/////////////////////////////////////////////////////////////////////////////
-// Name: string.cpp
+// Name: src/common/string.cpp
// Purpose: wxString class
-// Author: Vadim Zeitlin
+// Author: Vadim Zeitlin, Ryan Norton
// Modified by:
// Created: 29/01/98
// RCS-ID: $Id$
// Copyright: (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
-// Licence: wxWindows license
+// (c) 2004 Ryan Norton <wxprojects@comcast.net>
+// Licence: wxWindows licence
/////////////////////////////////////////////////////////////////////////////
-#ifdef __GNUG__
- #pragma implementation "string.h"
-#endif
-
-/*
- * About ref counting:
- * 1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
- * 2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
- * 3) Unlock() decrements nRefs and frees memory if it goes to 0
- */
-
// ===========================================================================
// headers, declarations, constants
// ===========================================================================
#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"
-#if wxUSE_THREADS
- #include <wx/thread.h>
-#endif
+ #include "wx/string.h"
+ #include "wx/wxcrtvararg.h"
#endif
#include <ctype.h>
-#include <string.h>
-#include <stdlib.h>
-#ifdef __SALFORDC__
- #include <clib.h>
+#ifndef __WXWINCE__
+ #include <errno.h>
#endif
-#if wxUSE_WCSRTOMBS
- #include <wchar.h> // for wcsrtombs(), see comments where it's used
-#endif // GNU
+#include <string.h>
+#include <stdlib.h>
-#ifdef WXSTRING_IS_WXOBJECT
- IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
-#endif //WXSTRING_IS_WXOBJECT
+#include "wx/hashmap.h"
+
+// string handling functions used by wxString:
+#if wxUSE_UNICODE_UTF8
+ #define wxStringMemcpy memcpy
+ #define wxStringMemcmp memcmp
+ #define wxStringMemchr memchr
+ #define wxStringStrlen strlen
+#else
+ #define wxStringMemcpy wxTmemcpy
+ #define wxStringMemcmp wxTmemcmp
+ #define wxStringMemchr wxTmemchr
+ #define wxStringStrlen wxStrlen
+#endif
-// allocating extra space for each string consumes more memory but speeds up
-// the concatenation operations (nLen is the current string's length)
-// NB: EXTRA_ALLOC must be >= 0!
-#define EXTRA_ALLOC (19 - nLen % 16)
// ---------------------------------------------------------------------------
// static class variables definition
// ---------------------------------------------------------------------------
-#ifdef wxSTD_STRING_COMPATIBILITY
- const size_t wxString::npos = wxSTRING_MAXLEN;
-#endif // wxSTD_STRING_COMPATIBILITY
+//According to STL _must_ be a -1 size_t
+const size_t wxString::npos = (size_t) -1;
// ----------------------------------------------------------------------------
-// static data
+// global functions
// ----------------------------------------------------------------------------
-// for an empty string, GetStringData() will return this address: this
-// structure has the same layout as wxStringData and it's data() method will
-// return the empty string (dummy pointer)
-static const struct
-{
- wxStringData data;
- wxChar dummy;
-} g_strEmpty = { {-1, 0, 0}, _T('\0') };
-
-// empty C style string: points to 'string data' byte of g_strEmpty
-extern const wxChar WXDLLEXPORT *g_szNul = &g_strEmpty.dummy;
+#if wxUSE_STD_IOSTREAM
-// ----------------------------------------------------------------------------
-// conditional compilation
-// ----------------------------------------------------------------------------
+#include <iostream>
-// we want to find out if the current platform supports vsnprintf()-like
-// function: for Unix this is done with configure, for Windows we test the
-// compiler explicitly.
-#ifdef __WXMSW__
- #ifdef __VISUALC__
- #define wxVsnprintf _vsnprintf
- #endif
-#else // !Windows
- #ifdef HAVE_VSNPRINTF
- #define wxVsnprintf vsnprintf
- #endif
-#endif // Windows/!Windows
-
-#ifndef wxVsnprintf
- // in this case we'll use vsprintf() (which is ANSI and thus should be
- // always available), but it's unsafe because it doesn't check for buffer
- // size - so give a warning
- #define wxVsnprintf(buffer,len,format,argptr) vsprintf(buffer,format, argptr)
-
- #if defined(__VISUALC__)
- #pragma message("Using sprintf() because no snprintf()-like function defined")
- #elif defined(__GNUG__) && !defined(__UNIX__)
- #warning "Using sprintf() because no snprintf()-like function defined"
- #elif defined(__MWERKS__)
- #warning "Using sprintf() because no snprintf()-like function defined"
- #endif //compiler
-#endif // no vsnprintf
-
-#ifdef _AIX
- // AIX has vsnprintf, but there's no prototype in the system headers.
- extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
+{
+#if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8
+ return os << (const char *)str.AsCharBuf();
+#else
+ return os << str.AsInternal();
#endif
+}
-// ----------------------------------------------------------------------------
-// global functions
-// ----------------------------------------------------------------------------
-
-#ifdef wxSTD_STRING_COMPATIBILITY
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
+{
+ return os << str.c_str();
+}
-// 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!
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str)
+{
+ return os << str.data();
+}
-istream& operator>>(istream& is, wxString& WXUNUSED(str))
+#ifndef __BORLANDC__
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str)
{
-#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;
- }
+ return os << str.data();
+}
+#endif
- str += ch;
- if ( --w == 1 )
- break;
- }
- }
+#if wxUSE_UNICODE && defined(HAVE_WOSTREAM)
- is.isfx();
- if ( str.length() == 0 )
- is.setstate(ios::failbit);
-#endif
- return is;
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str)
+{
+ return wos << str.wc_str();
}
-#endif //std::string compatibility
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str)
+{
+ return wos << str.AsWChar();
+}
-// ----------------------------------------------------------------------------
-// private classes
-// ----------------------------------------------------------------------------
+wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str)
+{
+ return wos << str.data();
+}
-// this small class is used to gather statistics for performance tuning
-//#define WXSTRING_STATISTICS
-#ifdef WXSTRING_STATISTICS
- class Averager
- {
- public:
- Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
- ~Averager()
- { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
-
- void Add(size_t n) { m_nTotal += n; m_nCount++; }
-
- private:
- size_t m_nCount, m_nTotal;
- const char *m_sz;
- } 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
+#endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM)
+
+#endif // wxUSE_STD_IOSTREAM
// ===========================================================================
// wxString class core
// ===========================================================================
-// ---------------------------------------------------------------------------
-// construction
-// ---------------------------------------------------------------------------
+#if wxUSE_UNICODE_UTF8
-// constructs string of <nLength> copies of character <ch>
-wxString::wxString(wxChar ch, size_t nLength)
+void wxString::PosLenToImpl(size_t pos, size_t len,
+ size_t *implPos, size_t *implLen) const
{
- Init();
-
- if ( nLength > 0 ) {
- AllocBuffer(nLength);
+ if ( pos == npos )
+ *implPos = npos;
+ else
+ {
+ const_iterator i = begin() + pos;
+ *implPos = wxStringImpl::const_iterator(i.impl()) - m_impl.begin();
+ if ( len == npos )
+ *implLen = npos;
+ else
+ {
+ // too large length is interpreted as "to the end of the string"
+ // FIXME-UTF8: verify this is the case in std::string, assert
+ // otherwise
+ if ( pos + len > length() )
+ len = length() - pos;
-#if wxUSE_UNICODE
- // memset only works on char
- for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
-#else
- memset(m_pchData, ch, nLength);
-#endif
- }
+ *implLen = (i + len).impl() - i.impl();
+ }
+ }
}
-// takes nLength elements of psz starting at nPos
-void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
-{
- Init();
-
- wxASSERT( nPos <= wxStrlen(psz) );
+#endif // wxUSE_UNICODE_UTF8
- if ( nLength == wxSTRING_MAXLEN )
- nLength = wxStrlen(psz + nPos);
+// ----------------------------------------------------------------------------
+// wxCStrData converted strings caching
+// ----------------------------------------------------------------------------
- STATISTICS_ADD(InitialLength, nLength);
+// 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 ( nLength > 0 ) {
- // trailing '\0' is written in AllocBuffer()
- AllocBuffer(nLength);
- memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
- }
+// 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() )
+ {
+ free(i->second);
+ hash.erase(i);
+ }
}
-#ifdef wxSTD_STRING_COMPATIBILITY
+#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;
-// poor man's iterators are "void *" pointers
-wxString::wxString(const void *pStart, const void *pEnd)
+const char* wxCStrData::AsChar() const
{
- InitWith((const wxChar *)pStart, 0,
- (const wxChar *)pEnd - (const wxChar *)pStart);
-}
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsCharCache, m_str);
-#endif //std::string compatibility
+ // convert the string and keep it:
+ const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
+ m_str->mb_str().release();
-#if wxUSE_UNICODE
+ return s + m_offset;
+}
+#endif // wxUSE_UNICODE
+
+#if !wxUSE_UNICODE_WCHAR
+WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
+ wxStringWCharConversionCache);
+static wxStringWCharConversionCache gs_stringsWCharCache;
-// from multibyte string
-wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
+const wchar_t* wxCStrData::AsWChar() const
{
- // first get necessary size
- size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
+ // remove previously cache value, if any (see FIXMEs above):
+ DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
- // nLength is number of *Unicode* characters here!
- if ((nLen != (size_t)-1) && (nLen > nLength))
- nLen = nLength;
+ // convert the string and keep it:
+ const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
+ m_str->wc_str().release();
- // empty?
- if ( (nLen != 0) && (nLen != (size_t)-1) ) {
- AllocBuffer(nLen);
- conv.MB2WC(m_pchData, psz, nLen);
- }
- else {
- Init();
- }
+ return s + m_offset;
}
+#endif // !wxUSE_UNICODE_WCHAR
-#else
-
-#if wxUSE_WCHAR_T
-// from wide string
-wxString::wxString(const wchar_t *pwz)
+wxString::~wxString()
{
- // first get necessary size
- size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0;
-
- // empty?
- if ( (nLen != 0) && (nLen != (size_t)-1) ) {
- AllocBuffer(nLen);
- wxWC2MB(m_pchData, pwz, nLen);
- }
- else {
- Init();
- }
+#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
+#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+const char* wxCStrData::AsChar() const
+{
+#if wxUSE_UNICODE_UTF8
+ if ( wxLocaleIsUtf8 )
+ return AsInternal();
#endif
+ // under non-UTF8 locales, we have to convert the internal UTF-8
+ // representation using wxConvLibc and cache the result
+
+ wxString *str = wxConstCast(m_str, wxString);
+
+ // convert the string:
+ //
+ // FIXME-UTF8: we'd like to do the conversion in the existing buffer (if we
+ // have it) but it's unfortunately not obvious to implement
+ // because we don't know how big buffer do we need for the
+ // given string length (in case of multibyte encodings, e.g.
+ // ISO-2022-JP or UTF-8 when internal representation is wchar_t)
+ //
+ // One idea would be to store more than just m_convertedToChar
+ // in wxString: then we could record the length of the string
+ // which was converted the last time and try to reuse the same
+ // buffer if the current length is not greater than it (this
+ // could still fail because string could have been modified in
+ // place but it would work most of the time, so we'd do it and
+ // only allocate the new buffer if in-place conversion returned
+ // an error). We could also store a bit saying if the string
+ // was modified since the last conversion (and update it in all
+ // operation modifying the string, of course) to avoid unneeded
+ // consequential conversions. But both of these ideas require
+ // adding more fields to wxString and require profiling results
+ // to be sure that we really gain enough from them to justify
+ // doing it.
