main.cpp

#include <cstdlib>
#include <cstdio>
#include <fstream>
#include <iomanip>


#include <libwccl/values/strset.h>
#include <libwccl/parser/Parser.h>
#include <libcorpus2/tagsetmanager.h>

#include <boost/bind.hpp>
#include <boost/algorithm/string.hpp>
#include <boost/make_shared.hpp>
#include <boost/program_options.hpp>
#include <boost/filesystem.hpp>
#include <libcorpus2/io/xcesreader.h>
#include <boost/lexical_cast.hpp>

#include <antlr/NoViableAltException.hpp>
#include <antlr/MismatchedTokenException.hpp>

namespace {
	bool quiet = false;
	bool tabs = false;
	bool output_orths = true;
	bool output_variables = false;
	bool global_numbering = false;
	bool in_sentence_numbering = true;
}

class streamsave
{
public:
	streamsave(std::ostream& os) : os_(os), flags_(os.flags()) {}
	~streamsave() { os_.flags(flags_); }
private:
	std::ostream& os_;
	std::ios_base::fmtflags flags_;
};

class Runner
{
public:
	Runner(const Corpus2::Tagset& tagset)
	 : tagset_(tagset), parser_(tagset_), token_idx(0)
	{
	}

	bool load_more_operators(const std::string &filename);
	const std::vector< boost::shared_ptr<Wccl::FunctionalOperator> >& operators() const {
		return ops_;
	}

	void do_sentence(const boost::shared_ptr<Corpus2::Sentence>& sentence);
	void do_stream(std::istream& is, bool first);

private:
	const Corpus2::Tagset& tagset_;
	Wccl::Parser parser_;
	std::vector< boost::shared_ptr<Wccl::FunctionalOperator> > ops_;
	std::vector< std::string > op_names_;
	int token_idx;
};

bool Runner::load_more_operators(const std::string& filename)
{
	boost::shared_ptr<Wccl::FunctionalOperator> retOp;
	try {
		std::ifstream is(filename.c_str());
		if (!is.good()) {
			throw Wccl::FileNotFound(filename, "", __FUNCTION__);
		}

		retOp = parser_.parseAnyOperator(is);
		if (retOp) {
			boost::filesystem::path p(filename);
			op_names_.push_back(p.stem());
			ops_.push_back(retOp);
			return true;
		} else {
			std::cerr << "Problem while parsing -- "
				<< "parser returned NULL!" << std::endl;
		}
	} catch (antlr::MismatchedTokenException &e) {
		std::cerr << e.getFileLineColumnString()
				<< " " << e.getMessage() << std::endl;
	} catch(antlr::NoViableAltException &e) {
		std::cerr << e.getFileLineColumnString()
				<< " " << e.getMessage() << std::endl;
	} catch (Wccl::InvalidVariableName &e) {
		std::cerr << "Wccl::InvalidVariableName " << e.info() << std::endl;
	} catch (Wccl::VariableTypeMismatch &e) {
		std::cerr << "Wccl::VariableTypeMismatch " << e.info() << std::endl;
	} catch (Wccl::WcclError& e) {
		std::cerr << "Wccl::WcclError:" << e.info() << std::endl;
	} catch (PwrNlp::PwrNlpError& e) {
		std::cerr << "PwrNlp::PwrNlpError " << e.info() << std::endl;
	} catch (antlr::ANTLRException& e) {
		std::cerr << "Antlr error " << e.getMessage() << std::endl;
	}
	return false;
}

void Runner::do_sentence(const boost::shared_ptr<Corpus2::Sentence>& sentence)
{
	Wccl::SentenceContext sc(sentence);
	std::vector< std::vector< UnicodeString > > outputs;

	streamsave sv(std::cout);

	if (1) { //header
		outputs.resize(outputs.size() + 1);
		std::vector< UnicodeString >& out = outputs.back();
		if (global_numbering) {
			out.push_back(UnicodeString::fromUTF8("##"));
		}
		if (in_sentence_numbering) {
			out.push_back(UnicodeString::fromUTF8("#"));
		}
		if (output_orths) {
			out.push_back(UnicodeString::fromUTF8("orth"));
		}
		for (size_t i = 0; i < op_names_.size(); ++i) {
			out.push_back(UnicodeString::fromUTF8(op_names_[i]));
			if (output_variables) {
				boost::shared_ptr<Wccl::FunctionalOperator> o = ops_[i];
				foreach (const std::string& varname, o->valid_variable_names()) {
					const Wccl::Value& value = (*o)[varname];
					std::string label = "(" + op_names_[i] + ")" + value.make_var_repr(varname);
					out.push_back(UnicodeString::fromUTF8(label));
				}
			}
		}
	}
	for (size_t i = 0; i < sentence->size(); ++i) {
		outputs.resize(outputs.size() + 1);
		std::vector< UnicodeString >& out = outputs.back();

