Branch Coverage

nametag/nametag.cpp

Criterion	Covered	Total	%
branch	1235	6846	18.0

line	true	false	branch
179	4	0	if (bilou[bilou_type_O].probability > bilou[best].probability) best = bilou_type_O;
180	2	2	if (bilou[bilou_type_U].probability > bilou[best].probability) best = bilou_type_U;
187	10	0	if (prev.bilou[bilou_type_O].probability > best_LOU_prob) {
191	6	4	if (prev.bilou[bilou_type_U].probability > best_LOU_prob) {
199	0	10	if (prev.bilou[bilou_type_I].probability > best_BI_prob) {
205	0	10	if (best_BI_prob > best_LOU_prob) {
222	0	10	if (bilou[bilou_type_I].probability > bilou[best].probability) best = bilou_type_I;
227	0	10	if (bilou[bilou_type_L].probability > bilou[best].probability) best = bilou_type_L;
232	10	0	if (bilou[bilou_type_O].probability > bilou[best].probability) best = bilou_type_O;
237	4	6	if (bilou[bilou_type_U].probability > bilou[best].probability) best = bilou_type_U;
353	2	2	if (words.size() < size) words.resize(size);
354	2	2	if (features.size() < size) features.resize(size);
355	2	2	if (probabilities.size() < size) probabilities.resize(size);
356	2	2	if (previous_stage.size() < size) previous_stage.resize(size);
360	4	14	for (unsigned i = 0; i < size; i++)
	0	0	for (unsigned i = 0; i < size; i++)
365	14	4	for (unsigned i = 0; i < size; i++)
	0	0	for (unsigned i = 0; i < size; i++)
370	7	2	for (unsigned i = 0; i < size; i++) {
	0	0	for (unsigned i = 0; i < size; i++) {
380	4	0	if (last_bilou[bilou_type_O].probability > last_bilou[best].probability) best = bilou_type_O;
381	0	4	if (last_bilou[bilou_type_U].probability > last_bilou[best].probability) best = bilou_type_U;
385	10	4	for (unsigned i = size - 1; i; i--) {
392	14	4	for (unsigned i = 0; i < size; i++) {
462	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
	0	0	explicit binary_decoder_error(const char* description) : runtime_error(description) {}
465	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
	0	0	class binary_decoder {
490	7	0	buffer.resize(len);
498	0	313	if (data + 1 > data_end) throw binary_decoder_error("No more data in binary_decoder");
503	0	679	if (data + sizeof(uint16_t) > data_end) throw binary_decoder_error("No more data in binary_decoder");
511	0	516	if (data + sizeof(uint32_t) > data_end) throw binary_decoder_error("No more data in binary_decoder");
520	0	46	if (len == 255) len = next_4B();
525	0	116	if (data + sizeof(T) * elements > data_end) throw binary_decoder_error("No more data in binary_decoder");
	0	253	if (data + sizeof(T) * elements > data_end) throw binary_decoder_error("No more data in binary_decoder");
	0	2	if (data + sizeof(T) * elements > data_end) throw binary_decoder_error("No more data in binary_decoder");
	0	118	if (data + sizeof(T) * elements > data_end) throw binary_decoder_error("No more data in binary_decoder");
540	0	1	if (pos > buffer.size()) throw binary_decoder_error("Cannot seek past end of binary_decoder");
576	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
	0	0	return a.len == b.len && memcmp(a.str, b.str, a.len) == 0;
630	0	0	if (uint8_t(val) != val) runtime_failure("Should encode value " << val << " in one byte!");
654	0	0	if (!(str.len < 255)) add_4B(str.len);
687	0	0	class network_classifier {
	4	2	class network_classifier {
	0	0	class network_classifier {
	0	0	class network_classifier {
814	16	10	while (size) {
	0	0	while (size) {
	0	0	while (size) {
816	10	6	if (unaligned_load(first + step) < val) {
	0	0	if (unaligned_load(first + step) < val) {
	0	0	if (unaligned_load(first + step) < val) {
857	2	0	if (!compressor::load(is, data)) return false;
	2	0	if (!compressor::load(is, data)) return false;
861	2	0	load_matrix(data, indices);
862	2	0	missing_weight = unaligned_load(data.next(1));
863	2	0	load_matrix(data, weights);
868	2	0	hidden_layer.resize(data.next_2B());
	2	0	hidden_layer.resize(data.next_2B());
869	0	2	if (!hidden_layer.empty()) {
870	0	0	load_matrix(data, hidden_weights[0]);
871	0	0	load_matrix(data, hidden_weights[1]);
875	2	0	unsigned outcomes = data.next_2B();
876	2	0	output_layer.resize(outcomes);
877	2	0	output_error.resize(outcomes);
	0	0	output_error.resize(outcomes);
888	330	2	for (auto&& row : m) {
	330	2	for (auto&& row : m) {
898	0	0	if (features <= 0) { if (verbose) cerr << "There must be more than zero features!" << endl; return false; }
	0	0	if (features <= 0) { if (verbose) cerr << "There must be more than zero features!" << endl; return false; }
899	0	0	if (outcomes <= 0) { if (verbose) cerr << "There must be more than zero features!" << endl; return false; }
	0	0	if (outcomes <= 0) { if (verbose) cerr << "There must be more than zero features!" << endl; return false; }
900	0	0	if (train.empty()) { if (verbose) cerr << "No training data!" << endl; return false; }
	0	0	if (train.empty()) { if (verbose) cerr << "No training data!" << endl; return false; }
901	0	0	for (auto&& instance : train) {
902	0	0	if (instance.outcome >= outcomes) { if (verbose) cerr << "Training instances out of range!" << endl; return false; }
	0	0	if (instance.outcome >= outcomes) { if (verbose) cerr << "Training instances out of range!" << endl; return false; }
903	0	0	for(auto& feature : instance.features)
904	0	0	if (feature >= features) { if (verbose) cerr << "Training instances out of range!" << endl; return false; }
	0	0	if (feature >= features) { if (verbose) cerr << "Training instances out of range!" << endl; return false; }
906	0	0	for (auto&& instance : heldout)
907	0	0	for(auto& feature : instance.features)
908	0	0	if (feature >= features) { if (verbose) cerr << "Heldout instances out of range!" << endl; return false; }
	0	0	if (feature >= features) { if (verbose) cerr << "Heldout instances out of range!" << endl; return false; }
916	0	0	for (auto&& instance : train)
917	0	0	for (auto&& feature : instance.features)
920	0	0	for (auto&& row : indices) {
927	0	0	for (auto&& row : indices)
933	0	0	if (!hidden_layer.empty()) {
937	0	0	for (auto&& row : hidden_weights[0])
938	0	0	for (auto&& weight : row.resize(hidden_layer.size()), row)
942	0	0	for (auto&& row : hidden_weights[1])
943	0	0	for (auto&& weight : row.resize(outcomes), row)
956	0	0	for (unsigned i = 0; i < train.size(); i++)
959	0	0	for (int iteration = 0; iteration < parameters.iterations; iteration++) {
960	0	0	if (verbose) cerr << "Iteration " << iteration + 1 << ": ";
	0	0	if (verbose) cerr << "Iteration " << iteration + 1 << ": ";
	0	0	if (verbose) cerr << "Iteration " << iteration + 1 << ": ";
962	0	0	double learning_rate = parameters.final_learning_rate && parameters.iterations > 1 ?
964	0	0	parameters.initial_learning_rate;
970	0	0	for (auto&& train_index : permutation) {
981	0	0	if (verbose)
985	0	0	<< "%, ";
988	0	0	if (!heldout.empty()) {
990	0	0	for (auto&& instance : heldout) {
994	0	0	if (verbose) cerr << "heldout acc " << heldout_correct * 100. / heldout.size() << ", ";
	0	0	if (verbose) cerr << "heldout acc " << heldout_correct * 100. / heldout.size() << ", ";
996	0	0	if (verbose) cerr << "done." << endl;
1002	1	13	if (outcomes.size() != output_layer.size()) outcomes.resize(output_layer.size());
1003	0	14	if (buffer.size() != hidden_layer.size()) buffer.resize(hidden_layer.size());
1010	0	0	propagate(features, hidden_layer, output_layer);
	0	0	propagate(features, hidden_layer, output_layer);
1017	291	14	for (auto&& feature : features)
1018	291	0	if (feature < indices.size())
1019	414	291	for (unsigned i = 0; i < indices[feature].size(); i++)
1023	0	14	if (!hidden_layer.empty()) {
1024	0	0	for (auto&& weight : hidden_layer)
1028	0	0	for (auto&& feature : features)
1029	0	0	if (feature < hidden_weights[0].size())
1030	0	0	for (unsigned i = 0; i < hidden_layer.size(); i++) {
1035	0	0	for (auto&& weight : hidden_layer)
1039	0	0	for (unsigned h = 0; h < hidden_layer.size(); h++)
1040	0	0	for (unsigned i = 0; i < output_layer.size(); i++)
1046	126	14	for (unsigned i = 0; i < output_layer.size(); sum += output_layer[i], i++)
1049	126	14	for (unsigned i = 0; i < output_layer.size(); i++)
1055	0	0	for (unsigned i = 1; i < output_layer.size(); i++)
	0	0	for (unsigned i = 1; i < output_layer.size(); i++)
1056	0	0	if (output_layer[i] > output_layer[best])
	0	0	if (output_layer[i] > output_layer[best])
1064	0	0	for (unsigned i = 0; i < output_error.size(); i++)
1065	0	0	output_error[i] = (i == instance.outcome) - output_layer[i];
1068	0	0	for (auto&& feature : instance.features)
1069	0	0	for (unsigned i = 0; i < indices[feature].size(); i++)
1073	0	0	if (!hidden_layer.empty()) {
1075	0	0	for (unsigned h = 0; h < hidden_layer.size(); h++) {
1077	0	0	for (unsigned i = 0; i < output_layer.size(); i++)
1083	0	0	for (unsigned h = 0; h < hidden_layer.size(); h++)
1084	0	0	for (unsigned i = 0; i < output_layer.size(); i++)
1088	0	0	for (auto&& feature : instance.features)
1089	0	0	for (unsigned i = 0; i < hidden_layer.size(); i++)
1218	0	0	if (str == "trivial") return id = TRIVIAL, true;
1219	0	0	if (str == "external") return id = EXTERNAL, true;
1220	0	0	if (str == "morphodita") return id = MORPHODITA, true;
1310	4	95	if (it == map.end() && total_features) {
	4	0	if (it == map.end() && total_features) {
	0	99	if (it == map.end() && total_features) {
1314	95	4	return it != map.end() ? it->second : ner_feature_unknown;
1342	0	0	if (window < 0) return false;
1343	0	0	if (!total_features) return false;
1348	0	0	lookup(string(), total_features); // Always add an empty string to the map
1358	35	8	for (unsigned i = data.next_4B(); i > 0; i--) {
1360	35	0	data.next_str(key);
1361	35	0	map.emplace(key, data.next_4B());
1373	0	0	for (auto&& element : map_elements) {
1374	0	0	enc.add_str(element.first);
1447	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	9	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	4	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	16	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	3	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	129	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	4	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	4	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	4	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	4	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	14	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	38	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	54	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
	0	0	return chr < CHARS ? 1 << category_block[category_index[chr >> 8]][chr & 0xFF] : DEFAULT_CAT;
1451	4	0	if (chr < CHARS) {
1453	4	0	if ((othercase & 0xFF) == othercase_type::LOWER_ONLY) return othercase >> 8;
1454	0	0	if ((othercase & 0xFF) == othercase_type::LOWER_THEN_UPPER) return othercase >> 8;
1455	0	0	if ((othercase & 0xFF) == othercase_type::TITLE_THEN_LOWER) return othercase_block[othercase_index[(othercase >> 8) >> 8]][(othercase >> 8) & 0xFF] >> 8;
1461	0	0	if (chr < CHARS) {
1463	0	0	if ((othercase & 0xFF) == othercase_type::UPPERTITLE_ONLY) return othercase >> 8;
1464	0	0	if ((othercase & 0xFF) == othercase_type::UPPER_ONLY) return othercase >> 8;
1465	0	0	if ((othercase & 0xFF) == othercase_type::UPPER_THEN_TITLE) return othercase >> 8;
1466	0	0	if ((othercase & 0xFF) == othercase_type::LOWER_THEN_UPPER) return othercase_block[othercase_index[(othercase >> 8) >> 8]][(othercase >> 8) & 0xFF] >> 8;
1560	58	10	if (((unsigned char)*str) < 0x80) return (unsigned char)*str++;
1561	0	10	else if (((unsigned char)*str) < 0xC0) return ++str, REPLACEMENT_CHAR;
1562	10	0	else if (((unsigned char)*str) < 0xE0) {
1564	10	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	10	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1566	0	0	} else if (((unsigned char)*str) < 0xF0) {
1568	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1570	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1572	0	0	} else if (((unsigned char)*str) < 0xF8) {
1574	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1576	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1578	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1584	218	0	if (!len) return 0;
1586	195	23	if (((unsigned char)*str) < 0x80) return (unsigned char)*str++;
1587	0	23	else if (((unsigned char)*str) < 0xC0) return ++str, REPLACEMENT_CHAR;
1588	23	0	else if (((unsigned char)*str) < 0xE0) {
1590	23	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	23	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	23	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1592	0	0	} else if (((unsigned char)*str) < 0xF0) {
1594	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1596	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1598	0	0	} else if (((unsigned char)*str) < 0xF8) {
1600	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1602	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1604	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
	0	0	if (len <= 0 || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return REPLACEMENT_CHAR;
1629	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
	0	0	iterator& operator++() { if (next) { codepoint = decode(next); if (!codepoint) next = nullptr; } return *this; }
1695	4	0	if (chr < 0x80) str += chr;
1696	0	0	else if (chr < 0x800) { str += 0xC0 + (chr >> 6); str += 0x80 + (chr & 0x3F); }
1697	0	0	else if (chr < 0x10000) { str += 0xE0 + (chr >> 12); str += 0x80 + ((chr >> 6) & 0x3F); str += 0x80 + (chr & 0x3F); }
1698	0	0	else if (chr < 0x200000) { str += 0xF0 + (chr >> 18); str += 0x80 + ((chr >> 12) & 0x3F); str += 0x80 + ((chr >> 6) & 0x3F); str += 0x80 + (chr & 0x3F); }
1705	0	0	for (char32_t chr; (chr = decode(str)); )
1712	0	0	while (len)
1758	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
1763	0	0	if (str.len && (str.str[0] == '+' || str.str[0] == '-')) {
	0	0	if (str.len && (str.str[0] == '+' || str.str[0] == '-')) {
1769	0	0	if (!str.len) return error.assign("Cannot parse ").append(value_name).append(" int value '").append(original.str, original.len).append("': empty string."), false;
1773	0	0	while (str.len && str.str[0] >= '0' && str.str[0] <= '9') {
	0	0	while (str.len && str.str[0] >= '0' && str.str[0] <= '9') {
	0	0	while (str.len && str.str[0] >= '0' && str.str[0] <= '9') {
1774	0	0	if (positive) {
1775	0	0	if (value > (numeric_limits::max() - (str.str[0] - '0')) / 10)
1779	0	0	if (value < (numeric_limits::min() + (str.str[0] - '0')) / 10)
1787	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
	0	0	while (str.len && (str.str[0] == ' ' || str.str[0] == '\f' || str.str[0] == '\n' || str.str[0] == '\r' || str.str[0] == '\t' || str.str[0] == '\v'))
1791	0	0	if (str.len) return error.assign("Cannot parse ").append(value_name).append(" int value '").append(original.str, original.len).append("': non-digit character found."), false;
1800	0	0	if (!parse_int(str, value_name, result, error))
	0	0	if (!parse_int(str, value_name, result, error))
1935	0	0	if (text.empty()) return;
1938	0	0	for (string::size_type next; (next = text.find(sep, index)) != string::npos; index = next + 1)
2030	0	0	if (!feature_processor::parse(window, args, entities, total_features, pipeline)) return false;
2031	0	0	if (args.size() < 1) return cerr << "BrownCluster requires a cluster file as the first argument!" << endl, false;
2034	0	0	if (!in.is_open()) return cerr << "Cannot open Brown clusters file '" << args[0] << "'!" << endl, false;
2037	0	0	substrings.emplace_back(string::npos);
2038	0	0	for (unsigned i = 1; i < args.size(); i++) {
2039	0	0	int len = parse_int(args[i].c_str(), "BrownCluster_prefix_length");
2040	0	0	if (len <= 0)
2041	0	0	return cerr << "Wrong prefix length '" << len << "' in BrownCluster specification!" << endl, false;
2043	0	0	substrings.emplace_back(len);
2051	0	0	while (getline(in, line)) {
	0	0	while (getline(in, line)) {
2052	0	0	split(line, '\t', tokens);
2053	0	0	if (tokens.size() != 2) return cerr << "Wrong line '" << line << "' in Brown cluster file '" << args[0] << "'!" << endl, false;
2057	0	0	if (it == cluster_map.end()) {
2059	0	0	clusters.emplace_back();
2060	0	0	for (auto&& substring : substrings)
2061	0	0	if (substring == string::npos || substring < cluster.size())
	0	0	if (substring == string::npos || substring < cluster.size())
	0	0	if (substring == string::npos || substring < cluster.size())
2062	0	0	clusters.back().emplace_back(prefixes_map.emplace(cluster.substr(0, substring), *total_features + (2*window + 1) * (int)prefixes_map.size() + window).first->second);