+ wxCharBuffer buf(str->mb_str());
+
+ // if it failed, return empty string and not NULL to avoid crashes in code
+ // written with either wxWidgets 2 wxString or std::string behaviour in
+ // mind: neither of them ever returns NULL and so we shouldn't neither
+ if ( !buf )
+ return "";
+
+ if ( str->m_convertedToChar &&
+ strlen(buf) == strlen(str->m_convertedToChar) )
+ {
+ // keep the same buffer for as long as possible, so that several calls
+ // to c_str() in a row still work:
+ strcpy(str->m_convertedToChar, buf);
+ }
+ else
+ {
+ str->m_convertedToChar = buf.release();
+ }
+
+ // and keep it:
+ return str->m_convertedToChar + m_offset;
+}
+#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+
+#if !wxUSE_UNICODE_WCHAR
+const wchar_t* wxCStrData::AsWChar() const
+{
+ wxString *str = wxConstCast(m_str, wxString);
+
+ // convert the string:
+ wxWCharBuffer buf(str->wc_str());
+
+ // notice that here, unlike above in AsChar(), conversion can't fail as our
+ // internal UTF-8 is always well-formed -- or the string was corrupted and
+ // all bets are off anyhow
+
+ // FIXME-UTF8: do the conversion in-place in the existing buffer
+ if ( str->m_convertedToWChar &&
+ wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) )
+ {
+ // keep the same buffer for as long as possible, so that several calls
+ // to c_str() in a row still work:
+ memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf));
+ }
+ else
+ {
+ str->m_convertedToWChar = buf.release();
+ }
+
+ // and keep it:
+ return str->m_convertedToWChar + m_offset;
+}
+#endif // !wxUSE_UNICODE_WCHAR
+
+// ===========================================================================
+// wxString class core
+// ===========================================================================
// ---------------------------------------------------------------------------
-// memory allocation
+// construction and conversion
// ---------------------------------------------------------------------------
-// allocates memory needed to store a C string of length nLen
-void wxString::AllocBuffer(size_t nLen)
+#if wxUSE_UNICODE_WCHAR
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+ const wxMBConv& conv)
{
- wxASSERT( nLen > 0 ); //
- wxASSERT( nLen <= INT_MAX-1 ); // max size (enough room for 1 extra)
+ // anything to do?
+ if ( !psz || nLength == 0 )
+ return SubstrBufFromMB(L"", 0);
- STATISTICS_ADD(Length, nLen);
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
- // 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));
- pData->nRefs = 1;
- pData->nDataLength = nLen;
- pData->nAllocLength = nLen + EXTRA_ALLOC;
- m_pchData = pData->data(); // data starts after wxStringData
- m_pchData[nLen] = _T('\0');
+ size_t wcLen;
+ wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ if ( !wcLen )
+ return SubstrBufFromMB(_T(""), 0);
+ else
+ return SubstrBufFromMB(wcBuf, wcLen);
}
+#endif // wxUSE_UNICODE_WCHAR
-// must be called before changing this string
-void wxString::CopyBeforeWrite()
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+ const wxMBConv& conv)
{
- wxStringData* pData = GetStringData();
+ // anything to do?
+ if ( !psz || nLength == 0 )
+ return SubstrBufFromMB("", 0);
- if ( pData->IsShared() ) {
- pData->Unlock(); // memory not freed because shared
- size_t nLen = pData->nDataLength;
- AllocBuffer(nLen);
- memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
- }
+ // 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) )
+ {
+ return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
+ }
+ // else: do the roundtrip through wchar_t*
+ }
- wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
+
+ // first convert to wide string:
+ size_t wcLen;
+ wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+ if ( !wcLen )
+ return SubstrBufFromMB("", 0);
+
+ // and then to UTF-8:
+ SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8()));
+ // widechar -> UTF-8 conversion isn't supposed to ever fail:
+ wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") );
+
+ return buf;
}
+#endif // wxUSE_UNICODE_UTF8
-// must be called before replacing contents of this string
-void wxString::AllocBeforeWrite(size_t nLen)
+#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
+/* static */
+wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
+ const wxMBConv& conv)
{
- wxASSERT( nLen != 0 ); // doesn't make any sense
+ // anything to do?
+ if ( !pwz || nLength == 0 )
+ return SubstrBufFromWC("", 0);
- // must not share string and must have enough space
- wxStringData* pData = GetStringData();
- if ( pData->IsShared() || (nLen > pData->nAllocLength) ) {
- // can't work with old buffer, get new one
- pData->Unlock();
- AllocBuffer(nLen);
- }
- else {
- // update the string length
- pData->nDataLength = nLen;
- }
+ if ( nLength == npos )
+ nLength = wxNO_LEN;
- wxASSERT( !GetStringData()->IsShared() ); // we must be the only owner
+ size_t mbLen;
+ wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
+ if ( !mbLen )
+ return SubstrBufFromWC("", 0);
+ else
+ return SubstrBufFromWC(mbBuf, mbLen);
}
+#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
-// allocate enough memory for nLen characters
-void wxString::Alloc(size_t nLen)
-{
- wxStringData *pData = GetStringData();
- if ( pData->nAllocLength <= nLen ) {
- if ( pData->IsEmpty() ) {
- nLen += EXTRA_ALLOC;
- wxStringData* pData = (wxStringData*)
- malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
- pData->nRefs = 1;
- pData->nDataLength = 0;
- pData->nAllocLength = nLen;
- m_pchData = pData->data(); // data starts after wxStringData
- m_pchData[0u] = _T('\0');
- }
- else if ( pData->IsShared() ) {
- pData->Unlock(); // memory not freed because shared
- size_t nOldLen = pData->nDataLength;
- AllocBuffer(nLen);
- memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
- }
- else {
- nLen += EXTRA_ALLOC;
+#if wxUSE_UNICODE_WCHAR
- wxStringData *p = (wxStringData *)
- realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
+//Convert wxString in Unicode mode to a multi-byte string
+const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+{
+ return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);
+}
- if ( p == NULL ) {
- // @@@ what to do on memory error?
- return;
- }
+#elif wxUSE_UNICODE_UTF8
- // 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)
- p->nAllocLength = nLen;
- m_pchData = p->data();
- }
- }
- //else: we've already got enough
+const wxWCharBuffer wxString::wc_str() const
+{
+ return wxMBConvStrictUTF8().cMB2WC
+ (
+ m_impl.c_str(),
+ m_impl.length() + 1, // size, not length
+ NULL
+ );
}
-// shrink to minimal size (releasing extra memory)
-void wxString::Shrink()
+const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
{
- wxStringData *pData = GetStringData();
+ if ( conv.IsUTF8() )
+ return wxCharBuffer::CreateNonOwned(m_impl.c_str());
- // this variable is unused in release build, so avoid the compiler warning by
- // just not declaring it
-#ifdef __WXDEBUG__
- void *p =
-#endif
- realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
+ // FIXME-UTF8: use wc_str() here once we have buffers with length
- wxASSERT( p != NULL ); // can't free memory?
- wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
+ size_t wcLen;
+ wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC
+ (
+ m_impl.c_str(),
+ m_impl.length() + 1, // size
+ &wcLen
+ ));
+ if ( !wcLen )
+ return wxCharBuffer("");
+
+ return conv.cWC2MB(wcBuf, wcLen+1, NULL);
}
-// get the pointer to writable buffer of (at least) nLen bytes
-wxChar *wxString::GetWriteBuf(size_t nLen)
+#else // ANSI
+
+//Converts this string to a wide character string if unicode
+//mode is not enabled and wxUSE_WCHAR_T is enabled
+const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const
{
- AllocBeforeWrite(nLen);
+ return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL);
+}
+
+#endif // Unicode/ANSI
- wxASSERT( GetStringData()->nRefs == 1 );
- GetStringData()->Validate(FALSE);
+// shrink to minimal size (releasing extra memory)
+bool wxString::Shrink()
+{
+ wxString tmp(begin(), end());
+ swap(tmp);
+ return tmp.length() == length();
+}
- return m_pchData;
+// deprecated compatibility code:
+#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
+wxStringCharType *wxString::GetWriteBuf(size_t nLen)
+{
+ return DoGetWriteBuf(nLen);
}
-// put string back in a reasonable state after GetWriteBuf
void wxString::UngetWriteBuf()
{
- GetStringData()->nDataLength = wxStrlen(m_pchData);
- GetStringData()->Validate(TRUE);
+ DoUngetWriteBuf();
}
+void wxString::UngetWriteBuf(size_t nLen)
+{
+ DoUngetWriteBuf(nLen);
+}
+#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
+
+
// ---------------------------------------------------------------------------
// data access
// ---------------------------------------------------------------------------
// all functions are inline in string.h
// ---------------------------------------------------------------------------
-// assignment operators
+// concatenation operators
// ---------------------------------------------------------------------------
-// helper function: does real copy
-void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
-{
- if ( nSrcLen == 0 ) {
- Reinit();
- }
- else {
- AllocBeforeWrite(nSrcLen);
- memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
- GetStringData()->nDataLength = nSrcLen;
- m_pchData[nSrcLen] = _T('\0');
- }
-}
+/*
+ * concatenation functions come in 5 flavours:
+ * string + string
+ * char + string and string + char
+ * C str + string and string + C str
+ */
-// assigns one string to another
-wxString& wxString::operator=(const wxString& stringSrc)
+wxString operator+(const wxString& str1, const wxString& str2)
{
- wxASSERT( stringSrc.GetStringData()->IsValid() );
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str1.IsValid() );
+ wxASSERT( str2.IsValid() );
+#endif
- // don't copy string over itself
- if ( m_pchData != stringSrc.m_pchData ) {
- if ( stringSrc.GetStringData()->IsEmpty() ) {
- Reinit();
- }
- else {
- // adjust references
- GetStringData()->Unlock();
- m_pchData = stringSrc.m_pchData;
- GetStringData()->Lock();
- }
- }
+ wxString s = str1;
+ s += str2;
- return *this;
+ return s;
}
-// assigns a single character
-wxString& wxString::operator=(wxChar ch)
+wxString operator+(const wxString& str, wxUniChar ch)
{
- AssignCopy(1, &ch);
- return *this;
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
+
+ wxString s = str;
+ s += ch;
+
+ return s;
}
-// assigns C string
-wxString& wxString::operator=(const wxChar *psz)
+wxString operator+(wxUniChar ch, const wxString& str)
{
- AssignCopy(wxStrlen(psz), psz);
- return *this;
-}
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
-#if !wxUSE_UNICODE
+ wxString s = ch;
+ s += str;
-// same as 'signed char' variant
-wxString& wxString::operator=(const unsigned char* psz)
-{
- *this = (const char *)psz;
- return *this;
+ return s;
}
-#if wxUSE_WCHAR_T
-wxString& wxString::operator=(const wchar_t *pwz)
+wxString operator+(const wxString& str, const char *psz)
{
- wxString str(pwz);
- *this = str;
- return *this;
-}
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
#endif
-#endif
+ wxString s;
+ if ( !s.Alloc(strlen(psz) + str.length()) ) {
+ wxFAIL_MSG( _T("out of memory in wxString::operator+") );
+ }
+ s += str;
+ s += psz;
-// ---------------------------------------------------------------------------
-// string concatenation
-// ---------------------------------------------------------------------------
+ return s;
+}
-// add something to this string
-void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
+wxString operator+(const wxString& str, const wchar_t *pwz)
{
- STATISTICS_ADD(SummandLength, nSrcLen);
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- // concatenating an empty string is a NOP
- if ( nSrcLen > 0 ) {
- wxStringData *pData = GetStringData();
- size_t nLen = pData->nDataLength;
- size_t nNewLen = nLen + nSrcLen;
+ wxString s;
+ if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+ wxFAIL_MSG( _T("out of memory in wxString::operator+") );
+ }
+ s += str;
+ s += pwz;
- // alloc new buffer if current is too small
- if ( pData->IsShared() ) {
- STATISTICS_ADD(ConcatHit, 0);
+ return s;
+}
- // we have to allocate another buffer
- wxStringData* pOldData = GetStringData();
- AllocBuffer(nNewLen);
- memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
- pOldData->Unlock();
- }
- else if ( nNewLen > pData->nAllocLength ) {
- STATISTICS_ADD(ConcatHit, 0);
+wxString operator+(const char *psz, const wxString& str)
+{
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- // we have to grow the buffer
- Alloc(nNewLen);
+ wxString s;
+ if ( !s.Alloc(strlen(psz) + str.length()) ) {
+ wxFAIL_MSG( _T("out of memory in wxString::operator+") );
}
- else {
- STATISTICS_ADD(ConcatHit, 1);
+ s = psz;
+ s += str;
- // the buffer is already big enough
- }
+ return s;
+}
- // should be enough space
- wxASSERT( nNewLen <= GetStringData()->nAllocLength );
+wxString operator+(const wchar_t *pwz, const wxString& str)
+{
+#if !wxUSE_STL_BASED_WXSTRING
+ wxASSERT( str.IsValid() );
+#endif
- // fast concatenation - all is done in our buffer
- memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
+ wxString s;
+ if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+ wxFAIL_MSG( _T("out of memory in wxString::operator+") );
+ }
+ s = pwz;
+ s += str;
- m_pchData[nNewLen] = _T('\0'); // put terminating '\0'
- GetStringData()->nDataLength = nNewLen; // and fix the length
- }
- //else: the string to append was empty
+ return s;
}
-/*
- * concatenation functions come in 5 flavours:
- * string + string
- * char + string and string + char
- * C str + string and string + C str
- */
+// ---------------------------------------------------------------------------
+// string comparison
+// ---------------------------------------------------------------------------
-wxString operator+(const wxString& string1, const wxString& string2)
+bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const
{
- wxASSERT( string1.GetStringData()->IsValid() );
- wxASSERT( string2.GetStringData()->IsValid() );
+ return (length() == 1) && (compareWithCase ? GetChar(0u) == c
+ : wxToupper(GetChar(0u)) == wxToupper(c));
+}
+
+#ifdef HAVE_STD_STRING_COMPARE
- wxString s = string1;
- s += string2;
+// 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).
