liboce-foundation-dev 0.18.2-2build1 / usr / include / oce / NCollection_UtfString.lxx

// Created on: 2013-01-28
// Created by: Kirill GAVRILOV
// Copyright (c) 2013-2014 OPEN CASCADE SAS
//
// This file is part of Open CASCADE Technology software library.
//
// This library is free software; you can redistribute it and/or modify it under
// the terms of the GNU Lesser General Public License version 2.1 as published
// by the Free Software Foundation, with special exception defined in the file
// OCCT_LGPL_EXCEPTION.txt. Consult the file LICENSE_LGPL_21.txt included in OCCT
// distribution for complete text of the license and disclaimer of any warranty.
//
// Alternatively, this file may be used under the terms of Open CASCADE
// commercial license or contractual agreement.

// =======================================================================
// function : strGetAdvance
// purpose  : Compute advance for specified string.
// =======================================================================
template<typename Type> template<typename TypeFrom> inline
void NCollection_UtfString<Type>::strGetAdvance (const TypeFrom*        theStringUtf,
                                                   const Standard_Integer theLengthMax,
                                                   Standard_Integer&      theSizeBytes,
                                                   Standard_Integer&      theLength)
{
  theSizeBytes = 0;
  theLength    = 0;
  NCollection_UtfIterator<TypeFrom> anIter (theStringUtf);
  const Standard_Integer aLengthMax = (theLengthMax > 0) ? theLengthMax : IntegerLast();
  switch (sizeof(Type))
  {
    case sizeof(Standard_Utf8Char):
    {
      for (; *anIter != 0 && anIter.Index() < aLengthMax; ++anIter)
      {
        theSizeBytes += anIter.AdvanceBytesUtf8();
      }
      theLength = anIter.Index();
      return;
    }
    case sizeof(Standard_Utf16Char):
    {
      for (; *anIter != 0 && anIter.Index() < aLengthMax; ++anIter)
      {
        theSizeBytes += anIter.AdvanceBytesUtf16();
      }
      theLength = anIter.Index();
      return;
    }
    case sizeof(Standard_Utf32Char):
    {
      for (; *anIter != 0 && anIter.Index() < aLengthMax; ++anIter)
      {
        theSizeBytes += anIter.AdvanceBytesUtf32();
      }
      theLength = anIter.Index();
      return;
    }
    default: return;
  }
}

// =======================================================================
// function : GetChar
// purpose  :
// =======================================================================
template<typename Type>
Standard_Utf32Char NCollection_UtfString<Type>::GetChar (const Standard_Integer theCharIndex) const
{
  //Standard_ASSERT_SKIP (theCharIndex < myLength, "Out of range");
  NCollection_UtfIterator<Type> anIter (myString);
  for (; *anIter != 0; ++anIter)
  {
    if (anIter.Index() == theCharIndex)
    {
      return *anIter;
    }
  }
  return 0;
}

// =======================================================================
// function : GetCharBuffer
// purpose  :
// =======================================================================
template<typename Type>
const Type* NCollection_UtfString<Type>::GetCharBuffer (const Standard_Integer theCharIndex) const
{
  //Standard_ASSERT_SKIP(theCharIndex < myLength);
  NCollection_UtfIterator<Type> anIter (myString);
  for (; *anIter != 0; ++anIter)
  {
    if (anIter.Index() == theCharIndex)
    {
      return anIter.BufferHere();
    }
  }
  return NULL;
}

// =======================================================================
// function : Clear
// purpose  :
// =======================================================================
template<typename Type> inline
void NCollection_UtfString<Type>::Clear()
{
  strFree (myString);
  mySize   = 0;
  myLength = 0;
  myString = strAlloc (mySize);
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString()
: myString (strAlloc(0)),
  mySize   (0),
  myLength (0)
{
  //
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString (const NCollection_UtfString& theCopy)
: myString (strAlloc (theCopy.mySize)),
  mySize   (theCopy.mySize),
  myLength (theCopy.myLength)
{
  strCopy ((Standard_Byte* )myString, (const Standard_Byte* )theCopy.myString, mySize);
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString (const char*            theCopyUtf8,
                                                    const Standard_Integer theLength)
: myString (NULL),
  mySize   (0),
  myLength (0)
{
  FromUnicode (theCopyUtf8, theLength);
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString (const Standard_Utf16Char* theCopyUtf16,
                                                    const Standard_Integer    theLength)
: myString (NULL),
  mySize   (0),
  myLength (0)
{
  FromUnicode (theCopyUtf16, theLength);
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString (const Standard_Utf32Char* theCopyUtf32,
                                                    const Standard_Integer     theLength)
: myString (NULL),
  mySize   (0),
  myLength (0)
{
  FromUnicode (theCopyUtf32, theLength);
}

