string.cpp

来自「A*算法 A*算法 A*算法 A*算法A*算法A*算法」· C++ 代码 · 共 1,635 行 · 第 1/3 页

/////////////////////////////////////////////////////////////////////////////
// Name:        string.cpp
// Purpose:     wxString class
// Author:      Vadim Zeitlin, Ryan Norton
// Modified by:
// Created:     29/01/98
// RCS-ID:      $Id: string.cpp,v 1.258.2.1 2005/11/30 13:30:08 VZ Exp $
// Copyright:   (c) 1998 Vadim Zeitlin <[e-mail]>
//              (c) 2004 Ryan Norton <[e-mail]>
// Licence:     wxWindows licence
/////////////////////////////////////////////////////////////////////////////

/*
 * About ref counting:
 *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
 *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
 *  3) Unlock() decrements nRefs and frees memory if it goes to 0
 */

// ===========================================================================
// headers, declarations, constants
// ===========================================================================

// For compilers that support precompilation, includes "wx.h".
#include "wx/wxprec.h"

#ifdef __BORLANDC__
  #pragma hdrstop
#endif

#ifndef WX_PRECOMP
  #include "wx/defs.h"
  #include "wx/string.h"
  #include "wx/intl.h"
  #include "wx/thread.h"
#endif

#include <ctype.h>
#include <string.h>
#include <stdlib.h>

#ifdef __SALFORDC__
  #include <clib.h>
#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
// ---------------------------------------------------------------------------

#if !wxUSE_STL
  //According to STL _must_ be a -1 size_t
  const size_t wxStringBase::npos = (size_t) -1;
#endif

// ----------------------------------------------------------------------------
// static data
// ----------------------------------------------------------------------------

#if wxUSE_STL

extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T("");

#else

// 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}, wxT('\0') };

// empty C style string: points to 'string data' byte of g_strEmpty
extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy;

#endif

// ----------------------------------------------------------------------------
// global functions
// ----------------------------------------------------------------------------

#if wxUSE_STD_IOSTREAM

// 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!

#include <iostream>

wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(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;
      }

      str += ch;
      if ( --w == 1 )
        break;
    }
  }

  is.isfx();
  if ( str.length() == 0 )
    is.setstate(ios::failbit);
#endif
  return is;
}

wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
{
#ifdef __BORLANDC__
    os << str.mb_str();
#else
    os << str.c_str();
#endif
    return os;
}

#endif // wxUSE_STD_IOSTREAM

// ----------------------------------------------------------------------------
// private classes
// ----------------------------------------------------------------------------

// this small class is used to gather statistics for performance tuning
//#define WXSTRING_STATISTICS
#ifdef  WXSTRING_STATISTICS
  class Averager
  {
  public:
    Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
   ~Averager()
   { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }

    void Add(size_t n) { m_nTotal += n; m_nCount++; }

  private:
    size_t m_nCount, m_nTotal;
    const wxChar *m_sz;
  } g_averageLength("allocation size"),
    g_averageSummandLength("summand length"),
    g_averageConcatHit("hit probability in concat"),
    g_averageInitialLength("initial string length");

  #define STATISTICS_ADD(av, val) g_average##av.Add(val)
#else
  #define STATISTICS_ADD(av, val)
#endif // WXSTRING_STATISTICS

#if !wxUSE_STL

// ===========================================================================
// wxStringData class deallocation
// ===========================================================================

#if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
#  pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
void wxStringData::Free()
{
    free(this);
}
#endif

// ===========================================================================
// wxStringBase
// ===========================================================================

// takes nLength elements of psz starting at nPos
void wxStringBase::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
{
  Init();

  // if the length is not given, assume the string to be NUL terminated
  if ( nLength == npos ) {
    wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );

    nLength = wxStrlen(psz + nPos);
  }

  STATISTICS_ADD(InitialLength, nLength);

  if ( nLength > 0 ) {
    // trailing '\0' is written in AllocBuffer()
    if ( !AllocBuffer(nLength) ) {
      wxFAIL_MSG( _T("out of memory in wxStringBase::InitWith") );
      return;
    }
    wxTmemcpy(m_pchData, psz + nPos, nLength);
  }
}

// poor man's iterators are "void *" pointers
wxStringBase::wxStringBase(const void *pStart, const void *pEnd)
{
  InitWith((const wxChar *)pStart, 0,
           (const wxChar *)pEnd - (const wxChar *)pStart);
}

wxStringBase::wxStringBase(size_type n, wxChar ch)
{
  Init();
  append(n, ch);
}