- return s;
+int wxString::compare(const wxString& str) const
+{
+ return m_impl.compare(str.m_impl);
+}
+
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str) const
+{
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+ return m_impl.compare(pos, len, str.m_impl);
}
-wxString operator+(const wxString& string, wxChar ch)
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str,
+ size_t nStart2, size_t nLen2) const
{
- wxASSERT( string.GetStringData()->IsValid() );
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- wxString s = string;
- s += ch;
+ size_t pos2, len2;
+ str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
- return s;
+ return m_impl.compare(pos, len, str.m_impl, pos2, len2);
}
-wxString operator+(wxChar ch, const wxString& string)
+int wxString::compare(const char* sz) const
{
- wxASSERT( string.GetStringData()->IsValid() );
-
- wxString s = ch;
- s += string;
+ return m_impl.compare(ImplStr(sz));
+}
- return s;
+int wxString::compare(const wchar_t* sz) const
+{
+ return m_impl.compare(ImplStr(sz));
}
-wxString operator+(const wxString& string, const wxChar *psz)
+int wxString::compare(size_t nStart, size_t nLen,
+ const char* sz, size_t nCount) const
{
- wxASSERT( string.GetStringData()->IsValid() );
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- wxString s;
- s.Alloc(wxStrlen(psz) + string.Len());
- s = string;
- s += psz;
+ SubstrBufFromMB str(ImplStr(sz, nCount));
- return s;
+ return m_impl.compare(pos, len, str.data, str.len);
}
-wxString operator+(const wxChar *psz, const wxString& string)
+int wxString::compare(size_t nStart, size_t nLen,
+ const wchar_t* sz, size_t nCount) const
{
- wxASSERT( string.GetStringData()->IsValid() );
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- wxString s;
- s.Alloc(wxStrlen(psz) + string.Len());
- s = psz;
- s += string;
+ SubstrBufFromWC str(ImplStr(sz, nCount));
- return s;
+ return m_impl.compare(pos, len, str.data, str.len);
}
-// ===========================================================================
-// other common string functions
-// ===========================================================================
+#else // !HAVE_STD_STRING_COMPARE
-// ---------------------------------------------------------------------------
-// simple sub-string extraction
-// ---------------------------------------------------------------------------
+static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
+ const wxStringCharType* s2, size_t l2)
+{
+ if( l1 == l2 )
+ return wxStringMemcmp(s1, s2, l1);
+ else if( l1 < l2 )
+ {
+ int ret = wxStringMemcmp(s1, s2, l1);
+ return ret == 0 ? -1 : ret;
+ }
+ else
+ {
+ int ret = wxStringMemcmp(s1, s2, l2);
+ return ret == 0 ? +1 : ret;
+ }
+}
-// helper function: clone the data attached to this string
-void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
+int wxString::compare(const wxString& str) const
{
- if ( nCopyLen == 0 ) {
- dest.Init();
- }
- else {
- dest.AllocBuffer(nCopyLen);
- memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
- }
+ return ::wxDoCmp(m_impl.data(), m_impl.length(),
+ str.m_impl.data(), str.m_impl.length());
}
-// extract string of length nCount starting at nFirst
-wxString wxString::Mid(size_t nFirst, size_t nCount) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wxString& str) const
{
- wxStringData *pData = GetStringData();
- size_t nLen = pData->nDataLength;
+ wxASSERT(nStart <= length());
+ size_type strLen = length() - nStart;
+ nLen = strLen < nLen ? strLen : nLen;
- // default value of nCount is wxSTRING_MAXLEN and means "till the end"
- if ( nCount == wxSTRING_MAXLEN )
- {
- nCount = nLen - nFirst;
- }
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
- // out-of-bounds requests return sensible things
- if ( nFirst + nCount > nLen )
- {
- nCount = nLen - nFirst;
- }
+ return ::wxDoCmp(m_impl.data() + pos, len,
+ str.m_impl.data(), str.m_impl.length());
+}
- if ( nFirst > nLen )
- {
- // AllocCopy() will return empty string
- nCount = 0;
- }
+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());
+ size_type strLen = length() - nStart,
+ strLen2 = str.length() - nStart2;
+ nLen = strLen < nLen ? strLen : nLen;
+ nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
- wxString dest;
- AllocCopy(dest, nCount, nFirst);
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+ size_t pos2, len2;
+ str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
- return dest;
+ return ::wxDoCmp(m_impl.data() + pos, len,
+ str.m_impl.data() + pos2, len2);
}
-// extract nCount last (rightmost) characters
-wxString wxString::Right(size_t nCount) const
+int wxString::compare(const char* sz) const
{
- if ( nCount > (size_t)GetStringData()->nDataLength )
- nCount = GetStringData()->nDataLength;
-
- wxString dest;
- AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
- return dest;
+ 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);
}
-// get all characters after the last occurence of ch
-// (returns the whole string if ch not found)
-wxString wxString::AfterLast(wxChar ch) const
+int wxString::compare(const wchar_t* sz) const
{
- wxString str;
- int iPos = Find(ch, TRUE);
- if ( iPos == wxNOT_FOUND )
- str = *this;
- else
- str = c_str() + iPos + 1;
-
- return str;
+ 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);
}
-// extract nCount first (leftmost) characters
-wxString wxString::Left(size_t nCount) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const char* sz, size_t nCount) const
{
- if ( nCount > (size_t)GetStringData()->nDataLength )
- nCount = GetStringData()->nDataLength;
+ wxASSERT(nStart <= length());
+ size_type strLen = length() - nStart;
+ nLen = strLen < nLen ? strLen : nLen;
- wxString dest;
- AllocCopy(dest, nCount, 0);
- return dest;
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+
+ SubstrBufFromMB str(ImplStr(sz, nCount));
+ if ( str.len == npos )
+ str.len = wxStringStrlen(str.data);
+
+ return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
}
-// get all characters before the first occurence of ch
-// (returns the whole string if ch not found)
-wxString wxString::BeforeFirst(wxChar ch) const
+int wxString::compare(size_t nStart, size_t nLen,
+ const wchar_t* sz, size_t nCount) const
{
- wxString str;
- for ( const wxChar *pc = m_pchData; *pc != _T('\0') && *pc != ch; pc++ )
- str += *pc;
+ wxASSERT(nStart <= length());
+ size_type strLen = length() - nStart;
+ nLen = strLen < nLen ? strLen : nLen;
- return str;
+ size_t pos, len;
+ PosLenToImpl(nStart, nLen, &pos, &len);
+
+ SubstrBufFromWC str(ImplStr(sz, nCount));
+ if ( str.len == npos )
+ str.len = wxStringStrlen(str.data);
+
+ return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
}
-/// get all characters before the last occurence of ch
-/// (returns empty string if ch not found)
-wxString wxString::BeforeLast(wxChar ch) const
-{
- wxString str;
- int iPos = Find(ch, TRUE);
- if ( iPos != wxNOT_FOUND && iPos != 0 )
- str = wxString(c_str(), iPos);
-
- return str;
-}
-
-/// get all characters after the first occurence of ch
-/// (returns empty string if ch not found)
-wxString wxString::AfterFirst(wxChar ch) const
-{
- wxString str;
- int iPos = Find(ch);
- if ( iPos != wxNOT_FOUND )
- str = c_str() + iPos + 1;
-
- 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)
-{
- size_t uiCount = 0; // count of replacements made
+#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
- size_t uiOldLen = wxStrlen(szOld);
- wxString strTemp;
- const wxChar *pCurrent = m_pchData;
- const wxChar *pSubstr;
- while ( *pCurrent != _T('\0') ) {
- pSubstr = wxStrstr(pCurrent, szOld);
- if ( pSubstr == NULL ) {
- // strTemp is unused if no replacements were made, so avoid the copy
- if ( uiCount == 0 )
- return 0;
-
- strTemp += pCurrent; // copy the rest
- break; // exit the loop
- }
- else {
- // take chars before match
- strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
- strTemp += szNew;
- pCurrent = pSubstr + uiOldLen; // restart after match
-
- uiCount++;
-
- // stop now?
- if ( !bReplaceAll ) {
- strTemp += pCurrent; // copy the rest
- break; // exit the loop
- }
- }
- }
+// ---------------------------------------------------------------------------
+// find_{first,last}_[not]_of functions
+// ---------------------------------------------------------------------------
- // only done if there were replacements, otherwise would have returned above
- *this = strTemp;
+#if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
- return uiCount;
-}
+// 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.