	0	0	clusters.back().emplace_back(prefixes_map.emplace(cluster.substr(0, substring), *total_features + (2*window + 1) * (int)prefixes_map.size() + window).first->second);
2065	0	0	if (!map.emplace(form, it->second).second) return cerr << "Form '" << form << "' is present twice in Brown cluster file '" << args[0] << "'!" << endl, false;
2076	0	0	for (auto&& cluster : clusters) {
2078	0	0	for (auto&& feature : cluster)
2087	0	0	for (auto&& cluster : clusters) {
2089	0	0	for (auto&& feature : cluster)
2095	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2097	0	0	if (it != map.end()) {
2099	0	0	for (auto&& feature : cluster)
2100	0	0	apply_in_window(i, feature);
	0	0	apply_in_window(i, feature);
2113	0	0	if (window) return cerr << "CzechAddContainers cannot have non-zero window!" << endl, false;
2121	0	0	for (unsigned i = 0; i < entities.size(); i++) {
2123	0	0	if (entities[i].type.compare("pf") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("pf") != 0)) {
	0	0	if (entities[i].type.compare("pf") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("pf") != 0)) {
	0	0	if (entities[i].type.compare("pf") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("pf") != 0)) {
	0	0	if (entities[i].type.compare("pf") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("pf") != 0)) {
	0	0	if (entities[i].type.compare("pf") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("pf") != 0)) {
2125	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("pf") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("pf") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("pf") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("pf") == 0) j++;
2126	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) {
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) {
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) {
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) {
2128	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) j++;
	0	0	while (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ps") == 0) j++;
2134	0	0	if (entities[i].type.compare("td") == 0 && i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("tm") == 0) {
	0	0	if (entities[i].type.compare("td") == 0 && i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("tm") == 0) {
	0	0	if (entities[i].type.compare("td") == 0 && i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("tm") == 0) {
	0	0	if (entities[i].type.compare("td") == 0 && i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("tm") == 0) {
	0	0	if (entities[i].type.compare("td") == 0 && i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("tm") == 0) {
2136	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ty") == 0) j++;
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ty") == 0) j++;
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ty") == 0) j++;
	0	0	if (j < entities.size() && entities[j].start == entities[j-1].start + entities[j-1].length && entities[j].type.compare("ty") == 0) j++;
2140	0	0	if (entities[i].type.compare("tm") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("td") != 0))
	0	0	if (entities[i].type.compare("tm") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("td") != 0))
	0	0	if (entities[i].type.compare("tm") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("td") != 0))
	0	0	if (entities[i].type.compare("tm") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("td") != 0))
	0	0	if (entities[i].type.compare("tm") == 0 && (!i || entities[i-1].start + entities[i-1].length < entities[i].start || entities[i-1].type.compare("td") != 0))
2141	0	0	if (i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("ty") == 0)
	0	0	if (i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("ty") == 0)
	0	0	if (i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("ty") == 0)
	0	0	if (i+1 < entities.size() && entities[i+1].start == entities[i].start + entities[i].length && entities[i+1].type.compare("ty") == 0)
2147	0	0	if (buffer.size() > entities.size()) entities = buffer;
2159	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2160	0	0	for (unsigned pos = 0; pos + 2 < sentence.words[i].lemma_comments.size(); pos++)
2161	0	0	if (sentence.words[i].lemma_comments[pos] == '_' && sentence.words[i].lemma_comments[pos+1] == ';') {
	0	0	if (sentence.words[i].lemma_comments[pos] == '_' && sentence.words[i].lemma_comments[pos+1] == ';') {
	0	0	if (sentence.words[i].lemma_comments[pos] == '_' && sentence.words[i].lemma_comments[pos+1] == ';') {
2163	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
2173	14	4	for (unsigned i = 0; i < sentence.size; i++)
2174	12	2	apply_in_window(i, lookup(sentence.words[i].form, total_features));
	40	12	apply_in_window(i, lookup(sentence.words[i].form, total_features));
2176	4	0	apply_outer_words_in_window(lookup_empty());
	8	4	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
2190	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2195	0	0	for (bool first = true; (chr = utf8::decode(form)); first = false) {
2197	0	0	was_upper = was_upper || category & unicode::Lut;
	0	0	was_upper = was_upper || category & unicode::Lut;
2198	0	0	was_lower = was_lower || category & unicode::Ll;
	0	0	was_lower = was_lower || category & unicode::Ll;
2200	0	0	if (first && was_upper) apply_in_window(i, fst_cap);
	0	0	if (first && was_upper) apply_in_window(i, fst_cap);
	0	0	if (first && was_upper) apply_in_window(i, fst_cap);
2202	0	0	if (was_upper && !was_lower) apply_in_window(i, all_cap);
	0	0	if (was_upper && !was_lower) apply_in_window(i, all_cap);
	0	0	if (was_upper && !was_lower) apply_in_window(i, all_cap);
2203	0	0	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
	0	0	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
	0	0	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
2214	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2216	0	0	for (auto&& chr : utf8::decoder(sentence.words[i].form))
2217	0	0	utf8::append(buffer, buffer.empty() ? chr : unicode::lowercase(chr));
2218	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
2221	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
2234	0	0	if (!feature_processor::parse(window, args, entities, total_features, pipeline)) return false;
2237	0	0	for (auto&& arg : args) {
2238	0	0	ifstream in(path_from_utf8(arg).c_str());
2239	0	0	if (!in.is_open()) return cerr << "Cannot open gazetteers file '" << arg << "'!" << endl, false;
2245	0	0	while (getline(in, line)) {
	0	0	while (getline(in, line)) {
2246	0	0	split(line, ' ', tokens);
2247	0	0	for (unsigned i = 0; i < tokens.size(); i++)
2248	0	0	if (!tokens[i][0])
2250	0	0	if (tokens.size() > longest) longest = tokens.size();
2253	0	0	for (unsigned i = 0; i < tokens.size(); i++) {
2254	0	0	if (i) gazetteer += ' ';
2257	0	0	if (it->second == gazetteers_info.size()) gazetteers_info.emplace_back();
	0	0	if (it->second == gazetteers_info.size()) gazetteers_info.emplace_back();
2259	0	0	if (i + 1 < tokens.size())
2262	0	0	if (find(info.features.begin(), info.features.end(), *total_features + window) == info.features.end())
2263	0	0	info.features.emplace_back(*total_features + window);
2266	0	0	*total_features += (2*window + 1) * (longest == 0 ? 0 : longest == 1 ? U+1 : longest == 2 ? L+1 : I+1);
	0	0	*total_features += (2*window + 1) * (longest == 0 ? 0 : longest == 1 ? U+1 : longest == 2 ? L+1 : I+1);
	0	0	*total_features += (2*window + 1) * (longest == 0 ? 0 : longest == 1 ? U+1 : longest == 2 ? L+1 : I+1);
2276	0	0	for (auto&& gazetteer : gazetteers_info) {
2279	0	0	for (auto&& feature : gazetteer.features)
2288	0	0	for (auto&& gazetteer : gazetteers_info) {
2291	0	0	for (auto&& feature : gazetteer.features)
2297	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2299	0	0	if (it == map.end()) continue;
2302	0	0	for (auto&& feature : gazetteers_info[it->second].features) {
2303	0	0	apply_in_window(i, feature + G * (2*window + 1));
	0	0	apply_in_window(i, feature + G * (2*window + 1));
2304	0	0	apply_in_window(i, feature + U * (2*window + 1));
	0	0	apply_in_window(i, feature + U * (2*window + 1));
2307	0	0	for (unsigned j = i + 1; gazetteers_info[it->second].prefix_of_longer && j < sentence.size; j++) {
	0	0	for (unsigned j = i + 1; gazetteers_info[it->second].prefix_of_longer && j < sentence.size; j++) {
	0	0	for (unsigned j = i + 1; gazetteers_info[it->second].prefix_of_longer && j < sentence.size; j++) {
2308	0	0	if (j == i + 1) buffer.assign(sentence.words[i].raw_lemma);
2312	0	0	if (it == map.end()) break;
2315	0	0	for (auto&& feature : gazetteers_info[it->second].features)
2316	0	0	for (unsigned g = i; g <= j; g++) {
2317	0	0	apply_in_window(g, feature + G * (2*window + 1));
	0	0	apply_in_window(g, feature + G * (2*window + 1));
2318	0	0	apply_in_window(g, feature + (g == i ? B : g == j ? L : I) * (2*window + 1));
	0	0	apply_in_window(g, feature + (g == i ? B : g == j ? L : I) * (2*window + 1));
	0	0	apply_in_window(g, feature + (g == i ? B : g == j ? L : I) * (2*window + 1));
	0	0	apply_in_window(g, feature + (g == i ? B : g == j ? L : I) * (2*window + 1));
2339	0	0	if (!feature_processor::parse(window, args, entities, total_features, pipeline)) return false;
2344	0	0	if (args.size() < 4) return cerr << "Not enough parameters to GazetteersEnhanced!" << endl, false;
2345	0	0	if (args.size() & 1) return cerr << "Odd number of parameters to GazetteersEnhanced!" << endl, false;
2347	0	0	if (args[0] == "form") match = MATCH_FORM;
2348	0	0	else if (args[0] == "rawlemma") match = MATCH_RAWLEMMA;
2349	0	0	else if (args[0] == "rawlemmas") match = MATCH_RAWLEMMAS;
2352	0	0	if (args[1] == "embed_in_model") embed = EMBED_IN_MODEL;
2353	0	0	else if (args[1] == "out_of_model") embed = OUT_OF_MODEL;
2356	0	0	for (unsigned i = 2; i < args.size(); i += 2) {
2360	0	0	gazetteer_metas.back().entity = args[i + 1] == "NONE" ? -1 : entities.parse(args[i + 1].c_str(), true);
2364	0	0	for (entity_type i = 0; i < entities.size(); i++)
2367	0	0	if (!load_gazetteer_lists(pipeline, embed == EMBED_IN_MODEL)) return false;
2379	0	0	for (auto&& gazetteer_meta : gazetteer_metas) {
2386	0	0	for (auto&& gazetteer_list : gazetteer_lists) {
2388	0	0	for (auto&& gazetteer : gazetteer_list.gazetteers)
2396	0	0	for (auto&& entity : entity_list)
2408	0	0	for (auto&& gazetteer_meta : gazetteer_metas) {
2414	0	0	if (embed == EMBED_IN_MODEL) {
2416	0	0	for (auto&& gazetteer_list : gazetteer_lists) {
2418	0	0	for (auto&& gazetteer : gazetteer_list.gazetteers)
2429	0	0	for (auto&& entity : entity_list)
2435	0	0	vector> features(sentence.size);
2437	0	0	vector> recased_match_sources(sentence.size);
2438	0	0	for (unsigned i = 0; i < sentence.size; i++)
2439	0	0	recase_match_source(sentence.words[i], RECASE_ANY, recased_match_sources[i]);
2441	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2445	0	0	for (unsigned j = i; j < sentence.size && !nodes.empty(); j++) {
	0	0	for (unsigned j = i; j < sentence.size && !nodes.empty(); j++) {
	0	0	for (unsigned j = i; j < sentence.size && !nodes.empty(); j++) {
2447	0	0	for (auto&& node : nodes)
2448	0	0	if (!gazetteers_trie[node].children.empty())
2449	0	0	for (auto&& match_source : recased_match_sources[j]) {
2451	0	0	for (auto&& it = range.first; it != range.second; it++)
2452	0	0	append_unless_exists(new_nodes, it->second);
2455	0	0	hard_pre_possible = hard_pre_possible && !sentence.probabilities[j].local_filled;
	0	0	hard_pre_possible = hard_pre_possible && !sentence.probabilities[j].local_filled;
2456	0	0	if (hard_pre_possible)
2457	0	0	for (auto&& node : new_nodes)
2458	0	0	if (gazetteers_trie[node].mode == HARD_PRE &&
	0	0	if (gazetteers_trie[node].mode == HARD_PRE &&
	0	0	if (gazetteers_trie[node].mode == HARD_PRE &&
2459	0	0	((j - i + 1) > hard_pre_length || node < hard_pre_node))
2463	0	0	for (auto&& node : new_nodes)
2464	0	0	for (auto&& feature : gazetteers_trie[node].features)
2465	0	0	for (unsigned k = i; k <= j; k++) {
2466	0	0	bilou_type type = j == i ? bilou_type_U : k == i ? bilou_type_B : k == j ? bilou_type_L : bilou_type_I;
	0	0	bilou_type type = j == i ? bilou_type_U : k == i ? bilou_type_B : k == j ? bilou_type_L : bilou_type_I;
	0	0	bilou_type type = j == i ? bilou_type_U : k == i ? bilou_type_B : k == j ? bilou_type_L : bilou_type_I;
2467	0	0	append_unless_exists(features[k], feature + G * (2 * window + 1));
2468	0	0	append_unless_exists(features[k], feature + type * (2 * window + 1));
2474	0	0	if (hard_pre_length)
2475	0	0	for (unsigned j = i; j < i + hard_pre_length; j++) {
2476	0	0	for (auto&& bilou : sentence.probabilities[j].local.bilou) {
2481	0	0	j == i ? bilou_type_B : j + 1 == i + hard_pre_length ? bilou_type_L : bilou_type_I;
	0	0	j == i ? bilou_type_B : j + 1 == i + hard_pre_length ? bilou_type_L : bilou_type_I;
	0	0	j == i ? bilou_type_B : j + 1 == i + hard_pre_length ? bilou_type_L : bilou_type_I;
2489	0	0	for (unsigned i = 0; i < sentence.size; i++)
2490	0	0	for (auto&& feature : features[i])
2491	0	0	apply_in_window(i, feature);
	0	0	apply_in_window(i, feature);
	0	0	apply_in_window(i, feature);
2497	0	0	vector> recased_match_sources(sentence.size);
2498	0	0	for (unsigned i = 0; i < sentence.size; i++)
2499	0	0	recase_match_source(sentence.words[i], RECASE_ANY, recased_match_sources[i]);
2503	0	0	for (unsigned i = 0, e = 0; i < sentence.size; i++) {
2504	0	0	while (e < entities.size() && entities[e].start == i) {
	0	0	while (e < entities.size() && entities[e].start == i) {
	0	0	while (e < entities.size() && entities[e].start == i) {
2505	0	0	if (i + entities[e].length > entity_until)
2507	0	0	buffer.push_back(entities[e++]);
2510	0	0	if (entity_until <= i) {
2512	0	0	unsigned free_until = e < entities.size() ? entities[e].start : sentence.size;
2516	0	0	for (unsigned j = i; j < free_until && !nodes.empty(); j++) {
	0	0	for (unsigned j = i; j < free_until && !nodes.empty(); j++) {
	0	0	for (unsigned j = i; j < free_until && !nodes.empty(); j++) {
2518	0	0	for (auto&& node : nodes)
2519	0	0	if (!gazetteers_trie[node].children.empty())
2520	0	0	for (auto&& match_source : recased_match_sources[j]) {
2522	0	0	for (auto&& it = range.first; it != range.second; it++)
2523	0	0	append_unless_exists(new_nodes, it->second);
2526	0	0	for (auto&& node : new_nodes)
2527	0	0	if (gazetteers_trie[node].mode == HARD_POST &&
	0	0	if (gazetteers_trie[node].mode == HARD_POST &&
	0	0	if (gazetteers_trie[node].mode == HARD_POST &&
2528	0	0	((j - i + 1) > hard_post_length || node < hard_post_node))
2534	0	0	if (hard_post_length) {
2535	0	0	buffer.emplace_back(i, hard_post_length, entity_list[gazetteers_trie[hard_post_node].entity]);
2541	0	0	if (buffer.size() != entities.size())
2546	0	0	for (auto&& gazetteer_list : gazetteer_lists)
2547	0	0	for (auto&& gazetteer : gazetteer_list.gazetteers) {
2548	0	0	gazetteers.push_back(gazetteer);
2549	0	0	if (gazetteer_types) gazetteer_types->push_back(gazetteer_list.entity);
	0	0	if (gazetteer_types) gazetteer_types->push_back(gazetteer_list.entity);
2590	0	0	for (i = array.size(); i; i--)
2591	0	0	if (array[i - 1] == value)
2594	0	0	if (!i)
2602	0	0	for (auto&& gazetteer_meta : gazetteer_metas)
2603	0	0	for (int mode = 0; mode < MODES_TOTAL; mode++) {
2606	0	0	ifstream file(path_from_utf8(file_name).c_str());
2607	0	0	if (!file.is_open()) {
2608	0	0	if (mode == SOFT && files_must_exist)
2613	0	0	gazetteer_lists.emplace_back();
2618	0	0	while (getline(file, line))
	0	0	while (getline(file, line))
2619	0	0	if (!line.empty() && line[0] != '#')
	0	0	if (!line.empty() && line[0] != '#')
	0	0	if (!line.empty() && line[0] != '#')
2620	0	0	gazetteer_lists.back().gazetteers.push_back(line);
2625	0	0	vector gazetteer_tokens, gazetteer_tokens_additional, gazetteer_token(1);
2630	0	0	gazetteers_trie.emplace_back();
2631	0	0	for (auto&& gazetteer_list : gazetteer_lists)
2632	0	0	for (auto&& gazetteer : gazetteer_list.gazetteers) {
2633	0	0	pipeline.tokenizer->set_text(gazetteer);
2634	0	0	if (!pipeline.tokenizer->next_sentence(&gazetteer_tokens, nullptr)) continue;
	0	0	if (!pipeline.tokenizer->next_sentence(&gazetteer_tokens, nullptr)) continue;
2635	0	0	while (pipeline.tokenizer->next_sentence(&gazetteer_tokens_additional, nullptr))
	0	0	while (pipeline.tokenizer->next_sentence(&gazetteer_tokens_additional, nullptr))
2636	0	0	gazetteer_tokens.insert(gazetteer_tokens.end(), gazetteer_tokens_additional.begin(), gazetteer_tokens_additional.end());
2640	0	0	for (unsigned token = 0; token < gazetteer_tokens.size(); token++) {
2641	0	0	if (token) prefix.push_back('\t');
