringBase::rfind(wxChar ch, size_t nStart) const
+// deprecated compatibility code:
+#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
+wxStringCharType *wxString::GetWriteBuf(size_t nLen)
{
- if ( nStart == npos )
- {
- nStart = length();
- }
- else
- {
- wxASSERT( nStart <= length() );
- }
-
- const wxChar *actual;
- for ( actual = c_str() + ( nStart == npos ? length() : nStart + 1 );
- actual > c_str(); --actual )
- {
- if ( *(actual - 1) == ch )
- return (actual - 1) - c_str();
- }
-
- return npos;
+ return DoGetWriteBuf(nLen);
}
-size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart) const
+void wxString::UngetWriteBuf()
{
- wxASSERT(nStart <= length());
-
- size_t len = wxStrlen(sz);
-
- size_t i;
- for(i = nStart; i < this->length(); ++i)
- {
- if (wxTmemchr(sz, *(c_str() + i), len))
- break;
- }
-
- if(i == this->length())
- return npos;
- else
- return i;
+ DoUngetWriteBuf();
}
-size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart,
- size_t n) const
+void wxString::UngetWriteBuf(size_t nLen)
{
- return find_first_of(wxStringBase(sz, n), nStart);
+ DoUngetWriteBuf(nLen);
}
+#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
-size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart) const
-{
- if ( nStart == npos )
- {
- nStart = length() - 1;
- }
- else
- {
- wxASSERT_MSG( nStart <= length(),
- _T("invalid index in find_last_of()") );
- }
- size_t len = wxStrlen(sz);
+// ---------------------------------------------------------------------------
+// data access
+// ---------------------------------------------------------------------------
- for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p )
- {
- if ( wxTmemchr(sz, *p, len) )
- return p - c_str();
- }
+// all functions are inline in string.h
- return npos;
-}
+// ---------------------------------------------------------------------------
+// concatenation operators
+// ---------------------------------------------------------------------------
-size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart,
- size_t n) const
-{
- return find_last_of(wxStringBase(sz, n), nStart);
-}
+/*
+ * concatenation functions come in 5 flavours:
+ * string + string
+ * char + string and string + char
+ * C str + string and string + C str
+ */
-size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart) const
+wxString operator+(const wxString& str1, const wxString& str2)
{
- if ( nStart == npos )
- {
- nStart = length();
- }
- else
- {
- wxASSERT( nStart <= length() );
- }
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str1.IsValid() );
+ wxASSERT( str2.IsValid() );
+#endif
- size_t len = wxStrlen(sz);
+ wxString s = str1;
+ s += str2;
- size_t i;
- for(i = nStart; i < this->length(); ++i)
- {
- if (!wxTmemchr(sz, *(c_str() + i), len))
- break;
- }
-
- if(i == this->length())
- return npos;
- else
- return i;
+ return s;
}
-size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart,
- size_t n) const
+wxString operator+(const wxString& str, wxUniChar ch)
{
- return find_first_not_of(wxStringBase(sz, n), nStart);
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
+
+ wxString s = str;
+ s += ch;
+
+ return s;
}
-size_t wxStringBase::find_first_not_of(wxChar ch, size_t nStart) const
+wxString operator+(wxUniChar ch, const wxString& str)
{
- wxASSERT( nStart <= length() );
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- for ( const wxChar *p = c_str() + nStart; *p; p++ )
- {
- if ( *p != ch )
- return p - c_str();
- }
+ wxString s = ch;
+ s += str;
- return npos;
+ return s;
}
-size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart) const
+wxString operator+(const wxString& str, const char *psz)
{
- if ( nStart == npos )
- {
- nStart = length() - 1;
- }
- else
- {
- wxASSERT( nStart <= length() );
- }
-
- size_t len = wxStrlen(sz);
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p )
- {
- if ( !wxTmemchr(sz, *p,len) )
- return p - c_str();
+ wxString s;
+ if ( !s.Alloc(strlen(psz) + str.length()) ) {
+ wxFAIL_MSG( wxT("out of memory in wxString::operator+") );
}
+ s += str;
+ s += psz;
- return npos;
-}
-
-size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart,
- size_t n) const
-{
- return find_last_not_of(wxStringBase(sz, n), nStart);
+ return s;
}
-size_t wxStringBase::find_last_not_of(wxChar ch, size_t nStart) const
+wxString operator+(const wxString& str, const wchar_t *pwz)
{
- if ( nStart == npos )
- {
- nStart = length() - 1;
- }
- else
- {
- wxASSERT( nStart <= length() );
- }
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- for ( const wxChar *p = c_str() + nStart; p >= c_str(); --p )
- {
- if ( *p != ch )
- return p - c_str();
+ wxString s;
+ if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+ wxFAIL_MSG( wxT("out of memory in wxString::operator+") );
}
+ s += str;
+ s += pwz;
- return npos;
+ return s;
}
-wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
- const wxChar *sz)
+wxString operator+(const char *psz, const wxString& str)
{
- wxASSERT_MSG( nStart <= length(),
- _T("index out of bounds in wxStringBase::replace") );
- size_t strLen = length() - nStart;
- nLen = strLen < nLen ? strLen : nLen;
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- wxStringBase strTmp;
- strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs
+ wxString s;
+ if ( !s.Alloc(strlen(psz) + str.length()) ) {
+ wxFAIL_MSG( wxT("out of memory in wxString::operator+") );
+ }
+ s = psz;
+ s += str;
- //This is kind of inefficient, but its pretty good considering...
- //we don't want to use character access operators here because on STL
- //it will freeze the reference count of strTmp, which means a deep copy
- //at the end when swap is called
- //
- //Also, we can't use append with the full character pointer and must
- //do it manually because this string can contain null characters
- for(size_t i1 = 0; i1 < nStart; ++i1)
- strTmp.append(1, this->c_str()[i1]);
+ return s;
+}
- //its safe to do the full version here because
- //sz must be a normal c string
- strTmp.append(sz);
+wxString operator+(const wchar_t *pwz, const wxString& str)
+{
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- for(size_t i2 = nStart + nLen; i2 < length(); ++i2)
- strTmp.append(1, this->c_str()[i2]);
+ wxString s;
+ if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+ wxFAIL_MSG( wxT("out of memory in wxString::operator+") );
+ }
+ s = pwz;
+ s += str;
- swap(strTmp);
- return *this;
+ return s;
}
-wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
- size_t nCount, wxChar ch)
-{
- return replace(nStart, nLen, wxStringBase(nCount, ch).c_str());
-}
+// ---------------------------------------------------------------------------
+// string comparison
+// ---------------------------------------------------------------------------
-wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
- const wxStringBase& str,
- size_t nStart2, size_t nLen2)
+bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const
{
- return replace(nStart, nLen, str.substr(nStart2, nLen2));
+ return (length() == 1) && (compareWithCase ? GetChar(0u) == c
+ : wxToupper(GetChar(0u)) == wxToupper(c));
}
-wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
- const wxChar* sz, size_t nCount)
+#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 replace(nStart, nLen, wxStringBase(sz, nCount).c_str());
+ return m_impl.compare(str.m_impl);
}
-wxStringBase wxStringBase::substr(size_t nStart, size_t nLen) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str) const
{
- if ( nLen == npos )
- nLen = length() - nStart;
- return wxStringBase(*this, nStart, nLen);
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+ return m_impl.compare(pos, len, str.m_impl);
}
-// assigns one string to another
-wxStringBase& wxStringBase::operator=(const wxStringBase& stringSrc)
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str,
+ size_t nStart2, size_t nLen2) const
{
- wxASSERT( stringSrc.GetStringData()->IsValid() );
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- // 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();
- }
- }
+ size_t pos2, len2;
+ str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
- return *this;
+ return m_impl.compare(pos, len, str.m_impl, pos2, len2);
}
-// assigns a single character
-wxStringBase& wxStringBase::operator=(wxChar ch)
+int wxString::compare(const char* sz) const
{
- if ( !AssignCopy(1, &ch) ) {
- wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(wxChar)") );
- }
- return *this;
+ return m_impl.compare(ImplStr(sz));
}
-// assigns C string
-wxStringBase& wxStringBase::operator=(const wxChar *psz)
+int wxString::compare(const wchar_t* sz) const
{
- if ( !AssignCopy(wxStrlen(psz), psz) ) {
- wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(const wxChar *)") );
- }
- return *this;
+ return m_impl.compare(ImplStr(sz));
}
-// helper function: does real copy
-bool wxStringBase::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
+int wxString::compare(size_t nStart, size_t nLen,
+ const char* sz, size_t nCount) const
{
- if ( nSrcLen == 0 ) {
- Reinit();
- }
- else {
- if ( !AllocBeforeWrite(nSrcLen) ) {
- // allocation failure handled by caller
- return false;
- }
- memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
- GetStringData()->nDataLength = nSrcLen;
- m_pchData[nSrcLen] = wxT('\0');
- }
- return true;
-}
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
-// ---------------------------------------------------------------------------
-// string concatenation
-// ---------------------------------------------------------------------------
-
-// add something to this string
-bool wxStringBase::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData,
- size_t nMaxLen)
-{
- STATISTICS_ADD(SummandLength, nSrcLen);
+ SubstrBufFromMB str(ImplStr(sz, nCount));
- nSrcLen = nSrcLen < nMaxLen ? nSrcLen : nMaxLen;
-
- // concatenating an empty string is a NOP
- if ( nSrcLen > 0 ) {
- wxStringData *pData = GetStringData();
- size_t nLen = pData->nDataLength;
- size_t nNewLen = nLen + nSrcLen;
-
- // alloc new buffer if current is too small
- if ( pData->IsShared() ) {
- STATISTICS_ADD(ConcatHit, 0);
-
- // we have to allocate another buffer
- wxStringData* pOldData = GetStringData();
- if ( !AllocBuffer(nNewLen) ) {
- // allocation failure handled by caller
- return false;
- }
- memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
- pOldData->Unlock();
- }
- else if ( nNewLen > pData->nAllocLength ) {
- STATISTICS_ADD(ConcatHit, 0);
-
- reserve(nNewLen);
- // we have to grow the buffer
- if ( capacity() < nNewLen ) {
- // allocation failure handled by caller
- return false;
- }
- }
- else {
- STATISTICS_ADD(ConcatHit, 1);
-
- // the buffer is already big enough
- }
-
- // should be enough space
- wxASSERT( nNewLen <= GetStringData()->nAllocLength );
-
- // fast concatenation - all is done in our buffer
- memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
-
- m_pchData[nNewLen] = wxT('\0'); // put terminating '\0'
- GetStringData()->nDataLength = nNewLen; // and fix the length
- }
- //else: the string to append was empty
- return true;
+ return m_impl.compare(pos, len, str.data, str.len);
}
-// ---------------------------------------------------------------------------
-// simple sub-string extraction
-// ---------------------------------------------------------------------------
-
-// helper function: clone the data attached to this string
-bool wxStringBase::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wchar_t* sz, size_t nCount) const
{
- if ( nCopyLen == 0 ) {
- dest.Init();
- }
- else {
- if ( !dest.AllocBuffer(nCopyLen) ) {
- // allocation failure handled by caller
- return false;
- }
- memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
- }
- return true;
-}
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
-#endif // !wxUSE_STL
+ SubstrBufFromWC str(ImplStr(sz, nCount));
-#if !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
+ return m_impl.compare(pos, len, str.data, str.len);
+}
-#if !wxUSE_STL
- #define STRINGCLASS wxStringBase
-#else
- #define STRINGCLASS wxString
-#endif
+#else // !HAVE_STD_STRING_COMPARE
-static inline int wxDoCmp(const wxChar* s1, size_t l1,
- const wxChar* s2, size_t l2)
+static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
+ const wxStringCharType* s2, size_t l2)
{
if( l1 == l2 )
- return wxTmemcmp(s1, s2, l1);
+ return wxStringMemcmp(s1, s2, l1);
else if( l1 < l2 )
{
- int ret = wxTmemcmp(s1, s2, l1);
+ int ret = wxStringMemcmp(s1, s2, l1);
return ret == 0 ? -1 : ret;
}
else
{
- int ret = wxTmemcmp(s1, s2, l2);
+ int ret = wxStringMemcmp(s1, s2, l2);
return ret == 0 ? +1 : ret;
}
}
-int STRINGCLASS::compare(const wxStringBase& str) const
+int wxString::compare(const wxString& str) const
{
- return ::wxDoCmp(data(), length(), str.data(), str.length());
+ return ::wxDoCmp(m_impl.data(), m_impl.length(),
+ str.m_impl.data(), str.m_impl.length());
}
-int STRINGCLASS::compare(size_t nStart, size_t nLen,
- const wxStringBase& str) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str) const
{
wxASSERT(nStart <= length());
size_type strLen = length() - nStart;
nLen = strLen < nLen ? strLen : nLen;
- return ::wxDoCmp(data() + nStart, nLen, str.data(), str.length());
+
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+
+ return ::wxDoCmp(m_impl.data() + pos, len,
+ str.m_impl.data(), str.m_impl.length());
}
-int STRINGCLASS::compare(size_t nStart, size_t nLen,
- const wxStringBase& str,
- size_t nStart2, size_t nLen2) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str,
+ size_t nStart2, size_t nLen2) const
{
wxASSERT(nStart <= length());
wxASSERT(nStart2 <= str.length());
strLen2 = str.length() - nStart2;
nLen = strLen < nLen ? strLen : nLen;
nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
- return ::wxDoCmp(data() + nStart, nLen, str.data() + nStart2, 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);
}
-int STRINGCLASS::compare(const wxChar* sz) const
+int wxString::compare(const char* sz) const
{
- size_t nLen = wxStrlen(sz);
- return ::wxDoCmp(data(), length(), sz, nLen);
+ SubstrBufFromMB str(ImplStr(sz, npos));
+ if ( str.len == npos )
+ str.len = wxStringStrlen(str.data);
+ return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
}
-int STRINGCLASS::compare(size_t nStart, size_t nLen,
- const wxChar* sz, size_t nCount) const
+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);
+}
+
+int wxString::compare(size_t nStart, size_t nLen,
+ const char* sz, size_t nCount) const
{
wxASSERT(nStart <= length());
size_type strLen = length() - nStart;
nLen = strLen < nLen ? strLen : nLen;
- if( nCount == npos )
- nCount = wxStrlen(sz);
-
- return ::wxDoCmp(data() + nStart, nLen, sz, nCount);
-}
-#undef STRINGCLASS
-
-#endif // !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
-
-// ===========================================================================
-// wxString class core
-// ===========================================================================
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
-// ---------------------------------------------------------------------------
-// construction and conversion
-// ---------------------------------------------------------------------------
+ SubstrBufFromMB str(ImplStr(sz, nCount));
+ if ( str.len == npos )
+ str.len = wxStringStrlen(str.data);
-#if wxUSE_UNICODE
+ return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
+}
-// from multibyte string
-wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
+int wxString::compare(size_t nStart, size_t nLen,
+ const wchar_t* sz, size_t nCount) const
{
- // anything to do?
- if ( psz && nLength != 0 )
- {
- if ( nLength == npos )
- {
- nLength = (size_t)-1;
- }
- else if ( nLength == length() )
- {
- // this is important to avoid copying the string in cMB2WC: we're
- // already NUL-terminated so we can pass this NUL with the data
- nLength++;
- }
+ wxASSERT(nStart <= length());
+ size_type strLen = length() - nStart;
+ nLen = strLen < nLen ? strLen : nLen;
- size_t nLenWide;
- wxWCharBuffer wbuf = conv.cMB2WC(psz, nLength, &nLenWide);
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- if ( nLenWide )
- assign(wbuf, nLenWide);
- }
-}
+ SubstrBufFromWC str(ImplStr(sz, nCount));
+ if ( str.len == npos )
+ str.len = wxStringStrlen(str.data);
-//Convert wxString in Unicode mode to a multi-byte string
-const wxCharBuffer wxString::mb_str(wxMBConv& conv) const
-{
- return conv.cWC2MB(c_str(), length() + 1 /* size, not length */, NULL);
+ return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
}
-#else // ANSI
+#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
-#if wxUSE_WCHAR_T
-// from wide string
-wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
-{
- // anything to do?