-bool wxString::IsAscii() const
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!isascii(*s)) return(FALSE);
- s++;
- }
- return(TRUE);
+ return find_first_of(sz, nStart, wxStrlen(sz));
}
-bool wxString::IsWord() const
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!wxIsalpha(*s)) return(FALSE);
- s++;
- }
- return(TRUE);
+ return find_first_not_of(sz, nStart, wxStrlen(sz));
}
-bool wxString::IsNumber() const
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
{
- const wxChar *s = (const wxChar*) *this;
- while(*s){
- if(!wxIsdigit(*s)) return(FALSE);
- s++;
- }
- return(TRUE);
-}
+ wxASSERT_MSG( nStart <= length(), _T("invalid index") );
-wxString wxString::Strip(stripType w) const
-{
- wxString s = *this;
- if ( w & leading ) s.Trim(FALSE);
- if ( w & trailing ) s.Trim(TRUE);
- return s;
-}
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( wxTmemchr(sz, *i, n) )
+ return idx;
+ }
-// ---------------------------------------------------------------------------
-// case conversion
-// ---------------------------------------------------------------------------
+ return npos;
+}
-wxString& wxString::MakeUpper()
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
{
- CopyBeforeWrite();
+ wxASSERT_MSG( nStart <= length(), _T("invalid index") );
- for ( wxChar *p = m_pchData; *p; p++ )
- *p = (wxChar)wxToupper(*p);
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( !wxTmemchr(sz, *i, n) )
+ return idx;
+ }
- return *this;
+ return npos;
}
-wxString& wxString::MakeLower()
-{
- CopyBeforeWrite();
- for ( wxChar *p = m_pchData; *p; p++ )
- *p = (wxChar)wxTolower(*p);
-
- return *this;
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
+{
+ return find_last_of(sz, nStart, wxStrlen(sz));
}
-// ---------------------------------------------------------------------------
-// trimming and padding
-// ---------------------------------------------------------------------------
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
+{
+ return find_last_not_of(sz, nStart, wxStrlen(sz));
+}
-// trims spaces (in the sense of isspace) from left or right side
-wxString& wxString::Trim(bool bFromRight)
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const
{
- // first check if we're going to modify the string at all
- if ( !IsEmpty() &&
- (
- (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
- (!bFromRight && wxIsspace(GetChar(0u)))
- )
- )
- {
- // ok, there is at least one space to trim
- CopyBeforeWrite();
+ size_t len = length();
- if ( bFromRight )
+ if ( nStart == npos )
{
- // find last non-space character
- wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
- while ( wxIsspace(*psz) && (psz >= m_pchData) )
- psz--;
-
- // truncate at trailing space start
- *++psz = _T('\0');
- GetStringData()->nDataLength = psz - m_pchData;
+ nStart = len - 1;
}
else
{
- // find first non-space character
- const wxChar *psz = m_pchData;
- while ( wxIsspace(*psz) )
- psz++;
-
- // fix up data and length
- int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
- memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
- GetStringData()->nDataLength = nDataLength;
+ wxASSERT_MSG( nStart <= len, _T("invalid index") );
}
- }
- return *this;
+ 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;
}
-// adds nCount characters chPad to the string from either side
-wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const
{
- wxString s(chPad, nCount);
+ size_t len = length();
- if ( bFromRight )
- *this += s;
- else
- {
- s += *this;
- *this = s;
- }
+ if ( nStart == npos )
+ {
+ nStart = len - 1;
+ }
+ else
+ {
+ wxASSERT_MSG( nStart <= len, _T("invalid index") );
+ }
- return *this;
+ 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;
}
-// truncate the string
-wxString& wxString::Truncate(size_t uiLen)
+size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const
{
- if ( uiLen < Len() ) {
- CopyBeforeWrite();
+ wxASSERT_MSG( nStart <= length(), _T("invalid index") );
- *(m_pchData + uiLen) = _T('\0');
- GetStringData()->nDataLength = uiLen;
- }
- //else: nothing to do, string is already short enough
+ size_t idx = nStart;
+ for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+ {
+ if ( *i != ch )
+ return idx;
+ }
- return *this;
+ return npos;
}
-// ---------------------------------------------------------------------------
-// finding (return wxNOT_FOUND if not found and index otherwise)
-// ---------------------------------------------------------------------------
-
-// find a character
-int wxString::Find(wxChar ch, bool bFromEnd) const
+size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const
{
- const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
+ size_t len = length();
- return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
-}
+ if ( nStart == npos )
+ {
+ nStart = len - 1;
+ }
+ else
+ {
+ wxASSERT_MSG( nStart <= len, _T("invalid index") );
+ }
-// find a sub-string (like strstr)
-int wxString::Find(const wxChar *pszSub) const
-{
- const wxChar *psz = wxStrstr(m_pchData, pszSub);
+ size_t idx = nStart;
+ for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+ i != rend(); --idx, ++i )
+ {
+ if ( *i != ch )
+ return idx;
+ }
- return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
+ return npos;
}
-// ---------------------------------------------------------------------------
-// stream-like operators
-// ---------------------------------------------------------------------------
-wxString& wxString::operator<<(int i)
-{
- wxString res;
- res.Printf(_T("%d"), i);
-
- return (*this) << res;
-}
+// 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& wxString::operator<<(float f)
-{
- wxString res;
- res.Printf(_T("%f"), f);
+size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const
+ { return find_first_of(STRCONV(sz), nStart); }
+
+size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart,
+ size_t n) const
+ { return find_first_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const
+ { return find_last_of(STRCONV(sz), nStart); }
+size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart,
+ size_t n) const
+ { return find_last_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const
+ { return find_first_not_of(STRCONV(sz), nStart); }
+size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart,
+ size_t n) const
+ { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const
+ { return find_last_not_of(STRCONV(sz), nStart); }
+size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart,
+ size_t n) const
+ { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); }
+
+#undef wxOtherCharType
+#undef STRCONV
+
+#endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
- return (*this) << res;
-}
+// ===========================================================================
+// other common string functions
+// ===========================================================================
-wxString& wxString::operator<<(double d)
+int wxString::CmpNoCase(const wxString& s) const
{
- wxString res;
- res.Printf(_T("%g"), d);
-
- return (*this) << res;
-}
+ // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
-// ---------------------------------------------------------------------------
-// formatted output
-// ---------------------------------------------------------------------------
-int wxString::Printf(const wxChar *pszFormat, ...)
-{
- va_list argptr;
- va_start(argptr, pszFormat);
+ const_iterator i1 = begin();
+ const_iterator end1 = end();
+ const_iterator i2 = s.begin();
+ const_iterator end2 = s.end();
- int iLen = PrintfV(pszFormat, argptr);
+ for ( ; i1 != end1 && i2 != end2; ++i1, ++i2 )
+ {
+ wxUniChar lower1 = (wxChar)wxTolower(*i1);
+ wxUniChar lower2 = (wxChar)wxTolower(*i2);
+ if ( lower1 != lower2 )
+ return lower1 < lower2 ? -1 : 1;
+ }
- va_end(argptr);
+ size_t len1 = length();
+ size_t len2 = s.length();
- return iLen;
+ if ( len1 < len2 )
+ return -1;
+ else if ( len1 > len2 )
+ return 1;
+ return 0;
}
-int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
-{
- // static buffer to avoid dynamic memory allocation each time
- static char s_szScratch[1024];
-#if wxUSE_THREADS
- // protect the static buffer
- static wxCriticalSection critsect;
- wxCriticalSectionLocker lock(critsect);
-#endif
-#if 1 // the new implementation
-
- Reinit();
- for (size_t n = 0; pszFormat[n]; n++)
- if (pszFormat[n] == _T('%')) {
- static char s_szFlags[256] = "%";
- size_t flagofs = 1;
- bool adj_left = FALSE, in_prec = FALSE,
- prec_dot = FALSE, done = FALSE;
- int ilen = 0;
- size_t min_width = 0, max_width = wxSTRING_MAXLEN;
- do {
-#define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
- switch (pszFormat[++n]) {
- case _T('\0'):
- done = TRUE;
- break;
- case _T('%'):
- *this += _T('%');
- done = TRUE;
- break;
- case _T('#'):
- case _T('0'):
- case _T(' '):
- case _T('+'):
- case _T('\''):
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('-'):
- CHECK_PREC
- adj_left = TRUE;
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('.'):
- CHECK_PREC
- in_prec = TRUE;
- prec_dot = FALSE;
- max_width = 0;
- // dot will be auto-added to s_szFlags if non-negative number follows
- break;
- case _T('h'):
- ilen = -1;
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('l'):
- ilen = 1;
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('q'):
- case _T('L'):
- ilen = 2;
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('Z'):
- ilen = 3;
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- break;
- case _T('*'):
- {
- int len = va_arg(argptr, int);
- if (in_prec) {
- if (len<0) break;
- CHECK_PREC
- max_width = len;
- } else {
- if (len<0) {
- adj_left = !adj_left;
- s_szFlags[flagofs++] = '-';
- len = -len;
- }
- min_width = len;
- }
- flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
- }
- break;
- case _T('1'): case _T('2'): case _T('3'):
- case _T('4'): case _T('5'): case _T('6'):
- case _T('7'): case _T('8'): case _T('9'):
- {
- int len = 0;
- CHECK_PREC
- while ((pszFormat[n]>=_T('0')) && (pszFormat[n]<=_T('9'))) {
- s_szFlags[flagofs++] = pszFormat[n];
- len = len*10 + (pszFormat[n] - _T('0'));
- n++;
- }
- if (in_prec) max_width = len;
- else min_width = len;
- n--; // the main loop pre-increments n again
- }
- break;
- case _T('d'):
- case _T('i'):
- case _T('o'):
- case _T('u'):
- case _T('x'):
- case _T('X'):
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- s_szFlags[flagofs] = '\0';
- if (ilen == 0 ) {
- int val = va_arg(argptr, int);
- ::sprintf(s_szScratch, s_szFlags, val);
- }
- else if (ilen == -1) {
- short int val = va_arg(argptr, short int);
- ::sprintf(s_szScratch, s_szFlags, val);
- }
- else if (ilen == 1) {
- long int val = va_arg(argptr, long int);
- ::sprintf(s_szScratch, s_szFlags, val);
- }
- else if (ilen == 2) {
-#if SIZEOF_LONG_LONG
- long long int val = va_arg(argptr, long long int);
- ::sprintf(s_szScratch, s_szFlags, val);
-#else
- long int val = va_arg(argptr, long int);
- ::sprintf(s_szScratch, s_szFlags, val);
-#endif
- }
- else if (ilen == 3) {
- size_t val = va_arg(argptr, size_t);
- ::sprintf(s_szScratch, s_szFlags, val);
- }
- *this += wxString(s_szScratch);
- done = TRUE;
- break;
- case _T('e'):
- case _T('E'):
- case _T('f'):
- case _T('g'):
- case _T('G'):
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- s_szFlags[flagofs] = '\0';
- if (ilen == 2) {
- long double val = va_arg(argptr, long double);
- ::sprintf(s_szScratch, s_szFlags, val);
- } else {
- double val = va_arg(argptr, double);
- ::sprintf(s_szScratch, s_szFlags, val);
- }
- *this += wxString(s_szScratch);
- done = TRUE;
- break;
- case _T('p'):
- {
- void *val = va_arg(argptr, void *);
- CHECK_PREC
- s_szFlags[flagofs++] = pszFormat[n];
- s_szFlags[flagofs] = '\0';
- ::sprintf(s_szScratch, s_szFlags, val);
- *this += wxString(s_szScratch);
- done = TRUE;
- }
- break;
- case _T('c'):
- {
- wxChar val = va_arg(argptr, int);
- // we don't need to honor padding here, do we?
- *this += val;
- done = TRUE;
- }
- break;
- case _T('s'):
- if (ilen == -1) {
- // wx extension: we'll let %hs mean non-Unicode strings
- char *val = va_arg(argptr, char *);
#if wxUSE_UNICODE
- // ASCII->Unicode constructor handles max_width right
- wxString s(val, wxConvLibc, max_width);
-#else
- size_t len = wxSTRING_MAXLEN;
- if (val) {
- for (len = 0; val[len] && (len<max_width); len++);
- } else val = _T("(null)");
- wxString s(val, len);
-#endif
- if (s.Len() < min_width)
- s.Pad(min_width - s.Len(), _T(' '), adj_left);
- *this += s;
- } else {
- wxChar *val = va_arg(argptr, wxChar *);
- size_t len = wxSTRING_MAXLEN;
- if (val) {
- for (len = 0; val[len] && (len<max_width); len++);
- } else val = _T("(null)");
- wxString s(val, len);
- if (s.Len() < min_width)
- s.Pad(min_width - s.Len(), _T(' '), adj_left);
- *this += s;
- }
- done = TRUE;
- break;
- case _T('n'):
- if (ilen == 0) {
- int *val = va_arg(argptr, int *);
- *val = Len();
- }
- else if (ilen == -1) {
- short int *val = va_arg(argptr, short int *);
- *val = Len();
- }
- else if (ilen >= 1) {
- long int *val = va_arg(argptr, long int *);
- *val = Len();
- }
- done = TRUE;
- break;
- default:
- if (wxIsalpha(pszFormat[n]))
- // probably some flag not taken care of here yet
- s_szFlags[flagofs++] = pszFormat[n];
- else {
- // bad format
- *this += _T('%'); // just to pass the glibc tst-printf.c
- n--;
- done = TRUE;
- }
- break;
- }
-#undef CHECK_PREC
- } while (!done);
- } else *this += pszFormat[n];
-#else
- // NB: wxVsnprintf() may return either less than the buffer size or -1 if there
- // is not enough place depending on implementation
- int iLen = wxVsnprintf(s_szScratch, WXSIZEOF(s_szScratch), pszFormat, argptr);
- char *buffer;
- if ( iLen < (int)WXSIZEOF(s_szScratch) ) {
- buffer = s_szScratch;
- }
- else {
- int size = WXSIZEOF(s_szScratch) * 2;
- buffer = (char *)malloc(size);
- while ( buffer != NULL ) {
- iLen = wxVsnprintf(buffer, WXSIZEOF(s_szScratch), pszFormat, argptr);
- if ( iLen < size ) {
- // ok, there was enough space
- break;
- }
-
- // still not enough, double it again
- buffer = (char *)realloc(buffer, size *= 2);
- }
-
- if ( !buffer ) {
- // out of memory
- return -1;
- }
- }
-
- wxString s(buffer);
- *this = s;
-
- if ( buffer != s_szScratch )
- free(buffer);
+#ifdef __MWERKS__
+#ifndef __SCHAR_MAX__
+#define __SCHAR_MAX__ 127
+#endif
#endif
- return Len();
-}
-
-// ----------------------------------------------------------------------------
-// misc other operations
-// ----------------------------------------------------------------------------
-bool wxString::Matches(const wxChar *pszMask) const
+wxString wxString::FromAscii(const char *ascii, size_t len)
{
- // check char by char
- const wxChar *pszTxt;
- for ( pszTxt = c_str(); *pszMask != _T('\0'); pszMask++, pszTxt++ ) {
- switch ( *pszMask ) {
- case _T('?'):
- if ( *pszTxt == _T('\0') )
- return FALSE;
+ if (!ascii || len == 0)
+ return wxEmptyString;
- pszTxt++;
- pszMask++;
- break;
+ wxString res;
- case _T('*'):
+ {
+ wxStringInternalBuffer buf(res, len);
+ wxStringCharType *dest = buf;
+
+ for ( ; len > 0; --len )
{
- // ignore special chars immediately following this one
- while ( *pszMask == _T('*') || *pszMask == _T('?') )
- pszMask++;
+ unsigned char c = (unsigned char)*ascii++;
+ wxASSERT_MSG( c < 0x80,
+ _T("Non-ASCII value passed to FromAscii().") );
- // if there is nothing more, match
- if ( *pszMask == _T('\0') )
- return TRUE;
+ *dest++ = (wchar_t)c;
+ }
+ }
- // are there any other metacharacters in the mask?