// =======================================================================
// function : NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::NCollection_UtfString (const Standard_WideChar* theCopyUtfWide,
                                                    const Standard_Integer   theLength)
: myString (NULL),
  mySize   (0),
  myLength (0)
{
  FromUnicode (theCopyUtfWide, theLength);
}

// =======================================================================
// function : ~NCollection_UtfString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>::~NCollection_UtfString()
{
  strFree (myString);
}

// =======================================================================
// function : operator=
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Type>& NCollection_UtfString<Type>::operator= (const NCollection_UtfString<Type>& theOther)
{
  if (this == &theOther)
  {
    return (*this);
  }

  strFree (myString);
  mySize   = theOther.mySize;
  myLength = theOther.myLength;
  myString = strAlloc (mySize);
  strCopy ((Standard_Byte* )myString, (const Standard_Byte* )theOther.myString, mySize);
  return (*this);
}

// =======================================================================
// function : FromUnicode
// purpose  :
// =======================================================================
template<typename Type> template<typename TypeFrom>
void NCollection_UtfString<Type>::FromUnicode (const TypeFrom*        theStringUtf,
                                               const Standard_Integer theLength)
{
  Type* anOldBuffer = myString; // necessary in case of self-copying
  NCollection_UtfIterator<TypeFrom> anIterRead (theStringUtf);
  if (*anIterRead == 0)
  {
    // special case
    Clear();
    return;
  }

  switch (sizeof(TypeFrom)) // use switch() rather than if() to shut up msvc compiler
  {
    case sizeof(Type):
    {
      if (theLength > 0)
      {
        // optimized copy
        for(; *anIterRead != 0 && anIterRead.Index() < theLength; ++anIterRead) {}

        mySize   = Standard_Integer((Standard_Byte* )anIterRead.BufferHere() - (Standard_Byte* )theStringUtf);
        myLength = anIterRead.Index();
        myString = strAlloc (mySize);
        strCopy ((Standard_Byte* )myString, (const Standard_Byte* )theStringUtf, mySize);
        strFree (anOldBuffer);
        return;
      }
    }
    default: break;
  }

  strGetAdvance (theStringUtf, theLength, mySize, myLength);
  myString = strAlloc (mySize);
  // reset iterator
  anIterRead.Init (theStringUtf);
  Type* anIterWrite = myString;
  for (; *anIterRead != 0 && anIterRead.Index() < myLength; ++anIterRead)
  {
    anIterWrite = anIterRead.GetUtf (anIterWrite);
  }
  strFree (anOldBuffer);
}

// =======================================================================
// function : FromLocale
// purpose  :
// =======================================================================
template<typename Type> inline
void NCollection_UtfString<Type>::FromLocale (const char*            theString,
                                              const Standard_Integer theLength)
{
  #ifdef _MSC_VER 
  // use WinAPI
  int aWideSize = MultiByteToWideChar (CP_ACP, MB_PRECOMPOSED, theString, -1, NULL, 0);
  if (aWideSize <= 0)
  {
    Clear();
    return;
  }
  wchar_t* aWideBuffer = new wchar_t[aWideSize + 1];
  MultiByteToWideChar (CP_ACP, MB_PRECOMPOSED, theString, -1, aWideBuffer, aWideSize);
  aWideBuffer[aWideSize] = L'\0';
  FromUnicode (aWideBuffer, theLength);
  delete[] aWideBuffer;
#else
  // this is size in bytes but should probably be enough to store string in wide chars
  // notice that these functions are sensitive to locale set by application!
  int aMbLen = mblen (theString, MB_CUR_MAX);
  if (aMbLen <= 0)
  {
    Clear();
    return;
  }
  wchar_t* aWideBuffer = new wchar_t[aMbLen + 1];
  mbstowcs (aWideBuffer, theString, aMbLen);
  aWideBuffer[aMbLen] = L'\0';
  FromUnicode (aWideBuffer, theLength);
  delete[] aWideBuffer;
#endif
}