	0	0	if (token) prefix.push_back('\t');
2642	0	0	prefix.append(gazetteer_tokens[token].str, gazetteer_tokens[token].len);
2644	0	0	if (prefix_it == gazetteer_prefixes.end()) {
2646	0	0	gazetteers_trie.emplace_back();
2650	0	0	pipeline.tagger->tag(gazetteer_token, gazetteer_token_tagged);
2651	0	0	recase_match_source(gazetteer_token_tagged.words[0], RECASE_NATIVE, gazetteer_recased_match_sources);
2652	0	0	for (auto&& match_source : gazetteer_recased_match_sources)
2661	0	0	append_unless_exists(gazetteers_trie[node].features, gazetteer_list.feature);
2662	0	0	if ((gazetteer_list.mode == HARD_PRE && gazetteers_trie[node].mode != HARD_PRE) ||
	0	0	if ((gazetteer_list.mode == HARD_PRE && gazetteers_trie[node].mode != HARD_PRE) ||
	0	0	if ((gazetteer_list.mode == HARD_PRE && gazetteers_trie[node].mode != HARD_PRE) ||
	0	0	if ((gazetteer_list.mode == HARD_PRE && gazetteers_trie[node].mode != HARD_PRE) ||
2663	0	0	(gazetteer_list.mode == HARD_POST && gazetteers_trie[node].mode == SOFT)) {
2678	0	0	if (mode == TO_UPPER)
2680	0	0	else if (mode == TO_LOWER)
2682	0	0	else if (mode == TO_TITLE)
2683	0	0	for (auto&& chr : utf8::decoder(text))
2684	0	0	utf8::append(recased.back(), recased.back().empty() ? unicode::uppercase(chr) : unicode::lowercase(chr));
2692	0	0	for (auto&& chr : utf8::decoder(word.form)) {
2693	0	0	any_lower = any_lower || (unicode::category(chr) & unicode::Ll);
	0	0	any_lower = any_lower || (unicode::category(chr) & unicode::Ll);
2694	0	0	if (first) first_uc = unicode::category(chr) & unicode::Lut;
2700	0	0	for (int perform = 0; perform < TO_TOTAL; perform++) {
2701	0	0	if (mode == RECASE_NATIVE) {
2702	0	0	if (perform == TO_UPPER && !(first_uc && !any_lower)) continue;
	0	0	if (perform == TO_UPPER && !(first_uc && !any_lower)) continue;
2703	0	0	if (perform == TO_TITLE && !(first_uc && any_lower)) continue;
	0	0	if (perform == TO_TITLE && !(first_uc && any_lower)) continue;
2704	0	0	if (perform == TO_LOWER && first_uc) continue;
2706	0	0	if (mode == RECASE_ANY) {
2707	0	0	if (perform == TO_UPPER && !(first_uc && !any_lower)) continue;
	0	0	if (perform == TO_UPPER && !(first_uc && !any_lower)) continue;
2708	0	0	if (perform == TO_TITLE && !first_uc) continue;
2711	0	0	if (match == MATCH_FORM)
2713	0	0	else if (match == MATCH_RAWLEMMA)
2715	0	0	else if (match == MATCH_RAWLEMMAS)
2716	0	0	for (auto&& raw_lemma : word.raw_lemmas_all)
2721	0	3	const vector gazetteers_enhanced::basename_suffixes = {".txt", ".hard_pre.txt", ".hard_post.txt"};
	9	3	const vector gazetteers_enhanced::basename_suffixes = {".txt", ".hard_pre.txt", ".hard_post.txt"};
	0	0	const vector gazetteers_enhanced::basename_suffixes = {".txt", ".hard_pre.txt", ".hard_post.txt"};
2727	14	4	for (unsigned i = 0; i < sentence.size; i++)
2728	14	0	apply_in_window(i, lookup(sentence.words[i].lemma_id, total_features));
	46	14	apply_in_window(i, lookup(sentence.words[i].lemma_id, total_features));
2730	4	0	apply_outer_words_in_window(lookup_empty());
	8	4	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
2745	14	4	for (unsigned i = 0; i < sentence.size; i++) {
2750	14	0	for (digit = false, num = 0; *form; form++) {
2751	0	14	if (*form < '0' || *form > '9') break;
2755	0	14	if (digit && !*form) {
	0	0	if (digit && !*form) {
2757	0	0	if (num < 24) apply_in_window(i, hour);
	0	0	if (num < 24) apply_in_window(i, hour);
	0	0	if (num < 24) apply_in_window(i, hour);
2758	0	0	if (num < 60) apply_in_window(i, minute);
	0	0	if (num < 60) apply_in_window(i, minute);
	0	0	if (num < 60) apply_in_window(i, minute);
2759	0	0	if (num >= 1 && num <= 31) apply_in_window(i, day);
	0	0	if (num >= 1 && num <= 31) apply_in_window(i, day);
	0	0	if (num >= 1 && num <= 31) apply_in_window(i, day);
2760	0	0	if (num >= 1 && num <= 12) apply_in_window(i, month);
	0	0	if (num >= 1 && num <= 12) apply_in_window(i, month);
	0	0	if (num >= 1 && num <= 12) apply_in_window(i, month);
2761	0	0	if (num >= 1000 && num <= 2200) apply_in_window(i, year);;
	0	0	if (num >= 1000 && num <= 2200) apply_in_window(i, year);;
	0	0	if (num >= 1000 && num <= 2200) apply_in_window(i, year);;
2763	0	14	if (digit && num < 24 && (*form == '.' || *form == ':')) {
	0	0	if (digit && num < 24 && (*form == '.' || *form == ':')) {
2765	0	0	for (digit = false, num = 0, form++; *form; form++) {
2766	0	0	if (*form < '0' || *form > '9') break;
2770	0	0	if (digit && !*form && num < 60) apply_in_window(i, time);
	0	0	if (digit && !*form && num < 60) apply_in_window(i, time);
	0	0	if (digit && !*form && num < 60) apply_in_window(i, time);
	0	0	if (digit && !*form && num < 60) apply_in_window(i, time);
	0	0	if (digit && !*form && num < 60) apply_in_window(i, time);
2780	14	4	for (unsigned i = 0; i < sentence.size; i++)
2781	7	7	if (sentence.previous_stage[i].bilou != bilou_type_unknown) {
2786	7	0	apply_in_range(i, lookup(buffer, total_features), 1, window);
	9	7	apply_in_range(i, lookup(buffer, total_features), 1, window);
2792	4	10	if (value < 0) {
2796	11	14	for (; value; value >>= 4)
2805	14	4	for (unsigned i = 0; i < sentence.size; i++)
2806	14	0	apply_in_window(i, lookup(sentence.words[i].raw_lemma, total_features));
	46	14	apply_in_window(i, lookup(sentence.words[i].raw_lemma, total_features));
2808	4	0	apply_outer_words_in_window(lookup_empty());
	8	4	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
2822	14	4	for (unsigned i = 0; i < sentence.size; i++) {
2827	54	14	for (bool first = true; (chr = utf8::decode(raw_lemma)); first = false) {
2829	54	0	was_upper = was_upper || category & unicode::Lut;
	54	0	was_upper = was_upper || category & unicode::Lut;
2830	14	40	was_lower = was_lower || category & unicode::Ll;
	4	10	was_lower = was_lower || category & unicode::Ll;
2832	54	0	if (first && was_upper) apply_in_window(i, fst_cap);
	0	0	if (first && was_upper) apply_in_window(i, fst_cap);
	0	0	if (first && was_upper) apply_in_window(i, fst_cap);
2834	0	14	if (was_upper && !was_lower) apply_in_window(i, all_cap);
	0	0	if (was_upper && !was_lower) apply_in_window(i, all_cap);
	0	0	if (was_upper && !was_lower) apply_in_window(i, all_cap);
2835	0	14	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
	0	0	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
	0	0	if (was_upper && was_lower) apply_in_window(i, mixed_cap);
2846	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2848	0	0	for (auto&& chr : utf8::decoder(sentence.words[i].raw_lemma))
2849	0	0	utf8::append(buffer, buffer.empty() ? chr : unicode::lowercase(chr));
2850	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
2853	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
2866	0	0	if (!feature_processor::parse(window, args, entities, total_features, pipeline)) return false;
2867	0	0	if (args.size() != 2) return cerr << "*Suffix features require exactly two arguments -- shortest and longest suffix length!" << endl, false;
2870	0	0	if (!parse_int(args[0], "*Suffix shortest length", shortest, error)) return cerr << error << endl, false;
	0	0	if (!parse_int(args[0], "*Suffix shortest length", shortest, error)) return cerr << error << endl, false;
2871	0	0	if (!parse_int(args[1], "*Suffix longest length", longest, error)) return cerr << error << endl, false;
	0	0	if (!parse_int(args[1], "*Suffix longest length", longest, error)) return cerr << error << endl, false;
2893	0	0	for (unsigned i = 0; i < sentence.size; i++) {
2895	0	0	for (auto&& chr : utf8::decoder(source == SUFFIX_SOURCE_FORM ? sentence.words[i].form : sentence.words[i].raw_lemma))
	0	0	for (auto&& chr : utf8::decoder(source == SUFFIX_SOURCE_FORM ? sentence.words[i].form : sentence.words[i].raw_lemma))
2896	0	0	chrs.push_back((casing == SUFFIX_CASE_ORIGINAL || chrs.empty()) ? chr : unicode::lowercase(chr));
	0	0	chrs.push_back((casing == SUFFIX_CASE_ORIGINAL || chrs.empty()) ? chr : unicode::lowercase(chr));
2899	0	0	for (int s = 1; s <= longest && s <= int(chrs.size()); s++) {
	0	0	for (int s = 1; s <= longest && s <= int(chrs.size()); s++) {
	0	0	for (int s = 1; s <= longest && s <= int(chrs.size()); s++) {
2900	0	0	utf8::append(buffer, chrs[chrs.size() - s]);
2901	0	0	if (s >= shortest) {
2902	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
	0	0	apply_in_window(i, lookup(buffer, total_features));
2907	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
	0	0	apply_outer_words_in_window(lookup_empty());
2919	14	4	for (unsigned i = 0; i < sentence.size; i++)
2920	12	2	apply_in_window(i, lookup(sentence.words[i].tag, total_features));
	40	12	apply_in_window(i, lookup(sentence.words[i].tag, total_features));
2922	4	0	apply_outer_words_in_window(lookup_empty());
	8	4	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
	8	0	apply_outer_words_in_window(lookup_empty());
	12	8	apply_outer_words_in_window(lookup_empty());
2931	0	0	if (!feature_processor::parse(window, args, entities, total_features, pipeline)) return false;
2932	0	0	if (args.size() != 2) return cerr << "URLEmailDetector requires exactly two arguments -- named entity types for URL and email!" << endl, false;
2937	0	0	if (url == entity_type_unknown || email == entity_type_unknown)
	0	0	if (url == entity_type_unknown || email == entity_type_unknown)
2957	14	4	for (unsigned i = 0; i < sentence.size; i++) {
2959	0	14	if (type == url_detector::NO_URL || sentence.probabilities[i].local_filled) continue;
	0	0	if (type == url_detector::NO_URL || sentence.probabilities[i].local_filled) continue;
	0	14	if (type == url_detector::NO_URL || sentence.probabilities[i].local_filled) continue;
2962	0	0	for (auto&& bilou : sentence.probabilities[i].local.bilou) {
2967	0	0	sentence.probabilities[i].local.bilou[bilou_type_U].entity = type == url_detector::EMAIL ? email : url;
2982	0	8	if (name.compare("BrownClusters") == 0) return new brown_clusters();
2983	0	8	if (name.compare("CzechAddContainers") == 0) return new czech_add_containers();
2984	0	8	if (name.compare("CzechLemmaTerm") == 0) return new czech_lemma_term();
2985	1	7	if (name.compare("Form") == 0) return new form();
2986	0	7	if (name.compare("FormCapitalization") == 0) return new form_capitalization();
2987	0	7	if (name.compare("FormCaseNormalized") == 0) return new form_case_normalized();
2988	0	7	if (name.compare("FormCaseNormalizedSuffix") == 0) return new suffix(SUFFIX_SOURCE_FORM, SUFFIX_CASE_NORMALIZED);
2989	0	7	if (name.compare("FormSuffix") == 0) return new suffix(SUFFIX_SOURCE_FORM, SUFFIX_CASE_ORIGINAL);
2990	0	7	if (name.compare("Gazetteers") == 0) return new feature_processors::gazetteers();
2991	0	7	if (name.compare("GazetteersEnhanced") == 0) return new gazetteers_enhanced();
2992	1	6	if (name.compare("Lemma") == 0) return new lemma();
2993	1	5	if (name.compare("NumericTimeValue") == 0) return new number_time_value();
2994	1	4	if (name.compare("PreviousStage") == 0) return new previous_stage();
2995	1	3	if (name.compare("RawLemma") == 0) return new raw_lemma();
2996	1	2	if (name.compare("RawLemmaCapitalization") == 0) return new raw_lemma_capitalization();
2997	0	2	if (name.compare("RawLemmaCaseNormalized") == 0) return new raw_lemma_case_normalized();
2998	0	2	if (name.compare("RawLemmaCaseNormalizedSuffix") == 0) return new suffix(SUFFIX_SOURCE_RAWLEMMA, SUFFIX_CASE_NORMALIZED);
2999	0	2	if (name.compare("RawLemmaSuffix") == 0) return new suffix(SUFFIX_SOURCE_RAWLEMMA, SUFFIX_CASE_ORIGINAL);
3000	1	1	if (name.compare("Tag") == 0) return new tag();
3001	1	0	if (name.compare("URLEmailDetector") == 0) return new url_email_detector();
3058	1	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
3061	1	0	total_features = data.next_4B();
3064	1	0	for (unsigned i = data.next_4B(); i; i--) {
	8	1	for (unsigned i = data.next_4B(); i; i--) {
3066	8	0	data.next_str(name);
3069	8	0	auto* processor = feature_processor::create(name);
3070	8	0	if (processor) {
3071	8	0	processor->load(data, pipeline);
3072	8	0	processors.emplace_back(name, processor);
3078	0	0	}
3088	14	4	for (unsigned i = 0; i < sentence.size; i++) {
3094	32	4	for (auto&& processor : processors)
3095	0	32	processor.processor->process_sentence(sentence, adding_features ? &total_features : nullptr, buffer);
3099	16	2	for (auto&& processor : processors)
	0	0	for (auto&& processor : processors)
3108	0	0	for (auto&& processor : processors)
	0	0	for (auto&& processor : processors)
3395	0	0	format_tagged_lemma(result);
3400	0	0	for (auto&& lemma : lemmas)
3403	0	0	if (lemmas.size() > 1)
3411	0	0	if (converter) converter->convert(lemma);
3415	0	0	if (converter) converter->convert_analyzed(lemmas);
3428	0	0	for (derivated_lemma parent; derinet->parent(lemma.lemma, parent); )
	0	0	for (derivated_lemma parent; derinet->parent(lemma.lemma, parent); )
3430	0	0	if (converter) converter->convert(lemma);
3438	0	0	return derinet ? new root_derivation_formatter(derinet) : nullptr;
	0	0	return derinet ? new root_derivation_formatter(derinet) : nullptr;
3447	0	0	if (converter) converter->convert(lemma);
	0	0	if (converter) converter->convert(lemma);
3448	0	0	for (derivated_lemma parent; derinet->parent(current.lemma, parent); current.lemma.swap(parent.lemma)) {
	0	0	for (derivated_lemma parent; derinet->parent(current.lemma, parent); current.lemma.swap(parent.lemma)) {
3449	0	0	tagged_lemma parrent_lemma(parent.lemma, current.tag);
3450	0	0	if (converter) converter->convert(parrent_lemma);
	0	0	if (converter) converter->convert(parrent_lemma);
3451	0	0	lemma.lemma.append(" ").append(parrent_lemma.lemma);
3460	0	0	return derinet ? new path_derivation_formatter(derinet) : nullptr;
	0	0	return derinet ? new path_derivation_formatter(derinet) : nullptr;
3469	0	0	if (converter) converter->convert(lemma);
	0	0	if (converter) converter->convert(lemma);
3470	0	0	for (derivated_lemma parent; derinet->parent(root, parent); root.swap(parent.lemma)) {}
	0	0	for (derivated_lemma parent; derinet->parent(root, parent); root.swap(parent.lemma)) {}
3471	0	0	format_tree(root, tag, lemma, converter);
3477	0	0	if (converter) {
3478	0	0	tagged_lemma current(root, tag);
3479	0	0	converter->convert(current);
3480	0	0	tree.lemma.append(" ").append(current.lemma);
3482	0	0	tree.lemma.append(" ").append(root);
3485	0	0	if (derinet->children(root, children))
	0	0	if (derinet->children(root, children))
3486	0	0	for (auto&& child : children)
3487	0	0	format_tree(child.lemma, tag, tree, converter);
3488	0	0	tree.lemma.push_back(' ');
3496	0	0	return derinet ? new tree_derivation_formatter(derinet) : nullptr;
	0	0	return derinet ? new tree_derivation_formatter(derinet) : nullptr;
3500	0	0	if (name == "none") return new_none_derivation_formatter();
3501	0	0	if (name == "root") return new_root_derivation_formatter(derinet);
3502	0	0	if (name == "path") return new_path_derivation_formatter(derinet);
3503	0	0	if (name == "tree") return new_tree_derivation_formatter(derinet);
3533	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	108	6	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	31	3	while (len--)
	136	26	while (len--)
	0	0	while (len--)
	118	10	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
3534	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	30	78	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	14	17	if (*a++ != *b++)
	124	12	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	98	20	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
	0	0	if (*a++ != *b++)
3543	0	0	while (len--)
	236	108	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	4	24	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	56	9	while (len--)
3684	52	16	while (mask < num)
3686	16	0	hash.resize(mask + 1);
3690	137	0	uint32_t size = data.next_4B();
3692	137	0	hash.resize(size);
3693	137	0	memcpy(hash.data(), data.next(size), size * sizeof(uint32_t));
3695	137	0	size = data.next_4B();
3696	137	0	this->data.resize(size);
3697	68	69	if (size) memcpy(this->data.data(), data.next(size), size);
	68	0	if (size) memcpy(this->data.data(), data.next(size), size);
3701	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	30	0	if (len <= 0) return 0;
	30	18	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	0	0	if (len <= 0) return 0;
	108	0	if (len <= 0) return 0;
	41	0	if (len <= 0) return 0;
	9	0	if (len <= 0) return 0;
	9	0	if (len <= 0) return 0;
3702	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	4	26	if (len == 1) return unaligned_load(data);
	18	12	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	0	if (len == 1) return unaligned_load(data);
	0	108	if (len == 1) return unaligned_load(data);