- size_t uiLenMask;
- const wxChar *pEndMask = wxStrpbrk(pszMask, _T("*?"));
+ return res;
+}
- if ( pEndMask != NULL ) {
- // we have to match the string between two metachars
- uiLenMask = pEndMask - pszMask;
- }
- else {
- // we have to match the remainder of the string
- uiLenMask = wxStrlen(pszMask);
- }
+wxString wxString::FromAscii(const char *ascii)
+{
+ return FromAscii(ascii, wxStrlen(ascii));
+}
- wxString strToMatch(pszMask, uiLenMask);
- const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
- if ( pMatch == NULL )
- return FALSE;
+wxString wxString::FromAscii(char ascii)
+{
+ // What do we do with '\0' ?
- // -1 to compensate "++" in the loop
- pszTxt = pMatch + uiLenMask - 1;
- pszMask += uiLenMask - 1;
- }
- break;
+ unsigned char c = (unsigned char)ascii;
- default:
- if ( *pszMask != *pszTxt )
- return FALSE;
- break;
- }
- }
+ wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") );
- // match only if nothing left
- return *pszTxt == _T('\0');
+ // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value
+ return wxString(wxUniChar((wchar_t)c));
}
-// Count the number of chars
-int wxString::Freq(wxChar ch) const
+const wxCharBuffer wxString::ToAscii() const
{
- int count = 0;
- int len = Len();
- for (int i = 0; i < len; i++)
+ // this will allocate enough space for the terminating NUL too
+ wxCharBuffer buffer(length());
+ char *dest = buffer.data();
+
+ for ( const_iterator i = begin(); i != end(); ++i )
{
- if (GetChar(i) == ch)
- count ++;
+ wxUniChar c(*i);
+ // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?')
+ *dest++ = c.IsAscii() ? (char)c : '_';
+
+ // the output string can't have embedded NULs anyhow, so we can safely
+ // stop at first of them even if we do have any
+ if ( !c )
+ break;
}
- return count;
+
+ return buffer;
}
-// convert to upper case, return the copy of the string
-wxString wxString::Upper() const
-{ wxString s(*this); return s.MakeUpper(); }
+#endif // wxUSE_UNICODE
-// convert to lower case, return the copy of the string
-wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
+// extract string of length nCount starting at nFirst
+wxString wxString::Mid(size_t nFirst, size_t nCount) const
+{
+ size_t nLen = length();
-int wxString::sprintf(const wxChar *pszFormat, ...)
- {
- va_list argptr;
- va_start(argptr, pszFormat);
- int iLen = PrintfV(pszFormat, argptr);
- va_end(argptr);
- return iLen;
- }
+ // default value of nCount is npos and means "till the end"
+ if ( nCount == npos )
+ {
+ nCount = nLen - nFirst;
+ }
-// ---------------------------------------------------------------------------
-// standard C++ library string functions
-// ---------------------------------------------------------------------------
-#ifdef wxSTD_STRING_COMPATIBILITY
+ // out-of-bounds requests return sensible things
+ if ( nFirst + nCount > nLen )
+ {
+ nCount = nLen - nFirst;
+ }
-wxString& wxString::insert(size_t nPos, const wxString& str)
-{
- wxASSERT( str.GetStringData()->IsValid() );
- wxASSERT( nPos <= Len() );
+ if ( nFirst > nLen )
+ {
+ // AllocCopy() will return empty string
+ return wxEmptyString;
+ }
- if ( !str.IsEmpty() ) {
- wxString strTmp;
- wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
- wxStrncpy(pc, c_str(), nPos);
- wxStrcpy(pc + nPos, str);
- wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
- strTmp.UngetWriteBuf();
- *this = strTmp;
- }
+ wxString dest(*this, nFirst, nCount);
+ if ( dest.length() != nCount )
+ {
+ wxFAIL_MSG( _T("out of memory in wxString::Mid") );
+ }
- return *this;
+ return dest;
}
-size_t wxString::find(const wxString& str, size_t nStart) const
+// check that the string starts with prefix and return the rest of the string
+// in the provided pointer if it is not NULL, otherwise return false
+bool wxString::StartsWith(const wxString& prefix, wxString *rest) const
{
- wxASSERT( str.GetStringData()->IsValid() );
- wxASSERT( nStart <= Len() );
+ if ( compare(0, prefix.length(), prefix) != 0 )
+ return false;
- const wxChar *p = wxStrstr(c_str() + nStart, str);
+ if ( rest )
+ {
+ // put the rest of the string into provided pointer
+ rest->assign(*this, prefix.length(), npos);
+ }
- return p == NULL ? npos : p - c_str();
+ return true;
}
-// VC++ 1.5 can't cope with the default argument in the header.
-#if !defined(__VISUALC__) || defined(__WIN32__)
-size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
-{
- return find(wxString(sz, n == npos ? 0 : n), nStart);
-}
-#endif // VC++ 1.5
-// Gives a duplicate symbol (presumably a case-insensitivity problem)
-#if !defined(__BORLANDC__)
-size_t wxString::find(wxChar ch, size_t nStart) const
+// 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
{
- wxASSERT( nStart <= Len() );
+ int start = length() - suffix.length();
- const wxChar *p = wxStrchr(c_str() + nStart, ch);
+ if ( start < 0 || compare(start, npos, suffix) != 0 )
+ return false;
- return p == NULL ? npos : p - c_str();
+ if ( rest )
+ {
+ // put the rest of the string into provided pointer
+ rest->assign(*this, 0, start);
+ }
+
+ return true;
}
-#endif
-size_t wxString::rfind(const wxString& str, size_t nStart) const
+
+// extract nCount last (rightmost) characters
+wxString wxString::Right(size_t nCount) const
{
- wxASSERT( str.GetStringData()->IsValid() );
- wxASSERT( nStart <= Len() );
+ if ( nCount > length() )
+ nCount = length();
- // # could be quicker than that
- const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
- while ( p >= c_str() + str.Len() ) {
- if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
- return p - str.Len() - c_str();
- p--;
+ wxString dest(*this, length() - nCount, nCount);
+ if ( dest.length() != nCount ) {
+ wxFAIL_MSG( _T("out of memory in wxString::Right") );
}
-
- return npos;
+ return dest;
}
-// VC++ 1.5 can't cope with the default argument in the header.
-#if !defined(__VISUALC__) || defined(__WIN32__)
-size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
+// get all characters after the last occurence of ch
+// (returns the whole string if ch not found)
+wxString wxString::AfterLast(wxUniChar ch) const
{
- return rfind(wxString(sz, n == npos ? 0 : n), nStart);
+ wxString str;
+ int iPos = Find(ch, true);
+ if ( iPos == wxNOT_FOUND )
+ str = *this;
+ else
+ str = wx_str() + iPos + 1;
+
+ return str;
}
-size_t wxString::rfind(wxChar ch, size_t nStart) const
+// extract nCount first (leftmost) characters
+wxString wxString::Left(size_t nCount) const
{
- wxASSERT( nStart <= Len() );
+ if ( nCount > length() )
+ nCount = length();
- const wxChar *p = wxStrrchr(c_str() + nStart, ch);
-
- return p == NULL ? npos : p - c_str();
+ wxString dest(*this, 0, nCount);
+ if ( dest.length() != nCount ) {
+ wxFAIL_MSG( _T("out of memory in wxString::Left") );
+ }
+ return dest;
}
-#endif // VC++ 1.5
-wxString wxString::substr(size_t nStart, size_t nLen) const
+// get all characters before the first occurence of ch
+// (returns the whole string if ch not found)
+wxString wxString::BeforeFirst(wxUniChar ch) const
{
- // npos means 'take all'
- if ( nLen == npos )
- nLen = 0;
-
- wxASSERT( nStart + nLen <= Len() );
-
- return wxString(c_str() + nStart, nLen == npos ? 0 : nLen);
+ int iPos = Find(ch);
+ if ( iPos == wxNOT_FOUND ) iPos = length();
+ return wxString(*this, 0, iPos);
}
-wxString& wxString::erase(size_t nStart, size_t nLen)
+/// get all characters before the last occurence of ch
+/// (returns empty string if ch not found)
+wxString wxString::BeforeLast(wxUniChar ch) const
{
- wxString strTmp(c_str(), nStart);
- if ( nLen != npos ) {
- wxASSERT( nStart + nLen <= Len() );
-
- strTmp.append(c_str() + nStart + nLen);
- }
+ wxString str;
+ int iPos = Find(ch, true);
+ if ( iPos != wxNOT_FOUND && iPos != 0 )
+ str = wxString(c_str(), iPos);
- *this = strTmp;
- return *this;
+ return str;
}
-wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
+/// get all characters after the first occurence of ch
+/// (returns empty string if ch not found)
+wxString wxString::AfterFirst(wxUniChar ch) const
{
- wxASSERT( nStart + nLen <= wxStrlen(sz) );
-
- wxString strTmp;
- if ( nStart != 0 )
- strTmp.append(c_str(), nStart);
- strTmp += sz;
- strTmp.append(c_str() + nStart + nLen);
-
- *this = strTmp;
- return *this;
-}
+ wxString str;
+ int iPos = Find(ch);
+ if ( iPos != wxNOT_FOUND )
+ str = wx_str() + iPos + 1;
-wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
-{
- return replace(nStart, nLen, wxString(ch, nCount));
+ return str;
}
-wxString& wxString::replace(size_t nStart, size_t nLen,
- const wxString& str, size_t nStart2, size_t nLen2)
+// replace first (or all) occurences of some substring with another one
+size_t wxString::Replace(const wxString& strOld,
+ const wxString& strNew, bool bReplaceAll)
{
- return replace(nStart, nLen, str.substr(nStart2, nLen2));
-}
+ // if we tried to replace an empty string we'd enter an infinite loop below
+ wxCHECK_MSG( !strOld.empty(), 0,
+ _T("wxString::Replace(): invalid parameter") );
-wxString& wxString::replace(size_t nStart, size_t nLen,
- const wxChar* sz, size_t nCount)
-{
- return replace(nStart, nLen, wxString(sz, nCount));
-}
+ size_t uiCount = 0; // count of replacements made
-#endif //std::string compatibility
+ size_t uiOldLen = strOld.length();
+ size_t uiNewLen = strNew.length();
-// ============================================================================
-// ArrayString
-// ============================================================================
+ size_t dwPos = 0;
-// size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
-#define ARRAY_MAXSIZE_INCREMENT 4096
-#ifndef ARRAY_DEFAULT_INITIAL_SIZE // also defined in dynarray.h
- #define ARRAY_DEFAULT_INITIAL_SIZE (16)
-#endif
+ while ( (*this)[dwPos] != wxT('\0') )
+ {
+ //DO NOT USE STRSTR HERE
+ //this string can contain embedded null characters,
+ //so strstr will function incorrectly
+ dwPos = find(strOld, dwPos);
+ if ( dwPos == npos )
+ break; // exit the loop
+ else
+ {
+ //replace this occurance of the old string with the new one
+ replace(dwPos, uiOldLen, strNew, uiNewLen);
-#define STRING(p) ((wxString *)(&(p)))
+ //move up pos past the string that was replaced
+ dwPos += uiNewLen;
-// ctor
-wxArrayString::wxArrayString()
-{
- m_nSize =
- m_nCount = 0;
- m_pItems = (wxChar **) NULL;
+ //increase replace count
+ ++uiCount;
+
+ // stop now?
