Re: Cheap (simple, efficient) solution for literal strings in templates?
From: Alf P. Steinbach (alfps_at_start.no)
Date: 03/24/05
- Next message: Marco Cassiani: "Re: Big O and algorithm to decide string A contains string B?"
- Previous message: Gernot Frisch: "Re: Big O and algorithm to decide string A contains string B?"
- In reply to: Ioannis Vranos: "Re: Cheap (simple, efficient) solution for literal strings in templates?"
- Next in thread: Pete Becker: "Re: Cheap (simple, efficient) solution for literal strings in templates?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 24 Mar 2005 12:24:49 GMT
* Ioannis Vranos:
> * Alf P. Steinbach:
>
> > Depending on a template type argument Char I'd like a literal string,
> > and for that matter also literal char constants, to be 'char' or 'wchar_t'
> > base type, i.e.
> >
> > template< typename Char >
> > bool isPassword( std::basic_string<Char> const& s )
> > {
> > return s == "password"; // Cheap solution for Char === wchar_t?
> > }
> >
> > The cheapest solution I could make to work involved a class + macro
> > like
> >
> > struct literal_string
> > {
> > wchar_t const* myWide;
> > char const* myNarrow;
> >
> > literal_string( wchar_t const* w, char const* n )
> > : myWide( w ), myNarrow( n )
> > {}
> >
> > operator wchar_t const* () const { return myWide; }
> > operator char const* () const { return myNarrow; }
> > };
> >
> > #define S( a ) literal_string( a, L ## a )
> >
> > I couldn't make that work with the strings as template arguments.
>
>
> I did not understood what exactly you want to do, however you can
> "store" string literals like this:
>
>
> #include <vector>
>
> int main()
> {
> using namespace std;
>
> vector<char *> charLiterals;
>
> charLiterals.push_back("literal1");
> charLiterals.push_back("whatever");
>
> vector<wchar_t *> wcharLiterals;
>
> wcharLiterals.push_back(L"literal1");
> wcharLiterals.push_back(L"whatever");
>
>
> char *charlits[5]= {"lit1", "lit2" };
>
> charlits[3]= "whatever";
>
> wchar_t *wcharlits[5];
>
> wcharlits[0]= L"wide character literal";
> }
Well, best of all I'd like the ability to just write a string directly
where it's used, and that's what the solution I sketched above provided:
if the context requires 'char', then it's 'char', and ditto for 'wchar_t'.
For efficiency, however, and also for the ability to use as template
arguments and other compile-time stuff, it would be nice to declare the
strings at namespace scope, even in a header file.
Here's what I came up with for that; it's ugly, isn't it? :-)
================================================================================
struct dummy_type {};
template< template<typename Char> class LiteralStringDefiner >
struct cameleon_string_
{
operator char const* () { return LiteralStringDefiner<char>::s; }
operator wchar_t const* () { return LiteralStringDefiner<wchar_t>::s; }
};
#define DEFINE_CAMEL( name, literal_string ) \
namespace literal \
{ \
template< typename Dummy> struct name ## _char { static char const s[]; }; \
template< typename Dummy> char const name ## _char<Dummy>::s[] = "s"; \
\
template< typename Dummy> struct name ## _wchar_t { static wchar_t const s[]; };\
template< typename Dummy> wchar_t const name ## _wchar_t<Dummy>::s[] = L"s"; \
\
template< typename Char> struct name ## _; \
\
template<> struct name ## _<char>: name ## _char<dummy_type> {}; \
template<> struct name ## _<wchar_t>: name ## _wchar_t<dummy_type> {}; \
\
typedef cameleon_string_<name ## _> name; \
}
DEFINE_CAMEL( huh, "duh" )
================================================================================
With that I can write code like
================================================================================
template< typename Char >
void display( Char const* )
{
}
template< typename Char >
bool is_password( std::basic_string<Char> const& s )
{
display<Char>( literal::huh() );
return s == literal::huh_<Char>::s;
}
int main()
{
is_password<char>( "birthdate" );
is_password<wchar_t>( L"spouse" );
}
================================================================================
-- A: Because it messes up the order in which people normally read text. Q: Why is it such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?
- Next message: Marco Cassiani: "Re: Big O and algorithm to decide string A contains string B?"
- Previous message: Gernot Frisch: "Re: Big O and algorithm to decide string A contains string B?"
- In reply to: Ioannis Vranos: "Re: Cheap (simple, efficient) solution for literal strings in templates?"
- Next in thread: Pete Becker: "Re: Cheap (simple, efficient) solution for literal strings in templates?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
