How should I modify this boost::spirit code to parse for any combination of ranges or lists?
Note that this question is somewhat different from my previous question. In this case, I don’t have a special pre-tag like RANGE: and LIST: that helps me parse, so I’m not sure if we require some kind of look-ahead here.
I do need to keep the parses separate, though, as doing so will help me capture the results into different data structures.
objective get all four test-cases to pass
EDIT ANSWER that seems to work with more complicated expressions
// #define BOOST_SPIRIT_DEBUG
#include <boost/foreach.hpp>
#include <boost/spirit/include/qi.hpp>
#include <boost/fusion/include/tuple.hpp>
#include <boost/tuple/tuple.hpp>
#include <boost/fusion/include/boost_tuple.hpp>
#include <iostream>
namespace qi = boost::spirit::qi;
typedef std::vector<boost::tuple<std::string,std::vector<int>>> MY_TYPE;
template <typename Iterator, typename Skipper>
struct my_grammar :
qi::grammar<Iterator, MY_TYPE(), Skipper>
{
my_grammar() :
my_grammar::base_type( entries )
{
entries %=
*(
+(qi::char_ - '-')
>> qi::lit( "->" )
>> ( comma | range )
)
;
range %= '{' >> qi::int_ >> ':' >> qi::int_ >> ':' >> qi::int_ >> '}';
comma %= '{' >> (qi::int_ % ',') >> '}';
BOOST_SPIRIT_DEBUG_NODE( range );
BOOST_SPIRIT_DEBUG_NODE( comma );
BOOST_SPIRIT_DEBUG_NODE( entries );
}
qi::rule<Iterator, std::vector<int>(), Skipper>
range, comma;
qi::rule<Iterator, MY_TYPE(), Skipper>
entries;
};
// -----------------------------------------------------------------------------
static void TryParse( const std::string& input, const std::string& label )
{
MY_TYPE entries;
auto it(input.cbegin()), end( input.cend() );
my_grammar<decltype(it), qi::space_type> entry_grammar;
if (qi::phrase_parse(it, end, entry_grammar, qi::space, entries)
&& it == end)
{
std::cerr << label << " SUCCESS" << std::endl;
}
else
{
std::cerr << label << " FAIL" << std::endl;
}
}
int
main( int argc, char* argv[] )
{
std::string range_first = "foo -> {1:9:1}\nbar -> {1,2,3,4,5}\n";
std::string comma_first = "foo -> {1,2,3,4,5}\nbar -> {1:9:1}\n";
std::string comma_only = "foo -> {1,2,3,4,5}\n";
std::string range_only = "foo -> {1:9:1}\n";
TryParse( range_first, "RANGE FIRST" );
TryParse( comma_first, "COMMA FIRST" );
TryParse( range_only, "RANGE ONLY" );
TryParse( comma_only, "COMMA ONLY" );
}
EDIT NEW OUTPUT (STILL CAN’T HANDLE HETEROGENEOUS LIST)
/tmp$ g++ -g -std=c++11 sandbox.cpp -o sandbox && ./sandbox
COMMA FIRST FAIL
RANGE FIRST FAIL
COMMA ONLY SUCCESS
RANGE ONLY SUCCESS
/tmp$
EDIT IMPROVED CODE (STILL CAN’T HANDLE HETEROGENEOUS LIST)
// #define BOOST_SPIRIT_DEBUG
#include <boost/foreach.hpp>
#include <boost/spirit/include/qi.hpp>
#include <iostream>
namespace qi = boost::spirit::qi;
template <typename Iterator, typename Skipper>
struct my_grammar :
qi::grammar<Iterator, std::vector<int>(), Skipper>
{
my_grammar() :
my_grammar::base_type(entries)
{
entries %=
*('{' >> qi::int_
>> ( range_tail | comma_tail ))
;
range_tail %= ':' >> qi::int_ >> ':' >> qi::int_ >> '}';
comma_tail %= *( ',' >> qi::int_ ) >> '}';