	7	34	if (len == 1) return unaligned_load(data);
	0	9	if (len == 1) return unaligned_load(data);
	0	9	if (len == 1) return unaligned_load(data);
3703	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	26	0	if (len == 2) return unaligned_load(data);
	0	12	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	0	0	if (len == 2) return unaligned_load(data);
	20	88	if (len == 2) return unaligned_load(data);
	30	4	if (len == 2) return unaligned_load(data);
	9	0	if (len == 2) return unaligned_load(data);
	9	0	if (len == 2) return unaligned_load(data);
3706	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	142	26	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	0	0	while (len--)
	60	20	while (len--)
	132	30	while (len--)
	56	9	while (len--)
	56	9	while (len--)
3720	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	48	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
	0	0	if (unsigned(len) >= hashes.size()) return nullptr;
3726	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
	0	48	if (len <= 2)
	0	0	if (len <= 2)
	0	0	if (len <= 2)
3727	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	30	18	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
	0	0	return data != end ? data + len : nullptr;
3729	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
3730	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
	0	0	if (small_memeq(str, data, len)) return data + len;
3741	108	0	if (unsigned(len) >= hashes.size()) return nullptr;
	41	55	if (unsigned(len) >= hashes.size()) return nullptr;
3747	20	88	if (len <= 2)
	30	11	if (len <= 2)
3748	52	36	return data != end ? (const T*)(data + len) : nullptr;
	11	0	return data != end ? (const T*)(data + len) : nullptr;
3750	20	17	while (data < end) {
	38	4	while (data < end) {
3751	3	17	if (small_memeq(str, data, len)) return (const T*)(data + len);
	26	12	if (small_memeq(str, data, len)) return (const T*)(data + len);
3760	0	0	if (unsigned(len) >= hashes.size()) return;
	30	0	if (unsigned(len) >= hashes.size()) return;
	0	0	if (unsigned(len) >= hashes.size()) return;
	0	0	if (unsigned(len) >= hashes.size()) return;
	0	0	if (unsigned(len) >= hashes.size()) return;
	0	0	if (unsigned(len) >= hashes.size()) return;
3766	0	0	while (data < end) {
	30	30	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
3776	3	1	for (unsigned len = 0; len < hashes.size(); len++) {
	0	0	for (unsigned len = 0; len < hashes.size(); len++) {
	0	0	for (unsigned len = 0; len < hashes.size(); len++) {
3780	13	3	while (data < end) {
	0	0	while (data < end) {
	0	0	while (data < end) {
3794	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
	0	0	return unsigned(len) < hashes.size() ? hashes[len].data.data() : nullptr;
3798	2	14	if (hashes.size() == 0) hashes.emplace_back(1);
3799	2	12	else if (hashes.size() == 1) hashes.emplace_back(1<<8);
3800	2	10	else if (hashes.size() == 2) hashes.emplace_back(1<<16);
3805	9	0	if (unsigned(str_len) < hashes.size())
3810	16	2	for (auto&& hash : hashes) {
3812	131616	16	for (auto&& len : hash.hash) total += len, len = total - len;
3818	9	0	if (unsigned(str_len) < hashes.size()) {
3829	16	2	for (auto&& hash : hashes)
3830	131616	16	for (int i = hash.hash.size() - 1; i >= 0; i--)
3831	131600	16	hash.hash[i] = i > 0 ? hash.hash[i-1] : 0;
3838	137	48	for (unsigned i = 0; i < sizes; i++)
3890	0	0	if (dictionary) lemma.len = dictionary->lemma_id_len(lemma);
3897	0	0	if (lemma_data) {
3899	0	0	if (parent_encoded) {
3903	0	0	if (parent_data[parent_len])
3913	0	0	if (dictionary) lemma.len = dictionary->lemma_id_len(lemma);
3920	0	0	if (lemma_data) {
3923	0	0	if (children_len) {
3925	0	0	for (unsigned i = 0; i < children_len; i++) {
3929	0	0	if (child_data[child_len])
3941	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
3944	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
3945	0	0	derinet.resize(data.next_4B());
	0	0	derinet.resize(data.next_4B());
3949	0	0	for (int pass = 1; pass <= 3; pass++) {
3950	0	0	if (pass > 1) data.seek(data_position);
	0	0	if (pass > 1) data.seek(data_position);
3953	0	0	for (int i = data.next_4B(); i > 0; i--) {
	0	0	for (int i = data.next_4B(); i > 0; i--) {
3954	0	0	lemma.resize(lemma.size() - data.next_1B());
	0	0	lemma.resize(lemma.size() - data.next_1B());
3955	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
3956	0	0	lemma.push_back(data.next_1B());
3958	0	0	unsigned char lemma_comment_len = data.next_1B();
3959	0	0	const char* lemma_comment = lemma_comment_len ? data.next(lemma_comment_len) : nullptr;
	0	0	const char* lemma_comment = lemma_comment_len ? data.next(lemma_comment_len) : nullptr;
3961	0	0	unsigned children = data.next_2B();
3963	0	0	if (pass == 3) parent.clear();
3965	0	0	int operations = data.next_1B();
3966	0	0	if (operations) {
3967	0	0	int remove_start = operations & REMOVE_START ? data.next_1B() : 0;
	0	0	int remove_start = operations & REMOVE_START ? data.next_1B() : 0;
3968	0	0	int remove_end = operations & REMOVE_END ? data.next_1B() : 0;
	0	0	int remove_end = operations & REMOVE_END ? data.next_1B() : 0;
3969	0	0	if (operations & ADD_START) {
3970	0	0	int add_start = data.next_1B();
3971	0	0	const char* str = data.next(add_start);
3972	0	0	if (pass == 3) parent.assign(str, str + add_start);
3974	0	0	if (pass == 3) parent.insert(parent.end(), lemma.begin() + remove_start, lemma.end() - remove_end);
	0	0	if (pass == 3) parent.insert(parent.end(), lemma.begin() + remove_start, lemma.end() - remove_end);
3975	0	0	if (operations & ADD_END) {
3976	0	0	int add_end = data.next_1B();
3977	0	0	const char* str = data.next(add_end);
3978	0	0	if (pass == 3) parent.insert(parent.end(), str, str + add_end);
3982	0	0	if (pass == 1) {
3984	0	0	} else if (pass == 2) {
3987	0	0	while (lemma_comment_len--) *lemma_data++ = *lemma_comment++;
3990	0	0	if (children) unaligned_store(((uint32_t*)lemma_data) + children - 1, 0);
3991	0	0	} else if (pass == 3 && !parent.empty()) {
	0	0	} else if (pass == 3 && !parent.empty()) {
	0	0	} else if (pass == 3 && !parent.empty()) {
4002	0	0	assert(lemma_data && parent_data);
4005	0	0	assert(parent.size() < (1<<8) && parent_offset < (1<<24));
	0	0	assert(parent.size() < (1<<8) && parent_offset < (1<<24));
4009	0	0	assert(lemma.size() < (1<<8) && lemma_offset < (1<<24));
	0	0	assert(lemma.size() < (1<<8) && lemma_offset < (1<<24));
4014	0	0	if (child_index+1 < children_len)
4019	0	0	if (pass == 1)
4020	0	0	derinet.done_adding();
4021	0	0	if (pass == 2)
4023	0	0	}
4056	38	14	while (form_tmp.len && !rest_has_Lut)
	38	0	while (form_tmp.len && !rest_has_Lut)
4065	4	10	if (first_Lut && !rest_has_Lut) { // common case allowing fast execution
4070	0	10	} else if (!first_Lut && rest_has_Lut) {
4073	0	10	} else if (first_Lut && rest_has_Lut) {
4080	0	0	while (form_tmp.len) {
4121	0	0	for (unsigned len = 1; len < lemma.len; len++)
4122	0	0	if (lemma.str[len] == '`' || lemma.str[len] == '_' ||
	0	0	if (lemma.str[len] == '`' || lemma.str[len] == '_' ||
4123	0	0	(lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9'))
	0	0	(lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9'))
	0	0	(lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9'))
4130	0	0	for (unsigned len = 1; len < lemma.len; len++) {
4131	0	0	if (lemma.str[len] == '`' || lemma.str[len] == '_')
4133	0	0	if (lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9') {
	0	0	if (lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9') {
	0	0	if (lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9') {
	0	0	if (lemma.str[len] == '-' && len+1 < lemma.len && lemma.str[len+1] >= '0' && lemma.str[len+1] <= '9') {
4135	0	0	while (len < lemma.len && lemma.str[len] >= '0' && lemma.str[len] <= '9') len++;
	0	0	while (len < lemma.len && lemma.str[len] >= '0' && lemma.str[len] <= '9') len++;
	0	0	while (len < lemma.len && lemma.str[len] >= '0' && lemma.str[len] <= '9') len++;
4145	0	0	if (addinfo_len) {
4146	0	0	res.reserve(addinfo_len + 4);
4147	0	0	if (addinfo[0] != 255) {
4152	0	0	for (int i = 1; i < addinfo_len; i++)
4160	0	0	for (int i = 1; i + 2 < addinfo_len; i++)
4161	0	0	if (addinfo[i] == '_' && addinfo[i+1] == ',' && addinfo[i+2] == 'x')
	0	0	if (addinfo[i] == '_' && addinfo[i+1] == ',' && addinfo[i+2] == 'x')
	0	0	if (addinfo[i] == '_' && addinfo[i+1] == ',' && addinfo[i+2] == 'x')
4171	0	0	if (lemma_info < lemma.str + lemma.len) {
4175	0	0	if (*lemma_info == '-') {
4178	0	0	lemma_additional_info < lemma.str + lemma.len && (*lemma_additional_info >= '0' && *lemma_additional_info <= '9');
	0	0	lemma_additional_info < lemma.str + lemma.len && (*lemma_additional_info >= '0' && *lemma_additional_info <= '9');
4182	0	0	if (lemma_additional_info == lemma_info + 1 || (lemma_additional_info < lemma.str + lemma.len && *lemma_additional_info != '`' && *lemma_additional_info != '_') || lemma_num >= 255) {
	0	0	if (lemma_additional_info == lemma_info + 1 || (lemma_additional_info < lemma.str + lemma.len && *lemma_additional_info != '`' && *lemma_additional_info != '_') || lemma_num >= 255) {
	0	0	if (lemma_additional_info == lemma_info + 1 || (lemma_additional_info < lemma.str + lemma.len && *lemma_additional_info != '`' && *lemma_additional_info != '_') || lemma_num >= 255) {
	0	0	if (lemma_additional_info == lemma_info + 1 || (lemma_additional_info < lemma.str + lemma.len && *lemma_additional_info != '`' && *lemma_additional_info != '_') || lemma_num >= 255) {
	0	0	if (lemma_additional_info == lemma_info + 1 || (lemma_additional_info < lemma.str + lemma.len && *lemma_additional_info != '`' && *lemma_additional_info != '_') || lemma_num >= 255) {
4183	0	0	if (die_on_failure)
4190	0	0	while (lemma_additional_info < lemma.str + lemma.len)
4193	0	0	if (data.size() > 255) {
4194	0	0	if (die_on_failure)
4205	0	0	if (data.empty()) return true;
4206	0	0	if (data[0] != 255 && (!other_addinfo_len || other_addinfo[0] != data[0])) return false;
	0	0	if (data[0] != 255 && (!other_addinfo_len || other_addinfo[0] != data[0])) return false;
	0	0	if (data[0] != 255 && (!other_addinfo_len || other_addinfo[0] != data[0])) return false;
	0	0	if (data[0] != 255 && (!other_addinfo_len || other_addinfo[0] != data[0])) return false;
4250	0	0	if (filters.empty()) return true;
4253	0	0	for (auto&& filter : filters) {
4255	0	0	while (tag_pos < filter.pos)
4256	0	0	if (!tag[tag_pos++])
4258	0	0	if (!tag[tag_pos])
4263	0	0	for (int i = 1; i < filter.chars_len && ((!matched) ^ filter.negate); i++)
	0	0	for (int i = 1; i < filter.chars_len && ((!matched) ^ filter.negate); i++)
4265	0	0	if (!matched) return false;
4305	7	1	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
4307	9	1	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
4313	1	0	vector root(max(lemmas.max_length(), roots.max_length()));
	0	0	vector root(max(lemmas.max_length(), roots.max_length()));
	0	0	vector root(max(lemmas.max_length(), roots.max_length()));
4315	2	1	for (int pass = 1; pass <= 2; pass++) {
	0	0	for (int pass = 1; pass <= 2; pass++) {
	0	0	for (int pass = 1; pass <= 2; pass++) {
4316	1	1	if (pass > 1) data.seek(data_position);
	1	0	if (pass > 1) data.seek(data_position);
	0	0	if (pass > 1) data.seek(data_position);
	0	0	if (pass > 1) data.seek(data_position);
	0	0	if (pass > 1) data.seek(data_position);
	0	0	if (pass > 1) data.seek(data_position);
4321	2	0	for (int i = data.next_4B(); i > 0; i--) {
	4	2	for (int i = data.next_4B(); i > 0; i--) {
	0	0	for (int i = data.next_4B(); i > 0; i--) {
	0	0	for (int i = data.next_4B(); i > 0; i--) {
	0	0	for (int i = data.next_4B(); i > 0; i--) {
	0	0	for (int i = data.next_4B(); i > 0; i--) {
4322	4	0	lemma_len -= data.next_1B();
	0	0	lemma_len -= data.next_1B();
	0	0	lemma_len -= data.next_1B();
4323	4	0	for (int i = data.next_1B(); i > 0; i--)
	24	4	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
	0	0	for (int i = data.next_1B(); i > 0; i--)
4324	24	0	lemma[lemma_len++] = data.next_1B();
	0	0	lemma[lemma_len++] = data.next_1B();
	0	0	lemma[lemma_len++] = data.next_1B();
4325	4	0	unsigned char lemma_info_len = data.next_1B();
	0	0	unsigned char lemma_info_len = data.next_1B();
	0	0	unsigned char lemma_info_len = data.next_1B();
4326	0	4	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
	0	0	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
	0	0	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
	0	0	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
	0	0	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
	0	0	const char* lemma_info = lemma_info_len ? data.next(lemma_info_len) : nullptr;
4327	4	0	unsigned lemma_roots = data.next_1B();
	0	0	unsigned lemma_roots = data.next_1B();
	0	0	unsigned lemma_roots = data.next_1B();
4332	2	2	if (pass == 1) {
	0	0	if (pass == 1) {
	0	0	if (pass == 1) {
4339	0	2	if (lemma_info_len) small_memcpy(lemma_data, lemma_info, lemma_info_len), lemma_data += lemma_info_len;
	0	0	if (lemma_info_len) small_memcpy(lemma_data, lemma_info, lemma_info_len), lemma_data += lemma_info_len;
	0	0	if (lemma_info_len) small_memcpy(lemma_data, lemma_info, lemma_info_len), lemma_data += lemma_info_len;
4344	14	4	for (unsigned i = 0; i < lemma_roots; i++) {
	0	0	for (unsigned i = 0; i < lemma_roots; i++) {
	0	0	for (unsigned i = 0; i < lemma_roots; i++) {
4346	14	0	int operations = data.next_1B();
	0	0	int operations = data.next_1B();
	0	0	int operations = data.next_1B();
4347	0	14	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
	0	0	if (operations & REMOVE_START) { int from = data.next_1B(), to = 0; while (from < root_len) root[to++] = root[from++]; root_len = to; }
4348	12	2	if (operations & REMOVE_END) root_len -= data.next_1B();
	12	0	if (operations & REMOVE_END) root_len -= data.next_1B();
	0	0	if (operations & REMOVE_END) root_len -= data.next_1B();
	0	0	if (operations & REMOVE_END) root_len -= data.next_1B();
	0	0	if (operations & REMOVE_END) root_len -= data.next_1B();
	0	0	if (operations & REMOVE_END) root_len -= data.next_1B();
4349	0	14	if (operations & ADD_START) {
	0	0	if (operations & ADD_START) {
	0	0	if (operations & ADD_START) {
4350	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
	0	0	int from = root_len, to = from + data.next_1B(); while (from > 0) root[--to] = root[--from]; root_len += to;
4351	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
	0	0	for (int i = 0; i < to; i++) root[i] = data.next_1B();
4353	14	0	if (operations & ADD_END)
	0	0	if (operations & ADD_END)
	0	0	if (operations & ADD_END)
4354	14	0	for (int len = data.next_1B(); len > 0; len--)
	28	14	for (int len = data.next_1B(); len > 0; len--)
	0	0	for (int len = data.next_1B(); len > 0; len--)
	0	0	for (int len = data.next_1B(); len > 0; len--)
	0	0	for (int len = data.next_1B(); len > 0; len--)
	0	0	for (int len = data.next_1B(); len > 0; len--)
4355	28	0	root[root_len++] = data.next_1B();
	0	0	root[root_len++] = data.next_1B();
	0	0	root[root_len++] = data.next_1B();
4356	14	0	uint16_t clas = data.next_2B();
	0	0	uint16_t clas = data.next_2B();
	0	0	uint16_t clas = data.next_2B();
4358	7	7	if (pass == 1) { // for each root
	0	0	if (pass == 1) { // for each root
	0	0	if (pass == 1) { // for each root
4367	0	7	assert(uint8_t(lemma_len) == lemma_len);
	0	0	assert(uint8_t(lemma_len) == lemma_len);
	0	0	assert(uint8_t(lemma_len) == lemma_len);
4372	0	7	assert(uint8_t(root_len) == root_len);
	0	0	assert(uint8_t(root_len) == root_len);
	0	0	assert(uint8_t(root_len) == root_len);
4377	1	1	if (pass == 1) { // after the whole pass
	0	0	if (pass == 1) { // after the whole pass
	0	0	if (pass == 1) { // after the whole pass
4378	1	0	lemmas.done_adding();
	0	0	lemmas.done_adding();
	0	0	lemmas.done_adding();
4379	1	0	roots.done_adding();
	0	0	roots.done_adding();
	0	0	roots.done_adding();
4387	1	0	tags.resize(data.next_2B());
	1	0	tags.resize(data.next_2B());
	0	0	tags.resize(data.next_2B());
	0	0	tags.resize(data.next_2B());
	0	0	tags.resize(data.next_2B());
	0	0	tags.resize(data.next_2B());
4388	20	1	for (auto&& tag : tags) {
	0	0	for (auto&& tag : tags) {
	0	0	for (auto&& tag : tags) {
4389	20	0	tag.resize(data.next_1B());
	0	0	tag.resize(data.next_1B());
	0	0	tag.resize(data.next_1B());
4390	60	20	for (unsigned i = 0; i < tag.size(); i++)
	0	0	for (unsigned i = 0; i < tag.size(); i++)
	0	0	for (unsigned i = 0; i < tag.size(); i++)
4391	60	0	tag[i] = data.next_1B();
	0	0	tag[i] = data.next_1B();
	0	0	tag[i] = data.next_1B();
4395	1	0	suffixes.load(data);
	0	0	suffixes.load(data);
	0	0	suffixes.load(data);
4398	1	0	suffixes.iter_all([this](const char* suffix, int len, pointer_decoder& data) mutable {
