New divvun-suggest mode -a/--autocorrect to auto-apply corrections

Implements #37 - Autocorrect mode for divvun-checker

Author: unhammer

src/main_suggest.cpp

@@ -31,6 +31,7 @@ int main(int argc, char ** argv)
 
 		options.add_options()
 			("j,json", "Use JSON output format (default: CG)")
+			("a,autocorrect", "Use Autocorrect output format (default: CG)")
 			("g,generator", "Generator (HFSTOL format)", cxxopts::value<std::string>(), "BIN")
 #ifdef HAVE_LIBPUGIXML
 			("m,messages", "ERROR messages (XML format)", cxxopts::value<std::string>(), "FILE")
@@ -84,7 +85,17 @@ int main(int argc, char ** argv)
 		}
 
 		const auto& genfile = options["generator"].as<std::string>();
-		bool json = options.count("j");
+		divvun::RunMode mode = divvun::RunCG;
+		if(options.count("j")) {
+			mode = divvun::RunJson;
+			if(options.count("a")) {
+			  std::cerr << argv[0] << " ERROR: Pick just one of --json/--autocorrect" << std::endl;
+                          return (EXIT_FAILURE);
+                        }
+                };
+		if(options.count("a")) {
+			mode = divvun::RunAutoCorrect;
+		};
 		bool genall = options.count("A");
 		bool verbose = options.count("v");
 
@@ -123,7 +134,7 @@ int main(int argc, char ** argv)
 #endif
 
 		divvun::Suggest s(t, m, locale, verbose, genall);
-		s.run(std::cin, std::cout, json);
+		s.run(std::cin, std::cout, mode);
 	}
 	catch (const cxxopts::OptionException& e)
 	{

src/pipeline.cpp

@@ -130,7 +130,7 @@ SuggestCmd::SuggestCmd (const string& gen_path, const string& msg_path, const st
 }
 void SuggestCmd::run(stringstream& input, stringstream& output) const
 {
-	suggest->run(input, output, true);
+	suggest->run(input, output, RunJson);
 }
 vector<Err> SuggestCmd::run_errs(stringstream& input) const
 {

src/suggest.cpp

@@ -879,6 +879,39 @@ RunState Suggest::run_json(std::istream& is, std::ostream& os)
 	return sentence.runstate;
 }
 
+RunState Suggest::run_autocorrect(std::istream& is, std::ostream& os)
+{
+	json::sanity_test();
+	Sentence sentence = run_sentence(is, *generator, msgs, generate_all_readings);
+	vector<Err> errs = mk_errs(sentence);
+
+	size_t offset = 0;
+	u16string text = fromUtf8(sentence.text.str());
+	for(const auto& e : errs) {
+		if(e.beg > offset) {
+			os << toUtf8(text.substr(offset, e.beg - offset));
+		}
+		bool printed = false;
+		for(const auto& r : e.rep) {
+			os << toUtf8(r);
+			printed = true;
+			break;
+		}
+		if(!printed) {
+			os << toUtf8(e.form);
+		}
+		offset = e.end;
+	}
+	os << toUtf8(text.substr(offset));
+
+	if(sentence.runstate == flushing) {
+		os << '\0';
+		os.flush();
+		os.clear();
+	}
+	return sentence.runstate;
+}
+
 
 void print_cg_reading(const Casing& inputCasing, const string& readinglines, std::ostream& os, const hfst::HfstTransducer& t, bool generate_all_readings) {
 	os << readinglines;
@@ -938,7 +971,7 @@ void run_cg(std::istream& is, std::ostream& os, const hfst::HfstTransducer& t, b
 	}
 }
 
-void Suggest::run(std::istream& is, std::ostream& os, bool json)
+void Suggest::run(std::istream& is, std::ostream& os, RunMode mode)
 {
 	try {
 		auto _old = std::locale::global(std::locale(""));
@@ -947,11 +980,17 @@ void Suggest::run(std::istream& is, std::ostream& os, bool json)
 	{
 		std::cerr << "WARNING: Couldn't set global locale \"\" (locale-specific native environment): " << e.what() << std::endl;
 	}
-	if(json) {
-		while(run_json(is, os) == flushing);
-	}
-	else {
-		run_cg(is, os, *generator, generate_all_readings); // ignores ignores
+	switch(mode) {
+		case RunJson:
+			while(run_json(is, os) == flushing);
+			break;
+		case RunAutoCorrect:
+			while(run_autocorrect(is, os) == flushing);
+			break;
+		case RunCG:
+		default:
+			run_cg(is, os, *generator, generate_all_readings); // ignores ignores
+			break;
 	}
 }
 

src/suggest.hpp

@@ -73,6 +73,12 @@ enum RunState {
 	eof
 };
 
+enum RunMode {
+	RunCG,
+	RunJson,
+	RunAutoCorrect
+};
+
 using rel_id = size_t;
 using relations = std::unordered_map<string, rel_id>;
 
@@ -187,7 +193,7 @@ class Suggest {
 		Suggest(const string& gen_path, const string& locale, bool verbose);
 		~Suggest() = default;
 
-		void run(std::istream& is, std::ostream& os, bool json);
+		void run(std::istream& is, std::ostream& os, RunMode mode);
 
 		vector<Err> run_errs(std::istream& is);
 		void setIgnores(const std::set<ErrId>& ignores);
@@ -200,6 +206,7 @@ class Suggest {
 	private:
 		const SortedMsgLangs sortedmsglangs; // invariant: contains all and only the keys of msgs
 		RunState run_json(std::istream& is, std::ostream& os);
+		RunState run_autocorrect(std::istream& is, std::ostream& os);
 		std::unique_ptr<const hfst::HfstTransducer> generator;
 		std::set<ErrId> ignores;
 		bool generate_all_readings = false;