/*=============================================================================
Boost.Wave: A Standard compliant C++ preprocessor library
http://www.boost.org/
Copyright (c) 2001-2005 Hartmut Kaiser. Distributed under the Boost
Software License, Version 1.0. (See accompanying file
LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
=============================================================================*/
#if !defined(CPP_CHLIT_GRAMMAR_HPP_9527D349_6592_449A_A409_42A001E6C64C_INCLUDED)
#define CPP_CHLIT_GRAMMAR_HPP_9527D349_6592_449A_A409_42A001E6C64C_INCLUDED
#include <limits> // sid::numeric_limits
#include <boost/static_assert.hpp>
#include <boost/cstdint.hpp>
#include <boost/spirit/core.hpp>
#include <boost/spirit/attribute/closure.hpp>
#include <boost/spirit/dynamic/if.hpp>
#include <boost/spirit/phoenix/operators.hpp>
#include <boost/spirit/phoenix/primitives.hpp>
#include <boost/spirit/phoenix/statements.hpp>
#include <boost/spirit/phoenix/functions.hpp>
#include <boost/wave/wave_config.hpp>
#include <boost/wave/cpp_exceptions.hpp>
#include <boost/wave/grammars/cpp_literal_grammar_gen.hpp>
///////////////////////////////////////////////////////////////////////////////
//
// Reusable grammar to parse a C++ style character literal
//
///////////////////////////////////////////////////////////////////////////////
namespace boost {
namespace wave {
namespace grammars {
namespace closures {
struct chlit_closure
: boost::spirit::closure<chlit_closure, boost::uint32_t, bool>
{
member1 value;
member2 long_lit;
};
}
namespace impl {
///////////////////////////////////////////////////////////////////////////////
//
// compose a multibyte character literal
//
///////////////////////////////////////////////////////////////////////////////
struct compose_character_literal {
template <typename ResultT, typename A1, typename A2, typename A3>
struct result {
typedef void type;
};
void
operator()(boost::uint32_t& value, bool long_lit, bool& overflow,
boost::uint32_t character) const
{
// The following assumes that wchar_t is max. 32 Bit
BOOST_STATIC_ASSERT(sizeof(wchar_t) <= 4);
static boost::uint32_t masks[] = {
0x000000ff, 0x0000ffff, 0x00ffffff, 0xffffffff
};
static boost::uint32_t overflow_masks[] = {
0xff000000, 0xffff0000, 0xffffff00, 0xffffffff
};
if (long_lit) {
// make sure no overflow will occur below
if ((value & overflow_masks[sizeof(wchar_t)-1]) != 0) {
overflow |= true;
}
else {
// calculate the new value (avoiding a warning regarding
// shifting count >= size of the type)
value <<= 8 * (sizeof(wchar_t)-1);
value <<= 8;
value |= character & masks[sizeof(wchar_t)-1];
}
}
else {
// make sure no overflow will occur below
if ((value & overflow_masks[sizeof(char)-1]) != 0) {
overflow |= true;
}
else {
// calculate the new value
value <<= 8 * sizeof(char);
value |= character & masks[sizeof(char)-1];
}
}
}
};
phoenix::function<compose_character_literal> const compose;
} // namespace impl
///////////////////////////////////////////////////////////////////////////////
// define, whether the rule's should generate some debug output
#define TRACE_CHLIT_GRAMMAR \
bool(BOOST_SPIRIT_DEBUG_FLAGS_CPP & BOOST_SPIRIT_DEBUG_FLAGS_CHLIT_GRAMMAR) \
/**/
struct chlit_grammar :
public boost::spirit::grammar<chlit_grammar,
closures::chlit_closure::context_t>
{
chlit_grammar()
: overflow(false)
{
BOOST_SPIRIT_DEBUG_TRACE_GRAMMAR_NAME(*this, "chlit_grammar",
TRACE_CHLIT_GRAMMAR);
}
template <typename ScannerT>
struct definition
{
typedef
boost::spirit::rule<ScannerT, closures::chlit_closure::context_t>
rule_t;
rule_t ch_lit;
definition(chlit_grammar const &self)
{
using namespace boost::spirit;
using namespace phoenix;
// special parsers for '\x..' and L'\x....'
