我正在尝试为数学表达式编写一个解析器,其中命名变量是boost::spirit
(版本1_51_0)中的nullaries,我完全是新的。我定义typedef boost::function<double()> Value
,我的规则将如此声明:qi::rule<Iterator, Value()> expression, term, others, ...;
我使用此宏
在nullaries上定义二元运算符#define BINARY_FUNCTOR(name, op) \
struct name \
{ \
name(Value x, Value y): x_(x), y_(y) {} \
double operator()() { return x_() op y_(); } \
Value x_, y_; \
};
并拥有ADD
,SUB
等。从我看到的示例中,我希望规则的定义如下:
expression = term
>> *( (lit('+') >> term[ADD(_val, _1)])
| (lit('-') >> term[SUB(_val, _1)])
);
但这似乎不是正确的语法,因为我收到错误
boost/spirit/home/support/action_dispatch.hpp:162: error: no match for call to ‘(const<unnamed>::SUB) (boost::function<double ()()>&, boost::spirit::context<boost::fusion::cons<boost::function<double ()()>&, boost::fusion::nil>, boost::fusion::vector0<void> >&, bool&)’
SRParser.cpp:38: note: candidates are: double<unnamed>::SUB::operator()()
我认为_1
并不像我期望的那样,即与下一个术语相关联的Value
。定义这样的规则的正确语法是什么?
答案 0 :(得分:5)
解析器表达式看起来没问题。
你感到困惑的是构建AST。显然,你已经决定使用Semantic Actions来做到这一点,但你的努力太过粗略,我无法看到如何(甚至决定你基于什么样的样本)。
本质上:你想用“ADD”/“SUB”实例做什么,你似乎魔法“将”进入你的规则?
现在,您只需直接使用实例作为语义操作。这会导致显示的错误消息直接告诉您实例无效作为语义操作。
我假设您确实希望使用Phoenix赋值将二进制操作分配给公开的属性。这看起来像:
expression = term
>> *( (lit('+') >> term[ _val = phx::construct<ADD>(_val, _1)])
| (lit('-') >> term[ _val = phx::construct<SUB>(_val, _1)])
);
你会发现这与传统的表达式语法更为匹配。
为了好玩,我根据您的Value
类型修改了一个完整的表达式解析器,并创建了这个工作演示: http://liveworkspace.org/code/3kgPJR$0
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/karma.hpp>
#include <boost/spirit/include/phoenix.hpp>
namespace qi = boost::spirit::qi;
namespace karma = boost::spirit::karma;
namespace phx = boost::phoenix;
typedef std::function<double()> Value;
#define BINARY_FUNCTOR(name, op) \
struct name \
{ \
name(Value x, Value y): x_(x), y_(y) {} \
double operator()() { return x_() op y_(); } \
Value x_, y_; \
};
BINARY_FUNCTOR(ADD, +)
BINARY_FUNCTOR(SUB, -)
BINARY_FUNCTOR(MUL, *)
BINARY_FUNCTOR(DIV, /)
struct LIT
{
LIT(double x): x_(x) {}
double operator()() { return x_; }
double x_;
};
struct NEG
{
NEG(Value x): x_(x) {}
double operator()() { return -x_(); }
Value x_;
};
template <typename It, typename Skipper = qi::space_type>
struct parser : qi::grammar<It, Value(), Skipper>
{
parser() : parser::base_type(expression)
{
using namespace qi;
expression =
term [_val = _1]
>> *( ('+' >> term [_val = phx::construct<ADD>(_val, _1)])
| ('-' >> term [_val = phx::construct<SUB>(_val, _1)])
);
term =
factor [_val = _1]
>> *( ('*' >> factor [_val = phx::construct<MUL>(_val, _1)])
| ('/' >> factor [_val = phx::construct<DIV>(_val, _1)])
);
factor =
double_ [_val = phx::construct<LIT>(_1)]
| '(' >> expression [_val = _1] >> ')'
| ('-' >> factor [_val = phx::construct<NEG>(_1)])
| ('+' >> factor [_val = _1]);
BOOST_SPIRIT_DEBUG_NODE(expression);
BOOST_SPIRIT_DEBUG_NODE(term);
BOOST_SPIRIT_DEBUG_NODE(factor);
}
private:
qi::rule<It, Value(), Skipper> expression, term, factor;
};
Value doParse(const std::string& input)
{
typedef std::string::const_iterator It;
parser<It, qi::space_type> p;
Value eval;
auto f(begin(input)), l(end(input));
if (!qi::phrase_parse(f,l,p,qi::space,eval))
std::cerr << "parse failed: '" << std::string(f,l) << "'\n";
if (f!=l)
std::cerr << "trailing unparsed: '" << std::string(f,l) << "'\n";
return eval;
}
int main()
{
auto area = doParse("2 * (3.1415927 * (10*10))");
std::cout << "Area of a circle r=10: " << area() << "\n";
}
会打印
Area of a circle r=10: 628.319