+ if ( !bReplaceAll )
+ break; // exit the loop
+ }
+ }
+
+ return uiCount;
}
-// copy ctor
-wxArrayString::wxArrayString(const wxArrayString& src)
+bool wxString::IsAscii() const
{
- m_nSize =
- m_nCount = 0;
- m_pItems = (wxChar **) NULL;
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !(*i).IsAscii() )
+ return false;
+ }
- *this = src;
+ return true;
}
-// assignment operator
-wxArrayString& wxArrayString::operator=(const wxArrayString& src)
+bool wxString::IsWord() const
{
- if ( m_nSize > 0 )
- Clear();
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( !wxIsalpha(*i) )
+ return false;
+ }
- if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
- Alloc(src.m_nCount);
+ return true;
+}
- // we can't just copy the pointers here because otherwise we would share
- // the strings with another array
- for ( size_t n = 0; n < src.m_nCount; n++ )
- Add(src[n]);
+bool wxString::IsNumber() const
+{
+ if ( empty() )
+ return true;
- if ( m_nCount != 0 )
- memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(wxChar *));
+ const_iterator i = begin();
- return *this;
-}
+ if ( *i == _T('-') || *i == _T('+') )
+ ++i;
-// grow the array
-void wxArrayString::Grow()
-{
- // only do it if no more place
- if( m_nCount == m_nSize ) {
- if( m_nSize == 0 ) {
- // was empty, alloc some memory
- m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
- m_pItems = new wxChar *[m_nSize];
- }
- else {
- // otherwise when it's called for the first time, nIncrement would be 0
- // and the array would never be expanded
- wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
-
- // add 50% but not too much
- size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
- ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
- if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
- nIncrement = ARRAY_MAXSIZE_INCREMENT;
- 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;
+ for ( ; i != end(); ++i )
+ {
+ if ( !wxIsdigit(*i) )
+ return false;
}
- }
+
+ return true;
}
-void wxArrayString::Free()
+wxString wxString::Strip(stripType w) const
{
- for ( size_t n = 0; n < m_nCount; n++ ) {
- STRING(m_pItems[n])->GetStringData()->Unlock();
- }
+ wxString s = *this;
+ if ( w & leading ) s.Trim(false);
+ if ( w & trailing ) s.Trim(true);
+ return s;
}
-// deletes all the strings from the list
-void wxArrayString::Empty()
+// ---------------------------------------------------------------------------
+// case conversion
+// ---------------------------------------------------------------------------
+
+wxString& wxString::MakeUpper()
{
- Free();
+ for ( iterator it = begin(), en = end(); it != en; ++it )
+ *it = (wxChar)wxToupper(*it);
- m_nCount = 0;
+ return *this;
}
-// as Empty, but also frees memory
-void wxArrayString::Clear()
+wxString& wxString::MakeLower()
{
- Free();
-
- m_nSize =
- m_nCount = 0;
+ for ( iterator it = begin(), en = end(); it != en; ++it )
+ *it = (wxChar)wxTolower(*it);
- wxDELETEA(m_pItems);
+ return *this;
}
-// dtor
-wxArrayString::~wxArrayString()
-{
- Free();
+// ---------------------------------------------------------------------------
+// trimming and padding
+// ---------------------------------------------------------------------------
- wxDELETEA(m_pItems);
-}
+// some compilers (VC++ 6.0 not to name them) return true for a call to
+// isspace('\xEA') in the C locale which seems to be broken to me, but we have
+// to live with this by checking that the character is a 7 bit one - even if
+// this may fail to detect some spaces (I don't know if Unicode doesn't have
+// space-like symbols somewhere except in the first 128 chars), it is arguably
+// still better than trimming away accented letters
+inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
-// pre-allocates memory (frees the previous data!)
-void wxArrayString::Alloc(size_t nSize)
+// trims spaces (in the sense of isspace) from left or right side
+wxString& wxString::Trim(bool bFromRight)
{
- wxASSERT( nSize > 0 );
+ // first check if we're going to modify the string at all
+ if ( !empty() &&
+ (
+ (bFromRight && wxSafeIsspace(GetChar(length() - 1))) ||
+ (!bFromRight && wxSafeIsspace(GetChar(0u)))
+ )
+ )
+ {
+ if ( bFromRight )
+ {
+ // find last non-space character
+ reverse_iterator psz = rbegin();
+ while ( (psz != rend()) && wxSafeIsspace(*psz) )
+ ++psz;
- // only if old buffer was not big enough
- if ( nSize > m_nSize ) {
- Free();
- wxDELETEA(m_pItems);
- m_pItems = new wxChar *[nSize];
- m_nSize = nSize;
- }
+ // truncate at trailing space start
+ erase(psz.base(), end());
+ }
+ else
+ {
+ // find first non-space character
+ iterator psz = begin();
+ while ( (psz != end()) && wxSafeIsspace(*psz) )
+ ++psz;
- m_nCount = 0;
+ // fix up data and length
+ erase(begin(), psz);
+ }
+ }
+
+ return *this;
}
-// minimizes the memory usage by freeing unused memory
-void wxArrayString::Shrink()
+// adds nCount characters chPad to the string from either side
+wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight)
{
- // 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];
+ wxString s(chPad, nCount);
- // copy data to new location
- memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
- delete [] m_pItems;
- m_pItems = pNew;
- }
+ if ( bFromRight )
+ *this += s;
+ else
+ {
+ s += *this;
+ swap(s);
+ }
+
+ return *this;
}
-// searches the array for an item (forward or backwards)
-int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
+// truncate the string
+wxString& wxString::Truncate(size_t uiLen)
{
- 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;
+ if ( uiLen < length() )
+ {
+ erase(begin() + uiLen, end());
}
- }
+ //else: nothing to do, string is already short enough
- return wxNOT_FOUND;
+ return *this;
}
-// add item at the end
-void wxArrayString::Add(const wxString& str)
-{
- wxASSERT( str.GetStringData()->IsValid() );
+// ---------------------------------------------------------------------------
+// finding (return wxNOT_FOUND if not found and index otherwise)
+// ---------------------------------------------------------------------------
- Grow();
+// find a character
+int wxString::Find(wxUniChar ch, bool bFromEnd) const
+{
+ size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
- // the string data must not be deleted!
- str.GetStringData()->Lock();
- m_pItems[m_nCount++] = (wxChar *)str.c_str();
+ return (idx == npos) ? wxNOT_FOUND : (int)idx;
}
-// add item at the given position
-void wxArrayString::Insert(const wxString& str, size_t nIndex)
-{
- wxASSERT( str.GetStringData()->IsValid() );
-
- wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Insert") );
+// ----------------------------------------------------------------------------
+// conversion to numbers
+// ----------------------------------------------------------------------------
- Grow();
+// 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.
- memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
- (m_nCount - nIndex)*sizeof(wxChar *));
+#ifndef __WXWINCE__
+ #define DO_IF_NOT_WINCE(x) x
+#else
+ #define DO_IF_NOT_WINCE(x)
+#endif
- str.GetStringData()->Lock();
- m_pItems[nIndex] = (wxChar *)str.c_str();
+#define WX_STRING_TO_INT_TYPE(out, base, func, T) \
+ wxCHECK_MSG( out, false, _T("NULL output pointer") ); \
+ wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \
+ \
+ DO_IF_NOT_WINCE( errno = 0; ) \
+ \
+ const wxStringCharType *start = wx_str(); \
+ wxStringCharType *end; \
+ T val = func(start, &end, base); \
+ \
+ /* return true only if scan was stopped by the terminating NUL and */ \
+ /* if the string was not empty to start with and no under/overflow */ \
+ /* occurred: */ \
+ if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \
+ return false; \
+ *out = val; \
+ return true
- m_nCount++;
+bool wxString::ToLong(long *pVal, int base) const
+{
+ WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long);
}
-// removes item from array (by index)
-void wxArrayString::Remove(size_t nIndex)
+bool wxString::ToULong(unsigned long *pVal, int base) const
{
- wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Remove") );
+ WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long);
+}
- // release our lock
- Item(nIndex).GetStringData()->Unlock();
+bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const
+{
+ WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t);
+}
- memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
- (m_nCount - nIndex - 1)*sizeof(wxChar *));
- m_nCount--;
+bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const
+{
+ WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t);
}
-// removes item from array (by value)
-void wxArrayString::Remove(const wxChar *sz)
+bool wxString::ToDouble(double *pVal) const
{
- int iIndex = Index(sz);
+ wxCHECK_MSG( pVal, false, _T("NULL output pointer") );
- wxCHECK_RET( iIndex != wxNOT_FOUND,
- _("removing inexistent element in wxArrayString::Remove") );
+ DO_IF_NOT_WINCE( errno = 0; )
- Remove(iIndex);
-}
+ const wxChar *start = c_str();
+ wxChar *end;
+ double val = wxStrtod(start, &end);
-// ----------------------------------------------------------------------------
-// sorting
-// ----------------------------------------------------------------------------
+ // return true only if scan was stopped by the terminating NUL and if the
+ // string was not empty to start with and no under/overflow occurred
+ if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) )
+ return false;
-// 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
-static int 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 = wxStrcmp(strFirst->c_str(), strSecond->c_str());
+ *pVal = val;
- return gs_sortAscending ? result : -result;
- }
+ return true;
}
-// sort array elements using passed comparaison function
-void wxArrayString::Sort(CompareFunction compareFunction)
+// ---------------------------------------------------------------------------
+// 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
{
- START_SORT();
+ va_list argptr;
+ va_start(argptr, format);
- wxASSERT( !gs_compareFunction ); // must have been reset to NULL
- gs_compareFunction = compareFunction;
+ wxString s;
+ s.PrintfV(format, argptr);
- DoSort();
+ va_end(argptr);
- END_SORT();
+ return s;
}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
-void wxArrayString::Sort(bool reverseOrder)
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString wxString::DoFormatUtf8(const char *format, ...)
{
- START_SORT();
+ va_list argptr;
+ va_start(argptr, format);
- wxASSERT( !gs_compareFunction ); // must have been reset to NULL
- gs_sortAscending = !reverseOrder;
+ wxString s;
+ s.PrintfV(format, argptr);
- DoSort();
+ va_end(argptr);
- END_SORT();
+ return s;
}
+#endif // wxUSE_UNICODE_UTF8
-void wxArrayString::DoSort()
+/* static */
+wxString wxString::FormatV(const wxString& format, va_list argptr)
{
- // 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);
+ wxString s;
+ s.PrintfV(format, argptr);
+ return s;
}
-// ============================================================================
-// MBConv
-// ============================================================================
-
-WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
-#if !wxUSE_WCHAR_T
-WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
+#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);
-#if wxUSE_WCHAR_T
+#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
-// ----------------------------------------------------------------------------
-// standard libc conversion
-// ----------------------------------------------------------------------------
+ int iLen = str->PrintfV(format, argptr);
-WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
+ va_end(argptr);
-size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
-{
- return wxMB2WC(buf, psz, n);
+ return iLen;
}
+#endif // !wxUSE_UTF8_LOCALE_ONLY
-size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+#if wxUSE_UNICODE_UTF8
+int wxString::DoPrintfUtf8(const char *format, ...)
