Yet another CString replacement (only much more practical)

• Download source files - 10 Kb

Introduction

Its been almost 5 months since I'm involved in a very large Visual C++ project. The thing that I missed the most was a really practical string class.

<Getting political>

A good programmer is one that gets things done on time. Getting things done on time depends on the language you use and the available libraries. It's obvious that the language we use is C++, more exactly Visual C++. C++ is probably the most difficult language to master these days but we can't do anything about it, that's the standard and we have to live with it, and so we have a -(minus) on the "gets things done in time", which is almost always true from the point of view of management people. The only thing that can save us, besides years of experience, is good library support. The specific library support I will talk about in this article is, of course, string related and the keyword is practical.

On the MFC side I have CString, it's a decent string class implementation but I'm not very happy with it mainly for three reasons:

1. I can use it only in MFC projects
   
2. I cannot chain operations on a CString instance. Example:

   CString str( "  05 09 1990 ");    
   
   // I want to convert to "05/09/1990"
   
   str.TrimLeft();
   str.TrimRight();
   str.Replace( " ", "/" );

   By chaining operations I mean the following:

   str.TrimLeft().TrimRight().Replace( " ", "/" );

   See the difference? I don't know about you but I definitely do and I don't want to go back.

3. Numeric operations support. This is something I used even in the smallest projects I've done. There's almost no way I can escape these conversion operations, and again CString doesn't do a good job to help me out. Of course CString provides Format and FormatV methods but they're not at all nice with me.

On the non-MFC side I have ( if you're a STL lover stop reading here ) well...almost nothing. Yes, I know there is std::string but I can't do much with it. It suffers the same drawbacks ( maybe more ) all STL does: too much academicism as oposed to practical, everything is so in place and standards aligned that I am nothing but delighted when I look at it. Unfortunately that's all about it.

</Getting political>

Using the code

The class name is CStr. I'll start directly with examples, I think the class interface is self-explanatory:

Chaining operations: The above CString example using the CStr class:

CStr str( "  05 09 1990 " );    
str.Trim().Replace( " ", "/" ); 

Numerical operations:

CStr str;
long lValue;
str = 123; 
lValue = str; // lValue -> 123;

        
// or if you don't like automatic conversion operators

double dValue;
str.SetDouble( 123.45 );
dValue = str.GetDouble(); // dValue -> 123.45;

Append operator ( denoted by << ) : Using this operator you can chain concatenation like this:

CStr str;
str << "Value=" << 12 << .5 << " seconds"; // str -> "Value=12.5 seconds";

Observations:

• It is not the perfect string class but I like it, there will be bugs for sure in this implementation, so please let me know. If you think there are methods that must be added/changed also let me know, and if I like the proposals I'll change the interface.
• In many cases are performed operations that modify the string length, and thus the size of the internal buffer, and to avoid some of memory reallocation cycles each new allocation adds an extra buffer size of CSTR_EXTRA_SIZE (defined as static member). The default value is 16, if you don't want to use this feature and want to keep the buffer as small as possible set CSTR_EXTRA_SIZE value to 0.
• Numerical conversion regarding double values is done using _vsntprintf and setting type field to "%g".
• Though I wrote this class with full Unicode support I didn't have a chance to test it using Unicode, so it'll be great if someone can test it, before I'll use it in real projects ;).
• The demo project is an empty MFC project that is full of ASSERTs that wrap CStr methods for testing, something like unit testing.

