String Conversions

This article describes a set of classes that are used to perform string conversions between the Unicode and ANSI character sets. Why is it needed? There are several reasons:

1. If you are developing for COM, then all strings passed via the COM interfaces are wide strings (unless they are part of the VARIANT structure).
   

2. If you are developing communication applications then (most probably) you need to send/receive data to/from the communication channel in ANSI character set (and your application is either Unicode or ANSI).
   
3. If you want to have a portable application and you need to use either ANSI or Unicode strings for any reason.
   
4. You are developing a DLL, service or applications that are not MFC based and you cannot use CString.

In addition, one should consider the following:

1. When passing wide strings via the COM interfaces, memory for these strings must be allocated with CoTaskMemAlloc() and released with CoTaskMemFree().
   

2. In any other situation, it is allowed to use new and delete.

Besides ANSI and Unicode strings, there is also a BSTR string. Microsoft class _bstr_t encapsulates the BSTR strings.

Summary

The small class library presented in this article contains the following classes:

where

All classes are implemented inline so it is enough to include the file in your project and use the classes.

Short Description

All classes are designed to create a new string based on the supplied one. The buffer for the created string is automatically deleted in class destructor (unless auto delete flag is set to FALSE in the call to the constructor). This is important in COM. According to COM memory management rules, all OUT arguments must be allocated by the callee and deallocated by the caller using the standard COM memory allocator (CoTaskMemAlloc/CoTaskMemFree).

All other classes have exactly the same interface. The only difference is the type casting operator and the type of string argument in constructor.

For example, the _totchar class declaration looks like following:

<FONT
    COLOR="BLUE">class</FONT> _totchar { <BR><FONT
    COLOR="BLUE">private</FONT>: <BR>    BOOL m_bAutoDelete; <BR>    LPTSTR m_tszBuffer; <BR><FONT
    COLOR="BLUE">public</FONT>: <BR>    _totchar(LPCSTR szText, BOOL bAutoDelete = TRUE);<BR>    _totchar(LPCWSTR wszText, BOOL bAutoDelete = TRUE);<BR>    ~_totchar();<BR><FONT
    COLOR="BLUE">    operator</FONT> LPTSTR();<BR><FONT
    COLOR="BLUE">    operator</FONT> LPCTSTR();<BR>}; 

The first constructor takes a const char * argument while the second constructor takes a const wchar_t * argument. Both constructors convert the string argument to LPTSTR using "new" (the _cotchar class has exactly the same functionality but uses CoTaskMemAlloc for memory allocation).

By default, the internal auto delete flag is set to TRUE. This means that the memory for internally created LPTSTR string will be deallocated in destructor. By setting auto delete to FALSE, memory allocated for internal LPTSTR is left intact and you have to delete it manually. This is most often used with class _cowchar because it creates a UNICODE string that can be passed via COM.

In order to access the internal LPTSTR string, class supplies 2 type casting operators.

Examples

Following are several examples that use some of the classes:

Example 1:

You are developing a COM server and one interface method receives a string as an argument. This is the code to convert it to a portable string:

HRESULT STDMETHODCALLTYPE <BR>IOPCGroupStateMgt::SetName( /* [string][in] */ LPCWSTR szName)<BR>{<BR>    _totchar c1(szName);<BR>    printf(_T("%s"), c1);<BR><FONT
    COLOR="BLUE">    return</FONT> S_OK;<BR>} 

Example 2:

You are developing a COM server and one interface method requires you to return a string as an argument.

HRESULT STDMETHODCALLTYPE <BR><BR>IOPCGroupStateMgt::GetName( /* [string][out] */ LPWSTR *ppName)  <BR>{<BR>    LPCTSTR szName = _T("Test Name");<BR>    _cowchar c1(szName, FALSE);<BR>    ppName = c1;<BR>    <FONT
    COLOR="BLUE">return</FONT> S_OK;<BR>}

This example converts a portable string (LPCTSTR) to Unicode string and does not deallocate memory because the converted string is assigned to an out argument of the COM interface method.

Example 3:

You are developing a portable application and need to send an ANSI string to serial port.

<FONT
    COLOR="BLUE">void</FONT> TSerialPort::Write(LPCTSTR szText)<BR>{    _tochar c1(szText);<BR>    writeString(c1, strlen(c1));<BR>}<BR><BR><FONT
    COLOR="BLUE">void</FONT> TSerialPort::writeString(<FONT COLOR="BLUE">const</FONT> <FONT COLOR="BLUE">char</FONT> *szString)<BR>{<BR>    ...<BR>}

That's all. I hope that you will find it as useful as I have.