// =======================================================================
// function : ToLocale
// purpose  :
// =======================================================================
template<typename Type> inline
bool NCollection_UtfString<Type>::ToLocale (char*                  theBuffer,
                                            const Standard_Integer theSizeBytes) const
{
  NCollection_UtfString<wchar_t> aWideCopy (myString, myLength);
#ifdef _MSC_VER 
  int aMbBytes = WideCharToMultiByte (CP_ACP, 0, aWideCopy.ToCString(), -1, theBuffer, theSizeBytes, NULL, NULL);
#else
  std::size_t aMbBytes = std::wcstombs (theBuffer, aWideCopy.ToCString(), theSizeBytes);
#endif
  if (aMbBytes <= 0)
  {
    *theBuffer = '\0';
    return false;
  }
  return true;
}

// =======================================================================
// function : operator=
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Type>& NCollection_UtfString<Type>::operator= (const char* theStringUtf8)
{
  FromUnicode (theStringUtf8);
  return (*this);
}

// =======================================================================
// function : operator=
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Type>& NCollection_UtfString<Type>::operator= (const Standard_WideChar* theStringUtfWide)
{
  FromUnicode (theStringUtfWide);
  return (*this);
}

// =======================================================================
// function : IsEqual
// purpose  :
// =======================================================================
template<typename Type> inline
bool NCollection_UtfString<Type>::IsEqual (const NCollection_UtfString& theCompare) const
{
  return this == &theCompare
      || strAreEqual (myString, mySize, theCompare.myString, theCompare.mySize);
}

// =======================================================================
// function : operator!=
// purpose  :
// =======================================================================
template<typename Type> inline
bool NCollection_UtfString<Type>::operator!= (const NCollection_UtfString& theCompare) const
{
  return (!NCollection_UtfString::operator== (theCompare));
}

// =======================================================================
// function : operator+=
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type>& NCollection_UtfString<Type>::operator+= (const NCollection_UtfString<Type>& theAppend)
{
  if (theAppend.IsEmpty())
  {
    return (*this);
  }

  // create new string
  Standard_Integer aSize = mySize + theAppend.mySize;
  Type* aString = strAlloc (aSize);
  strCopy ((Standard_Byte* )aString,          (const Standard_Byte* )myString,           mySize);
  strCopy ((Standard_Byte* )aString + mySize, (const Standard_Byte* )theAppend.myString, theAppend.mySize);

  strFree (myString);
  mySize   = aSize;
  myString = aString;
  myLength += theAppend.myLength;
  return (*this);
}

// =======================================================================
// function : SubString
// purpose  :
// =======================================================================
template<typename Type> inline
NCollection_UtfString<Type> NCollection_UtfString<Type>::SubString (const Standard_Integer theStart,
                                                                    const Standard_Integer theEnd) const
{
  if (theStart >= theEnd)
  {
    return NCollection_UtfString<Type>();
  }
  for (NCollection_UtfIterator<Type> anIter(myString); *anIter != 0; ++anIter)
  {
    if (anIter.Index() >= theStart)
    {
      return NCollection_UtfString<Type> (anIter.BufferHere(), theEnd - theStart);
    }
  }
  return NCollection_UtfString<Type>();
}

// =======================================================================
// function : ToUtf8
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Standard_Utf8Char> NCollection_UtfString<Type>::ToUtf8() const
{
  NCollection_UtfString<Standard_Utf8Char> aCopy;
  aCopy.FromUnicode (myString);
  return aCopy;
}

// =======================================================================
// function : ToUtf16
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Standard_Utf16Char> NCollection_UtfString<Type>::ToUtf16() const
{
  NCollection_UtfString<Standard_Utf16Char> aCopy;
  aCopy.FromUnicode (myString);
  return aCopy;
}

// =======================================================================
// function : ToUtf32
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Standard_Utf32Char> NCollection_UtfString<Type>::ToUtf32() const
{
  NCollection_UtfString<Standard_Utf32Char> aCopy;
  aCopy.FromUnicode (myString);
  return aCopy;
}

// =======================================================================
// function : ToUtfWide
// purpose  :
// =======================================================================
template<typename Type> inline
const NCollection_UtfString<Standard_WideChar> NCollection_UtfString<Type>::ToUtfWide() const
{
  NCollection_UtfString<Standard_WideChar> aCopy;
  aCopy.FromUnicode (myString);
  return aCopy;
}