	0	0	suffixes.iter_all([this](const char* suffix, int len, pointer_decoder& data) mutable {
	0	0	suffixes.iter_all([this](const char* suffix, int len, pointer_decoder& data) mutable {
4405	15	13	for (unsigned i = 0; i < classes_len; i++) {
	0	0	for (unsigned i = 0; i < classes_len; i++) {
	0	0	for (unsigned i = 0; i < classes_len; i++) {
4407	5	10	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
	5	0	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
	0	0	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
	0	0	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
	0	0	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
	0	0	if (classes_ptr_i >= classes.size()) classes.resize(classes_ptr_i + 1);
4408	15	0	classes[classes_ptr_i].emplace_back(suffix_str, vector());
	0	0	classes[classes_ptr_i].emplace_back(suffix_str, vector());
	0	0	classes[classes_ptr_i].emplace_back(suffix_str, vector());
4409	20	15	for (const uint16_t* ptr = tags_ptr + unaligned_load(indices_ptr + i),
	0	0	for (const uint16_t* ptr = tags_ptr + unaligned_load(indices_ptr + i),
	0	0	for (const uint16_t* ptr = tags_ptr + unaligned_load(indices_ptr + i),
4412	20	0	classes[classes_ptr_i].back().second.emplace_back(unaligned_load(ptr));
	0	0	classes[classes_ptr_i].back().second.emplace_back(unaligned_load(ptr));
	0	0	classes[classes_ptr_i].back().second.emplace_back(unaligned_load(ptr));
4422	0	18	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
	0	0	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
	0	0	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
	0	0	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
	0	0	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
	0	0	uint16_t** suff = max_suffix_len <= 16 ? suff_stack : (suff_heap.resize(max_suffix_len), suff_heap.data());
4424	48	0	for (int i = form.len; i >= 0 && suff_len < max_suffix_len; i--, suff_len++) {
	0	0	for (int i = form.len; i >= 0 && suff_len < max_suffix_len; i--, suff_len++) {
	0	0	for (int i = form.len; i >= 0 && suff_len < max_suffix_len; i--, suff_len++) {
4432	30	18	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	30	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	30	18	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
	0	0	for (int root_len = int(form.len) - --suff_len; suff_len >= 0 && root_len < int(roots.max_length()); suff_len--, root_len++)
4433	30	0	if (unaligned_load(suff[suff_len])) {
	0	0	if (unaligned_load(suff[suff_len])) {
	0	0	if (unaligned_load(suff[suff_len])) {
4437	30	0	roots.iter(form.str, root_len, [&](const char* root, pointer_decoder& root_data) {
	0	0	roots.iter(form.str, root_len, [&](const char* root, pointer_decoder& root_data) {
	0	0	roots.iter(form.str, root_len, [&](const char* root, pointer_decoder& root_data) {
4442	10	20	if (small_memeq(form.str, root, root_len)) {
	0	0	if (small_memeq(form.str, root, root_len)) {
	0	0	if (small_memeq(form.str, root, root_len)) {
4444	10	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	10	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	10	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
	0	0	if (suffix_class_ptr < suff_data + suff_classes && unaligned_load(suffix_class_ptr) == root_class) {
4447	0	10	if (lemma_data[lemma_len]) lemma += LemmaAddinfo::format(lemma_data + lemma_len + 1, lemma_data[lemma_len]);
	0	0	if (lemma_data[lemma_len]) lemma += LemmaAddinfo::format(lemma_data + lemma_len + 1, lemma_data[lemma_len]);
	0	0	if (lemma_data[lemma_len]) lemma += LemmaAddinfo::format(lemma_data + lemma_len + 1, lemma_data[lemma_len]);
	0	0	if (lemma_data[lemma_len]) lemma += LemmaAddinfo::format(lemma_data + lemma_len + 1, lemma_data[lemma_len]);
4451	18	10	for (unsigned i = unaligned_load(suff_tag_indices + (suffix_class_ptr - suff_data));
	0	0	for (unsigned i = unaligned_load(suff_tag_indices + (suffix_class_ptr - suff_data));
	0	0	for (unsigned i = unaligned_load(suff_tag_indices + (suffix_class_ptr - suff_data));
4453	18	0	lemmas.emplace_back(lemma, tags[unaligned_load(suff_tags + i)]);
	0	0	lemmas.emplace_back(lemma, tags[unaligned_load(suff_tags + i)]);
	0	0	lemmas.emplace_back(lemma, tags[unaligned_load(suff_tags + i)]);
4463	0	0	int raw_lemma_len = addinfo.parse(lemma);
	0	0	int raw_lemma_len = addinfo.parse(lemma);
4466	0	0	lemmas.iter(lemma.str, raw_lemma_len, [&](const char* lemma_str, pointer_decoder& data) {
	0	0	lemmas.iter(lemma.str, raw_lemma_len, [&](const char* lemma_str, pointer_decoder& data) {
	0	0	lemmas.iter(lemma.str, raw_lemma_len, [&](const char* lemma_str, pointer_decoder& data) {
4472	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
	0	0	if (small_memeq(lemma.str, lemma_str, raw_lemma_len) && addinfo.match_lemma_id(lemma_info, lemma_info_len) && LemmaAddinfo::generatable(lemma_info, lemma_info_len)) {
4477	0	0	for (unsigned i = 0; i < lemma_roots_len; i++) {
	0	0	for (unsigned i = 0; i < lemma_roots_len; i++) {
	0	0	for (unsigned i = 0; i < lemma_roots_len; i++) {
4483	0	0	for (auto&& suffix : classes[clas]) {
	0	0	for (auto&& suffix : classes[clas]) {
	0	0	for (auto&& suffix : classes[clas]) {
4485	0	0	for (auto&& tag : suffix.second)
	0	0	for (auto&& tag : suffix.second)
	0	0	for (auto&& tag : suffix.second)
4486	0	0	if (filter.matches(tags[tag].c_str())) {
	0	0	if (filter.matches(tags[tag].c_str())) {
	0	0	if (filter.matches(tags[tag].c_str())) {
4487	0	0	if (!forms) {
	0	0	if (!forms) {
	0	0	if (!forms) {
4488	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
	0	0	lemmas_forms.emplace_back(string(lemma.str, raw_lemma_len) + LemmaAddinfo::format(lemma_info, lemma_info_len));
4492	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
	0	0	if (root_with_suffix.empty() && root_len + suffix.first.size()) {
4493	0	0	root_with_suffix.reserve(root_len + suffix.first.size());
	0	0	root_with_suffix.reserve(root_len + suffix.first.size());
	0	0	root_with_suffix.reserve(root_len + suffix.first.size());
4498	0	0	forms->emplace_back(root_with_suffix, tags[tag]);
	0	0	forms->emplace_back(root_with_suffix, tags[tag]);
	0	0	forms->emplace_back(root_with_suffix, tags[tag]);
4545	0	0	for (unsigned tag_filters_len = data.next_1B(); tag_filters_len; tag_filters_len--) {
4549	0	0	tag_filters.emplace_back(tag_filter.c_str());
4560	0	0	if (!form.len) return;
4564	0	0	middle_masks.reserve(form.len);
4566	0	0	for (unsigned initial = 0; initial < form.len; initial++) {
4569	0	0	if (initial) {
4571	0	0	if (!found) break;
4576	0	0	if (initial_mask) {
4577	0	0	middle_masks.resize(initial);
4578	0	0	middle_masks.emplace_back(initial_mask);
4579	0	0	for (unsigned middle = initial; middle < middle_masks.size(); middle++) {
4580	0	0	if (!middle_masks[middle]) continue;
4582	0	0	for (unsigned i = middle + 1; i < form.len; i++) {
4584	0	0	if (!found) break;
4585	0	0	if (unaligned_load(found)) {
4586	0	0	if (i + 1 > middle_masks.size()) middle_masks.resize(i + 1);
	0	0	if (i + 1 > middle_masks.size()) middle_masks.resize(i + 1);
4592	0	0	if (middle > initial && middle < form.len ) {
	0	0	if (middle > initial && middle < form.len ) {
4593	0	0	if (initial) {
4594	0	0	if (form_tmp.empty()) form_tmp.assign(form.str, form.str + form.len);
4598	0	0	dictionary.analyze(string_piece((initial ? form_tmp.data() : form.str) + middle - initial, form.len - middle + initial), lemmas);
	0	0	dictionary.analyze(string_piece((initial ? form_tmp.data() : form.str) + middle - initial, form.len - middle + initial), lemmas);
4600	0	0	for (unsigned i = lemmas_ori_size; i < lemmas.size(); i++) {
4601	0	0	for (unsigned filter = 0; filter < tag_filters.size(); filter++)
4602	0	0	if ((middle_masks[middle] & (1<
	0	0	if ((middle_masks[middle] & (1<
	0	0	if ((middle_masks[middle] & (1<
4603	0	0	if (i == lemmas_new_size) {
4606	0	0	lemmas[lemmas_new_size].lemma.reserve(lemmas[i].lemma.size() + middle - initial);
4615	0	0	if (lemmas_new_size < lemmas.size()) lemmas.erase(lemmas.begin() + lemmas_new_size, lemmas.end());
4740	22	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	5	17	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	37	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	5	32	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
	0	0	return chr.chr < ragel_map.size() && ragel_map[chr.chr] != 128 ? ragel_map[chr.chr] : 128 + (uint32_t(chr.cat) * uint32_t(0x077CB531U) >> 27);
4800	0	0	czech_morpho(morpho_language language, unsigned version) : language(language), version(version) {}
	0	0	czech_morpho(morpho_language language, unsigned version) : language(language), version(version) {}
	0	0	czech_morpho(morpho_language language, unsigned version) : language(language), version(version) {}
4843	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
4847	0	0	unsigned tag_length = data.next_1B();
4848	0	0	if (tag_length < unknown_tag.size()) unknown_tag.erase(tag_length);
	0	0	if (tag_length < unknown_tag.size()) unknown_tag.erase(tag_length);
4849	0	0	if (tag_length < number_tag.size()) number_tag.erase(tag_length);
	0	0	if (tag_length < number_tag.size()) number_tag.erase(tag_length);
4850	0	0	if (tag_length < punctuation_tag.size()) punctuation_tag.erase(tag_length);
	0	0	if (tag_length < punctuation_tag.size()) punctuation_tag.erase(tag_length);
4853	0	0	dictionary.load(data);
4857	0	0	if (data.next_1B()) {
	0	0	if (data.next_1B()) {
4858	0	0	prefix_guesser.reset(new morpho_prefix_guesser(dictionary));
4859	0	0	prefix_guesser->load(data);
4864	0	0	if (data.next_1B()) {
	0	0	if (data.next_1B()) {
4865	0	0	statistical_guesser.reset(new morpho_statistical_guesser());
4866	0	0	statistical_guesser->load(data);
4867	0	0	}
4878	0	0	if (form.len) {
4882	0	0	generate_casing_variants(form, form_uclc, form_lc);
4885	0	0	dictionary.analyze(form, lemmas);
4886	0	0	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
	0	0	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
4887	0	0	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
	0	0	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
4888	0	0	if (!lemmas.empty()) return NO_GUESSER;
4891	0	0	analyze_special(form, lemmas);
4892	0	0	if (!lemmas.empty()) return NO_GUESSER;
4895	0	0	if (guesser == GUESSER && prefix_guesser)
	0	0	if (guesser == GUESSER && prefix_guesser)
	0	0	if (guesser == GUESSER && prefix_guesser)
4896	0	0	prefix_guesser->analyze(form_lc.empty() ? form : form_lc, lemmas);
	0	0	prefix_guesser->analyze(form_lc.empty() ? form : form_lc, lemmas);
4900	0	0	if (guesser == GUESSER && statistical_guesser) {
	0	0	if (guesser == GUESSER && statistical_guesser) {
	0	0	if (guesser == GUESSER && statistical_guesser) {
4901	0	0	if (form_uclc.empty() && form_lc.empty())
	0	0	if (form_uclc.empty() && form_lc.empty())
	0	0	if (form_uclc.empty() && form_lc.empty())
4902	0	0	statistical_guesser->analyze(form, lemmas, nullptr);
4904	0	0	morpho_statistical_guesser::used_rules used_rules; used_rules.reserve(3);
4905	0	0	statistical_guesser->analyze(form, lemmas, &used_rules);
4906	0	0	if (!form_uclc.empty()) statistical_guesser->analyze(form_uclc, lemmas, &used_rules);
	0	0	if (!form_uclc.empty()) statistical_guesser->analyze(form_uclc, lemmas, &used_rules);
4907	0	0	if (!form_lc.empty()) statistical_guesser->analyze(form_lc, lemmas, &used_rules);
	0	0	if (!form_lc.empty()) statistical_guesser->analyze(form_lc, lemmas, &used_rules);
4913	0	0	if (prefix_guesser_guesses) {
4916	0	0	return lemma_compare < 0 || (lemma_compare == 0 && a.tag < b.tag);
4919	0	0	return a.lemma == b.lemma && a.tag == b.tag;
	0	0	return a.lemma == b.lemma && a.tag == b.tag;
4921	0	0	if (lemmas_end != lemmas.end()) lemmas.erase(lemmas_end, lemmas.end());
4924	0	0	if (!lemmas.empty()) return GUESSER;
4927	0	0	lemmas.emplace_back(string(form.str, form.len), unknown_tag);
4936	0	0	if (lemma.len) {
4937	0	0	if (dictionary.generate(lemma, filter, forms))
	0	0	if (dictionary.generate(lemma, filter, forms))
4940	0	0	if (guesser == GUESSER && prefix_guesser)
	0	0	if (guesser == GUESSER && prefix_guesser)
4961	0	0	return new czech_tokenizer(language, version, this);
4992	0	0	if (!form.len) return;
5000	0	0	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(form.str, form.len);
5001	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(form.str, form.len);
5002	0	0	if ((codepoint == '.' && form.len) || codepoint == ',') codepoint = utf8::decode(form.str, form.len);
	0	0	if ((codepoint == '.' && form.len) || codepoint == ',') codepoint = utf8::decode(form.str, form.len);
	0	0	if ((codepoint == '.' && form.len) || codepoint == ',') codepoint = utf8::decode(form.str, form.len);
5003	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(form.str, form.len);
5004	0	0	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
	0	0	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
5006	0	0	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(form.str, form.len);
5008	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(form.str, form.len);
5011	0	0	if (any_digit && !form.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !form.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !form.len && (!codepoint || codepoint == '.')) {
5012	0	0	lemmas.emplace_back(string(form_ori.str, form_ori.len), number_tag);
5013	0	0	} else if ((first < sizeof(punctuation_additional) && punctuation_additional[first]) ||
	0	0	} else if ((first < sizeof(punctuation_additional) && punctuation_additional[first]) ||
	0	0	} else if ((first < sizeof(punctuation_additional) && punctuation_additional[first]) ||
	0	0	} else if ((first < sizeof(punctuation_additional) && punctuation_additional[first]) ||
5014	0	0	((unicode::category(first) & unicode::P) && (first >= sizeof(punctuation_exceptions) || !punctuation_exceptions[first])))
	0	0	((unicode::category(first) & unicode::P) && (first >= sizeof(punctuation_exceptions) || !punctuation_exceptions[first])))
5015	0	0	lemmas.emplace_back(string(form_ori.str, form_ori.len), punctuation_tag);
5053	0	0	for (unsigned len = 1; len < lemma.len; len++) {
5054	0	0	if (len + 1 == lemma.len && (lemma.str[len] == '^' || lemma.str[len] == '+'))
	0	0	if (len + 1 == lemma.len && (lemma.str[len] == '^' || lemma.str[len] == '+'))
5056	0	0	if (len + 1 < lemma.len && lemma.str[len] == '^') {
	0	0	if (len + 1 < lemma.len && lemma.str[len] == '^') {
5058	0	0	for (unsigned i = len + 1; ok && i < lemma.len; i++)
	0	0	for (unsigned i = len + 1; ok && i < lemma.len; i++)
5059	0	0	ok &= (lemma.str[i] >= 'A' && lemma.str[i] <= 'Z') ||
5060	0	0	(lemma.str[i] >= 'a' && lemma.str[i] <= 'z') ||
	0	0	(lemma.str[i] >= 'a' && lemma.str[i] <= 'z') ||
5061	0	0	(i > len + 1 && lemma.str[i] == '-');
5062	0	0	if (ok) return len;
5085	0	0	for (size_t i = len; i < lemma.len; i++)
5092	0	0	if (data.empty()) return true;
5093	0	0	if (data.size() == 1 && data[0] == '^') return other_addinfo_len > 0 && other_addinfo[0] == '^';
	0	0	if (data.size() == 1 && data[0] == '^') return other_addinfo_len > 0 && other_addinfo[0] == '^';
	0	0	if (data.size() == 1 && data[0] == '^') return other_addinfo_len > 0 && other_addinfo[0] == '^';
	0	0	if (data.size() == 1 && data[0] == '^') return other_addinfo_len > 0 && other_addinfo[0] == '^';
	0	0	if (data.size() == 1 && data[0] == '^') return other_addinfo_len > 0 && other_addinfo[0] == '^';
5094	0	0	if (data.size() == 1 && data[0] == '+') return other_addinfo_len == 0;
	0	0	if (data.size() == 1 && data[0] == '+') return other_addinfo_len == 0;
	0	0	if (data.size() == 1 && data[0] == '+') return other_addinfo_len == 0;
5095	0	0	return data.size() == size_t(other_addinfo_len) && small_memeq(data.data(), other_addinfo, other_addinfo_len);
	0	0	return data.size() == size_t(other_addinfo_len) && small_memeq(data.data(), other_addinfo, other_addinfo_len);
5115	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
	0	0	class english_morpho_guesser {
5163	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
	0	0	english_morpho(unsigned version) : version(version) {}
5237	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
5240	0	0	dictionary.load(data);
5241	0	0	morpho_guesser.load(data);
	0	0	morpho_guesser.load(data);
5252	0	0	if (form.len) {
5256	0	0	generate_casing_variants(form, form_uclc, form_lc);
5259	0	0	dictionary.analyze(form, lemmas);
5260	0	0	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
	0	0	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
5261	0	0	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
	0	0	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
5262	0	0	if (!lemmas.empty())