CStr public Interface

                static int CSTR_EXTRA_SIZE;

                CStr() throw();
                ~CStr();
                CStr( const CStr &str ) throw();
                CStr( LPCTSTR pszStr ) throw();
                CStr( int allocSize ) throw();

CStr&           operator = ( const CStr& str ) throw();
CStr&           operator = ( LPCTSTR pszStr ) throw();
CStr&           operator = ( int iValue ) throw();
CStr&           operator = ( long lValue ) throw();
CStr&           operator = ( unsigned long ulValue ) throw();
CStr&           operator = ( double dValue ) throw();
            
// internal buffer access

TCHAR*          Buffer() const;
TCHAR*          Buffer();
int             BufferSize() const;
BOOL            Realloc( int size ) throw();
void            Compact() throw();

// attributes

int             Length() const;
BOOL            IsEmpty() const;

// comparision

BOOL            operator == ( LPCTSTR pszStr ) const;
BOOL            operator != ( const LPCTSTR& str ) const;
int             Compare( LPCTSTR pszStr ) const;
int             CompareNoCase( LPCTSTR pszStr ) const;

// accessors

               operator LPCTSTR () const;
               operator int () const;
               operator long () const;
               operator unsigned long () const;
               operator double () const;

TCHAR&         operator [] ( int pos );
const TCHAR&   operator [] ( int pos ) const;
    
CStr           Left( int size ) const throw();
CStr           Right( int size ) const throw();
CStr           Mid( int start, int size ) const throw();
    
               // number of occurences

int            FindCount( LPCTSTR pszStr ) const; 
               // same as above only starting at startPos

int            FindCount( int startPos, LPCTSTR pszStr ) const; 
               // zero-based index of the 1st occurence

int            Find( LPCTSTR pszStr ) const;
               // zero-based index of the nth occurence

int            FindNth( int nth, LPCTSTR pszStr ) const;
               // zero-based index of the 1st occurence starting

               // from the end of the string

int            ReverseFind( LPCTSTR pszStr ) const;
               // zero-based index of the nth occurence starting

               // from the end of the string

int            ReverseFindNth( int nth, LPCTSTR pszStr ) const;

// operations

CStr&          Empty();
CStr&          Fill( const TCHAR chr );
CStr&          Trim();
CStr&          TrimLeft();
CStr&          TrimRight();
CStr&          Lower();
CStr&          Upper();
CStr&          Insert( int pos, LPCTSTR pszStr ) throw();
CStr&          Prepend( LPCTSTR pszStr ) throw();
CStr&          Append( LPCTSTR pszStr ) throw();
CStr&          Remove( int pos, int len ) throw();
CStr&          Trunc( int pos );
CStr&          Replace( LPCTSTR pszOld, LPCTSTR pszNew ) throw();
CStr&          Replace( int startPos, LPCTSTR pszOld, LPCTSTR pszNew ) throw();
CStr&          Format( LPCTSTR pszFormat, ... ) throw();
CStr&          FormatV( LPCTSTR pszFormat, va_list args ) throw();
    
// numerics ( get/set )

int            GetInt() const;
long           GetLong() const;
unsigned long  GetULong() const;
double         GetDouble() const;
CStr&          SetInt( const int iValue ) throw();
CStr&          SetLong( const long lValue ) throw();
CStr&          SetULong( const unsigned long ulValue ) throw();
CStr&          SetDouble( const double dValue ) throw();

// << concatenation operator

friend CStr&   operator << ( CStr &str, LPCTSTR pszStr ) throw();
friend CStr&   operator << ( CStr &str, int iValue ) throw();
friend CStr&   operator << ( CStr &str, long lValue ) throw();
friend CStr&   operator << ( CStr &str, unsigned long ulValue ) throw();
friend CStr&   operator << ( CStr &str, double dValue ) throw();

That's it. Have fun !

History

June 21, 2003

• Replaced new/delete operators with malloc/free for internal buffer allocation/deallocation. It seems malloc is faster for large arrays.
• If memory allocation fails the std::bad_alloc exception will be thrown. The methods that may throw this exception are marked with throw() in the class declaration.
• As requested two new functions were adeded: FindNth( int nth, LPCTSTR pszStr ) const which returns the zero-based index of the nth occurence of pszStr. ReverseFindNth( int nth, LPCTSTR pszStr) const does the same thing only doing reverse search starting from the end of the string.
• Contains method was renamed to the more suggestive FindCount.
• All #defines replaced with static members, including CSTR_EXTRA_SIZE.
• The CStr implementation was backed up by _ASSERTs.
• Other small cleanups.