{
- return wxWC2MB(buf, psz, n);
-}
+ va_list argptr;
+ va_start(argptr, format);
-// ----------------------------------------------------------------------------
-// standard file conversion
-// ----------------------------------------------------------------------------
+ int iLen = PrintfV(format, argptr);
-WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
+ va_end(argptr);
-// just use the libc conversion for now
-size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
-{
- return wxMB2WC(buf, psz, n);
+ return iLen;
}
+#endif // wxUSE_UNICODE_UTF8
-size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+/*
+ Uses wxVsnprintf and places the result into the this string.
+
+ In ANSI build, wxVsnprintf is effectively vsnprintf but in Unicode build
+ it is vswprintf. Due to a discrepancy between vsnprintf and vswprintf in
+ the ISO C99 (and thus SUSv3) standard the return value for the case of
+ an undersized buffer is inconsistent. For conforming vsnprintf
+ implementations the function must return the number of characters that
+ would have been printed had the buffer been large enough. For conforming
+ vswprintf implementations the function must return a negative number
+ and set errno.
+
+ What vswprintf sets errno to is undefined but Darwin seems to set it to
+ EOVERFLOW. The only expected errno are EILSEQ and EINVAL. Both of
+ those are defined in the standard and backed up by several conformance
+ statements. Note that ENOMEM mentioned in the manual page does not
+ apply to swprintf, only wprintf and fwprintf.
+
+ Official manual page:
+ http://www.opengroup.org/onlinepubs/009695399/functions/swprintf.html
+
+ Some conformance statements (AIX, Solaris):
+ http://www.opengroup.org/csq/view.mhtml?RID=ibm%2FSD1%2F3
+ http://www.theopengroup.org/csq/view.mhtml?norationale=1&noreferences=1&RID=Fujitsu%2FSE2%2F10
+
+ Since EILSEQ and EINVAL are rather common but EOVERFLOW is not and since
+ EILSEQ and EINVAL are specifically defined to mean the error is other than
+ an undersized buffer and no other errno are defined we treat those two
+ as meaning hard errors and everything else gets the old behavior which
+ is to keep looping and increasing buffer size until the function succeeds.
+
+ In practice it's impossible to determine before compilation which behavior
+ may be used. The vswprintf function may have vsnprintf-like behavior or
+ vice-versa. Behavior detected on one release can theoretically change
+ with an updated release. Not to mention that configure testing for it
+ would require the test to be run on the host system, not the build system
+ which makes cross compilation difficult. Therefore, we make no assumptions
+ about behavior and try our best to handle every known case, including the
+ case where wxVsnprintf returns a negative number and fails to set errno.
+
+ There is yet one more non-standard implementation and that is our own.
+ Fortunately, that can be detected at compile-time.
+
+ On top of all that, ISO C99 explicitly defines snprintf to write a null
+ character to the last position of the specified buffer. That would be at
+ at the given buffer size minus 1. It is supposed to do this even if it
+ turns out that the buffer is sized too small.
+
+ Darwin (tested on 10.5) follows the C99 behavior exactly.
+
+ Glibc 2.6 almost follows the C99 behavior except vswprintf never sets
+ errno even when it fails. However, it only seems to ever fail due
+ to an undersized buffer.
+*/
+#if wxUSE_UNICODE_UTF8
+template<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)
{
- return wxWC2MB(buf, psz, n);
-}
+ int size = 1024;
-// ----------------------------------------------------------------------------
-// standard gdk conversion
-// ----------------------------------------------------------------------------
+ for ( ;; )
+ {
+#if wxUSE_UNICODE_UTF8
+ BufferType tmp(str, size + 1);
+ typename BufferType::CharType *buf = tmp;
+#else
+ wxStringBuffer tmp(str, size + 1);
+ wxChar *buf = tmp;
+#endif
+
+ if ( !buf )
+ {
+ // out of memory
-#ifdef __WXGTK12__
-WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
+ // in UTF-8 build, leaving uninitialized junk in the buffer
+ // could result in invalid non-empty UTF-8 string, so just
+ // reset the string to empty on failure:
+ buf[0] = '\0';
+ return -1;
+ }
-#include <gdk/gdk.h>
+ // wxVsnprintf() may modify the original arg pointer, so pass it
+ // only a copy
+ va_list argptrcopy;
+ wxVaCopy(argptrcopy, argptr);
-size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
-{
- if (buf) {
- return gdk_mbstowcs((GdkWChar *)buf, psz, n);
- } else {
- GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
- size_t len = gdk_mbstowcs(nbuf, psz, n);
- delete [] nbuf;
- return len;
- }
+#ifndef __WXWINCE__
+ // Set errno to 0 to make it determinate if wxVsnprintf fails to set it.
+ errno = 0;
+#endif
+ int len = wxVsnprintf(buf, size, format, argptrcopy);
+ va_end(argptrcopy);
+
+ // some implementations of vsnprintf() don't NUL terminate
+ // the string if there is not enough space for it so
+ // always do it manually
+ // FIXME: This really seems to be the wrong and would be an off-by-one
+ // bug except the code above allocates an extra character.
+ buf[size] = _T('\0');
+
+ // vsnprintf() may return either -1 (traditional Unix behaviour) or the
+ // total number of characters which would have been written if the
+ // buffer were large enough (newer standards such as Unix98)
+ if ( len < 0 )
+ {
+ // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or
+ // wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF
+ // is true if *both* of them use our own implementation,
+ // otherwise we can't be sure
+#if wxUSE_WXVSNPRINTF
+ // we know that our own implementation of wxVsnprintf() returns -1
+ // only for a format error - thus there's something wrong with
+ // the user's format string
+ buf[0] = '\0';
+ return -1;
+#else // possibly using system version
+ // assume it only returns error if there is not enough space, but
+ // as we don't know how much we need, double the current size of
+ // the buffer
+#ifndef __WXWINCE__
+ if( (errno == EILSEQ) || (errno == EINVAL) )
+ // If errno was set to one of the two well-known hard errors
+ // then fail immediately to avoid an infinite loop.
+ return -1;
+ else
+#endif // __WXWINCE__
+ // still not enough, as we don't know how much we need, double the
+ // current size of the buffer
+ size *= 2;
+#endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
+ }
+ else if ( len >= size )
+ {
+#if wxUSE_WXVSNPRINTF
+ // we know that our own implementation of wxVsnprintf() returns
+ // size+1 when there's not enough space but that's not the size
+ // of the required buffer!
+ size *= 2; // so we just double the current size of the buffer
+#else
+ // some vsnprintf() implementations NUL-terminate the buffer and
+ // some don't in len == size case, to be safe always add 1
+ // FIXME: I don't quite understand this comment. The vsnprintf
+ // function is specifically defined to return the number of
+ // characters printed not including the null terminator.
+ // So OF COURSE you need to add 1 to get the right buffer size.
+ // The following line is definitely correct, no question.
+ size = len + 1;
+#endif
+ }
+ else // ok, there was enough space
+ {
+ break;
+ }
+ }
+
+ // we could have overshot
+ str.Shrink();
+
+ return str.length();
}
-size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+int wxString::PrintfV(const wxString& format, va_list argptr)
{
- char *mbstr = gdk_wcstombs((GdkWChar *)psz);
- size_t len = mbstr ? strlen(mbstr) : 0;
- if (buf) {
- if (len > n) len = n;
- memcpy(buf, psz, len);
- if (len < n) buf[len] = 0;
- }
- return len;
+#if wxUSE_UNICODE_UTF8
+ #if wxUSE_STL_BASED_WXSTRING
+ typedef wxStringTypeBuffer<char> Utf8Buffer;
+ #else
+ typedef wxStringInternalBuffer Utf8Buffer;
+ #endif
+#endif
+
+#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
}
-#endif // GTK > 1.0
// ----------------------------------------------------------------------------
-// UTF-7
+// misc other operations
// ----------------------------------------------------------------------------
-WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
+// returns true if the string matches the pattern which may contain '*' and
+// '?' metacharacters (as usual, '?' matches any character and '*' any number
+// of them)
+bool wxString::Matches(const wxString& mask) const
+{
+ // I disable this code as it doesn't seem to be faster (in fact, it seems
+ // to be much slower) than the old, hand-written code below and using it
+ // here requires always linking with libregex even if the user code doesn't
+ // use it
+#if 0 // wxUSE_REGEX
+ // first translate the shell-like mask into a regex
+ wxString pattern;
+ pattern.reserve(wxStrlen(pszMask));
+
+ pattern += _T('^');
+ while ( *pszMask )
+ {
+ switch ( *pszMask )
+ {
+ case _T('?'):
+ pattern += _T('.');
+ break;
+
+ case _T('*'):
+ pattern += _T(".*");
+ break;
+
+ case _T('^'):
+ case _T('.'):
+ case _T('$'):
+ case _T('('):
+ case _T(')'):
+ case _T('|'):
+ case _T('+'):
+ case _T('\\'):
+ // these characters are special in a RE, quote them
+ // (however note that we don't quote '[' and ']' to allow
+ // using them for Unix shell like matching)
+ pattern += _T('\\');
+ // fall through
+
+ default:
+ pattern += *pszMask;
+ }
-#if 0
-static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
- "abcdefghijklmnopqrstuvwxyz"
- "0123456789'(),-./:?";
-static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
-static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
- "abcdefghijklmnopqrstuvwxyz"
- "0123456789+/";
+ pszMask++;
+ }
+ pattern += _T('$');
+
+ // and now use it
+ return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
+#else // !wxUSE_REGEX
+ // TODO: this is, of course, awfully inefficient...
+
+ // FIXME-UTF8: implement using iterators, remove #if
+#if wxUSE_UNICODE_UTF8
+ wxWCharBuffer maskBuf = mask.wc_str();
+ wxWCharBuffer txtBuf = wc_str();
+ const wxChar *pszMask = maskBuf.data();
+ const wxChar *pszTxt = txtBuf.data();
+#else
+ const wxChar *pszMask = mask.wx_str();
+ // the char currently being checked
+ const wxChar *pszTxt = wx_str();
#endif
-// TODO: write actual implementations of UTF-7 here
-size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
- const char * WXUNUSED(psz),
- size_t WXUNUSED(n)) const
-{
- return 0;
-}
+ // the last location where '*' matched
+ const wxChar *pszLastStarInText = NULL;
+ const wxChar *pszLastStarInMask = NULL;
-size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
- const wchar_t * WXUNUSED(psz),
- size_t WXUNUSED(n)) const
-{
- return 0;
-}
+match:
+ for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
+ switch ( *pszMask ) {
+ case wxT('?'):
+ if ( *pszTxt == wxT('\0') )
+ return false;
-// ----------------------------------------------------------------------------
-// UTF-8
-// ----------------------------------------------------------------------------
+ // pszTxt and pszMask will be incremented in the loop statement
-WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
-
-static unsigned long utf8_max[]={0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff,0xffffffff};
-
-size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
-{
- size_t len = 0;
-
- while (*psz && ((!buf) || (len<n))) {
- unsigned char cc=*psz++, fc=cc;
- unsigned cnt;
- for (cnt=0; fc&0x80; cnt++) fc<<=1;
- if (!cnt) {
- // plain ASCII char
- if (buf) *buf++=cc;
- len++;
- } else {
- cnt--;
- if (!cnt) {
- // invalid UTF-8 sequence
- return (size_t)-1;
- } else {
- unsigned ocnt=cnt-1;
- unsigned long res=cc&(0x3f>>cnt);
- while (cnt--) {
- cc = *psz++;
- if ((cc&0xC0)!=0x80) {
- // invalid UTF-8 sequence
- return (size_t)-1;
- }
- res=(res<<6)|(cc&0x3f);
- }
- if (res<=utf8_max[ocnt]) {
- // illegal UTF-8 encoding
- return (size_t)-1;
- }
- if (buf) *buf++=res;
- len++;
- }
- }
- }
- if (buf && (len<n)) *buf = 0;
- return len;
-}
-
-size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
-{
- size_t len = 0;
-
- while (*psz && ((!buf) || (len<n))) {
- unsigned long cc=(*psz++)&0x7fffffff;
- unsigned cnt;
- for (cnt=0; cc>utf8_max[cnt]; cnt++);
- if (!cnt) {
- // plain ASCII char
- if (buf) *buf++=cc;
- len++;
- } else {
- len+=cnt+1;
- if (buf) {
- *buf++=(-128>>cnt)|((cc>>(cnt*6))&(0x3f>>cnt));
- while (cnt--)
- *buf++=0x80|((cc>>(cnt*6))&0x3f);
- }
- }
- }
- if (buf && (len<n)) *buf = 0;
- return len;
-}
+ break;
-// ----------------------------------------------------------------------------
-// specified character set
-// ----------------------------------------------------------------------------
+ case wxT('*'):
+ {
+ // remember where we started to be able to backtrack later
+ pszLastStarInText = pszTxt;
+ pszLastStarInMask = pszMask;
-class wxCharacterSet
-{
-public:
- wxArrayString names;
- wchar_t *data;
-};
+ // ignore special chars immediately following this one
+ // (should this be an error?)