- if ( pwz && nLength != 0 )
- {
- if ( nLength == npos )
- {
- nLength = (size_t)-1;
- }
- else if ( nLength == length() )
- {
- // this is important to avoid copying the string in cMB2WC: we're
- // already NUL-terminated so we can pass this NUL with the data
- nLength++;
- }
+// ---------------------------------------------------------------------------
+// find_{first,last}_[not]_of functions
+// ---------------------------------------------------------------------------
- size_t nLenMB;
- wxCharBuffer buf = conv.cWC2MB(pwz, nLength, &nLenMB);
+#if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
- if ( nLenMB )
- assign(buf, nLenMB);
- }
-}
+// 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.
-//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(wxMBConv& conv) const
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
{
- return conv.cMB2WC(c_str(), length() + 1 /* size, not length */, NULL);
+ return find_first_of(sz, nStart, wxStrlen(sz));
}
-#endif // wxUSE_WCHAR_T
-
-#endif // Unicode/ANSI
-
-// shrink to minimal size (releasing extra memory)
-bool wxString::Shrink()
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
{
- wxString tmp(begin(), end());
- swap(tmp);
- return tmp.length() == length();
+ return find_first_not_of(sz, nStart, wxStrlen(sz));
}
-#if !wxUSE_STL
-// get the pointer to writable buffer of (at least) nLen bytes
-wxChar *wxString::GetWriteBuf(size_t nLen)
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
{
- if ( !AllocBeforeWrite(nLen) ) {
- // allocation failure handled by caller
- return NULL;
- }
-
- wxASSERT( GetStringData()->nRefs == 1 );
- GetStringData()->Validate(false);
+ wxASSERT_MSG( nStart <= length(), wxT("invalid index") );
- return m_pchData;
-}
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( wxTmemchr(sz, *i, n) )
+ return idx;
+ }
-// put string back in a reasonable state after GetWriteBuf
-void wxString::UngetWriteBuf()
-{
- GetStringData()->nDataLength = wxStrlen(m_pchData);
- GetStringData()->Validate(true);
+ return npos;
}
-void wxString::UngetWriteBuf(size_t nLen)
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
{
- GetStringData()->nDataLength = nLen;
- GetStringData()->Validate(true);
-}
-#endif
-
-// ---------------------------------------------------------------------------
-// data access
-// ---------------------------------------------------------------------------
+ wxASSERT_MSG( nStart <= length(), wxT("invalid index") );
-// all functions are inline in string.h
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( !wxTmemchr(sz, *i, n) )
+ return idx;
+ }
-// ---------------------------------------------------------------------------
-// assignment operators
-// ---------------------------------------------------------------------------
+ return npos;
+}
-#if !wxUSE_UNICODE
-// same as 'signed char' variant
-wxString& wxString::operator=(const unsigned char* psz)
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
{
- *this = (const char *)psz;
- return *this;
+ return find_last_of(sz, nStart, wxStrlen(sz));
}
-#if wxUSE_WCHAR_T
-wxString& wxString::operator=(const wchar_t *pwz)
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
{
- wxString str(pwz);
- swap(str);
- return *this;
+ return find_last_not_of(sz, nStart, wxStrlen(sz));
}
-#endif
-
-#endif
-/*
- * concatenation functions come in 5 flavours:
- * string + string
- * char + string and string + char
- * C str + string and string + C str
- */
-
-wxString operator+(const wxString& str1, const wxString& str2)
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const
{
-#if !wxUSE_STL
- wxASSERT( str1.GetStringData()->IsValid() );
- wxASSERT( str2.GetStringData()->IsValid() );
-#endif
+ size_t len = length();
- wxString s = str1;
- s += str2;
+ if ( nStart == npos )
+ {
+ nStart = len - 1;
+ }
+ else
+ {
+ wxASSERT_MSG( nStart <= len, wxT("invalid index") );
+ }
- return s;
+ size_t idx = nStart;
+ for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+ i != rend(); --idx, ++i )
+ {
+ if ( wxTmemchr(sz, *i, n) )
+ return idx;
+ }
+
+ return npos;
}
-wxString operator+(const wxString& str, wxChar ch)
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const
{
-#if !wxUSE_STL
- wxASSERT( str.GetStringData()->IsValid() );
-#endif
+ size_t len = length();
- wxString s = str;
- s += ch;
+ if ( nStart == npos )
+ {
+ nStart = len - 1;
+ }
+ else
+ {
+ wxASSERT_MSG( nStart <= len, wxT("invalid index") );
+ }
- return s;
+ size_t idx = nStart;
+ for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+ i != rend(); --idx, ++i )
+ {
+ if ( !wxTmemchr(sz, *i, n) )
+ return idx;
+ }
+
+ return npos;
}
-wxString operator+(wxChar ch, const wxString& str)
+size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const
{
-#if !wxUSE_STL
- wxASSERT( str.GetStringData()->IsValid() );
-#endif
+ wxASSERT_MSG( nStart <= length(), wxT("invalid index") );
- wxString s = ch;
- s += str;
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( *i != ch )
+ return idx;
+ }
- return s;
+ return npos;
}
-wxString operator+(const wxString& str, const wxChar *psz)
-{
-#if !wxUSE_STL
- wxASSERT( str.GetStringData()->IsValid() );
-#endif
+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, wxT("invalid index") );
+ }
- wxString s;
- if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
- wxFAIL_MSG( _T("out of memory in wxString::operator+") );
- }
- s += str;
- s += psz;
+ size_t idx = nStart;
+ for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+ i != rend(); --idx, ++i )
+ {
+ if ( *i != ch )
+ return idx;
+ }
- return s;
+ return npos;
}
-wxString operator+(const wxChar *psz, const wxString& str)
-{
-#if !wxUSE_STL
- wxASSERT( str.GetStringData()->IsValid() );
+// 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
- wxString s;
- if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
- wxFAIL_MSG( _T("out of memory in wxString::operator+") );
- }
- s = psz;
- s += str;
+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); }
- return s;
-}
+#undef wxOtherCharType
+#undef STRCONV
+
+#endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
// ===========================================================================
// other common string functions
// ===========================================================================
-int wxString::Cmp(const wxString& s) const
+int wxString::CmpNoCase(const wxString& s) const
{
- return compare(s);
-}
+#if !wxUSE_UNICODE_UTF8
+ // We compare NUL-delimited chunks of the strings inside the loop. We will
+ // do as many iterations as there are embedded NULs in the string, i.e.
+ // usually we will run it just once.
-int wxString::Cmp(const wxChar* psz) const
-{
- return compare(psz);
-}
+ typedef const wxStringImpl::value_type *pchar_type;
+ const pchar_type thisBegin = m_impl.c_str();
+ const pchar_type thatBegin = s.m_impl.c_str();
-static inline int wxDoCmpNoCase(const wxChar* s1, size_t l1,
- const wxChar* s2, size_t l2)
-{
- size_t i;
+ const pchar_type thisEnd = thisBegin + m_impl.length();
+ const pchar_type thatEnd = thatBegin + s.m_impl.length();
- if( l1 == l2 )
- {
- for(i = 0; i < l1; ++i)
- {
- if(wxTolower(s1[i]) != wxTolower(s2[i]))
- break;
- }
- return i == l1 ? 0 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1;
- }
- else if( l1 < l2 )
+ pchar_type thisCur = thisBegin;
+ pchar_type thatCur = thatBegin;
+
+ int rc;
+ for ( ;; )
{
- for(i = 0; i < l1; ++i)
+ // Compare until the next NUL, if the strings differ this is the final
+ // result.
+ rc = wxStricmp(thisCur, thatCur);
+ if ( rc )
+ break;
+
+ const size_t lenChunk = wxStrlen(thisCur);
+ thisCur += lenChunk;
+ thatCur += lenChunk;
+
+ // Skip all the NULs as wxStricmp() doesn't handle them.
+ for ( ; !*thisCur; thisCur++, thatCur++ )
{
- if(wxTolower(s1[i]) != wxTolower(s2[i]))
- break;
+ // Check if we exhausted either of the strings.
+ if ( thisCur == thisEnd )
+ {
+ // This one is exhausted, is the other one too?