// ---------------------------------------------------------------------------
// memory allocation
// ---------------------------------------------------------------------------

// allocates memory needed to store a C string of length nLen
bool wxStringBase::AllocBuffer(size_t nLen)
{
  // allocating 0 sized buffer doesn't make sense, all empty strings should
  // reuse g_strEmpty
  wxASSERT( nLen >  0 );

  // make sure that we don't overflow
  wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
                   (sizeof(wxStringData) + EXTRA_ALLOC + 1) );

  STATISTICS_ADD(Length, nLen);

  // allocate memory:
  // 1) one extra character for '\0' termination
  // 2) sizeof(wxStringData) for housekeeping info
  wxStringData* pData = (wxStringData*)
    malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));

  if ( pData == NULL ) {
    // allocation failures are handled by the caller
    return false;
  }

  pData->nRefs        = 1;
  pData->nDataLength  = nLen;
  pData->nAllocLength = nLen + EXTRA_ALLOC;
  m_pchData           = pData->data();  // data starts after wxStringData
  m_pchData[nLen]     = wxT('\0');
  return true;
}

// must be called before changing this string
bool wxStringBase::CopyBeforeWrite()
{
  wxStringData* pData = GetStringData();

  if ( pData->IsShared() ) {
    pData->Unlock();                // memory not freed because shared
    size_t nLen = pData->nDataLength;
    if ( !AllocBuffer(nLen) ) {
      // allocation failures are handled by the caller
      return false;
    }
    wxTmemcpy(m_pchData, pData->data(), nLen);
  }

  wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner

  return true;
}

// must be called before replacing contents of this string
bool wxStringBase::AllocBeforeWrite(size_t nLen)
{
  wxASSERT( nLen != 0 );  // doesn't make any sense

  // must not share string and must have enough space
  wxStringData* pData = GetStringData();
  if ( pData->IsShared() || pData->IsEmpty() ) {
    // can't work with old buffer, get new one
    pData->Unlock();
    if ( !AllocBuffer(nLen) ) {
      // allocation failures are handled by the caller
      return false;
    }
  }
  else {
    if ( nLen > pData->nAllocLength ) {
      // realloc the buffer instead of calling malloc() again, this is more
      // efficient
      STATISTICS_ADD(Length, nLen);

      nLen += EXTRA_ALLOC;

      pData = (wxStringData*)
          realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));

      if ( pData == NULL ) {
        // allocation failures are handled by the caller
        // keep previous data since reallocation failed
        return false;
      }

      pData->nAllocLength = nLen;
      m_pchData = pData->data();
    }
  }

  wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner

  // it doesn't really matter what the string length is as it's going to be
  // overwritten later but, for extra safety, set it to 0 for now as we may
  // have some junk in m_pchData
  GetStringData()->nDataLength = 0;

  return true;
}

wxStringBase& wxStringBase::append(size_t n, wxChar ch)
{
    size_type len = length();

    if ( !Alloc(len + n) || !CopyBeforeWrite() ) {
      wxFAIL_MSG( _T("out of memory in wxStringBase::append") );
    }
    GetStringData()->nDataLength = len + n;
    m_pchData[len + n] = '\0';
    for ( size_t i = 0; i < n; ++i )
        m_pchData[len + i] = ch;
    return *this;
}

void wxStringBase::resize(size_t nSize, wxChar ch)
{
    size_t len = length();

    if ( nSize < len )
    {
        erase(begin() + nSize, end());
    }
    else if ( nSize > len )
    {
        append(nSize - len, ch);
    }
    //else: we have exactly the specified length, nothing to do
}

// allocate enough memory for nLen characters
bool wxStringBase::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));

      if ( pData == NULL ) {
        // allocation failure handled by caller
        return false;
      }

      pData->nRefs = 1;
      pData->nDataLength = 0;
      pData->nAllocLength = nLen;
      m_pchData = pData->data();  // data starts after wxStringData
      m_pchData[0u] = wxT('\0');
    }
    else if ( pData->IsShared() ) {
      pData->Unlock();                // memory not freed because shared
      size_t nOldLen = pData->nDataLength;
      if ( !AllocBuffer(nLen) ) {
        // allocation failure handled by caller
        return false;
      }
      // +1 to copy the terminator, too
      memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxChar));
      GetStringData()->nDataLength = nOldLen;
    }
    else {
      nLen += EXTRA_ALLOC;

      pData = (wxStringData *)
        realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));

      if ( pData == NULL ) {
        // allocation failure handled by caller
        // keep previous data since reallocation failed
        return false;
      }

