Boost spirit revers parsing
I want to parse a file containing the following structure:
some
garbage *&%
section1 {
section_content
}
section2 {
section_content
}
Rule parsing is section_name1 { ... } section_name2 { ... }
already defined:
section_name_rule = lexeme[+char_("A-Za-z0-9_")];
section = section_name_rule > lit("{") > /*some complicated things*/... > lit("}");
sections %= +section;
So I need to skip any garbage until the rule is met sections
. Is there a way to do this? I tried seek[sections]
it but doesn't seem to work.
EDIT : I have localized the reason why the search is not working: if I use the next ( >>
) operator then it works. If the wait parser ( >
) is used, it throws an exception. Here's some sample code:
#define BOOST_SPIRIT_DEBUG
#include <boost/fusion/adapted/struct.hpp>
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/repository/include/qi_seek.hpp>
#include <boost/spirit/include/phoenix.hpp>
namespace qi = boost::spirit::qi;
using boost::phoenix::push_back;
struct section_t {
std::string name, contents;
friend std::ostream& operator<<(std::ostream& os, section_t const& s) { return os << "section_t[" << s.name << "] {" << s.contents << "}"; }
};
BOOST_FUSION_ADAPT_STRUCT(section_t, (std::string, name)(std::string, contents))
typedef std::vector<section_t> sections_t;
template <typename It, typename Skipper = qi::space_type>
struct grammar : qi::grammar<It, sections_t(), Skipper>
{
grammar() : grammar::base_type(start) {
using namespace qi;
using boost::spirit::repository::qi::seek;
section_name_rule = lexeme[+char_("A-Za-z0-9_")];
//Replacing '>> with '> throws an exception, while this works as expected!!
section = section_name_rule
>>
lit("{") >> lexeme[*~char_('}')] >> lit("}");
start = seek [ hold[section[push_back(qi::_val, qi::_1)]] ]
>> *(section[push_back(qi::_val, qi::_1)]);
}
private:
qi::rule<It, sections_t(), Skipper> start;
qi::rule<It, section_t(), Skipper> section;
qi::rule<It, std::string(), Skipper> section_name_rule;
};
int main() {
typedef std::string::const_iterator iter;
std::string storage("sdfsdf\n sd:fgdfg section1 {dummy } section2 {dummy } section3 {dummy }");
iter f(storage.begin()), l(storage.end());
sections_t sections;
if (qi::phrase_parse(f, l, grammar<iter>(), qi::space, sections))
{
for(auto& s : sections)
std::cout << "Parsed: " << s << "\n";
}
if (f != l)
std::cout << "Remaining unparsed: '" << std::string(f,l) << "'\n";
}
So, in a real-world example, my entire grammar is built with wait statements. Do I have to change everything to make the "search" work, or is there some other way (say, search for a simple "{" and return one name rule back) ??
source to share
Here's a demonstration of using Hamlet for inspiration: Live On Coliru
start = *seek [ no_skip[eol] >> hold [section] ];
Notes:
- remove waiting points
- optimize by requiring start of line before section name
Input example:
some
garbage *&%
section1 {
Claudius: ...But now, my cousin Hamlet, and my son —
Hamlet: A little more than kin, and less than kind.
}
WE CAN DO MOAR GARBAGE
section2 {
Claudius: How is it that the clouds still hang on you?
Hamlet: Not so my lord; I am too much i' the sun
}
Output:
Parsed: section_t[section1] {Claudius: ...But now, my cousin Hamlet, and my son —
Hamlet: A little more than kin, and less than kind.
}
Parsed: section_t[section2] {Claudius: How is it that the clouds still hang on you?
Hamlet: Not so my lord; I am too much i' the sun
}
List of links
// #define BOOST_SPIRIT_DEBUG
#include <boost/fusion/adapted/struct.hpp>
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/repository/include/qi_seek.hpp>
namespace qi = boost::spirit::qi;
struct section_t {
std::string name, contents;
friend std::ostream& operator<<(std::ostream& os, section_t const& s) { return os << "section_t[" << s.name << "] {" << s.contents << "}"; }
};
BOOST_FUSION_ADAPT_STRUCT(section_t, (std::string, name)(std::string, contents))
typedef std::vector<section_t> sections_t;
template <typename It, typename Skipper = qi::space_type>
struct grammar : qi::grammar<It, sections_t(), Skipper>
{
grammar() : grammar::base_type(start) {
using namespace qi;
using boost::spirit::repository::qi::seek;
section_name_rule = lexeme[+char_("A-Za-z0-9_")];
section = section_name_rule >> '{' >> lexeme[*~char_('}')] >> '}';
start = *seek [ no_skip[eol] >> hold [section] ];
BOOST_SPIRIT_DEBUG_NODES((start)(section)(section_name_rule))
}
private:
qi::rule<It, sections_t(), Skipper> start;
qi::rule<It, section_t(), Skipper> section;
qi::rule<It, std::string(), Skipper> section_name_rule;
};
int main() {
using It = boost::spirit::istream_iterator;
It f(std::cin >> std::noskipws), l;
sections_t sections;
if (qi::phrase_parse(f, l, grammar<It>(), qi::space, sections))
{
for(auto& s : sections)
std::cout << "Parsed: " << s << "\n";
}
if (f != l)
std::cout << "Remaining unparsed: '" << std::string(f,l) << "'\n";
}
source to share