Parsing null character with boost spirit qi

Question

I am trying to parse a string with boost spirit qi, which has the following form:

"\0help@masonlive.gmu.edu\0test\r\n"

with the following grammar: Here is the hpp:

class EmailGrammar :
    public boost::spirit::qi::grammar< const boost::uint8_t*,
        boost::tuple< boost::iterator_range< const boost::uint8_t*>,
                      boost::iterator_range< const boost::uint8_t*> >()>
{
public:
    const static EmailGrammar instance;

    EmailGrammar ();    

    /* omitting uninteresting stuff here i.e. constructors and assignment */

private:
    boost::spirit::qi::rule< const boost::uint8_t*,
        boost::tuple<
            boost::iterator_range< const boost::uint8_t*>,
            boost::iterator_range< const boost::uint8_t* >()> m_start;
};

and the cpp of the grammar looks like this:

EmailGrammar::EmailGrammar() :
    EmailGrammar::base_type(m_start),
    m_start()
{
    namespace qi = boost::spirit::qi;
    m_start = 
             (
             qi::lit('\0')
             >> (
                    qi::raw[*(qi::char_ - qi::lit('\0'))]
                )
             >> qi::lit('\0')
             >> (
                    qi::raw[*(qi::char_ - qi::eol)]
                )
             >> qi::eol >> qi::eoi
             );
}

Which I intend to use to parse the two strings and break them into two separate iterator ranges.

This is then called like so:

int main()
{
    typedef typename EmailGrammar::start_type::attr_type attr;

    std::string testStr("\0help@masonlive.gmu.edu\0test\r\n");

    // this is not done this way in the real code just as a test
    boost::iterator_range<const boost::uint8_t*> data =
        boost::make_iterator_range(
            reinterpret_cast< const boost::uint8_t* >(testStr.data()),
            reinterpret_cast< const boost::uint8_t* >(testStr.data() + testStr.length()));

    attr exposedAttribute;
    if (boost::spirit::qi::parse(data.begin(),
                                 data.end(),
                                 EmailGrammar::instance,
                                 exposedAttribute)
    {
        std::cout << "success" << std::endl;
    }
}

The problem appears to be in parsing the null terminator. I think this because when I add the debug(m_rule); to the code I get the xml output:

<unnamed-rule>
<try></try>
<fail/>
</unnamed-rule>

However. if I explicitly erase for example the first null terminator I get the output:

<unnamed-rule>
<try>help@masonlive.gmu.e</try>
<fail/>
</unnamed-rule>

Which leads to the questions:

• How does one parse null terminators with spirit I have look for documentation and haven't been able to find any information on it except a mention of null terminated strings at the very bottom of this page which mentions the default models in spirit.

• Does spirit do look ahead in a manner that if the parser sees in the look ahead it doesn't end correctly it automatically fails?

• Is there any documentation I am missing that I can use to read up on this sort of behavior?

Answer 1

More than likely the whole problem originates here:

std::string testStr("\0help@masonlive.gmu.edu\0test\r\n");

doesn't do what you think. It creates an empty string. Instead, specify the length of the raw literal/buffer:

std::string testStr("\0help@masonlive.gmu.edu\0test\r\n", 31);

BONUS

If you don't want to do the math/counting (you shouldn't!), make a helper:

template <typename Char, size_t N>
std::string bake(Char const (&p)[N], bool include_terminator = false) {
    return { p, p + N - (include_terminator?0:1) };
}

Which you can then use like:

std::string const testStr = bake("\0help@masonlive.gmu.edu\0test\r\n");

---

Live On Coliru

#define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/fusion/adapted/boost_tuple.hpp>
namespace qi = boost::spirit::qi;

using It = uint8_t const*;
using Range = boost::iterator_range<It>;
using Attribute = boost::tuple<Range, Range>;

class EmailGrammar : public qi::grammar<It, Attribute()> {
  public:
    const static EmailGrammar instance;

    EmailGrammar() : EmailGrammar::base_type(m_start)
    {
        using namespace qi;

        m_start = 
            '\0' >> raw[*(char_ - '\0')] >> 
            '\0' >> raw[*(char_ - eol)] >> 
            eol >> eoi
            ;

        BOOST_SPIRIT_DEBUG_NODES((m_start))
    }

  private:
    qi::rule<It, Attribute()> m_start;
};

const EmailGrammar EmailGrammar::instance {};

template <typename Char, size_t N>
std::string bake(Char const (&p)[N], bool include_terminator = false) {
    return { p, p + N - (include_terminator?0:1) };
}

int main() {
    std::string const testStr = bake("\0help@masonlive.gmu.edu\0test\r\n");

    It f = reinterpret_cast<It>(testStr.data()),
       l = f + testStr.length();

    Attribute exposedAttribute;
    if (boost::spirit::qi::parse(f, l, EmailGrammar::instance, exposedAttribute)) {
        std::cout << "success" << std::endl;
    }
}

Prints

<m_start>
  <try></try>
  <success></success>
  <attributes>[[[h, e, l, p, @, m, a, s, o, n, l, i, v, e, ., g, m, u, ., e, d, u], [t, e, s, t]]]</attributes>
</m_start>
success