How to use the "unsigned" representation of a number

[FabioAng]

Assuming I have this function (it's not complete):

template<typena me InputType, typename OutputIterator>
void to_utf8(InputTy pe input, OutputIterator result)
{
// trivial conversion
if (input < (1 << 7)) {
*result = input; ++result;
}else {
*result = ((input >6) | 0xc0); ++result;
*result = ((input & 0x3f) | 0x80); ++result;
}
}

Everything works fine if InputType is "unsigned" but it could not work if
InputType is "signed".
This problem can be generalized in every template function where a numbering
compare is applied.

template<typena me InputType>
bool is_valid_xml_ra nge(InputType input)
{
if ( ( input >= 0x20 ) && ( input <= 0xD7FF ) )
return true;
else
return false;
}

Is there a way to use always the "unsigned" representation, e.g.

template<typena me InputType>
bool is_valid_xml_ra nge(InputType input)
{
UnsignedInputTy pe uinput = input;

if ( ( uinput >= 0x20 ) && ( uinput <= 0xD7FF ) )
return true;
else
return false;
}

Regards,
Fabio

[Michael DOUBEZ]

FabioAng a écrit :

Assuming I have this function (it's not complete):

template<typena me InputType, typename OutputIterator>
void to_utf8(InputTy pe input, OutputIterator result)
{
// trivial conversion
if (input < (1 << 7)) {
*result = input; ++result;
}else {
*result = ((input >6) | 0xc0); ++result;
*result = ((input & 0x3f) | 0x80); ++result;
}
}

Everything works fine if InputType is "unsigned" but it could not work if
InputType is "signed".
This problem can be generalized in every template function where a numbering
compare is applied.
[snip]

Is there a way to use always the "unsigned" representation, e.g.
[snip]

What you want to known is in fact if there is ones above 7th bit:
if( (input>>7) != 0 ) ...

Michael

[FabioAng]

"Michael DOUBEZ" <mi************ @free.frwrote in message
news:46******** *************** @news.free.fr.. .

>
What you want to known is in fact if there is ones above 7th bit:
if( (input>>7) != 0 ) ...

Michael

I would like to solve the generic problem not the only one I posted

Fabio

[Gianni Mariani]

On May 14, 9:39 pm, "FabioAng" <fabioangNOS... @libero.itwrote :

"Michael DOUBEZ" <michael.dou... @free.frwrote in message

news:46******** *************** @news.free.fr.. .

What you want to known is in fact if there is ones above 7th bit:
if( (input>>7) != 0 ) ...

Michael

I would like to solve the generic problem not the only one I posted

Have you looked at this ?

http://www.unicode.org/Public/PROGRA...F/ConvertUTF.c

[Sylvester Hesp]

"FabioAng" <fa************ @libero.itwrote in message
news:f2******** **@nnrp-beta.newsland.i t...

Is there a way to use always the "unsigned" representation, e.g.

template<typena me InputType>
bool is_valid_xml_ra nge(InputType input)
{
UnsignedInputTy pe uinput = input;

if ( ( uinput >= 0x20 ) && ( uinput <= 0xD7FF ) )
return true;
else
return false;
}

You can do it via template specialization

template<class Tstruct unsigned_type { typedef T type; };
template<struct unsigned_type<c har{ typedef unsigned char type; };
template<struct unsigned_type<s igned char{ typedef unsigned char
type; };
template<struct unsigned_type<s hort{ typedef unsigned short type; };
template<struct unsigned_type<i nt{ typedef unsigned int type; };
template<struct unsigned_type<l ong{ typedef unsigned long type; };
template<typena me InputType>
bool is_valid_xml_ra nge(InputType input)
{
typedef unsigned_type<I nputType>::type UnsignedInputTy pe;
UnsignedInputTy pe uinput = input;

if ( ( uinput >= 0x20 ) && ( uinput <= 0xD7FF ) )
return true;
else
return false;
}
- Sylvester

[FabioAng]

I found a solution in an old boost thread

#include <boost/type_traits.hpp >
#include <boost/integer.hpp>

namespace detail
{
template <class T, bool is_signed>
struct remove_signedne ss_helper
{
typedef T type;
};

template <class T>
struct remove_signedne ss_helper<T, true>
{
typedef typename boost::uint_t<s izeof(T) * 8>::least type;
};
}

template <class T>
struct remove_signedne ss
{
typedef typename
detail::remove_ signedness_help er<T,boost::is_ signed<T>::valu e>::type type;
};

template<typena me InputType, typename OutputIterator>
void to_utf8(InputTy pe input, OutputIterator result)
{
remove_signedne ss<InputType>:: type uinput;
uinput = input;
// trivial conversion
if (uinput < (1 << 7)) {
*result = uinput; ++result;
} else {
*result = ((uinput >6) | 0xc0); ++result;
*result = ((uinput & 0x3f) | 0x80); ++result;
}
}

Is there anybody who sees something wrong in this piece if code ?

Regards
Fabio

[FabioAng]

"Sylvester Hesp" <s.****@oisyn.n lwrote in message
news:46******** *************@n ews.xs4all.nl.. .

>
"FabioAng" <fa************ @libero.itwrote in message
news:f2******** **@nnrp-beta.newsland.i t...

>Is there a way to use always the "unsigned" representation, e.g.

template<typen ame InputType>
bool is_valid_xml_ra nge(InputType input)
{
UnsignedInputTy pe uinput = input;

if ( ( uinput >= 0x20 ) && ( uinput <= 0xD7FF ) )
return true;
else
return false;
}

You can do it via template specialization

template<class Tstruct unsigned_type { typedef T type; };
template<struct unsigned_type<c har{ typedef unsigned char type; };
template<struct unsigned_type<s igned char{ typedef unsigned char
type; };
template<struct unsigned_type<s hort{ typedef unsigned short type; };
template<struct unsigned_type<i nt{ typedef unsigned int type; };
template<struct unsigned_type<l ong{ typedef unsigned long type; };
template<typena me InputType>
bool is_valid_xml_ra nge(InputType input)
{
typedef unsigned_type<I nputType>::type UnsignedInputTy pe;
UnsignedInputTy pe uinput = input;

if ( ( uinput >= 0x20 ) && ( uinput <= 0xD7FF ) )
return true;
else
return false;
}
- Sylvester

Great idea...the same which I found using boost library

Thanks
Fabio

[faceman28208]

On May 14, 5:35 am, "FabioAng" <fabioangNOS... @libero.itwrote :

if (input < (1 << 7)) {

if ( *(unsigned char *) &input < (1 << 7)) {

[Gianni Mariani]

FabioAng wrote:

I found a solution in an old boost thread

....

Is there anybody who sees something wrong in this piece if code ?

It's wrong ? It only works for is8859-1 ?