add --progress/-p in wccl-run, clean up progress output in wccl-match and wccl-rules

wccl-apps/wccl-match.cpp

@@ -17,7 +17,6 @@
 
 namespace {
 	bool quiet = false;
-	bool progress = false;
 
 	struct options {
 		bool first;
@@ -30,10 +29,18 @@ class MatchRunner
 {
 public:
 	MatchRunner(const Corpus2::Tagset& tagset)
-		: tagset_(tagset), parser_(tagset_)
+		: tagset_(tagset), parser_(tagset_), progress_(false)
 	{
 	}
 
+	void use_progress(bool use) {
+		progress_ = use;
+		if (use) {
+			Corpus2::TokenTimer& timer = Corpus2::global_timer();
+			timer.register_signal_handler();
+		}
+	}
+
 	bool load_more_rules(const std::string &filename);
 
 	bool load_operator_string(const std::string &op_string);
@@ -50,6 +57,7 @@ private:
 	Wccl::Parser parser_;
 	std::vector<std::string> rule_names_;
 	std::vector<boost::shared_ptr<Wccl::MatchRule> > rules_;
+	bool progress_;
 };
 
 bool MatchRunner::load_more_rules(const std::string& filename)
@@ -95,16 +103,13 @@ void MatchRunner::apply_rules(boost::shared_ptr<Corpus2::TokenReader> reader,
 			}
 
 			timer.count_sentence(*as);
-			if (progress) {
+			if (progress_) {
 				timer.check_slice();
 			}
 			//writer->write_sentence(*as);
 		}
 		writer->write_chunk(*c);
 	}
-	if (progress) {
-		timer.stats();
-	}
 }
 
 void usage(char* name)
@@ -121,6 +126,7 @@ int main(int argc, char** argv)
 	std::string tagset_load = "kipi";
 	std::string input_format;
 	std::string output_format;
+	bool progress = false;
 	options opts;
 	opts.first = false;
 	opts.until_done = false;
@@ -203,6 +209,7 @@ int main(int argc, char** argv)
 	try {
 		const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_load);
 		MatchRunner runner(tagset);
+		runner.use_progress(progress);
 		foreach (const std::string& file, ccl_files) {
 			runner.load_more_rules(file);
 		}
@@ -222,6 +229,10 @@ int main(int argc, char** argv)
 				reader->set_option("ann");
 				runner.apply_rules(reader, writer);
 			}
+			if (progress) {
+				Corpus2::TokenTimer& timer = Corpus2::global_timer();
+				timer.stats();
+			}
 		}
 	} catch (PwrNlp::PwrNlpError& e) {
 		std::cerr << e.info() << std::endl;

wccl-apps/wccl-rules.cpp

@@ -229,6 +229,10 @@ int main(int argc, char** argv)
 				reader->set_option("ann");
 				apply_rules(reader, writer, rules, opts);
 			}
+			if (progress) {
+				Corpus2::TokenTimer& timer = Corpus2::global_timer();
+				timer.stats();
+			}
 		}
 	} catch (PwrNlp::PwrNlpError& e) {
 		std::cerr << e.info() << std::endl;

wccl-apps/wccl-run.cpp

@@ -7,6 +7,7 @@
 #include <libwccl/values/strset.h>
 #include <libwccl/parser/Parser.h>
 #include <libcorpus2/tagsetmanager.h>
+#include <libcorpus2/util/tokentimer.h>
 
 #include <boost/bind.hpp>
 #include <boost/algorithm/string.hpp>
@@ -43,10 +44,18 @@ class Runner
 {
 public:
 	Runner(const Corpus2::Tagset& tagset)
-	 : tagset_(tagset), parser_(tagset_), token_idx(0)
+	 : tagset_(tagset), parser_(tagset_), token_idx(0), progress_(false)
 	{
 	}
 
+	void use_progress(bool use) {
+		progress_ = use;
+		if (use) {
+			Corpus2::TokenTimer& timer = Corpus2::global_timer();
+			timer.register_signal_handler();
+		}
+	}
+
 	bool load_more_operators(const std::string &filename);
 
 	bool load_operator_string(const std::string &op_string);
@@ -66,6 +75,7 @@ private:
 	std::vector< boost::shared_ptr<Wccl::FunctionalOperator> > ops_;
 	std::vector< std::string > op_names_;
 	int token_idx;
+	bool progress_;
 };
 
 bool Runner::load_more_operators(const std::string& filename)
@@ -219,8 +229,13 @@ void Runner::do_stream(std::istream& is, bool first)
 {
 	Corpus2::XcesReader xr(tagset_, is);
 	Corpus2::Sentence::Ptr s;
+	Corpus2::TokenTimer& timer = Corpus2::global_timer();
 	while ((s = xr.get_next_sentence())) {
 		do_sentence(s);
+		timer.count_sentence(*s);
+		if (progress_) {
+			timer.check_slice();
+		}
 		std::cout << "\n";
 		if (first) break;
 	}
@@ -229,7 +244,7 @@ void Runner::do_stream(std::istream& is, bool first)
 int main(int argc, char** argv)
 {
 	std::string tagset_load = "kipi";
-	bool first = false;
+	bool first = false, progress = false;
 	std::vector<std::string> corpora_files, files, operator_strings;
 	bool corpus_stdin = false;
 	using boost::program_options::value;
@@ -262,6 +277,8 @@ int main(int argc, char** argv)
 			 "Output operator variables")
 			("output-header,H", value(&output_header),
 			 "Output table header")
+			("progress,p", value(&progress)->zero_tokens(),
+			 "Show progress info")
 			("help,h", "Show help")
 			;
 	boost::program_options::variables_map vm;
@@ -303,6 +320,7 @@ int main(int argc, char** argv)
 	try {
 		const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_load);
 		Runner runner(tagset);
+		runner.use_progress(progress);
 		foreach (const std::string& f, operator_strings) {
 			if (boost::algorithm::ends_with(f, ".ccl")) {
 				size_t sz = runner.operators().size();
@@ -328,6 +346,10 @@ int main(int argc, char** argv)
 			if (corpus_stdin) {
 				runner.do_stream(std::cin, first);
 			}
+			if (progress) {
+				Corpus2::TokenTimer& timer = Corpus2::global_timer();
+				timer.stats();
+			}
 		}
 	} catch (PwrNlp::PwrNlpError& e) {
 		std::cerr << e.info() << std::endl;