Fix some edge-cases in the thesarus code

Handle missing parts-of-speach and trailing pipes on lines.

aegisub/libaegisub/common/thesaurus.cpp

@@ -55,15 +55,15 @@ Thesaurus::Thesaurus(agi::fs::path const& dat_path, agi::fs::path const& idx_pat
 
 Thesaurus::~Thesaurus() { }
 
-void Thesaurus::Lookup(std::string const& word, std::vector<Entry> *out) {
-	out->clear();
-	if (!dat.get()) return;
+std::vector<Thesaurus::Entry> Thesaurus::Lookup(std::string const& word) {
+	std::vector<Entry> out;
+	if (!dat.get()) return out;
 
-	std::map<std::string, int>::const_iterator it = offsets.find(word);
-	if (it == offsets.end()) return;
+	auto it = offsets.find(word);
+	if (it == offsets.end()) return out;
 
 	dat->seekg(it->second, std::ios::beg);
-	if (!dat->good()) return;
+	if (!dat->good()) return out;
 
 	// First line is the word and meaning count
 	std::string temp;
@@ -71,22 +71,36 @@ void Thesaurus::Lookup(std::string const& word, std::vector<Entry> *out) {
 	std::vector<std::string> header;
 	std::string converted(conv->Convert(temp));
 	boost::split(header, converted, _1 == '|');
-	if (header.size() != 2) return;
+	if (header.size() != 2) return out;
 	int meanings = atoi(header[1].c_str());
 
-	out->resize(meanings);
+	out.reserve(meanings);
 	for (int i = 0; i < meanings; ++i) {
-		std::vector<std::string> line;
 		getline(*dat, temp);
-		std::string converted(conv->Convert(temp));
+		auto converted = conv->Convert(temp);
+		std::vector<std::string> line;
 		boost::split(line, converted, _1 == '|');
 
-		// The "definition" is just the part of speech plus the word it's
-		// giving synonyms for (which may not be the passed word)
-		(*out)[i].first = line[0] + ' ' + line[1];
-		(*out)[i].second.reserve(line.size() - 2);
-		copy(line.begin() + 2, line.end(), back_inserter((*out)[i].second));
+		if (line.size() < 2)
+			continue;
+
+		Entry e;
+		// The "definition" is just the part of speech (which may be empty)
+		// plus the word it's giving synonyms for (which may not be the passed word)
+		if (!line[0].empty())
+			e.first = line[0] + ' ';
+		e.first += line[1];
+		e.second.reserve(line.size() - 2);
+
+		for (size_t i = 2; i < line.size(); ++i) {
+			if (line[i].size())
+				e.second.emplace_back(std::move(line[i]));
+		}
+
+		out.emplace_back(std::move(e));
 	}
+
+	return out;
 }
 
 }

aegisub/libaegisub/include/libaegisub/thesaurus.h

@@ -38,7 +38,7 @@ class Thesaurus {
 
 public:
 	/// A pair of a word and synonyms for that word
-	typedef std::pair<std::string, std::vector<std::string> > Entry;
+	typedef std::pair<std::string, std::vector<std::string>> Entry;
 
 	/// Constructor
 	/// @param dat_path Path to data file
@@ -48,8 +48,7 @@ public:
 
 	/// Look up synonyms for a word
 	/// @param word Word to look up
-	/// @param[out] out Vector to fill with word/synonym lists
-	void Lookup(std::string const& word, std::vector<Entry> *out);
+	std::vector<Entry> Lookup(std::string const& word);
 };
 
 }

aegisub/src/subs_edit_ctrl.cpp

@@ -387,8 +387,7 @@ void SubsTextEditCtrl::AddThesaurusEntries(wxMenu &menu) {
 	if (!thesaurus)
 		thesaurus = agi::util::make_unique<Thesaurus>();
 
-	std::vector<Thesaurus::Entry> results;
-	thesaurus->Lookup(currentWord, &results);
+	auto results = thesaurus->Lookup(currentWord);
 
 	thesSugs.clear();
 
@@ -398,7 +397,7 @@ void SubsTextEditCtrl::AddThesaurusEntries(wxMenu &menu) {
 		int curThesEntry = 0;
 		for (auto const& result : results) {
 			// Single word, insert directly
-			if (result.second.size() == 1) {
+			if (result.second.empty()) {
 				thesMenu->Append(EDIT_MENU_THESAURUS_SUGS+curThesEntry, to_wx(result.first));
 				thesSugs.push_back(result.first);
 				++curThesEntry;

aegisub/src/thesaurus.cpp

@@ -44,10 +44,10 @@ Thesaurus::~Thesaurus() {
 	// Explicit empty destructor needed for scoped_ptr with incomplete types
 }
 
-void Thesaurus::Lookup(std::string word, std::vector<Entry> *result) {
-	if (!impl.get()) return;
+std::vector<Thesaurus::Entry> Thesaurus::Lookup(std::string word) {
+	if (!impl.get()) return std::vector<Entry>();
 	boost::to_lower(word);
-	impl->Lookup(word, result);
+	return impl->Lookup(word);
 }
 
 std::vector<std::string> Thesaurus::GetLanguageList() const {

aegisub/src/thesaurus.h

@@ -53,8 +53,7 @@ public:
 
 	/// Get a list of synonyms for a word, grouped by possible meanings of the word
 	/// @param word Word to get synonyms for
-	/// @param[out] result Output list
-	void Lookup(std::string word, std::vector<Entry> *result);
+	std::vector<Entry> Lookup(std::string word);
 
 	/// Get a list of language codes which thesauri are available for
 	std::vector<std::string> GetLanguageList() const;