+ return thatCur == thatEnd ? 0 : -1;
+ }
+
+ if ( thatCur == thatEnd )
+ {
+ // Because of the test above we know that this one is not
+ // exhausted yet so it's greater than the other one that is.
+ return 1;
+ }
+
+ if ( *thatCur )
+ {
+ // Anything non-NUL is greater than NUL.
+ return -1;
+ }
}
- return i == l1 ? -1 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1;
}
- else
+
+ return rc;
+#else // wxUSE_UNICODE_UTF8
+ // CRT functions can't be used for case-insensitive comparison of UTF-8
+ // strings so do it in the naive, simple and inefficient way.
+
+ // 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 )
{
- for(i = 0; i < l2; ++i)
- {
- if(wxTolower(s1[i]) != wxTolower(s2[i]))
- break;
- }
- return i == l2 ? 1 : wxTolower(s1[i]) < wxTolower(s2[i]) ? -1 : 1;
+ wxUniChar lower1 = (wxChar)wxTolower(*i1);
+ wxUniChar lower2 = (wxChar)wxTolower(*i2);
+ if ( lower1 != lower2 )
+ return lower1 < lower2 ? -1 : 1;
}
-}
-
-int wxString::CmpNoCase(const wxString& s) const
-{
- return wxDoCmpNoCase(data(), length(), s.data(), s.length());
-}
-int wxString::CmpNoCase(const wxChar* psz) const
-{
- int nLen = wxStrlen(psz);
+ size_t len1 = length();
+ size_t len2 = s.length();
- return wxDoCmpNoCase(data(), length(), psz, nLen);
+ if ( len1 < len2 )
+ return -1;
+ else if ( len1 > len2 )
+ return 1;
+ return 0;
+#endif // !wxUSE_UNICODE_UTF8/wxUSE_UNICODE_UTF8
}
#endif
#endif
-wxString wxString::FromAscii(const char *ascii)
+wxString wxString::FromAscii(const char *ascii, size_t len)
{
- if (!ascii)
+ if (!ascii || len == 0)
return wxEmptyString;
- size_t len = strlen( ascii );
wxString res;
- if ( len )
{
- wxStringBuffer buf(res, len);
-
- wchar_t *dest = buf;
+ wxStringInternalBuffer buf(res, len);
+ wxStringCharType *dest = buf;
- for ( ;; )
+ for ( ; len > 0; --len )
{
- if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
- break;
+ unsigned char c = (unsigned char)*ascii++;
+ wxASSERT_MSG( c < 0x80,
+ wxT("Non-ASCII value passed to FromAscii().") );
+
+ *dest++ = (wchar_t)c;
}
}
return res;
}
-wxString wxString::FromAscii(const char ascii)
+wxString wxString::FromAscii(const char *ascii)
+{
+ return FromAscii(ascii, wxStrlen(ascii));
+}
+
+wxString wxString::FromAscii(char ascii)
{
// What do we do with '\0' ?
- wxString res;
- res += (wchar_t)(unsigned char) ascii;
+ unsigned char c = (unsigned char)ascii;
- return res;
+ wxASSERT_MSG( c < 0x80, wxT("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
+const wxScopedCharBuffer wxString::ToAscii() const
{
// this will allocate enough space for the terminating NUL too
wxCharBuffer buffer(length());
-
-
char *dest = buffer.data();
- const wchar_t *pwc = c_str();
- for ( ;; )
+ for ( const_iterator i = begin(); i != end(); ++i )
{
- *dest++ = (char)(*pwc > SCHAR_MAX ? wxT('_') : *pwc);
+ 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 ( !*pwc++ )
+ if ( !c )
break;
}
return buffer;
}
-#endif // Unicode
+#endif // wxUSE_UNICODE
// extract string of length nCount starting at nFirst
wxString wxString::Mid(size_t nFirst, size_t nCount) const
wxString dest(*this, nFirst, nCount);
if ( dest.length() != nCount )
{
- wxFAIL_MSG( _T("out of memory in wxString::Mid") );
+ wxFAIL_MSG( wxT("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 wxChar *prefix, wxString *rest) const
+bool wxString::StartsWith(const wxString& prefix, wxString *rest) const
{
- wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
+ if ( compare(0, prefix.length(), prefix) != 0 )
+ return false;
- // first check if the beginning of the string matches the prefix: note
- // that we don't have to check that we don't run out of this string as
- // when we reach the terminating NUL, either prefix string ends too (and
- // then it's ok) or we break out of the loop because there is no match
- const wxChar *p = c_str();
- while ( *prefix )
+ if ( rest )
{
- if ( *prefix++ != *p++ )
- {
- // no match
- return false;
- }
+ // 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
+{
+ int start = length() - suffix.length();
+
+ if ( start < 0 || compare(start, npos, suffix) != 0 )
+ return false;
+
if ( rest )
{
// put the rest of the string into provided pointer
- *rest = p;
+ rest->assign(*this, 0, start);
}
return true;
}
+
// extract nCount last (rightmost) characters
wxString wxString::Right(size_t nCount) const
{
wxString dest(*this, length() - nCount, nCount);
if ( dest.length() != nCount ) {
- wxFAIL_MSG( _T("out of memory in wxString::Right") );
+ wxFAIL_MSG( wxT("out of memory in wxString::Right") );
}
return dest;
}
-// get all characters after the last occurence of ch
+// get all characters after the last occurrence of ch
// (returns the whole string if ch not found)
-wxString wxString::AfterLast(wxChar ch) const
+wxString wxString::AfterLast(wxUniChar ch) const
{
wxString str;
int iPos = Find(ch, true);
if ( iPos == wxNOT_FOUND )
str = *this;
else
- str = c_str() + iPos + 1;
+ str.assign(*this, iPos + 1, npos);
return str;
}
wxString dest(*this, 0, nCount);
if ( dest.length() != nCount ) {
- wxFAIL_MSG( _T("out of memory in wxString::Left") );
+ wxFAIL_MSG( wxT("out of memory in wxString::Left") );
}
return dest;
}
-// get all characters before the first occurence of ch
+// get all characters before the first occurrence of ch
// (returns the whole string if ch not found)
-wxString wxString::BeforeFirst(wxChar ch) const
+wxString wxString::BeforeFirst(wxUniChar ch, wxString *rest) const
{
int iPos = Find(ch);
- if ( iPos == wxNOT_FOUND ) iPos = length();
+ if ( iPos == wxNOT_FOUND )
+ {
+ iPos = length();
+ if ( rest )
+ rest->clear();
+ }
+ else
+ {
+ if ( rest )
+ rest->assign(*this, iPos + 1, npos);
+ }
+
return wxString(*this, 0, iPos);
}
-/// get all characters before the last occurence of ch
+/// get all characters before the last occurrence of ch
/// (returns empty string if ch not found)
-wxString wxString::BeforeLast(wxChar ch) const
+wxString wxString::BeforeLast(wxUniChar ch, wxString *rest) const
{
wxString str;
int iPos = Find(ch, true);
- if ( iPos != wxNOT_FOUND && iPos != 0 )
- str = wxString(c_str(), iPos);
+ if ( iPos != wxNOT_FOUND )
+ {
+ if ( iPos != 0 )
+ str.assign(*this, 0, iPos);
+
+ if ( rest )
+ rest->assign(*this, iPos + 1, npos);
+ }
+ else
+ {
+ if ( rest )
+ *rest = *this;
+ }
return str;
}
-/// get all characters after the first occurence of ch
+/// get all characters after the first occurrence of ch
/// (returns empty string if ch not found)
-wxString wxString::AfterFirst(wxChar ch) const
+wxString wxString::AfterFirst(wxUniChar ch) const
{
wxString str;
int iPos = Find(ch);
if ( iPos != wxNOT_FOUND )
- str = c_str() + iPos + 1;
+ str.assign(*this, iPos + 1, npos);
return str;
}
-// replace first (or all) occurences of some substring with another one
-size_t wxString::Replace(const wxChar *szOld,
- const wxChar *szNew, bool bReplaceAll)
+// replace first (or all) occurrences of some substring with another one
+size_t wxString::Replace(const wxString& strOld,
+ const wxString& strNew, bool bReplaceAll)
{
// if we tried to replace an empty string we'd enter an infinite loop below
- wxCHECK_MSG( szOld && *szOld && szNew, 0,
- _T("wxString::Replace(): invalid parameter") );
+ wxCHECK_MSG( !strOld.empty(), 0,
+ wxT("wxString::Replace(): invalid parameter") );
+
+ wxSTRING_INVALIDATE_CACHE();
size_t uiCount = 0; // count of replacements made
- size_t uiOldLen = wxStrlen(szOld);
- size_t uiNewLen = wxStrlen(szNew);
+ // 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;
+
+ m_impl[pos++] = chNew;
- size_t dwPos = 0;
+ uiCount++;
- while ( this->c_str()[dwPos] != wxT('\0') )
+ if ( !bReplaceAll )
+ break;
+ }
+ }
+ else if ( !bReplaceAll)
{
- //DO NOT USE STRSTR HERE
- //this string can contain embedded null characters,
- //so strstr will function incorrectly
- dwPos = find(szOld, dwPos);
- if ( dwPos == npos )
- break; // exit the loop
- else
+ size_t pos = m_impl.find(strOld, 0);
+ if ( pos != npos )
{
- //replace this occurance of the old string with the new one
- replace(dwPos, uiOldLen, szNew, uiNewLen);
+ m_impl.replace(pos, strOld.m_impl.length(), strNew.m_impl);
+ uiCount = 1;
+ }
+ }
+ else // replace all occurrences
+ {
+ const size_t uiOldLen = strOld.m_impl.length();
+ const size_t uiNewLen = strNew.m_impl.length();
- //move up pos past the string that was replaced
- dwPos += uiNewLen;
+ // first scan the string to find all positions at which the replacement
+ // should be made
+ wxVector<size_t> replacePositions;
- //increase replace count
+ size_t pos;
+ for ( pos = m_impl.find(strOld.m_impl, 0);
+ pos != npos;
+ pos = m_impl.find(strOld.m_impl, pos + uiOldLen))
+ {
+ replacePositions.push_back(pos);
++uiCount;
+ }
- // stop now?
