How to get the substring matched by a boost.spirit numeric parser?

user_481267 0
Viewed 113

I want to use qi::int_parser<int64_t> to parse an integer value (it's really convenient how it automatically checks for overflows, handles the INT_MIN case, and so on). But, I also want to get the substring that the int_parser matched, because I want to print a warning message if it has extraneous characters (i.e., a plus sign, leading zeroes, or the case of -0).

I saw in another answer the suggestion to use qi::as_string, but it doesn't seem to work in this case. Here is some code that illustrates the issue:

#include <boost/phoenix/core.hpp>
#include <boost/phoenix/operator.hpp>
#include <boost/spirit/include/qi.hpp>
#include <cstdint>
#include <iostream>
#include <string>
int main() {
    namespace phx = boost::phoenix;
    namespace qi = boost::spirit::qi;
    using namespace std;

    std::string value_str;
    int64_t value;
    std::string test_str = "<+123>";

    const auto success = qi::parse(test_str.begin(), test_str.end(),
        qi::char_('<') >>
        qi::as_string[
            qi::int_parser<int64_t>{}[phx::ref(value) = qi::_1]
        ][phx::ref(value_str) = qi::_1] >>
        qi::char_('>')
    );

    std::cout << "success: " << success << '\n';
    std::cout << "value: " << value << '\n';
    std::cout << "matched substring: " << value_str << '\n';
}

The output I want is

success: 1
value: 123
matched substring: +123

The output I get is

success: 1
value: 123
matched substring: {

(or some other garbage). Parsing the int value works just fine, but I can't figure out how to get the substring.

boost-spirit-qi boost-spirit c++
2 Answers

Use qi::raw before passing it to qi::as_string:

qi::raw[qi::int_parser<int64_t>{}[phx::ref(value) = qi::_1]]

Result:

success: 1
value: 123
matched substring: +123
user_48993300

I'd simplify using a proper rule - so you don't need to spell out as_string in the parse expression.

There is something wrong with the way it works in this particular case (which should be reported as a bug to the library maintainers). However I could work around it by adding an eps parser inside the raw directive:

r %= '<' >> raw[int64_[phx::ref(value) = _1] >> eps] >> '>';

(Note also there is no need to parse the brackets with char_).

Live On Coliru

#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <iomanip>
using namespace std::string_literals;

int main()
{
    namespace phx = boost::phoenix;
    namespace qi = boost::spirit::qi;

    qi::rule<std::string::const_iterator, std::string()> r;

    int64_t value;
    {
        using namespace qi;
        static const int_parser<int64_t> int64_{};
        r %= '<' >> raw[int64_[phx::ref(value) = _1] >> eps] >> '>';
    }

    using Lim = std::numeric_limits<decltype(value)>;
    for (std::string const test_str : {
             "<+123>"s,
             "<0123>"s,
             "<0123>"s,
             "<123>"s,
             "<" + std::to_string(Lim::max()) + ">",
             "<" + std::to_string(Lim::min()) + ">",
         })
    {
        std::string value_str;

        auto success
            = qi::parse(test_str.begin(), test_str.end(), r, value_str);

        std::cout << "success: " << std::boolalpha  << success << "\n";
        std::cout << "value: "             << value                  << "\n";
        std::cout << "matched substring: " << std::quoted(value_str) << "\n";
    }
}

Prints

success: true
value: 123
matched substring: "+123"
success: true
value: 123
matched substring: "0123"
success: true
value: 123
matched substring: "0123"
success: true
value: 123
matched substring: "123"
success: true
value: 9223372036854775807
matched substring: "9223372036854775807"
success: true
value: -9223372036854775808
matched substring: "-9223372036854775808"

BONUS

Encapsulate the "value" param as well, so you don't use globals:

Live On Coliru

#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <iomanip>
using namespace std::string_literals;
namespace phx = boost::phoenix;
namespace qi = boost::spirit::qi;

using It = std::string::const_iterator;
using T = std::int64_t;

struct Parser : qi::grammar<It, std::string(T&)> {
    Parser() : Parser::base_type(r) {
        r %= '<' >> qi::raw[ int64_[qi::_r1 = qi::_1] >> qi::eps ] >> '>';
    }
  private:
    qi::int_parser<T> int64_;
    qi::rule<It, std::string(T&)> r;
};

int main()
{
    Parser const p;
    using Lim = std::numeric_limits<T>;
    for (std::string const test_str : {
             "<+123>"s,
             "<0123>"s,
             "<0123>"s,
             "<123>"s,
             "<" + std::to_string(Lim::max()) + ">",
             "<" + std::to_string(Lim::min()) + ">",
         })
    {
        std::string value_str;
        int64_t value;

        auto success
            = qi::parse(test_str.begin(), test_str.end(), p(phx::ref(value)), value_str);

        std::cout << "success: " << std::boolalpha  << success << "\n";
        std::cout << "value: "             << value                  << "\n";
        std::cout << "matched substring: " << std::quoted(value_str) << "\n";
    }
}

Printing:

success: true
value: 123
matched substring: "+123"
success: true
value: 123
matched substring: "0123"
success: true
value: 123
matched substring: "0123"
success: true
value: 123
matched substring: "123"
success: true
value: 9223372036854775807
matched substring: "9223372036854775807"
success: true
value: -9223372036854775808
matched substring: "-9223372036854775808"
user_853710
Related