Question

Let’s say I’m reading a line from a file:

{Parent{{ChildA}{ChildB}}}

More complex example:

{Parent{{ChildA{ChildC}{ChildD}}{ChildB{ChildE}{ChildF}}}}

Which is the grammar used to construct a tree.

Any name inside {} brackets is a node, and if within that bracket there are other nodes (brackets), those nodes are children.

I’m able to parse the first specific example using a counter, but only to find the text names of the nodes. How can I parse this so that I can determine what nodes are children of one another? I can’t seem to wrap my mind around the code I would use. I have a feeling I would use recursion.

Any help or advice would be appreciated.

C++ is preferred.

Thank you very much.

Answer 1

Spoiling the fun with an answer you can’t use anyway if it’s homework:

A minimal implementation with Boost Spirit Qi:

#include <boost/spirit/include/qi.hpp>
namespace qi = boost::spirit::qi;

typedef boost::make_recursive_variant<
    std::vector<boost::recursive_variant_>, 
    std::string>::type ast_t;

void dump(const ast_t&);

// adhoc parser rule:
static const qi::rule<std::string::iterator, ast_t()> node = 
    '{' >> *node >> '}' | +~qi::char_("{}");

int main()
{
     std::string input = "{Parent{{ChildA{ChildC}{ChildD}}{ChildB{ChildE}{ChildF}}}}";
     std::string::iterator f(input.begin()), l(input.end());

     ast_t tree;
     if (qi::parse(f, l, node, tree))
         dump(tree);
     else
         std::cerr << "Unparsed: " << std::string(f, l) << std::endl;
}

The implementation of dump is regrettably almost the equivalent amount of code 🙂

---

It will print:

{
    Parent
    {
        {
            ChildA
            {
                ChildC
            }
            {
                ChildD
            }
        }
        {
            ChildB
            {
                ChildE
            }
            {
                ChildF
            }
        }
    }
}

_{Here is the definition of dump(const ast_t&):}

struct dump_visitor : boost::static_visitor<>
{
    dump_visitor(int indent=0) : _indent(indent) {}

    void operator()(const std::string& s) const { print(s); }

    template <class V>
        void operator()(const V& vec) const
    {
        print("{");
        for(typename V::const_iterator it=vec.begin(); it!=vec.end(); it++)
            boost::apply_visitor(dump_visitor(_indent+1), *it);
        print("}");
    }

  private:
    template <typename T> void print(const T& v) const 
      { std::cout << std::string(_indent*4, ' ') << v << std::endl; }
    int _indent;
};

void dump(const ast_t& tree)
{
    boost::apply_visitor(dump_visitor(), tree);
}