- if ( !bReplaceAll )
- break; // exit the loop
+ if ( !uiCount )
+ return 0;
+
+ // allocate enough memory for the whole new string
+ wxString tmp;
+ tmp.m_impl.reserve(m_impl.length() + uiCount*(uiNewLen - uiOldLen));
+
+ // copy this string to tmp doing replacements on the fly
+ size_t replNum = 0;
+ for ( pos = 0; replNum < uiCount; replNum++ )
+ {
+ const size_t nextReplPos = replacePositions[replNum];
+
+ if ( pos != nextReplPos )
+ {
+ tmp.m_impl.append(m_impl, pos, nextReplPos - pos);
+ }
+
+ tmp.m_impl.append(strNew.m_impl);
+ pos = nextReplPos + uiOldLen;
+ }
+
+ if ( pos != m_impl.length() )
+ {
+ // append the rest of the string unchanged
+ tmp.m_impl.append(m_impl, pos, m_impl.length() - pos);
}
+
+ swap(tmp);
}
return uiCount;
bool wxString::IsAscii() const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!isascii(*s)) return(false);
- s++;
- }
- return(true);
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !(*i).IsAscii() )
+ return false;
+ }
+
+ return true;
}
bool wxString::IsWord() const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!wxIsalpha(*s)) return(false);
- s++;
- }
- return(true);
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !wxIsalpha(*i) )
+ return false;
+ }
+
+ return true;
}
bool wxString::IsNumber() const
{
- const wxChar *s = (const wxChar*) *this;
- if (wxStrlen(s))
- if ((s[0] == wxT('-')) || (s[0] == wxT('+'))) s++;
- while(*s){
- if(!wxIsdigit(*s)) return(false);
- s++;
- }
- return(true);
+ if ( empty() )
+ return true;
+
+ const_iterator i = begin();
+
+ if ( *i == wxT('-') || *i == wxT('+') )
+ ++i;
+
+ for ( ; i != end(); ++i )
+ {
+ if ( !wxIsdigit(*i) )
+ return false;
+ }
+
+ return true;
}
wxString wxString::Strip(stripType w) const
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('ê') in the C locale which seems to be broken to me, but we have to
-// live with this by checking that the character is a 7 bit one - even if this
-// may fail to detect some spaces (I don't know if Unicode doesn't have
+// 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)
{
- // first check if we're going to modify the string at all
- if ( !empty() &&
- (
- (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
- (!bFromRight && wxSafeIsspace(GetChar(0u)))
+ // 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
- iterator psz = begin() + length() - 1;
- while ( wxSafeIsspace(*psz) && (psz >= begin()) )
- psz--;
-
- // truncate at trailing space start
- *++psz = wxT('\0');
- erase(psz, end());
- }
- else
{
- // find first non-space character
- iterator psz = begin();
- while ( wxSafeIsspace(*psz) )
- psz++;
+ if ( bFromRight )
+ {
+ // find last non-space character
+ reverse_iterator psz = rbegin();
+ while ( (psz != rend()) && wxSafeIsspace(*psz) )
+ ++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;
- // fix up data and length
- erase(begin(), psz);
+ // 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, wxChar 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;
- swap(s);
- }
+ if ( bFromRight )
+ *this += s;
+ else
+ {
+ s += *this;
+ swap(s);
+ }
- return *this;
+ return *this;
}
// truncate the string
wxString& wxString::Truncate(size_t uiLen)
{
- if ( uiLen < Len() ) {
- erase(begin() + uiLen, end());
- }
- //else: nothing to do, string is already short enough
+ if ( uiLen < length() )
+ {
+ erase(begin() + uiLen, end());
+ }
+ //else: nothing to do, string is already short enough
- return *this;
+ return *this;
}
// ---------------------------------------------------------------------------
// ---------------------------------------------------------------------------
// find a character
-int wxString::Find(wxChar ch, bool bFromEnd) const
+int wxString::Find(wxUniChar ch, bool bFromEnd) const
+{
+ size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
+
+ return (idx == npos) ? wxNOT_FOUND : (int)idx;
+}
+
+// ----------------------------------------------------------------------------
+// conversion to numbers
+// ----------------------------------------------------------------------------
+
+// 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.
+
+#ifndef __WXWINCE__
+ #define DO_IF_NOT_WINCE(x) x
+#else
+ #define DO_IF_NOT_WINCE(x)
+#endif
+
+#define WX_STRING_TO_X_TYPE_START \
+ wxCHECK_MSG( pVal, false, wxT("NULL output pointer") ); \
+ DO_IF_NOT_WINCE( errno = 0; ) \
+ const wxStringCharType *start = wx_str(); \
+ wxStringCharType *end;
+
+// notice that we return false without modifying the output parameter at all if
+// nothing could be parsed but we do modify it and return false then if we did
+// parse something successfully but not the entire string
+#define WX_STRING_TO_X_TYPE_END \
+ if ( end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \
+ return false; \
+ *pVal = val; \
+ return !*end;
+
+bool wxString::ToLong(long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ long val = wxStrtol(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToULong(unsigned long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ unsigned long val = wxStrtoul(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ wxLongLong_t val = wxStrtoll(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+ wxULongLong_t val = wxStrtoull(start, &end, base);
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToDouble(double *pVal) const
+{
+ WX_STRING_TO_X_TYPE_START
+ double val = wxStrtod(start, &end);
+ WX_STRING_TO_X_TYPE_END
+}
+
+#if wxUSE_XLOCALE
+
+bool wxString::ToCLong(long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ long val = wxStrtol_lA(start, &end, base, wxCLocale);
+#else
+ long val = wxStrtol_l(start, &end, base, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToCULong(unsigned long *pVal, int base) const
+{
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), wxT("invalid base") );
+
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ unsigned long val = wxStrtoul_lA(start, &end, base, wxCLocale);
+#else
+ unsigned long val = wxStrtoul_l(start, &end, base, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
+}
+
+bool wxString::ToCDouble(double *pVal) const
+{
+ WX_STRING_TO_X_TYPE_START
+#if (wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE) && defined(wxHAS_XLOCALE_SUPPORT)
+ double val = wxStrtod_lA(start, &end, wxCLocale);
+#else
+ double val = wxStrtod_l(start, &end, wxCLocale);
+#endif
+ WX_STRING_TO_X_TYPE_END
+}
+
+#else // wxUSE_XLOCALE
+
+// Provide implementation of these functions even when wxUSE_XLOCALE is
+// disabled, we still need them in wxWidgets internal code.