BOOST_SPIRIT_DEBUG_NODE(entries );
BOOST_SPIRIT_DEBUG_NODE(comma_tail);
BOOST_SPIRIT_DEBUG_NODE(range_tail);
}
qi::rule<Iterator, std::vector<int>(), Skipper> entries, comma_tail, range_tail;
};
// -----------------------------------------------------------------------------
static void TryParse( const std::string& input, const std::string& label )
{
std::vector<int> entries;
auto it(input.cbegin()), end( input.cend() );
my_grammar<decltype(it), qi::blank_type> entry_grammar;
if (qi::phrase_parse(it, end, entry_grammar, qi::blank, entries)
&& it == end)
{
std::cerr << label << " SUCCESS" << std::endl;
}
else
{
std::cerr << label << " FAIL" << std::endl;
}
}
int
main( int argc, char* argv[] )
{
std::string range_first = "{3:5:7}\n{1,2,3,4,5}";
std::string comma_first = "{1,2,3,4,5}\n{3:5:7}";
std::string comma_only = "{1,2,3,4,5}";
std::string range_only = "{3:5:7}";
TryParse( comma_first, "COMMA FIRST" );
TryParse( range_first, "RANGE FIRST" );
TryParse( comma_only, "COMMA ONLY" );
TryParse( range_only, "RANGE ONLY" );
}
ORIGINAL CODE
compile and run
/tmp$ g++ -g -std=c++11 sandbox.cpp -o sandbox && ./sandbox
COMMA FIRST FAIL
RANGE FIRST FAIL
COMMA ONLY SUCCESS
RANGE ONLY FAIL
/tmp$
sandbox.cpp
// #define BOOST_SPIRIT_DEBUG
#include <boost/foreach.hpp>
#include <boost/spirit/include/qi.hpp>
#include <iostream>
namespace qi = boost::spirit::qi;
typedef int Entry;
template <typename Iterator, typename Skipper>
struct my_grammar :
qi::grammar<Iterator, std::vector<Entry>(), Skipper>
{
my_grammar() :
my_grammar::base_type(entries)
{
entries %=
*(
'{' >>
( comma_list | range_list )
>> '}'
)
;
comma_list %= qi::int_ % ',';
range_list %= qi::int_ >> ':' >> qi::int_ >> ':' >> qi::int_;
BOOST_SPIRIT_DEBUG_NODE(entries);
BOOST_SPIRIT_DEBUG_NODE(comma_list);
BOOST_SPIRIT_DEBUG_NODE(range_list);
}
qi::rule<Iterator, std::vector<Entry>(), Skipper> entries, comma_list, range_list;
};
// -----------------------------------------------------------------------------
static void TryParse( const std::string& input, const std::string& label )
{
std::vector<Entry> entries;
auto it(input.cbegin()), end( input.cend() );
my_grammar<decltype(it), qi::blank_type> entry_grammar;
if (qi::phrase_parse(it, end, entry_grammar, qi::blank, entries)
&& it == end)
{
std::cerr << label << " SUCCESS" << std::endl;
}
else
{
std::cerr << label << " FAIL" << std::endl;
}
}
int
main( int argc, char* argv[] )
{
std::string range_first = "{3:5:7}\n{1,2,3,4,5}";
std::string comma_first = "{1,2,3,4,5}\n{3:5:7}";
std::string comma_only = "{1,2,3,4,5}";
std::string range_only = "{3:5:7}";
TryParse( range_first, "COMMA FIRST" );
TryParse( comma_first, "RANGE FIRST" );
TryParse( comma_only, "COMMA ONLY" );
TryParse( range_only, "RANGE ONLY" );
}
Consume the leading number in common code. After that number have either a range tail or a list tail. List tails start with a , and range tails start with a :.
Possibly fold the } into the tail, so a list tail is (,number)*} and a range tail is :number:number}, which makes the empty list easier to parse.