      // it's not important if the pointer changed or not (the check for this
      // is not faster than assigning to m_pchData in all cases)
      pData->nAllocLength = nLen;
      m_pchData = pData->data();
    }
  }
  //else: we've already got enough
  return true;
}

wxStringBase::iterator wxStringBase::begin()
{
    if (length() > 0)
        CopyBeforeWrite();
    return m_pchData;
}

wxStringBase::iterator wxStringBase::end()
{
    if (length() > 0)
        CopyBeforeWrite();
    return m_pchData + length();
}

wxStringBase::iterator wxStringBase::erase(iterator it)
{
    size_type idx = it - begin();
    erase(idx, 1);
    return begin() + idx;
}

wxStringBase& wxStringBase::erase(size_t nStart, size_t nLen)
{
    wxASSERT(nStart <= length());
    size_t strLen = length() - nStart;
    // delete nLen or up to the end of the string characters
    nLen = strLen < nLen ? strLen : nLen;
    wxString strTmp(c_str(), nStart);
    strTmp.append(c_str() + nStart + nLen, length() - nStart - nLen);

    swap(strTmp);
    return *this;
}

wxStringBase& wxStringBase::insert(size_t nPos, const wxChar *sz, size_t n)
{
    wxASSERT( nPos <= length() );

    if ( n == npos ) n = wxStrlen(sz);
    if ( n == 0 ) return *this;

    if ( !Alloc(length() + n) || !CopyBeforeWrite() ) {
        wxFAIL_MSG( _T("out of memory in wxStringBase::insert") );
    }

    memmove(m_pchData + nPos + n, m_pchData + nPos,
            (length() - nPos) * sizeof(wxChar));
    memcpy(m_pchData + nPos, sz, n * sizeof(wxChar));
    GetStringData()->nDataLength = length() + n;
    m_pchData[length()] = '\0';

    return *this;
}

void wxStringBase::swap(wxStringBase& str)
{
    wxChar* tmp = str.m_pchData;
    str.m_pchData = m_pchData;
    m_pchData = tmp;
}

size_t wxStringBase::find(const wxStringBase& str, size_t nStart) const
{
    wxASSERT( str.GetStringData()->IsValid() );
    wxASSERT( nStart <= length() );

    //anchor
    const wxChar* p = (const wxChar*)wxTmemchr(c_str() + nStart,
                                               str.c_str()[0],
                                               length() - nStart);

    if(!p)
        return npos;

    while(p - c_str() + str.length() <= length() &&
          wxTmemcmp(p, str.c_str(), str.length()) )
    {
        //Previosly passed as the first argument to wxTmemchr,
        //but C/C++ standard does not specify evaluation order
        //of arguments to functions -
        //http://embedded.com/showArticle.jhtml?articleID=9900607
        ++p;

        //anchor again
        p = (const wxChar*)wxTmemchr(p,
                                     str.c_str()[0],
                                     length() - (p - c_str()));

        if(!p)
            return npos;
    }

    return (p - c_str() + str.length() <= length()) ? p - c_str() : npos;
}

size_t wxStringBase::find(const wxChar* sz, size_t nStart, size_t n) const
{
    return find(wxStringBase(sz, n), nStart);
}

size_t wxStringBase::find(wxChar ch, size_t nStart) const
{
    wxASSERT( nStart <= length() );

    const wxChar *p = (const wxChar*)wxTmemchr(c_str() + nStart, ch, length() - nStart);

    return p == NULL ? npos : p - c_str();
}

size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const
{
    wxASSERT( str.GetStringData()->IsValid() );
    wxASSERT( nStart == npos || nStart <= length() );

    if ( length() >= str.length() )
    {
        // avoids a corner case later
        if ( length() == 0 && str.length() == 0 )
            return 0;

        // "top" is the point where search starts from
        size_t top = length() - str.length();

        if ( nStart == npos )
            nStart = length() - 1;
        if ( nStart < top )
            top = nStart;

        const wxChar *cursor = c_str() + top;
        do
        {
            if ( wxTmemcmp(cursor, str.c_str(),
                        str.length()) == 0 )
            {
                return cursor - c_str();