+
+// For integers we just assume the current locale uses the same number
+// representation as the C one as there is nothing else we can do.
+bool wxString::ToCLong(long *pVal, int base) const
{
- size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
+ return ToLong(pVal, base);
+}
- return (idx == npos) ? wxNOT_FOUND : (int)idx;
+bool wxString::ToCULong(unsigned long *pVal, int base) const
+{
+ return ToULong(pVal, base);
}
-// find a sub-string (like strstr)
-int wxString::Find(const wxChar *pszSub) const
+// For floating point numbers we have to handle the problem of the decimal
+// point which is different in different locales.
+bool wxString::ToCDouble(double *pVal) const
{
- size_type idx = find(pszSub);
+ // Create a copy of this string using the decimal point instead of whatever
+ // separator the current locale uses.
+#if wxUSE_INTL
+ wxString sep = wxLocale::GetInfo(wxLOCALE_DECIMAL_POINT,
+ wxLOCALE_CAT_NUMBER);
+ if ( sep == "." )
+ {
+ // We can avoid an unnecessary string copy in this case.
+ return ToDouble(pVal);
+ }
+#else // !wxUSE_INTL
+ // We don't know what the current separator is so it might even be a point
+ // already, try to parse the string as a double:
+ if ( ToDouble(pVal) )
+ {
+ // It must have been the point, nothing else to do.
+ return true;
+ }
- return (idx == npos) ? wxNOT_FOUND : (int)idx;
+ // Try to guess the separator, using the most common alternative value.
+ wxString sep(",");
+#endif // wxUSE_INTL/!wxUSE_INTL
+ wxString cstr(*this);
+ cstr.Replace(".", sep);
+
+ return cstr.ToDouble(pVal);
}
+#endif // wxUSE_XLOCALE/!wxUSE_XLOCALE
+
// ----------------------------------------------------------------------------
-// conversion to numbers
+// number to string conversion
// ----------------------------------------------------------------------------
-bool wxString::ToLong(long *val, int base) const
+/* static */
+wxString wxString::FromDouble(double val, int precision)
{
- wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToLong") );
- wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
+ wxCHECK_MSG( precision >= -1, wxString(), "Invalid negative precision" );
- const wxChar *start = c_str();
- wxChar *end;
- *val = wxStrtol(start, &end, base);
+ wxString format;
+ if ( precision == -1 )
+ {
+ format = "%g";
+ }
+ else // Use fixed precision.
+ {
+ format.Printf("%%.%df", precision);
+ }
- // return true only if scan was stopped by the terminating NUL and if the
- // string was not empty to start with
- return !*end && (end != start);
+ return wxString::Format(format, val);
}
-bool wxString::ToULong(unsigned long *val, int base) const
+/* static */
+wxString wxString::FromCDouble(double val, int precision)
{
- wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToULong") );
- wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
-
- const wxChar *start = c_str();
- wxChar *end;
- *val = wxStrtoul(start, &end, base);
+ wxCHECK_MSG( precision >= -1, wxString(), "Invalid negative precision" );
- // return true only if scan was stopped by the terminating NUL and if the
- // string was not empty to start with
- return !*end && (end != start);
+#if wxUSE_STD_IOSTREAM && wxUSE_STD_STRING
+ // We assume that we can use the ostream and not wstream for numbers.
+ wxSTD ostringstream os;
+ if ( precision != -1 )
+ {
+ os.precision(precision);
+ os.setf(std::ios::fixed, std::ios::floatfield);
+ }
+
+ os << val;
+ return os.str();
+#else // !wxUSE_STD_IOSTREAM
+ // Can't use iostream locale support, fall back to the manual method
+ // instead.
+ wxString s = FromDouble(val, precision);
+#if wxUSE_INTL
+ wxString sep = wxLocale::GetInfo(wxLOCALE_DECIMAL_POINT,
+ wxLOCALE_CAT_NUMBER);
+#else // !wxUSE_INTL
+ // As above, this is the most common alternative value. Notice that here it
+ // doesn't matter if we guess wrongly and the current separator is already
+ // ".": we'll just waste a call to Replace() in this case.
+ wxString sep(",");
+#endif // wxUSE_INTL/!wxUSE_INTL
+
+ s.Replace(sep, ".");
+ return s;
+#endif // wxUSE_STD_IOSTREAM/!wxUSE_STD_IOSTREAM
}
-bool wxString::ToDouble(double *val) const
+// ---------------------------------------------------------------------------
+// formatted output
+// ---------------------------------------------------------------------------
+
+#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
{
- wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToDouble") );
+ va_list argptr;
+ va_start(argptr, format);
- const wxChar *start = c_str();
- wxChar *end;
- *val = wxStrtod(start, &end);
+ wxString s;
+ s.PrintfV(format, argptr);
- // return true only if scan was stopped by the terminating NUL and if the
- // string was not empty to start with
- return !*end && (end != start);
-}
+ va_end(argptr);
-// ---------------------------------------------------------------------------
-// formatted output
-// ---------------------------------------------------------------------------
+ return s;
+}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
+#if wxUSE_UNICODE_UTF8
/* static */
-wxString wxString::Format(const wxChar *pszFormat, ...)
+wxString wxString::DoFormatUtf8(const char *format, ...)
{
va_list argptr;
- va_start(argptr, pszFormat);
+ va_start(argptr, format);
wxString s;
- s.PrintfV(pszFormat, argptr);
+ s.PrintfV(format, argptr);
va_end(argptr);
return s;
}
+#endif // wxUSE_UNICODE_UTF8
/* static */
-wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
+wxString wxString::FormatV(const wxString& format, va_list argptr)
{
wxString s;
- s.PrintfV(pszFormat, argptr);
+ s.PrintfV(format, argptr);
return s;
}
-int wxString::Printf(const wxChar *pszFormat, ...)
+#if !wxUSE_UTF8_LOCALE_ONLY
+#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
+int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...)
+#else
+int wxString::DoPrintfWchar(const wxChar *format, ...)
+#endif
+{
+ va_list argptr;
+ va_start(argptr, format);
+
+#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<wxString*>(this);
+#else
+ wxString *str = this;
+#endif
+
+ int iLen = str->PrintfV(format, argptr);
+
+ va_end(argptr);
+
+ return iLen;
+}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
+
+#if wxUSE_UNICODE_UTF8
+int wxString::DoPrintfUtf8(const char *format, ...)
{
va_list argptr;
- va_start(argptr, pszFormat);
+ va_start(argptr, format);
- int iLen = PrintfV(pszFormat, argptr);
+ int iLen = PrintfV(format, argptr);
va_end(argptr);
return iLen;
}
+#endif // wxUSE_UNICODE_UTF8
-int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
+/*
+ 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 behaviour which
+ is to keep looping and increasing buffer size until the function succeeds.
+
+ In practice it's impossible to determine before compilation which behaviour
+ may be used. The vswprintf function may have vsnprintf-like behaviour or
+ vice-versa. Behaviour 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 behaviour 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 behaviour exactly.