+ while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
+ pszMask++;
-#ifndef WX_PRECOMP
- #include "wx/dynarray.h"
- #include "wx/filefn.h"
- #include "wx/textfile.h"
- #include "wx/tokenzr.h"
- #include "wx/utils.h"
-#endif
+ // if there is nothing more, match
+ if ( *pszMask == wxT('\0') )
+ return true;
+
+ // are there any other metacharacters in the mask?
+ size_t uiLenMask;
+ const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
+
+ if ( pEndMask != NULL ) {
+ // we have to match the string between two metachars
+ uiLenMask = pEndMask - pszMask;
+ }
+ else {
+ // we have to match the remainder of the string
+ uiLenMask = wxStrlen(pszMask);
+ }
+
+ wxString strToMatch(pszMask, uiLenMask);
+ const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
+ if ( pMatch == NULL )
+ return false;
+
+ // -1 to compensate "++" in the loop
+ pszTxt = pMatch + uiLenMask - 1;
+ pszMask += uiLenMask - 1;
+ }
+ break;
-WX_DECLARE_OBJARRAY(wxCharacterSet, wxCSArray);
-#include "wx/arrimpl.cpp"
-WX_DEFINE_OBJARRAY(wxCSArray);
-
-static wxCSArray wxCharsets;
-
-static void wxLoadCharacterSets(void)
-{
- static bool already_loaded = FALSE;
-
- if (already_loaded) return;
-
- already_loaded = TRUE;
-#if defined(__UNIX__)
- // search through files in /usr/share/i18n/charmaps
- wxString fname;
- for (fname = ::wxFindFirstFile(_T("/usr/share/i18n/charmaps/*"));
- !fname.IsEmpty();
- fname = ::wxFindNextFile()) {
- wxTextFile cmap(fname);
- if (cmap.Open()) {
- wxCharacterSet *cset = new wxCharacterSet;
- wxString comchar,escchar;
- bool in_charset = FALSE;
-
- // wxFprintf(stderr,_T("Loaded: %s\n"),fname.c_str());
-
- wxString line;
- for (line = cmap.GetFirstLine();
- !cmap.Eof();
- line = cmap.GetNextLine()) {
- // wxFprintf(stderr,_T("line contents: %s\n"),line.c_str());
- wxStringTokenizer token(line);
- wxString cmd = token.GetNextToken();
- if (cmd == comchar) {
- if (token.GetNextToken() == _T("alias"))
- cset->names.Add(token.GetNextToken());
- }
- else if (cmd == _T("<code_set_name>"))
- cset->names.Add(token.GetNextToken());
- else if (cmd == _T("<comment_char>"))
- comchar = token.GetNextToken();
- else if (cmd == _T("<escape_char>"))
- escchar = token.GetNextToken();
- else if (cmd == _T("<mb_cur_min>")) {
- delete cset;
- cset = (wxCharacterSet *) NULL;
- break; // we don't support multibyte charsets ourselves (yet)
- }
- else if (cmd == _T("CHARMAP")) {
- cset->data = (wchar_t *)calloc(256, sizeof(wchar_t));
- in_charset = TRUE;
- }
- else if (cmd == _T("END")) {
- if (token.GetNextToken() == _T("CHARMAP"))
- in_charset = FALSE;
- }
- else if (in_charset) {
- // format: <NUL> /x00 <U0000> NULL (NUL)
- // <A> /x41 <U0041> LATIN CAPITAL LETTER A
- wxString hex = token.GetNextToken();
- // skip whitespace (why doesn't wxStringTokenizer do this?)
- while (wxIsEmpty(hex) && token.HasMoreTokens()) hex = token.GetNextToken();
- wxString uni = token.GetNextToken();
- // skip whitespace again
- while (wxIsEmpty(uni) && token.HasMoreTokens()) uni = token.GetNextToken();
-
- if ((hex.Len() > 2) && (hex.GetChar(0) == escchar) && (hex.GetChar(1) == _T('x')) &&
- (uni.Left(2) == _T("<U"))) {
- hex.MakeUpper(); uni.MakeUpper();
- int pos = ::wxHexToDec(hex.Mid(2,2));
- if (pos>=0) {
- unsigned long uni1 = ::wxHexToDec(uni.Mid(2,2));
- unsigned long uni2 = ::wxHexToDec(uni.Mid(4,2));
- cset->data[pos] = (uni1 << 16) | uni2;
- // wxFprintf(stderr,_T("char %02x mapped to %04x (%c)\n"),pos,cset->data[pos],cset->data[pos]);
- }
- }
- }
- }
- if (cset) {
- cset->names.Shrink();
- wxCharsets.Add(cset);
- }
+ default:
+ if ( *pszMask != *pszTxt )
+ return false;
+ break;
}
}
-#endif
- wxCharsets.Shrink();
-}
-static wxCharacterSet *wxFindCharacterSet(const wxChar *charset)
-{
- if (!charset) return (wxCharacterSet *)NULL;
- wxLoadCharacterSets();
- for (size_t n=0; n<wxCharsets.GetCount(); n++)
- if (wxCharsets[n].names.Index(charset) != wxNOT_FOUND)
- return &(wxCharsets[n]);
- return (wxCharacterSet *)NULL;
-}
+ // match only if nothing left
+ if ( *pszTxt == wxT('\0') )
+ return true;
-WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
+ // if we failed to match, backtrack if we can
+ if ( pszLastStarInText ) {
+ pszTxt = pszLastStarInText + 1;
+ pszMask = pszLastStarInMask;
-wxCSConv::wxCSConv(const wxChar *charset)
-{
- m_name = (wxChar *) NULL;
- m_cset = (wxCharacterSet *) NULL;
- m_deferred = TRUE;
- SetName(charset);
-}
+ pszLastStarInText = NULL;
-wxCSConv::~wxCSConv()
-{
- if (m_name) free(m_name);
-}
+ // don't bother resetting pszLastStarInMask, it's unnecessary
-void wxCSConv::SetName(const wxChar *charset)
-{
- if (charset) {
-#ifdef __UNIX__
- // first, convert the character set name to standard form
- wxString codeset;
- if (wxString(charset,3).CmpNoCase(_T("ISO")) == 0) {
- // make sure it's represented in the standard form: ISO_8859-1
- codeset = _T("ISO_");
- charset += 3;
- if ((*charset == _T('-')) || (*charset == _T('_'))) charset++;
- if (wxStrlen(charset)>4) {
- if (wxString(charset,4) == _T("8859")) {
- codeset << _T("8859-");
- if (*charset == _T('-')) charset++;
- }
- }
- }
- codeset << charset;
- codeset.MakeUpper();
- m_name = wxStrdup(codeset.c_str());
- m_deferred = TRUE;
-#endif
+ goto match;
}
-}
-void wxCSConv::LoadNow()
-{
-// wxPrintf(_T("Conversion request\n"));
- if (m_deferred) {
- if (!m_name) {
-#ifdef __UNIX__
- wxChar *lang = wxGetenv(_T("LANG"));
- wxChar *dot = lang ? wxStrchr(lang, _T('.')) : (wxChar *)NULL;
- if (dot) SetName(dot+1);
-#endif
- }
- m_cset = wxFindCharacterSet(m_name);
- m_deferred = FALSE;
- }
+ return false;
+#endif // wxUSE_REGEX/!wxUSE_REGEX
}
-size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+// Count the number of chars
+int wxString::Freq(wxUniChar ch) const
{
- ((wxCSConv *)this)->LoadNow(); // discard constness
- if (buf) {
- if (m_cset) {
- for (size_t c=0; c<n; c++)
- buf[c] = m_cset->data[(unsigned char)(psz[c])];
- } else {
- // latin-1 (direct)
- for (size_t c=0; c<n; c++)
- buf[c] = (unsigned char)(psz[c]);
- }
- return n;
- }
- return strlen(psz);
-}
-
-size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
-{
- ((wxCSConv *)this)->LoadNow(); // discard constness
- if (buf) {
- if (m_cset) {
- for (size_t c=0; c<n; c++) {
- size_t n;
- for (n=0; (n<256) && (m_cset->data[n] != psz[c]); n++);
- buf[c] = (n>0xff) ? '?' : n;
- }
- } else {
- // latin-1 (direct)
- for (size_t c=0; c<n; c++)
- buf[c] = (psz[c]>0xff) ? '?' : psz[c];
+ int count = 0;
+ for ( const_iterator i = begin(); i != end(); ++i )
+ {
+ if ( *i == ch )
+ count ++;
}
- return n;
- }
- return wcslen(psz);
+ return count;
}
-#endif//wxUSE_WCHAR_T
+// 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(); }
-#if wxUSE_WCHAR_T
-const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
- {
- if (psz) {
- size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
- wxWCharBuffer buf(nLen);
- MB2WC(WCSTRINGCAST buf, psz, nLen);
- return buf;
- } else return wxWCharBuffer((wchar_t *) NULL);
- }
+// ----------------------------------------------------------------------------
+// wxUTF8StringBuffer
+// ----------------------------------------------------------------------------
-const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
- {
- if (psz) {
- size_t nLen = WC2MB((char *) NULL, psz, 0);
- wxCharBuffer buf(nLen);
- WC2MB(MBSTRINGCAST buf, psz, nLen);
- return buf;
- } else return wxCharBuffer((char *) NULL);
- }
+#if wxUSE_UNICODE_WCHAR
+wxUTF8StringBuffer::~wxUTF8StringBuffer()
+{
+ wxMBConvStrictUTF8 conv;
+ size_t wlen = conv.ToWChar(NULL, 0, m_buf);
+ wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" );
+
+ wxStringInternalBuffer wbuf(m_str, wlen);
+ conv.ToWChar(wbuf, wlen, m_buf);
+}
+
+wxUTF8StringBufferLength::~wxUTF8StringBufferLength()
+{
+ wxCHECK_RET(m_lenSet, "length not set");
-#endif//wxUSE_WCHAR_T
+ wxMBConvStrictUTF8 conv;
+ size_t wlen = conv.ToWChar(NULL, 0, m_buf, m_len);
+ wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" );
+ wxStringInternalBufferLength wbuf(m_str, wlen);
+ conv.ToWChar(wbuf, wlen, m_buf, m_len);
+ wbuf.SetLength(wlen);
+}
+#endif // wxUSE_UNICODE_WCHAR
projects / wxWidgets.git / blobdiff