5263	0	0	return guesser == NO_GUESSER || !morpho_guesser.analyze_proper_names(form, form_lc.empty() ? form : form_lc, lemmas) ? NO_GUESSER : GUESSER;
	0	0	return guesser == NO_GUESSER || !morpho_guesser.analyze_proper_names(form, form_lc.empty() ? form : form_lc, lemmas) ? NO_GUESSER : GUESSER;
	0	0	return guesser == NO_GUESSER || !morpho_guesser.analyze_proper_names(form, form_lc.empty() ? form : form_lc, lemmas) ? NO_GUESSER : GUESSER;
	0	0	return guesser == NO_GUESSER || !morpho_guesser.analyze_proper_names(form, form_lc.empty() ? form : form_lc, lemmas) ? NO_GUESSER : GUESSER;
5266	0	0	analyze_special(form, lemmas);
5267	0	0	if (!lemmas.empty()) return NO_GUESSER;
5270	0	0	if (guesser == GUESSER)
5271	0	0	morpho_guesser.analyze(form, form_lc.empty() ? form : form_lc, lemmas);
	0	0	morpho_guesser.analyze(form, form_lc.empty() ? form : form_lc, lemmas);
5272	0	0	if (!lemmas.empty()) return GUESSER;
5275	0	0	lemmas.emplace_back(string(form.str, form.len), unknown_tag);
5284	0	0	if (lemma.len) {
5285	0	0	if (dictionary.generate(lemma, filter, forms))
	0	0	if (dictionary.generate(lemma, filter, forms))
5305	0	0	return new english_tokenizer(version <= 2 ? 1 : 2);
5312	0	0	if (!form.len) return;
5315	0	0	if (form.len == 1)
5319	0	0	case '?': lemmas.emplace_back(string(form.str, form.len), dot_tag); return;
5320	0	0	case ',': lemmas.emplace_back(string(form.str, form.len), comma_tag); return;
5321	0	0	case '#': lemmas.emplace_back(string(form.str, form.len), hash_tag); return;
5322	0	0	case '$': lemmas.emplace_back(string(form.str, form.len), dollar_tag); return;
5323	0	0	case '[': lemmas.emplace_back(string(form.str, form.len), sym_tag); return;
5324	0	0	case ']': lemmas.emplace_back(string(form.str, form.len), sym_tag); return;
5325	0	0	case '%': lemmas.emplace_back(string(form.str, form.len), jj_tag);
5326	0	0	lemmas.emplace_back(string(form.str, form.len), nn_tag); return;
5327	0	0	case '&': lemmas.emplace_back(string(form.str, form.len), cc_tag);
5328	0	0	lemmas.emplace_back(string(form.str, form.len), sym_tag); return;
5329	0	0	case '*': lemmas.emplace_back(string(form.str, form.len), sym_tag);
5330	0	0	lemmas.emplace_back(string(form.str, form.len), nn_tag); return;
5331	0	0	case '@': lemmas.emplace_back(string(form.str, form.len), sym_tag);
5332	0	0	lemmas.emplace_back(string(form.str, form.len), in_tag); return;
5333	0	0	case '\'': lemmas.emplace_back(string(form.str, form.len), close_quotation_tag);
5334	0	0	lemmas.emplace_back(string(form.str, form.len), pos_tag); return;
5341	0	0	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(number.str, number.len);
5342	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
5343	0	0	while (codepoint == ',') {
5345	0	0	if (unicode::category(utf8::decode(group.str, group.len) & ~unicode::N)) break;
5346	0	0	if (unicode::category(utf8::decode(group.str, group.len) & ~unicode::N)) break;
5347	0	0	if (unicode::category(utf8::decode(group.str, group.len) & ~unicode::N)) break;
5352	0	0	if (codepoint == '.' && number.len) {
	0	0	if (codepoint == '.' && number.len) {
5354	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
5356	0	0	if (version >= 2 && any_digit && codepoint == 's' && !number.len) {
	0	0	if (version >= 2 && any_digit && codepoint == 's' && !number.len) {
	0	0	if (version >= 2 && any_digit && codepoint == 's' && !number.len) {
5357	0	0	lemmas.emplace_back(string(form.str, form.len), number_tag);
5358	0	0	lemmas.emplace_back(string(form.str, form.len - 1), nns_tag);
5361	0	0	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
	0	0	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
5363	0	0	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(number.str, number.len);
5365	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
5367	0	0	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
5368	0	0	lemmas.emplace_back(string(form.str, form.len), number_tag);
5369	0	0	lemmas.emplace_back(string(form.str, form.len), nnp_tag);
5370	0	0	if (form.len == 1 + (codepoint == '.') && *form.str >= '1' && *form.str <= '9')
	0	0	if (form.len == 1 + (codepoint == '.') && *form.str >= '1' && *form.str <= '9')
	0	0	if (form.len == 1 + (codepoint == '.') && *form.str >= '1' && *form.str <= '9')
	0	0	if (form.len == 1 + (codepoint == '.') && *form.str >= '1' && *form.str <= '9')
5371	0	0	lemmas.emplace_back(string(form.str, form.len), ls_tag);
5378	0	0	while ((symbol || any_punctuation) && punctuation.len) {
	0	0	while ((symbol || any_punctuation) && punctuation.len) {
5380	0	0	if (open_quotation) open_quotation = codepoint == '`' || unicode::category(codepoint) & unicode::Pi;
	0	0	if (open_quotation) open_quotation = codepoint == '`' || unicode::category(codepoint) & unicode::Pi;
	0	0	if (open_quotation) open_quotation = codepoint == '`' || unicode::category(codepoint) & unicode::Pi;
5381	0	0	if (close_quotation) close_quotation = codepoint == '\'' || codepoint == '"' || unicode::category(codepoint) & unicode::Pf;
	0	0	if (close_quotation) close_quotation = codepoint == '\'' || codepoint == '"' || unicode::category(codepoint) & unicode::Pf;
	0	0	if (close_quotation) close_quotation = codepoint == '\'' || codepoint == '"' || unicode::category(codepoint) & unicode::Pf;
5382	0	0	if (open_parenthesis) open_parenthesis = unicode::category(codepoint) & unicode::Ps;
5383	0	0	if (close_parenthesis) close_parenthesis = unicode::category(codepoint) & unicode::Pe;
5384	0	0	if (any_punctuation) any_punctuation = unicode::category(codepoint) & unicode::P;
5385	0	0	if (symbol) symbol = codepoint == '*' || unicode::category(codepoint) & unicode::S;
	0	0	if (symbol) symbol = codepoint == '*' || unicode::category(codepoint) & unicode::S;
	0	0	if (symbol) symbol = codepoint == '*' || unicode::category(codepoint) & unicode::S;
5387	0	0	if (!punctuation.len && open_quotation) { lemmas.emplace_back(string(form.str, form.len), open_quotation_tag); return; }
	0	0	if (!punctuation.len && open_quotation) { lemmas.emplace_back(string(form.str, form.len), open_quotation_tag); return; }
	0	0	if (!punctuation.len && open_quotation) { lemmas.emplace_back(string(form.str, form.len), open_quotation_tag); return; }
5388	0	0	if (!punctuation.len && close_quotation) { lemmas.emplace_back(string(form.str, form.len), close_quotation_tag); return; }
	0	0	if (!punctuation.len && close_quotation) { lemmas.emplace_back(string(form.str, form.len), close_quotation_tag); return; }
	0	0	if (!punctuation.len && close_quotation) { lemmas.emplace_back(string(form.str, form.len), close_quotation_tag); return; }
5389	0	0	if (!punctuation.len && open_parenthesis) { lemmas.emplace_back(string(form.str, form.len), open_parenthesis_tag); return; }
	0	0	if (!punctuation.len && open_parenthesis) { lemmas.emplace_back(string(form.str, form.len), open_parenthesis_tag); return; }
	0	0	if (!punctuation.len && open_parenthesis) { lemmas.emplace_back(string(form.str, form.len), open_parenthesis_tag); return; }
5390	0	0	if (!punctuation.len && close_parenthesis) { lemmas.emplace_back(string(form.str, form.len), close_parenthesis_tag); return; }
	0	0	if (!punctuation.len && close_parenthesis) { lemmas.emplace_back(string(form.str, form.len), close_parenthesis_tag); return; }
	0	0	if (!punctuation.len && close_parenthesis) { lemmas.emplace_back(string(form.str, form.len), close_parenthesis_tag); return; }
5391	0	0	if (!punctuation.len && symbol) { lemmas.emplace_back(string(form.str, form.len), sym_tag); return; }
	0	0	if (!punctuation.len && symbol) { lemmas.emplace_back(string(form.str, form.len), sym_tag); return; }
	0	0	if (!punctuation.len && symbol) { lemmas.emplace_back(string(form.str, form.len), sym_tag); return; }
5392	0	0	if (!punctuation.len && any_punctuation) { lemmas.emplace_back(string(form.str, form.len), punctuation_tag); return; }
	0	0	if (!punctuation.len && any_punctuation) { lemmas.emplace_back(string(form.str, form.len), punctuation_tag); return; }
	0	0	if (!punctuation.len && any_punctuation) { lemmas.emplace_back(string(form.str, form.len), punctuation_tag); return; }
5426	0	0	while (tags--) {
5428	0	0	exceptions_tags.emplace_back(string(data.next(len), len));
5564	0	0	for (unsigned len = data.next_1B(); len; len--) {
5570	0	0	if (exception) {
5573	0	0	for (unsigned len = data.next_1B(); len; len--) {
5576	0	0	for (unsigned tags = data.next_1B(); tags; tags--)
5577	0	0	lemmas.emplace_back(lemma, exceptions_tags[data.next_2B()]);
5584	0	0	for (unsigned prefix = 1; prefix <= form_lc.len; prefix++) {
5586	0	0	if (!found) break;
5587	0	0	if (found[NEGATION_LEN]) {
5588	0	0	if (form_lc.len - prefix >= found[TO_FOLLOW]) negation_len = found[NEGATION_LEN];
5594	0	0	add(JJ, lemma_lc, negation_len, lemmas);
5595	0	0	add(RB, lemma_lc, negation_len, lemmas);
5596	0	0	add(NN, lemma_lc, negation_len, lemmas);
5597	0	0	add_NNS(lemma_lc, negation_len, lemmas);
5614	0	0	if ( p == ( (form_lc.str + form_lc.len)) )
5621	0	0	if ( _klen > 0 ) {
5626	0	0	if ( _upper < _lower )
5630	0	0	if ( ( form_lc.str[form_lc.len - 1 - (p - form_lc.str)]) < *_mid )
5632	0	0	else if ( ( form_lc.str[form_lc.len - 1 - (p - form_lc.str)]) > *_mid )
5644	0	0	if ( _klen > 0 ) {
5649	0	0	if ( _upper < _lower )
5653	0	0	if ( ( form_lc.str[form_lc.len - 1 - (p - form_lc.str)]) < _mid[0] )
5655	0	0	else if ( ( form_lc.str[form_lc.len - 1 - (p - form_lc.str)]) > _mid[1] )
5669	0	0	if ( _tag_guesser_trans_actions[_trans] == 0 )
5674	0	0	while ( _nacts-- > 0 )
5679	0	0	{ if (!added_JJR_RBR) added_JJR_RBR = true, add_JJR_RBR(lemma_lc, negation_len, lemmas); }
	0	0	{ if (!added_JJR_RBR) added_JJR_RBR = true, add_JJR_RBR(lemma_lc, negation_len, lemmas); }
5682	0	0	{ if (!added_JJS_RBS) added_JJS_RBS = true, add_JJS_RBS(lemma_lc, negation_len, lemmas); }
	0	0	{ if (!added_JJS_RBS) added_JJS_RBS = true, add_JJS_RBS(lemma_lc, negation_len, lemmas); }
5685	0	0	{ add_VBG(lemma_lc, lemmas); }
5688	0	0	{ add_VBD_VBN(lemma_lc, lemmas); }
5691	0	0	{ add_VBZ(lemma_lc, lemmas); }
5697	0	0	{ if (!added_SYM) added_SYM = true, add(SYM, lemma_lc, lemmas); }
5700	0	0	{ if (!added_CD) added_CD = true, add(CD, lemma_lc, lemmas); }
5706	0	0	if ( ++p != ( (form_lc.str + form_lc.len)) )
5709	0	0	if ( p == ( (form_lc.str + form_lc.len)) )
5713	0	0	while ( __nacts-- > 0 ) {
5714	0	0	switch ( *__acts++ ) {
5716	0	0	{ if (!added_CD) added_CD = true, add(CD, lemma_lc, lemmas); }
5732	0	0	bool is_NNP = form.str != form_lc.str || (form.len && (*form.str == '\'' || (*form.str >= '0' && *form.str <= '9')));
	0	0	bool is_NNP = form.str != form_lc.str || (form.len && (*form.str == '\'' || (*form.str >= '0' && *form.str <= '9')));
	0	0	bool is_NNP = form.str != form_lc.str || (form.len && (*form.str == '\'' || (*form.str >= '0' && *form.str <= '9')));
	0	0	bool is_NNP = form.str != form_lc.str || (form.len && (*form.str == '\'' || (*form.str >= '0' && *form.str <= '9')));
5734	0	0	if (!is_NNP && !is_NNPS) return false;
5737	0	0	for (auto&& lemma : lemmas) {
5741	0	0	if (!((is_NNP && !was_NNP) || (is_NNPS && !was_NNPS))) return false;
	0	0	if (!((is_NNP && !was_NNP) || (is_NNPS && !was_NNPS))) return false;
5744	0	0	if (is_NNP && !was_NNP) add(NNP, lemma, lemmas);
5745	0	0	if (is_NNPS && !was_NNPS) add_NNPS(lemma, lemmas);
	0	0	if (is_NNPS && !was_NNPS) add_NNPS(lemma, lemmas);
5750	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
	0	0	lemmas.emplace_back(form, tag);
5759	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
	0	0	lemmas.emplace_back(negation_len ? form.substr(negation_len) + "^" + form.substr(0, negation_len) : form, tag);
5861	0	0	if ( p == ( (form.c_str() + form.size())) )
5870	0	0	if ( _klen > 0 ) {
5875	0	0	if ( _upper < _lower )
5879	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < *_mid )
5881	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > *_mid )
5893	0	0	if ( _klen > 0 ) {
5898	0	0	if ( _upper < _lower )
5902	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < _mid[0] )
5904	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > _mid[1] )
5918	0	0	if ( _NNS_trans_actions[_trans] == 0 )
5923	0	0	while ( _nacts-- > 0 )
5928	0	0	{ if (best > 'a') best = 'a', remove = 2, append = "an"; }
5931	0	0	{ if (best > 'b') best = 'b', remove = 1, append = nullptr; }
5934	0	0	{ if (best > 'c') best = 'c', remove = 3, append = "fe"; }
5937	0	0	{ if (best > 'd') best = 'd', remove = 2, append = nullptr; }
5940	0	0	{ if (best > 'e') best = 'e', remove = 1, append = nullptr; }
5943	0	0	{ if (best > 'f') best = 'f', remove = 2, append = nullptr; }
5946	0	0	{ if (best > 'g') best = 'g', remove = 1, append = nullptr; }
5949	0	0	{ if (best > 'h') best = 'h', remove = 2, append = nullptr; }
5952	0	0	{ if (best > 'i') best = 'i', remove = 1, append = nullptr; }
5955	0	0	{ if (best > 'j') best = 'j', remove = 1, append = nullptr; }
5958	0	0	{ if (best > 'k') best = 'k', remove = 2, append = nullptr; }
5961	0	0	{ if (best > 'l') best = 'l', remove = 3, append = "y"; }
5964	0	0	{ if (best > 'm') best = 'm', remove = 2, append = nullptr; }
5967	0	0	{ if (best > 'n') best = 'n', remove = 1, append = nullptr; }
5973	0	0	if ( cs == 0 )
5975	0	0	if ( ++p != ( (form.c_str() + form.size())) )
5981	0	0	add(NNS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(NNS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(NNS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
6107	0	0	if ( p == ( (form.c_str() + form.size())) )
6116	0	0	if ( _klen > 0 ) {
6121	0	0	if ( _upper < _lower )
6125	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < *_mid )
6127	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > *_mid )
6139	0	0	if ( _klen > 0 ) {
6144	0	0	if ( _upper < _lower )
6148	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < _mid[0] )
6150	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > _mid[1] )
6164	0	0	if ( _NNPS_trans_actions[_trans] == 0 )
6169	0	0	while ( _nacts-- > 0 )
6174	0	0	{ if (best > 'a') best = 'a', remove = 2, append = "AN"; }
6177	0	0	{ if (best > 'b') best = 'b', remove = 2, append = "an"; }
6180	0	0	{ if (best > 'c') best = 'c', remove = 1, append = nullptr; }
6183	0	0	{ if (best > 'd') best = 'd', remove = 3, append = "FE"; }
6186	0	0	{ if (best > 'e') best = 'e', remove = 3, append = "fe"; }
6189	0	0	{ if (best > 'f') best = 'f', remove = 2, append = nullptr; }
6192	0	0	{ if (best > 'g') best = 'g', remove = 1, append = nullptr; }
6195	0	0	{ if (best > 'h') best = 'h', remove = 2, append = nullptr; }
6198	0	0	{ if (best > 'i') best = 'i', remove = 1, append = nullptr; }
6201	0	0	{ if (best > 'j') best = 'j', remove = 2, append = nullptr; }
6204	0	0	{ if (best > 'k') best = 'k', remove = 1, append = nullptr; }
6207	0	0	{ if (best > 'l') best = 'l', remove = 1, append = nullptr; }
6210	0	0	{ if (best > 'm') best = 'm', remove = 2, append = nullptr; }
6213	0	0	{ if (best > 'n') best = 'n', remove = 3, append = "Y"; }
6216	0	0	{ if (best > 'o') best = 'o', remove = 3, append = "y"; }
6219	0	0	{ if (best > 'p') best = 'p', remove = 2, append = nullptr; }
6222	0	0	{ if (best > 'q') best = 'q', remove = 1, append = nullptr; }
6228	0	0	if ( cs == 0 )
6230	0	0	if ( ++p != ( (form.c_str() + form.size())) )
6236	0	0	add(NNPS, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
	0	0	add(NNPS, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
6536	0	0	if ( p == ( (form.c_str() + form.size())) )
6545	0	0	if ( _klen > 0 ) {
6550	0	0	if ( _upper < _lower )
6554	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < *_mid )
6556	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > *_mid )
6568	0	0	if ( _klen > 0 ) {
6573	0	0	if ( _upper < _lower )
6577	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < _mid[0] )
6579	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > _mid[1] )
6593	0	0	if ( _VBG_trans_actions[_trans] == 0 )
6598	0	0	while ( _nacts-- > 0 )
6603	0	0	{ if (best > 'a') best = 'a', remove = 3, append = nullptr; }
6606	0	0	{ if (best > 'b') best = 'b', remove = 3, append = "e"; }
6609	0	0	{ if (best > 'c') best = 'c', remove = 3, append = nullptr; }
6612	0	0	{ if (best > 'd') best = 'd', remove = 3, append = "e"; }
6615	0	0	{ if (best > 'e') best = 'e', remove = 3, append = nullptr; }
6618	0	0	{ if (best > 'f') best = 'f', remove = 3, append = "e"; }
6621	0	0	{ if (best > 'g') best = 'g', remove = 3, append = nullptr; }
6624	0	0	{ if (best > 'h') best = 'h', remove = 3, append = "e"; }
6627	0	0	{ if (best > 'i') best = 'i', remove = 3, append = nullptr; }
6630	0	0	{ if (best > 'j') best = 'j', remove = 3, append = "e"; }