typedef uint_parser<
unsigned int, 16, 1, 2 * sizeof(char)
> hex_char_parser_type;
typedef uint_parser<
unsigned int, 16, 1, 2 * sizeof(wchar_t)
> hex_wchar_parser_type;
// the rule for a character literal
ch_lit
= eps_p[self.value = val(0), self.long_lit = val(false)]
>> !ch_p('L')[self.long_lit = val(true)]
>> ch_p('\'')
>> +( (
ch_p('\\')
>> ( ch_p('a') // BEL
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x07))
]
| ch_p('b') // BS
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x08))
]
| ch_p('t') // HT
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x09))
]
| ch_p('n') // NL
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x0a))
]
| ch_p('v') // VT
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x0b))
]
| ch_p('f') // FF
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x0c))
]
| ch_p('r') // CR
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val(0x0d))
]
| ch_p('?')
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val('?'))
]
| ch_p('\'')
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val('\''))
]
| ch_p('\"')
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val('\"'))
]
| ch_p('\\')
[
impl::compose(self.value, self.long_lit,
var(self.overflow), val('\\'))
]
| ch_p('x')
>> if_p(self.long_lit)
[
hex_wchar_parser_type()
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
]
.else_p
[
hex_char_parser_type()
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
]
| ch_p('u')
>> uint_parser<unsigned int, 16, 4, 4>()
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
| ch_p('U')
>> uint_parser<unsigned int, 16, 8, 8>()
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
| uint_parser<unsigned int, 8, 1, 3>()
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
)
)
| ~eps_p(ch_p('\'')) >> anychar_p
[
impl::compose(self.value, self.long_lit,
var(self.overflow), arg1)
]
)
>> ch_p('\'')
;
BOOST_SPIRIT_DEBUG_TRACE_RULE(ch_lit, TRACE_CHLIT_GRAMMAR);
}
// start rule of this grammar
rule_t const& start() const
{ return ch_lit; }
};
// flag signaling integer overflow during value composition
mutable bool overflow;
};
#undef TRACE_CHLIT_GRAMMAR
///////////////////////////////////////////////////////////////////////////////
//
// The following function is defined here, to allow the separation of
// the compilation of the intlit_grammap from the function using it.
//
///////////////////////////////////////////////////////////////////////////////
#if BOOST_WAVE_SEPARATE_GRAMMAR_INSTANTIATION != 0
#define BOOST_WAVE_CHLITGRAMMAR_GEN_INLINE
#else
#define BOOST_WAVE_CHLITGRAMMAR_GEN_INLINE inline
#endif
template <typename TokenT>
BOOST_WAVE_CHLITGRAMMAR_GEN_INLINE
unsigned int
chlit_grammar_gen<TokenT>::evaluate(TokenT const &token)
{
using namespace boost::spirit;
chlit_grammar g;
boost::uint32_t result = 0;
typename TokenT::string_type const &token_val = token.get_value();
parse_info<typename TokenT::string_type::const_iterator> hit =
parse(token_val.begin(), token_val.end(), g[spirit_assign_actor(result)]);
if (!hit.hit) {
BOOST_WAVE_THROW(preprocess_exception, ill_formed_character_literal,
token_val.c_str(), token.get_position());
}
else {
// range check
if ('L' == token_val[0]) {
// recognised wide character
if (g.overflow ||
result > (unsigned long)(std::numeric_limits<wchar_t>::max)())
{
// out of range
BOOST_WAVE_THROW(preprocess_exception,
character_literal_out_of_range,
token_val.c_str(), token.get_position());
}
}
else {
// recognised narrow ('normal') character
if (g.overflow ||
result > (unsigned long)(std::numeric_limits<unsigned char>::max)())
{
// out of range
BOOST_WAVE_THROW(preprocess_exception,
character_literal_out_of_range,
token_val.c_str(), token.get_position());
}
}
}
return result;
}
#undef BOOST_WAVE_CHLITGRAMMAR_GEN_INLINE
///////////////////////////////////////////////////////////////////////////////
} // namespace grammars
} // namespace wave
} // namespace boost
#endif // !defined(CPP_CHLIT_GRAMMAR_HPP_9527D349_6592_449A_A409_42A001E6C64C_INCLUDED)