Subj : Re: How do I string compare two strings, one Unicode and the other
To   : borland.public.cpp.borlandcpp
From : maeder
Date : Wed Sep 28 2005 06:45 pm

"D" <mr_dharris@hotmail.com> writes:

> Can somebody tell me how to perform a string comparision of two
> strings, one string is Unicode and the other a char*?
> The following isn't working for me.
> Am I using the correct string comparison functions?
>   
>   wchar_t* name = (wchar_t*) "fred";

First of all, the type of "fred" is 'array of 5 char const'. Writing
over the elements of this array isn't possible directly; doing so
indirectly, e.g. after casting away the constness, has undefined
behavior. You should therefore avoid having a pointer to non-const
pointing at that array.

Then the expression (wchar_t*) "fred" is equivalent to
reinterpret_cast<wchar_t *>("fred"). The value of this expression is
implementation-defined, and almost certainly not what you need.


>   if (wcscmp(name, (wchar_t*) "fred") == 0)

The function wcscmp is used to compare two wide character strings
(cf. http://tinyurl.com/935o5 aka
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vclib/html/_crt_strcmp.2c_.wcscmp.2c_._mbscmp.asp
, which may wrap). Passing it the address of an array of no-wide
characters is thus lying. You'll get undefined behavior; with a little
luck on your side, the program will crash and you won't miss the
error.


>   {
>     // never enters here.

You are out of luck then.

>   }


You have to pass wcscmp() pointers into two arrays of wchar_t. E.g.:

if (wcscmp(name,L"fred")==0)

The prefix L in front of a string literal causes the literal's element
type to be wchar_t.


Whether this suggestion works depends on whether name and L"fred" have
the same encoding. You write that name's encoding is Unicode. The
encoding of L"fred" depends on the settings of the compiler and the OS
it is run on. If it's not Unicode, you'll first have to convert L"fred"
to Unicode.

.