6633	0	0	{ if (best > 'k') best = 'k', remove = 3, append = nullptr; }
6636	0	0	{ if (best > 'l') best = 'l', remove = 3, append = "e"; }
6639	0	0	{ if (best > 'm') best = 'm', remove = 3, append = nullptr; }
6642	0	0	{ if (best > 'n') best = 'n', remove = 3, append = "e"; }
6645	0	0	{ if (best > 'o') best = 'o', remove = 3, append = nullptr; }
6648	0	0	{ if (best > 'p') best = 'p', remove = 3, append = "e"; }
6651	0	0	{ if (best > 'q') best = 'q', remove = 3, append = nullptr; }
6654	0	0	{ if (best > 'r') best = 'r', remove = 3, append = "e"; }
6660	0	0	if ( cs == 0 )
6662	0	0	if ( ++p != ( (form.c_str() + form.size())) )
6665	0	0	if ( p == ( (form.c_str() + form.size())) )
6669	0	0	while ( __nacts-- > 0 ) {
6672	0	0	{ if (best > 'c') best = 'c', remove = 3, append = nullptr; }
6675	0	0	{ if (best > 'f') best = 'f', remove = 3, append = "e"; }
6678	0	0	{ if (best > 'p') best = 'p', remove = 3, append = "e"; }
6687	0	0	add(VBG, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
	0	0	add(VBG, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
6990	0	0	if ( p == ( (form.c_str() + form.size())) )
6999	0	0	if ( _klen > 0 ) {
7004	0	0	if ( _upper < _lower )
7008	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < *_mid )
7010	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > *_mid )
7022	0	0	if ( _klen > 0 ) {
7027	0	0	if ( _upper < _lower )
7031	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < _mid[0] )
7033	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > _mid[1] )
7047	0	0	if ( _VBD_VBN_trans_actions[_trans] == 0 )
7052	0	0	while ( _nacts-- > 0 )
7057	0	0	{ if (best > 'a') best = 'a', remove = 1, append = nullptr; }
7060	0	0	{ if (best > 'b') best = 'b', remove = 2, append = nullptr; }
7063	0	0	{ if (best > 'c') best = 'c', remove = 1, append = nullptr; }
7066	0	0	{ if (best > 'd') best = 'd', remove = 2, append = nullptr; }
7069	0	0	{ if (best > 'e') best = 'e', remove = 1, append = nullptr; }
7072	0	0	{ if (best > 'f') best = 'f', remove = 2, append = nullptr; }
7075	0	0	{ if (best > 'h') best = 'h', remove = 2, append = nullptr; }
7078	0	0	{ if (best > 'i') best = 'i', remove = 3, append = "y"; }
7081	0	0	{ if (best > 'j') best = 'j', remove = 1, append = nullptr; }
7084	0	0	{ if (best > 'k') best = 'k', remove = 2, append = nullptr; }
7087	0	0	{ if (best > 'l') best = 'l', remove = 1, append = nullptr; }
7090	0	0	{ if (best > 'm') best = 'm', remove = 2, append = nullptr; }
7093	0	0	{ if (best > 'n') best = 'n', remove = 1, append = nullptr; }
7096	0	0	{ if (best > 'o') best = 'o', remove = 2, append = nullptr; }
7099	0	0	{ if (best > 'p') best = 'p', remove = 1, append = nullptr; }
7102	0	0	{ if (best > 'q') best = 'q', remove = 2, append = nullptr; }
7105	0	0	{ if (best > 'r') best = 'r', remove = 1, append = nullptr; }
7111	0	0	if ( cs == 0 )
7113	0	0	if ( ++p != ( (form.c_str() + form.size())) )
7116	0	0	if ( p == ( (form.c_str() + form.size())) )
7120	0	0	while ( __nacts-- > 0 ) {
7123	0	0	{ if (best > 'd') best = 'd', remove = 2, append = nullptr; }
7126	0	0	{ if (best > 'g') best = 'g', remove = 1, append = nullptr; }
7129	0	0	{ if (best > 'j') best = 'j', remove = 1, append = nullptr; }
7138	0	0	add(VBD, VBN, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
	0	0	add(VBD, VBN, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
7217	0	0	if ( p == ( (form.c_str() + form.size())) )
7226	0	0	if ( _klen > 0 ) {
7231	0	0	if ( _upper < _lower )
7235	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < *_mid )
7237	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > *_mid )
7249	0	0	if ( _klen > 0 ) {
7254	0	0	if ( _upper < _lower )
7258	0	0	if ( ( form[form.size() - 1 - (p - form.c_str())]) < _mid[0] )
7260	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str())]) > _mid[1] )
7274	0	0	if ( _VBZ_trans_actions[_trans] == 0 )
7279	0	0	while ( _nacts-- > 0 )
7284	0	0	{ if (best > 'a') best = 'a', remove = 1, append = nullptr; }
7287	0	0	{ if (best > 'b') best = 'b', remove = 2, append = nullptr; }
7290	0	0	{ if (best > 'c') best = 'c', remove = 1, append = nullptr; }
7293	0	0	{ if (best > 'd') best = 'd', remove = 2, append = nullptr; }
7296	0	0	{ if (best > 'e') best = 'e', remove = 1, append = nullptr; }
7299	0	0	{ if (best > 'f') best = 'f', remove = 2, append = nullptr; }
7302	0	0	{ if (best > 'g') best = 'g', remove = 3, append = "y"; }
7305	0	0	{ if (best > 'h') best = 'h', remove = 2, append = nullptr; }
7308	0	0	{ if (best > 'i') best = 'i', remove = 1, append = nullptr; }
7314	0	0	if ( cs == 0 )
7316	0	0	if ( ++p != ( (form.c_str() + form.size())) )
7322	0	0	add(VBZ, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
	0	0	add(VBZ, form.substr(0, form.size() - remove).append(append ? append : ""), lemmas);
7448	0	0	if ( p == ( (form.c_str() + form.size())) )
7457	0	0	if ( _klen > 0 ) {
7462	0	0	if ( _upper < _lower )
7466	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < *_mid )
7468	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > *_mid )
7480	0	0	if ( _klen > 0 ) {
7485	0	0	if ( _upper < _lower )
7489	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < _mid[0] )
7491	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > _mid[1] )
7505	0	0	if ( _JJR_RBR_trans_actions[_trans] == 0 )
7510	0	0	while ( _nacts-- > 0 )
7515	0	0	{ if (best > 'a') best = 'a', remove = 2, append = nullptr; }
7518	0	0	{ if (best > 'b') best = 'b', remove = 3, append = nullptr; }
7521	0	0	{ if (best > 'c') best = 'c', remove = 3, append = "y"; }
7524	0	0	{ if (best > 'd') best = 'd', remove = 2, append = nullptr; }
7527	0	0	{ if (best > 'e') best = 'e', remove = 1, append = nullptr; }
7530	0	0	{ if (best > 'f') best = 'f', remove = 2, append = nullptr; }
7536	0	0	if ( cs == 0 )
7538	0	0	if ( ++p != ( (form.c_str() + form.size())) )
7544	0	0	add(JJR, RBR, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(JJR, RBR, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(JJR, RBR, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
7674	0	0	if ( p == ( (form.c_str() + form.size())) )
7683	0	0	if ( _klen > 0 ) {
7688	0	0	if ( _upper < _lower )
7692	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < *_mid )
7694	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > *_mid )
7706	0	0	if ( _klen > 0 ) {
7711	0	0	if ( _upper < _lower )
7715	0	0	if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) < _mid[0] )
7717	0	0	else if ( ( form[form.size() - 1 - (p - form.c_str() - negation_len)]) > _mid[1] )
7731	0	0	if ( _JJS_RBS_trans_actions[_trans] == 0 )
7736	0	0	while ( _nacts-- > 0 )
7741	0	0	{ if (best > 'a') best = 'a', remove = 3, append = nullptr; }
7744	0	0	{ if (best > 'b') best = 'b', remove = 4, append = nullptr; }
7747	0	0	{ if (best > 'c') best = 'c', remove = 4, append = "y"; }
7750	0	0	{ if (best > 'd') best = 'd', remove = 3, append = nullptr; }
7753	0	0	{ if (best > 'e') best = 'e', remove = 2, append = nullptr; }
7756	0	0	{ if (best > 'f') best = 'f', remove = 3, append = nullptr; }
7762	0	0	if ( cs == 0 )
7764	0	0	if ( ++p != ( (form.c_str() + form.size())) )
7770	0	0	add(JJS, RBS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(JJS, RBS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
	0	0	add(JJS, RBS, form.substr(0, form.size() - remove).append(append ? append : ""), negation_len, lemmas);
7853	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
7857	0	0	unsigned length = data.next_1B();
7858	0	0	unknown_tag.assign(data.next(length), length);
	0	0	unknown_tag.assign(data.next(length), length);
7869	0	0	if (form.len) {
7872	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
7873	0	0	if (lemmatags.len) lemmatags.len--, lemmatags.str++;
7876	0	0	while (lemmatags.len) {
7878	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
7879	0	0	if (!lemmatags.len) break;
7884	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
	0	0	while (lemmatags.len && *lemmatags.str != ' ') lemmatags.len--, lemmatags.str++;
7886	0	0	if (lemmatags.len) lemmatags.len--, lemmatags.str++;
7888	0	0	lemmas.emplace_back(string(lemma_start, lemma_len), string(tag_start, tag_len));
7891	0	0	if (!lemmas.empty()) return NO_GUESSER;
7894	0	0	lemmas.emplace_back(string(form.str, form.len), unknown_tag);
7903	0	0	if (lemma.len) {
7906	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
7908	0	0	if (formtags.len) formtags.len--, formtags.str++;
7912	0	0	while (formtags.len) {
7914	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
7915	0	0	if (!formtags.len) break;
7920	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
	0	0	while (formtags.len && *formtags.str != ' ') formtags.len--, formtags.str++;
7922	0	0	if (formtags.len) formtags.len--, formtags.str++;
7926	0	0	if (filter.matches(tag.c_str())) {
7927	0	0	if (forms.empty()) forms.emplace_back(string(real_lemma.str, real_lemma.len));
	0	0	if (forms.empty()) forms.emplace_back(string(real_lemma.str, real_lemma.len));
7928	0	0	forms.back().forms.emplace_back(string(form_start, form_len), tag);
7932	0	0	if (any_result) return NO_GUESSER;
7940	0	0	while (lemma_len < lemma.len && lemma.str[lemma_len] != ' ') lemma_len++;
	0	0	while (lemma_len < lemma.len && lemma.str[lemma_len] != ' ') lemma_len++;
7946	0	0	while (lemma_len < lemma.len && lemma.str[lemma_len] != ' ') lemma_len++;
	0	0	while (lemma_len < lemma.len && lemma.str[lemma_len] != ' ') lemma_len++;
7952	0	0	while (form_len < form.len && form.str[form_len] != ' ') form_len++;
	0	0	while (form_len < form.len && form.str[form_len] != ' ') form_len++;
8073	1	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
8077	1	0	unsigned length = data.next_1B();
8078	1	0	unknown_tag.assign(data.next(length), length);
8079	1	0	length = data.next_1B();
8080	1	0	number_tag.assign(data.next(length), length);
8081	1	0	length = data.next_1B();
8082	1	0	punctuation_tag.assign(data.next(length), length);
8083	1	0	length = data.next_1B();
8084	1	0	symbol_tag.assign(data.next(length), length);
8087	1	0	dictionary.load(data);
8091	1	0	if (data.next_1B()) {
	0	1	if (data.next_1B()) {
8092	0	0	statistical_guesser.reset(new morpho_statistical_guesser());
8093	0	0	statistical_guesser->load(data);
8094	0	0	}
8105	14	0	if (form.len) {
8109	14	0	generate_casing_variants(form, form_uclc, form_lc);
8112	14	0	dictionary.analyze(form, lemmas);
8113	0	14	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
	0	0	if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
8114	4	10	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
	4	0	if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
8115	4	10	if (!lemmas.empty()) return NO_GUESSER;
8118	4	0	analyze_special(form, lemmas);
8119	0	4	if (!lemmas.empty()) return NO_GUESSER;
8122	0	0	if (guesser == GUESSER && statistical_guesser) {
	0	0	if (guesser == GUESSER && statistical_guesser) {
	0	0	if (guesser == GUESSER && statistical_guesser) {
8123	0	0	if (form_uclc.empty() && form_lc.empty())
	0	0	if (form_uclc.empty() && form_lc.empty())
	0	0	if (form_uclc.empty() && form_lc.empty())
8124	0	0	statistical_guesser->analyze(form, lemmas, nullptr);
8126	0	0	morpho_statistical_guesser::used_rules used_rules; used_rules.reserve(3);
8127	0	0	statistical_guesser->analyze(form, lemmas, &used_rules);
8128	0	0	if (!form_uclc.empty()) statistical_guesser->analyze(form_uclc, lemmas, &used_rules);
	0	0	if (!form_uclc.empty()) statistical_guesser->analyze(form_uclc, lemmas, &used_rules);
8129	0	0	if (!form_lc.empty()) statistical_guesser->analyze(form_lc, lemmas, &used_rules);
	0	0	if (!form_lc.empty()) statistical_guesser->analyze(form_lc, lemmas, &used_rules);
8132	0	0	if (!lemmas.empty()) return GUESSER;
8135	0	0	lemmas.emplace_back(string(form.str, form.len), unknown_tag);
8144	0	0	if (lemma.len) {
8145	0	0	if (dictionary.generate(lemma, filter, forms))
	0	0	if (dictionary.generate(lemma, filter, forms))
8175	4	0	if (!form.len) return;
8183	0	4	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(number.str, number.len);
8184	0	4	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
8185	4	0	if ((codepoint == '.' && number.len) || codepoint == ',') codepoint = utf8::decode(number.str, number.len);
	4	0	if ((codepoint == '.' && number.len) || codepoint == ',') codepoint = utf8::decode(number.str, number.len);
	0	4	if ((codepoint == '.' && number.len) || codepoint == ',') codepoint = utf8::decode(number.str, number.len);
8186	0	4	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
8187	0	4	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
	0	0	if (any_digit && (codepoint == 'e' || codepoint == 'E')) {
8189	0	0	if (codepoint == '+' || codepoint == '-') codepoint = utf8::decode(number.str, number.len);
8191	0	0	while (unicode::category(codepoint) & unicode::N) any_digit = true, codepoint = utf8::decode(number.str, number.len);
8194	0	4	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
	0	0	if (any_digit && !number.len && (!codepoint || codepoint == '.')) {
8195	0	0	lemmas.emplace_back(string(form.str, form.len), number_tag);
8202	4	4	while (form.len) {
8204	4	0	punctuation = punctuation && unicode::category(codepoint) & unicode::P;
	0	4	punctuation = punctuation && unicode::category(codepoint) & unicode::P;
8205	0	4	symbol = symbol && unicode::category(codepoint) & unicode::S;
	0	4	symbol = symbol && unicode::category(codepoint) & unicode::S;
8207	4	0	if (punctuation)
8208	4	0	lemmas.emplace_back(string(form_ori.str, form_ori.len), punctuation_tag);
8209	0	0	else if (symbol)
8210	0	0	lemmas.emplace_back(string(form_ori.str, form_ori.len), symbol_tag);
8274	0	0	return unique_ptr(new T(std::forward(args)...));
	1	0	return unique_ptr(new T(std::forward(args)...));
	0	0	return unique_ptr(new T(std::forward(args)...));
8300	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
8309	0	0	3);
	0	0	3);
8310	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
8316	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
8322	1	0	if (res->load(is)) return res.release();
	1	0	if (res->load(is)) return res.release();
8328	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
8334	0	0	if (!derinet->load(is)) return nullptr;
	0	0	if (!derinet->load(is)) return nullptr;
8336	0	0	unique_ptr dictionary(load(is));
8337	0	0	if (!dictionary) return nullptr;
8348	0	0	ifstream f(path_from_utf8(fname).c_str(), ifstream::binary);
8349	0	0	if (!f) return nullptr;
8351	0	0	return load(f);
8378	0	0	for (auto&& tag : tags) {
8380	0	0	for (unsigned i = 0; i < tag.size(); i++)
8391	0	0	if (!used) return false;
8393	0	0	for (auto&& used_rule : *used)
8394	0	0	if (used_rule == rule)
8406	0	0	string rule_label; rule_label.reserve(12);
8408	0	0	for (; suffix_len < form.len; suffix_len++) {
8409	0	0	rule_label.push_back(form.str[form.len - (suffix_len + 1)]);
8410	0	0	if (!rules.at(rule_label.c_str(), rule_label.size(), [](pointer_decoder& data){ data.next(data.next_2B()); }))
8414	0	0	for (suffix_len++; suffix_len--; ) {
8416	0	0	rule_label.push_back(' ');
8420	0	0	for (unsigned prefix_len = 0; prefix_len + suffix_len <= form.len; prefix_len++) {
8421	0	0	if (prefix_len) rule_label.push_back(form.str[prefix_len - 1]);
	0	0	if (prefix_len) rule_label.push_back(form.str[prefix_len - 1]);
8423	0	0	if (!found) break;
8424	0	0	if (*(found += sizeof(uint16_t))) {
8430	0	0	if (rule) {
8432	0	0	if (rule_label.size() > 1 && !contains(used, rule_label)) { // ignore rule ' '
	0	0	if (rule_label.size() > 1 && !contains(used, rule_label)) { // ignore rule ' '
	0	0	if (rule_label.size() > 1 && !contains(used, rule_label)) { // ignore rule ' '
8433	0	0	if (used) used->push_back(rule_label);
	0	0	if (used) used->push_back(rule_label);
8434	0	0	for (int rules_len = *rule++; rules_len; rules_len--) {
8441	0	0	if (pref_del_len + suff_del_len > form.len ||
	0	0	if (pref_del_len + suff_del_len > form.len ||
8442	0	0	(pref_del_len && !small_memeq(pref_del, form.str, pref_del_len)) ||
	0	0	(pref_del_len && !small_memeq(pref_del, form.str, pref_del_len)) ||
8443	0	0	(suff_del_len && !small_memeq(suff_del, form.str + form.len - suff_del_len, suff_del_len)) ||
	0	0	(suff_del_len && !small_memeq(suff_del, form.str + form.len - suff_del_len, suff_del_len)) ||
	0	0	(suff_del_len && !small_memeq(suff_del, form.str + form.len - suff_del_len, suff_del_len)) ||
8448	0	0	lemma.reserve(form.len + pref_add_len - pref_del_len + suff_add_len - suff_del_len);
8449	0	0	if (pref_add_len) lemma.append(pref_add, pref_add_len);
	0	0	if (pref_add_len) lemma.append(pref_add, pref_add_len);