+
+ Glibc 2.6 almost follows the C99 behaviour 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<typename BufferType>
+#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)
{
int size = 1024;
for ( ;; )
{
- wxStringBuffer tmp(*this, size + 1);
+#if wxUSE_UNICODE_UTF8
+ BufferType tmp(str, size + 1);
+ typename BufferType::CharType *buf = tmp;
+#else
+ wxStringBuffer tmp(str, size + 1);
wxChar *buf = tmp;
+#endif
if ( !buf )
{
// only a copy
va_list argptrcopy;
wxVaCopy(argptrcopy, argptr);
- int len = wxVsnprintf(buf, size, pszFormat, argptrcopy);
+
+#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
- buf[size] = _T('\0');
+ // 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] = wxT('\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;
+ size *= 2;
+#endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
}
- else if ( len > size )
+ else if ( len >= size )
{
- size = len;
+#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
{
}
// we could have overshot
- Shrink();
+ str.Shrink();
+
+ return str.length();
+}
+
+int wxString::PrintfV(const wxString& format, va_list argptr)
+{
+#if wxUSE_UNICODE_UTF8
+ #if wxUSE_STL_BASED_WXSTRING
+ typedef wxStringTypeBuffer<char> Utf8Buffer;
+ #else
+ typedef wxStringInternalBuffer Utf8Buffer;
+ #endif
+#endif
- return Len();
+#if wxUSE_UTF8_LOCALE_ONLY
+ return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+#else
+ #if wxUSE_UNICODE_UTF8
+ if ( wxLocaleIsUtf8 )
+ return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+ else
+ // wxChar* version
+ return DoStringPrintfV<wxStringBuffer>(*this, format, argptr);
+ #else
+ return DoStringPrintfV(*this, format, argptr);
+ #endif // UTF8/WCHAR
+#endif
}
// ----------------------------------------------------------------------------
// returns true if the string matches the pattern which may contain '*' and
// '?' metacharacters (as usual, '?' matches any character and '*' any number
// of them)
-bool wxString::Matches(const wxChar *pszMask) const
+bool wxString::Matches(const wxString& mask) const
{
// I disable this code as it doesn't seem to be faster (in fact, it seems
// to be much slower) than the old, hand-written code below and using it
wxString pattern;
pattern.reserve(wxStrlen(pszMask));
- pattern += _T('^');
+ pattern += wxT('^');
while ( *pszMask )
{
switch ( *pszMask )
{
- case _T('?'):
- pattern += _T('.');
+ case wxT('?'):
+ pattern += wxT('.');
break;
- case _T('*'):
- pattern += _T(".*");
+ case wxT('*'):
+ pattern += wxT(".*");
break;
- case _T('^'):
- case _T('.'):
- case _T('$'):
- case _T('('):
- case _T(')'):
- case _T('|'):
- case _T('+'):
- case _T('\\'):
+ case wxT('^'):
+ case wxT('.'):
+ case wxT('$'):
+ case wxT('('):
+ case wxT(')'):
+ case wxT('|'):
+ case wxT('+'):
+ case wxT('\\'):
// 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('\\');
+ pattern += wxT('\\');
// fall through
default:
pszMask++;
}
- pattern += _T('$');
+ pattern += wxT('$');
// 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
+ const wxScopedWCharBuffer maskBuf = mask.wc_str();
+ const wxScopedWCharBuffer txtBuf = wc_str();
+ const wxChar *pszMask = maskBuf.data();
+ const wxChar *pszTxt = txtBuf.data();
+#else
+ const wxChar *pszMask = mask.wx_str();
// the char currently being checked
- const wxChar *pszTxt = c_str();
+ const wxChar *pszTxt = wx_str();
+#endif
// the last location where '*' matched
const wxChar *pszLastStarInText = NULL;
}
// Count the number of chars
-int wxString::Freq(wxChar ch) const
+int wxString::Freq(wxUniChar ch) const
{
int count = 0;
- int len = Len();
- for (int i = 0; i < len; i++)
+ for ( const_iterator i = begin(); i != end(); ++i )
{
- if (GetChar(i) == ch)
+ if ( *i == ch )
count ++;
}
return count;
}
-// convert to upper case, return the copy of the string
-wxString wxString::Upper() const
-{ wxString s(*this); return s.MakeUpper(); }
-
-// convert to lower case, return the copy of the string
-wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
-
-int wxString::sprintf(const wxChar *pszFormat, ...)
- {
- va_list argptr;
- va_start(argptr, pszFormat);
- int iLen = PrintfV(pszFormat, argptr);
- va_end(argptr);
- return iLen;
- }
-
-// ============================================================================
-// ArrayString
-// ============================================================================
-
-#include "wx/arrstr.h"
-
-wxArrayString::wxArrayString(size_t sz, const wxChar** a)
-{
-#if !wxUSE_STL
- Init(false);
-#endif
- for (size_t i=0; i < sz; i++)
- Add(a[i]);
-}
-
-wxArrayString::wxArrayString(size_t sz, const wxString* a)
-{
-#if !wxUSE_STL
- Init(false);
-#endif
- for (size_t i=0; i < sz; i++)
- Add(a[i]);
-}
-
-#if !wxUSE_STL
-
-// size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
-#define ARRAY_MAXSIZE_INCREMENT 4096
-
-#ifndef ARRAY_DEFAULT_INITIAL_SIZE // also defined in dynarray.h
-#define ARRAY_DEFAULT_INITIAL_SIZE (16)
-#endif
-
-#define STRING(p) ((wxString *)(&(p)))
-
-// ctor
-void wxArrayString::Init(bool autoSort)
-{
- m_nSize =
- m_nCount = 0;
- m_pItems = (wxChar **) NULL;
- m_autoSort = autoSort;
-}
-
-// copy ctor
-wxArrayString::wxArrayString(const wxArrayString& src)
-{
- Init(src.m_autoSort);
-
- *this = src;
-}
-
-// assignment operator
-wxArrayString& wxArrayString::operator=(const wxArrayString& src)
-{
- if ( m_nSize > 0 )
- Clear();
-
- Copy(src);
-
- m_autoSort = src.m_autoSort;
-
- return *this;
-}
-
-void wxArrayString::Copy(const wxArrayString& src)
-{
- if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
- Alloc(src.m_nCount);
-
- for ( size_t n = 0; n < src.m_nCount; n++ )
- Add(src[n]);
-}
-
-// grow the array
-void wxArrayString::Grow(size_t nIncrement)
-{
- // only do it if no more place
- if ( (m_nSize - m_nCount) < nIncrement ) {
- // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
- // be never resized!
- #if ARRAY_DEFAULT_INITIAL_SIZE == 0
- #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
- #endif
-
- if ( m_nSize == 0 ) {
- // was empty, alloc some memory
- m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
- if (m_nSize < nIncrement)
- m_nSize = nIncrement;
- m_pItems = new wxChar *[m_nSize];
- }
- else {
- // otherwise when it's called for the first time, nIncrement would be 0
- // and the array would never be expanded
- // add 50% but not too much
- size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
- ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
- if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
- ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
- if ( nIncrement < ndefIncrement )
- nIncrement = ndefIncrement;
- m_nSize += nIncrement;
- wxChar **pNew = new wxChar *[m_nSize];
-
- // copy data to new location
- memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
-
- // delete old memory (but do not release the strings!)
- wxDELETEA(m_pItems);
-
- m_pItems = pNew;
- }
- }
-}
-
-void wxArrayString::Free()
-{
- for ( size_t n = 0; n < m_nCount; n++ ) {
- STRING(m_pItems[n])->GetStringData()->Unlock();
- }
-}
-
-// deletes all the strings from the list
-void wxArrayString::Empty()
-{
- Free();
-
- m_nCount = 0;
-}
-
-// as Empty, but also frees memory
-void wxArrayString::Clear()
-{
- Free();
-
- m_nSize =
- m_nCount = 0;
-
- wxDELETEA(m_pItems);
-}
-
-// dtor
-wxArrayString::~wxArrayString()
-{
- Free();
-
- wxDELETEA(m_pItems);
-}
-
-void wxArrayString::reserve(size_t nSize)
-{
- Alloc(nSize);
-}
-
-// pre-allocates memory (frees the previous data!)
-void wxArrayString::Alloc(size_t nSize)
-{
- // only if old buffer was not big enough
- if ( nSize > m_nSize ) {
- Free();
- wxDELETEA(m_pItems);
- m_pItems = new wxChar *[nSize];
- m_nSize = nSize;
- }
-
- m_nCount = 0;
-}
-
-// minimizes the memory usage by freeing unused memory
-void wxArrayString::Shrink()
-{
- // only do it if we have some memory to free
- if( m_nCount < m_nSize ) {
- // allocates exactly as much memory as we need
- wxChar **pNew = new wxChar *[m_nCount];
-
- // copy data to new location
- memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
- delete [] m_pItems;
- m_pItems = pNew;
- }
-}
-
-#if WXWIN_COMPATIBILITY_2_4
-
-// return a wxString[] as required for some control ctors.
