mwe.cpp

#include "mwe.h"
#include <boost/algorithm/string.hpp>

namespace Corpus2{

LexicalUnit::LexicalUnit(const std::string &base,
						 LexicalUnit::BoolOpPtr condition,
						 LexicalUnit::BoolOpPtr head_cond,
						 LexicalUnit::strmap variables)
	: condition_(condition),
	  head_cond_(head_cond),
	  variables_(variables),
	  base_(base),
	  nowhere_(Wccl::Position())
{
	for(strmap::iterator iter = variables_.begin();
		iter != variables_.end(); ++iter)
		potential_bases_.insert(iter->second);
}

bool LexicalUnit::IsHere(const Wccl::SentenceContext &sc,
					std::set<size_t> &out_position) const
{
	boost::shared_ptr<const Wccl::Bool> pResult = condition_->apply(sc);
	if(pResult->get_value() == false)
		return false;

	foreach(const std::string&varname, condition_->valid_variable_names()){
		if(boost::algorithm::starts_with(varname, "Pos")){
			Wccl::Position pos = condition_->get<Wccl::Position>(varname);
			if(pos.equals(nowhere_)){
				std::string errmsg("Position for found MWE cannot be zero.");
				errmsg += " Offending unit: " + base_;
				throw Wccl::WcclError(errmsg);
			}
			out_position.insert( sc.get_abs_position(pos) );
		}
	}
	return true;
}

////////////////////////////////////////////////////////////////////
// TODO realy needed?

FixedLU::FixedLU(const std::string &base,
				boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
				boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
				std::map<std::string, std::string> variables)
		: LexicalUnit(base, condition, head_cond, variables)
{
}

FlexLU::FlexLU(const std::string &base,
				boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
				boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
				std::map<std::string, std::string> variables)
		: LexicalUnit(base, condition, head_cond, variables)
{
}

////////////////////////////////////////////////////////////////////

MWEIndex::MWEIndex() : index_(), empty_()
{
	// noop
}

void MWEIndex::add_lexicalunit(LexicalUnit::Ptr lu)
{
	foreach(const std::string& base, lu->get_potential_bases()){
		value_type::iterator find = index_.find(base);
		if(find == index_.end()){ // not found -> create new one
			luvec v;
			v.push_back(lu);
			index_.insert( std::make_pair(base, v));
		}else{// already exists -> add lu
			(find->second).push_back(lu);
		}
	}
}

const MWEIndex::luvec& MWEIndex::get_potential_lu(const std::string &base){
	value_type::iterator find = index_.find(base);
	if(find == index_.end()){ // not found -> return empty
		return empty_;
	}

	return (find->second);
}

}//ns Corpus2