		++token_idx;
		if (global_numbering) {
			out.push_back(UnicodeString::fromUTF8(boost::lexical_cast<std::string>(token_idx)));
		}
		if (in_sentence_numbering) {
			out.push_back(UnicodeString::fromUTF8(boost::lexical_cast<std::string>(i + 1)));
		}
		if (output_orths) {
			out.push_back(sentence->tokens()[i]->orth());
		}

		sc.set_position(i);
		foreach (const boost::shared_ptr<Wccl::FunctionalOperator>& o, ops_) {
			boost::shared_ptr<const Wccl::Value> v = o->base_apply(sc);
			UnicodeString vstr = v->to_string_u(tagset_);
			out.push_back(vstr);
			if (output_variables) {
				foreach (const std::string& varname, o->valid_variable_names()) {
					out.push_back((*o)[varname].to_string_u(tagset_));
				}
			}
		}
	}

	std::vector<int> lengths(outputs[0].size());
	foreach (const std::vector< UnicodeString >& line, outputs) {
		for (size_t i = 0; i < line.size(); ++i) {
			lengths[i] = std::max(lengths[i], line[i].length());
		}
	}

	foreach (const std::vector< UnicodeString >& line, outputs) {
		for (size_t i = 0; i < line.size(); ++i) {
			UnicodeString u = line[i];
			if (tabs) {
				std::cout << PwrNlp::to_utf8(line[i]) << "\t";
			} else {
				u.padTrailing(lengths[i] + 1);
				std::cout << PwrNlp::to_utf8(u);
			}
		}
		std::cout << "\n";
	}
}

void Runner::do_stream(std::istream& is, bool first)
{
	Corpus2::XcesReader xr(tagset_, is);
	Corpus2::Sentence::Ptr s;
	while ((s = xr.get_next_sentence())) {
		do_sentence(s);
		std::cout << "\n";
		if (first) break;
	}
}

int main(int argc, char** argv)
{
	std::string tagset_load = "kipi";
	bool first = false;
	std::vector<std::string> corpora_files, ccl_files, files;
	bool corpus_stdin = false;
	using boost::program_options::value;

	boost::program_options::options_description desc("Allowed options");
	desc.add_options()
			("tagset,t", value(&tagset_load),
			 "Tagset to use\n")
			("corpus,c", value(&corpora_files),
			 "Corpus file to load (XCES)\n")
			("ccl-file,C", value(&ccl_files),
			 "CCL query file\n")
			("files,f", value(&files),
			 "Files to load, looking at the extension to determine type\n")
			("corpus-from-stdin,I", value(&corpus_stdin)->zero_tokens(),
			 "Read corpus from stdin\n")
			("quiet,q", value(&quiet)->zero_tokens(),
			 "Suppress messages\n")
			("first-sentence-only,1", value(&first)->zero_tokens(),
			 "Only process first sentence\n")
			("tabs", value(&tabs)->zero_tokens(),
			 "Output a tab-separated file\n")
			("local-counts,l", value(&in_sentence_numbering),
			 "Output in-sentence token counts\n")
			("global-counts,g", value(&global_numbering),
			 "Output global counts\n")
			("output-orths,O", value(&output_orths),
			 "Output token orths\n")
			("output-variables,V", value(&output_variables),
			 "Output operator variables\n")
			("help,h", "Show help")
			;
	boost::program_options::variables_map vm;
	boost::program_options::positional_options_description p;
	p.add("files", -1);

	try {
		boost::program_options::store(
			boost::program_options::command_line_parser(argc, argv)
			.options(desc).positional(p).run(), vm);
	} catch (boost::program_options::error& e) {
		std::cerr << e.what() << std::endl;
		return 2;
	}
	boost::program_options::notify(vm);

	if (vm.count("help")) {
		std::cerr << "Usage " << argv[0] << " [OPTIONS] FILES\n"
			<< "Files ending with .xml are treated as corpora, otherwise \n"
			<< "as CCL files. Use - to read corpus from stdin (as with -I)";
		std::cout << desc << "\n";
		return 1;
	}

	foreach (const std::string& f, files) {
		if (f == "-") {
			corpus_stdin = true;
		} else if (boost::algorithm::ends_with(f, ".xml")) {
			corpora_files.push_back(f);
		} else {
			ccl_files.push_back(f);
		}
	}

	try {
		const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_load);
		Runner runner(tagset);
		foreach (const std::string& f, ccl_files) {
			size_t sz = runner.operators().size();
			if (!runner.load_more_operators(f)) {
				std::cerr << "Warning: error while parsing " << f << "\n";
			}
			if (runner.operators().size() == sz) {
				std::cerr << "Warning: no operators loaded from " << f << "\n";
			}
		}
		if (!runner.operators().empty()) {
			foreach (const std::string& f, corpora_files) {
				std::ifstream ifs(f.c_str());
				if (ifs.good()) {
					runner.do_stream(ifs, first);
				} else {
					std::cerr << "Error reading corpus from " << f << "\n";
				}
			}
			if (corpus_stdin) {
				runner.do_stream(std::cin, first);
			}
		}
	} catch (PwrNlp::PwrNlpError& e) {
		std::cerr << e.info() << std::endl;
		return 2;
	}

	return 0;
}