-wxString* wxArrayString::GetStringArray() const
-{
- wxString *array = 0;
-
- if( m_nCount > 0 )
- {
- array = new wxString[m_nCount];
- for( size_t i = 0; i < m_nCount; i++ )
- array[i] = m_pItems[i];
- }
-
- return array;
-}
-
-void wxArrayString::Remove(size_t nIndex, size_t nRemove)
-{
- RemoveAt(nIndex, nRemove);
-}
-
-#endif // WXWIN_COMPATIBILITY_2_4
-
-// searches the array for an item (forward or backwards)
-int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
-{
- if ( m_autoSort ) {
- // use binary search in the sorted array
- wxASSERT_MSG( bCase && !bFromEnd,
- wxT("search parameters ignored for auto sorted array") );
-
- size_t i,
- lo = 0,
- hi = m_nCount;
- int res;
- while ( lo < hi ) {
- i = (lo + hi)/2;
-
- res = wxStrcmp(sz, m_pItems[i]);
- if ( res < 0 )
- hi = i;
- else if ( res > 0 )
- lo = i + 1;
- else
- return i;
- }
-
- return wxNOT_FOUND;
- }
- else {
- // use linear search in unsorted array
- if ( bFromEnd ) {
- if ( m_nCount > 0 ) {
- size_t ui = m_nCount;
- do {
- if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
- return ui;
- }
- while ( ui != 0 );
- }
- }
- else {
- for( size_t ui = 0; ui < m_nCount; ui++ ) {
- if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
- return ui;
- }
- }
- }
-
- return wxNOT_FOUND;
-}
-
-// add item at the end
-size_t wxArrayString::Add(const wxString& str, size_t nInsert)
-{
- if ( m_autoSort ) {
- // insert the string at the correct position to keep the array sorted
- size_t i,
- lo = 0,
- hi = m_nCount;
- int res;
- while ( lo < hi ) {
- i = (lo + hi)/2;
-
- res = str.Cmp(m_pItems[i]);
- if ( res < 0 )
- hi = i;
- else if ( res > 0 )
- lo = i + 1;
- else {
- lo = hi = i;
- break;
- }
- }
-
- wxASSERT_MSG( lo == hi, wxT("binary search broken") );
-
- Insert(str, lo, nInsert);
-
- return (size_t)lo;
- }
- else {
- wxASSERT( str.GetStringData()->IsValid() );
-
- Grow(nInsert);
-
- for (size_t i = 0; i < nInsert; i++)
- {
- // the string data must not be deleted!
- str.GetStringData()->Lock();
-
- // just append
- m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
- }
- size_t ret = m_nCount;
- m_nCount += nInsert;
- return ret;
- }
-}
-
-// add item at the given position
-void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
-{
- wxASSERT( str.GetStringData()->IsValid() );
-
- wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
- wxCHECK_RET( m_nCount <= m_nCount + nInsert,
- wxT("array size overflow in wxArrayString::Insert") );
-
- Grow(nInsert);
-
- memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
- (m_nCount - nIndex)*sizeof(wxChar *));
-
- for (size_t i = 0; i < nInsert; i++)
- {
- str.GetStringData()->Lock();
- m_pItems[nIndex + i] = (wxChar *)str.c_str();
- }
- m_nCount += nInsert;
-}
-
-// range insert (STL 23.2.4.3)
-void
-wxArrayString::insert(iterator it, const_iterator first, const_iterator last)
-{
- const int idx = it - begin();
-
- // grow it once
- Grow(last - first);
-
- // reset "it" since it can change inside Grow()
- it = begin() + idx;
-
- while ( first != last )
- {
- it = insert(it, *first);
-
- // insert returns an iterator to the last element inserted but we need
- // insert the next after this one, that is before the next one
- ++it;
-
- ++first;
- }
-}
-
-// expand the array
-void wxArrayString::SetCount(size_t count)
-{
- Alloc(count);
-
- wxString s;
- while ( m_nCount < count )
- m_pItems[m_nCount++] = (wxChar *)s.c_str();
-}
-
-// removes item from array (by index)
-void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove)
-{
- wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
- wxCHECK_RET( nIndex + nRemove <= m_nCount,
- wxT("removing too many elements in wxArrayString::Remove") );
-
- // release our lock
- for (size_t i = 0; i < nRemove; i++)
- Item(nIndex + i).GetStringData()->Unlock();
-
- memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
- (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
- m_nCount -= nRemove;
-}
-
-// removes item from array (by value)
-void wxArrayString::Remove(const wxChar *sz)
-{
- int iIndex = Index(sz);
-
- wxCHECK_RET( iIndex != wxNOT_FOUND,
- wxT("removing inexistent element in wxArrayString::Remove") );
-
- RemoveAt(iIndex);
-}
-
-void wxArrayString::assign(const_iterator first, const_iterator last)
-{
- reserve(last - first);
- for(; first != last; ++first)
- push_back(*first);
-}
-
-// ----------------------------------------------------------------------------
-// sorting
-// ----------------------------------------------------------------------------
-
-// we can only sort one array at a time with the quick-sort based
-// implementation
-#if wxUSE_THREADS
- // need a critical section to protect access to gs_compareFunction and
- // gs_sortAscending variables
- static wxCriticalSection *gs_critsectStringSort = NULL;
-
- // call this before the value of the global sort vars is changed/after
- // you're finished with them
- #define START_SORT() wxASSERT( !gs_critsectStringSort ); \
- gs_critsectStringSort = new wxCriticalSection; \
- gs_critsectStringSort->Enter()
- #define END_SORT() gs_critsectStringSort->Leave(); \
- delete gs_critsectStringSort; \
- gs_critsectStringSort = NULL
-#else // !threads
- #define START_SORT()
- #define END_SORT()
-#endif // wxUSE_THREADS
-
-// function to use for string comparaison
-static wxArrayString::CompareFunction gs_compareFunction = NULL;
-
-// if we don't use the compare function, this flag tells us if we sort the
-// array in ascending or descending order
-static bool gs_sortAscending = true;
-
-// function which is called by quick sort
-extern "C" int wxC_CALLING_CONV // LINKAGEMODE
-wxStringCompareFunction(const void *first, const void *second)
-{
- wxString *strFirst = (wxString *)first;
- wxString *strSecond = (wxString *)second;
-
- if ( gs_compareFunction ) {
- return gs_compareFunction(*strFirst, *strSecond);
- }
- else {
- // maybe we should use wxStrcoll
- int result = strFirst->Cmp(*strSecond);
-
- return gs_sortAscending ? result : -result;
- }
-}
-
-// sort array elements using passed comparaison function
-void wxArrayString::Sort(CompareFunction compareFunction)
-{
- START_SORT();
-
- wxASSERT( !gs_compareFunction ); // must have been reset to NULL
- gs_compareFunction = compareFunction;
-
- DoSort();
-
- // reset it to NULL so that Sort(bool) will work the next time
- gs_compareFunction = NULL;
-
- END_SORT();
-}
-
-extern "C"
-{
- typedef int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first,
- const void *second);
-}
-
-void wxArrayString::Sort(CompareFunction2 compareFunction)
-{
- qsort(m_pItems, m_nCount, sizeof(wxChar *), (wxStringCompareFn)compareFunction);
-}
-
-void wxArrayString::Sort(bool reverseOrder)
-{
- Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending);
-}
-
-void wxArrayString::DoSort()
-{
- wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
-
- // just sort the pointers using qsort() - of course it only works because
- // wxString() *is* a pointer to its data
- qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
-}
-
-bool wxArrayString::operator==(const wxArrayString& a) const
-{
- if ( m_nCount != a.m_nCount )
- return false;
-
- for ( size_t n = 0; n < m_nCount; n++ )
- {
- if ( Item(n) != a[n] )
- return false;
- }
-
- return true;
-}
-
-#endif // !wxUSE_STL
-
-int wxCMPFUNC_CONV wxStringSortAscending(wxString* s1, wxString* s2)
-{
- return s1->Cmp(*s2);
-}
-
-int wxCMPFUNC_CONV wxStringSortDescending(wxString* s1, wxString* s2)
-{
- return -s1->Cmp(*s2);
-}
projects / wxWidgets.git / blobdiff