8450	0	0	if (pref_del_len + suff_del_len < form.len) lemma.append(form.str + pref_del_len, form.len - pref_del_len - suff_del_len);
	0	0	if (pref_del_len + suff_del_len < form.len) lemma.append(form.str + pref_del_len, form.len - pref_del_len - suff_del_len);
8451	0	0	if (suff_add_len) lemma.append(suff_add, suff_add_len);
	0	0	if (suff_add_len) lemma.append(suff_add, suff_add_len);
8452	0	0	while (tags_len--)
8453	0	0	lemmas.emplace_back(lemma, this->tags[unaligned_load_inc(tags)]);
8461	0	0	if (lemmas.size() == lemmas_initial_size)
8462	0	0	if (!contains(used, string())) {
8463	0	0	if (used) used->push_back(string());
8464	0	0	lemmas.emplace_back(string(form.str, form.len), tags[default_tag]);
8486	0	0	if (!filter) return;
8488	0	0	wildcard.assign(filter);
8491	0	0	for (int tag_pos = 0, filter_pos = 0; filter[filter_pos]; tag_pos++, filter_pos++) {
8492	0	0	if (filter[filter_pos] == '?') continue;
8493	0	0	if (filter[filter_pos] == '[') {
8497	0	0	if (filter[filter_pos] == '^') negate = true, filter_pos++;
8500	0	0	for (bool first = true; filter[filter_pos] && (first || filter[filter_pos] != ']'); first = false)
	0	0	for (bool first = true; filter[filter_pos] && (first || filter[filter_pos] != ']'); first = false)
	0	0	for (bool first = true; filter[filter_pos] && (first || filter[filter_pos] != ']'); first = false)
8503	0	0	filters.emplace_back(tag_pos, negate, chars_start, filter_pos - chars_start);
8504	0	0	if (!filter[filter_pos]) break;
8506	0	0	filters.emplace_back(tag_pos, false, filter_pos, 1);
8599	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	8	3	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	11	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	11	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	11	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	7	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	6	1	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	2	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	2	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
	0	0	return it ? unaligned_load(it) : elementary_feature_unknown;
8607	1	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
8610	1	0	maps.resize(data.next_1B());
	1	0	maps.resize(data.next_1B());
8611	26	1	for (auto&& map : maps)
8612	26	0	map.load(data);
	0	0	map.load(data);
8654	267	0	if (value < 0x80) *where++ = value;
8655	0	0	else if (value < 0x4000) *where++ = (value >> 7) | 0x80u, *where++ = value & 0x7Fu;
8656	0	0	else if (value < 0x200000) *where++ = (value >> 14) | 0x80u, *where++ = ((value >> 7) & 0x7Fu) | 0x80u, *where++ = value & 0x7Fu;
8657	0	0	else if (value < 0x10000000) *where++ = (value >> 21) | 0x80u, *where++ = ((value >> 14) & 0x7Fu) | 0x80u, *where++ = ((value >> 7) & 0x7Fu) | 0x80u, *where++ = value & 0x7Fu;
8705	0	0	class feature_sequences {
	1	0	class feature_sequences {
	0	0	class feature_sequences {
8734	0	0	return it ? unaligned_load(it) : 0;
	55	53	return it ? unaligned_load(it) : 0;
	0	0	return it ? unaligned_load(it) : 0;
8743	0	0	if (!elementary.load(is)) return false;
	1	0	if (!elementary.load(is)) return false;
	0	0	if (!elementary.load(is)) return false;
8746	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
	0	0	if (!compressor::load(is, data)) return false;
8749	0	0	sequences.resize(data.next_1B());
	0	0	sequences.resize(data.next_1B());
	1	0	sequences.resize(data.next_1B());
	1	0	sequences.resize(data.next_1B());
	0	0	sequences.resize(data.next_1B());
	0	0	sequences.resize(data.next_1B());
8750	0	0	for (auto&& sequence : sequences) {
	21	1	for (auto&& sequence : sequences) {
	0	0	for (auto&& sequence : sequences) {
8751	0	0	sequence.dependant_range = data.next_4B();
	21	0	sequence.dependant_range = data.next_4B();
	0	0	sequence.dependant_range = data.next_4B();
8752	0	0	sequence.elements.resize(data.next_1B());
	0	0	sequence.elements.resize(data.next_1B());
	21	0	sequence.elements.resize(data.next_1B());
	21	0	sequence.elements.resize(data.next_1B());
	0	0	sequence.elements.resize(data.next_1B());
	0	0	sequence.elements.resize(data.next_1B());
8753	0	0	for (auto&& element : sequence.elements) {
	45	21	for (auto&& element : sequence.elements) {
	0	0	for (auto&& element : sequence.elements) {
8754	0	0	element.type = elementary_feature_type(data.next_4B());
	45	0	element.type = elementary_feature_type(data.next_4B());
	0	0	element.type = elementary_feature_type(data.next_4B());
8755	0	0	element.elementary_index = data.next_4B();
	45	0	element.elementary_index = data.next_4B();
	0	0	element.elementary_index = data.next_4B();
8756	0	0	element.sequence_index = data.next_4B();
	45	0	element.sequence_index = data.next_4B();
	0	0	element.sequence_index = data.next_4B();
8760	0	0	scores.resize(data.next_1B());
	0	0	scores.resize(data.next_1B());
	1	0	scores.resize(data.next_1B());
	1	0	scores.resize(data.next_1B());
	0	0	scores.resize(data.next_1B());
	0	0	scores.resize(data.next_1B());
8761	0	0	for (auto&& score : scores)
	21	1	for (auto&& score : scores)
	0	0	for (auto&& score : scores)
8762	0	0	score.load(data);
	21	0	score.load(data);
	0	0	score.load(data);
	0	0	score.load(data);
	0	0	score.load(data);
	0	0	score.load(data);
8782	0	0	cache_element(int elements) : key(vli::max_length() * elements), key_size(0), score(0) {}
	0	0	cache_element(int elements) : key(vli::max_length() * elements), key_size(0), score(0) {}
	0	0	cache_element(int elements) : key(vli::max_length() * elements), key_size(0), score(0) {}
8790	0	0	caches.reserve(self.sequences.size());
	1	0	caches.reserve(self.sequences.size());
	0	0	caches.reserve(self.sequences.size());
8792	0	0	for (auto&& sequence : self.sequences) {
	21	1	for (auto&& sequence : self.sequences) {
	0	0	for (auto&& sequence : self.sequences) {
8793	0	0	caches.emplace_back(int(sequence.elements.size()));
	21	0	caches.emplace_back(int(sequence.elements.size()));
	0	0	caches.emplace_back(int(sequence.elements.size()));
8794	0	0	if (int(sequence.elements.size()) > max_sequence_elements) max_sequence_elements = sequence.elements.size();
	1	20	if (int(sequence.elements.size()) > max_sequence_elements) max_sequence_elements = sequence.elements.size();
	0	0	if (int(sequence.elements.size()) > max_sequence_elements) max_sequence_elements = sequence.elements.size();
8795	0	0	for (auto&& element : sequence.elements)
	45	21	for (auto&& element : sequence.elements)
	0	0	for (auto&& element : sequence.elements)
8796	0	0	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
	0	0	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
	26	19	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
	2	24	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
	0	0	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
	0	0	if (element.type == PER_TAG && 1 - element.sequence_index > max_window_size)
8799	0	0	key.resize(max_sequence_elements * vli::max_length());
	1	0	key.resize(max_sequence_elements * vli::max_length());
	0	0	key.resize(max_sequence_elements * vli::max_length());
8800	0	0	window.resize(max_window_size);
	1	0	window.resize(max_window_size);
	0	0	window.resize(max_window_size);
8811	0	0	if (forms.size() > c.elementary_per_form.size()) c.elementary_per_form.resize(forms.size() * 2);
	1	1	if (forms.size() > c.elementary_per_form.size()) c.elementary_per_form.resize(forms.size() * 2);
	0	0	if (forms.size() > c.elementary_per_form.size()) c.elementary_per_form.resize(forms.size() * 2);
8812	0	0	if (forms.size() > c.elementary_per_tag.size()) c.elementary_per_tag.resize(forms.size() * 2);
	1	1	if (forms.size() > c.elementary_per_tag.size()) c.elementary_per_tag.resize(forms.size() * 2);
	0	0	if (forms.size() > c.elementary_per_tag.size()) c.elementary_per_tag.resize(forms.size() * 2);
8813	0	0	for (unsigned i = 0; i < forms.size(); i++)
	7	2	for (unsigned i = 0; i < forms.size(); i++)
	0	0	for (unsigned i = 0; i < forms.size(); i++)
8814	0	0	if (analyses[i].size() > c.elementary_per_tag[i].size())
	5	2	if (analyses[i].size() > c.elementary_per_tag[i].size())
	0	0	if (analyses[i].size() > c.elementary_per_tag[i].size())
8822	0	0	for (auto&& cache : c.caches)
	42	2	for (auto&& cache : c.caches)
	0	0	for (auto&& cache : c.caches)
8828	0	0	elementary.compute_dynamic_features((*c.analyses)[form_index][tag_index], c.elementary_per_form[form_index], c.elementary_per_tag[form_index][tag_index], form_index > 0 ? prev_dynamic : nullptr, dynamic);
	20	2	elementary.compute_dynamic_features((*c.analyses)[form_index][tag_index], c.elementary_per_form[form_index], c.elementary_per_tag[form_index][tag_index], form_index > 0 ? prev_dynamic : nullptr, dynamic);
	0	0	elementary.compute_dynamic_features((*c.analyses)[form_index][tag_index], c.elementary_per_form[form_index], c.elementary_per_tag[form_index][tag_index], form_index > 0 ? prev_dynamic : nullptr, dynamic);
8834	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	54	16	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	2	52	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	52	18	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
	0	0	for (int i = 0; i < int(c.window.size()) && form_index - i >= 0; i++)
8839	0	0	for (unsigned i = 0; i < sequences.size(); i++) {
	190	7	for (unsigned i = 0; i < sequences.size(); i++) {
	0	0	for (unsigned i = 0; i < sequences.size(); i++) {
8840	0	0	if (tags_unchanged >= sequences[i].dependant_range)
	179	11	if (tags_unchanged >= sequences[i].dependant_range)
	0	0	if (tags_unchanged >= sequences[i].dependant_range)
8844	0	0	for (unsigned j = 0; j < sequences[i].elements.size(); j++) {
	332	114	for (unsigned j = 0; j < sequences[i].elements.size(); j++) {
	0	0	for (unsigned j = 0; j < sequences[i].elements.size(); j++) {
8850	0	0	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
	0	0	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
	91	4	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
	86	5	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
	0	0	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
	0	0	value = form_index + element.sequence_index < 0 || unsigned(form_index + element.sequence_index) >= c.forms->size() ? elementary_feature_empty : c.elementary_per_form[form_index + element.sequence_index].values[element.elementary_index];
8853	0	0	value = form_index + element.sequence_index < 0 ? elementary_feature_empty : c.window[-element.sequence_index]->values[element.elementary_index];
	235	2	value = form_index + element.sequence_index < 0 ? elementary_feature_empty : c.window[-element.sequence_index]->values[element.elementary_index];
	0	0	value = form_index + element.sequence_index < 0 ? elementary_feature_empty : c.window[-element.sequence_index]->values[element.elementary_index];
8860	0	0	if (value == elementary_feature_unknown) {
	65	267	if (value == elementary_feature_unknown) {
	0	0	if (value == elementary_feature_unknown) {
8869	0	0	if (!key_size) {
	65	114	if (!key_size) {
	0	0	if (!key_size) {
8872	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	84	30	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	78	6	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	108	6	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
	0	0	} else if (key_size != c.caches[i].key_size || !small_memeq(c.key.data(), c.caches[i].key.data(), key_size)) {
8933	0	0	cache(const viterbi& self) : features_cache(self.features) {}
	1	0	cache(const viterbi& self) : features_cache(self.features) {}
	0	0	cache(const viterbi& self) : features_cache(self.features) {}
8946	0	0	if (!forms.size()) return;
	2	0	if (!forms.size()) return;
	0	0	if (!forms.size()) return;
8950	0	0	for (unsigned i = 0, states = 1; i < forms.size(); i++) {
	7	2	for (unsigned i = 0, states = 1; i < forms.size(); i++) {
	0	0	for (unsigned i = 0, states = 1; i < forms.size(); i++) {
8951	0	0	if (analyses[i].empty()) return;
	7	0	if (analyses[i].empty()) return;
	0	0	if (analyses[i].empty()) return;
8952	0	0	states = (i+1 >= unsigned(decoding_order) ? states / analyses[i-decoding_order+1].size() : states) * analyses[i].size();
	3	4	states = (i+1 >= unsigned(decoding_order) ? states / analyses[i-decoding_order+1].size() : states) * analyses[i].size();
	0	0	states = (i+1 >= unsigned(decoding_order) ? states / analyses[i-decoding_order+1].size() : states) * analyses[i].size();
8955	0	0	if (nodes > c.nodes.size()) c.nodes.resize(nodes);
	2	0	if (nodes > c.nodes.size()) c.nodes.resize(nodes);
	0	0	if (nodes > c.nodes.size()) c.nodes.resize(nodes);
8961	0	0	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
	0	0	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
	0	2	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
	0	0	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
	0	0	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
	0	0	int* window = window_size <= 16 ? window_stack : (window_heap.resize(window_size), window_heap.data());
8967	0	0	for (unsigned i = 0; i < forms.size(); i++) {
	7	2	for (unsigned i = 0; i < forms.size(); i++) {
	0	0	for (unsigned i = 0; i < forms.size(); i++) {
8970	0	0	for (int j = 0; j < window_size; j++) window[j] = -1;
	7	21	for (int j = 0; j < window_size; j++) window[j] = -1;
	0	0	for (int j = 0; j < window_size; j++) window[j] = -1;
8971	0	0	for (int tag = 0; tag < int(analyses[i].size()); tag++)
	11	7	for (int tag = 0; tag < int(analyses[i].size()); tag++)
	0	0	for (int tag = 0; tag < int(analyses[i].size()); tag++)
8972	0	0	for (int prev = nodes_prev; prev < nodes_now; prev++) {
	22	11	for (int prev = nodes_prev; prev < nodes_now; prev++) {
	0	0	for (int prev = nodes_prev; prev < nodes_now; prev++) {
8976	0	0	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
	0	0	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
	45	14	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
	37	8	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
	0	0	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
	0	0	for (int p = prev, n = 1; p >= 0 && n < window_size; p = c.nodes[p].prev, n++) {
8977	0	0	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
	0	0	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
	15	22	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
	11	4	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
	0	0	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
	0	0	same_tags += same_tags == n && window[n] == c.nodes[p].tag;
8982	0	0	features.compute_dynamic_features(i, tag, prev >= 0 ? &c.nodes[prev].dynamic : nullptr, dynamic, c.features_cache);
	20	2	features.compute_dynamic_features(i, tag, prev >= 0 ? &c.nodes[prev].dynamic : nullptr, dynamic, c.features_cache);
	0	0	features.compute_dynamic_features(i, tag, prev >= 0 ? &c.nodes[prev].dynamic : nullptr, dynamic, c.features_cache);
8983	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	6	16	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	2	4	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	20	2	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
	0	0	score = (nodes_prev + 1 == nodes_now && analyses[i].size() == 1 ? 0 : features.score(i, window, same_tags, dynamic, c.features_cache)) +
8987	0	0	if (same_tags >= decoding_order-1) {
	4	18	if (same_tags >= decoding_order-1) {
	0	0	if (same_tags >= decoding_order-1) {
8988	0	0	if (score <= c.nodes[nodes_next-1].score) continue;
	4	0	if (score <= c.nodes[nodes_next-1].score) continue;
	0	0	if (score <= c.nodes[nodes_next-1].score) continue;
9003	0	0	for (int node = nodes_prev + 1; node < nodes_now; node++)
	3	2	for (int node = nodes_prev + 1; node < nodes_now; node++)
	0	0	for (int node = nodes_prev + 1; node < nodes_now; node++)
9004	0	0	if (c.nodes[node].score > c.nodes[best].score)
	1	2	if (c.nodes[node].score > c.nodes[best].score)
	0	0	if (c.nodes[node].score > c.nodes[best].score)
9007	0	0	for (int i = forms.size() - 1; i >= 0; i--, best = c.nodes[best].prev)
	7	2	for (int i = forms.size() - 1; i >= 0; i--, best = c.nodes[best].prev)
	0	0	for (int i = forms.size() - 1; i >= 0; i--, best = c.nodes[best].prev)
9056	0	0	maps.resize(MAP_TOTAL);
9105	0	0	for (unsigned i = forms.size(); i--;) {
9109	0	0	for (unsigned j = 0; j < analyses[i].size(); j++) {
9120	0	0	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == lemma ? per_tag[i][j-1].values[LEMMA] :
	0	0	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == lemma ? per_tag[i][j-1].values[LEMMA] :
9125	0	0	if (index == string::npos) index = tag.size();
9126	0	0	per_tag[i][j].values[TAG_UPOS] = maps[MAP_TAG_UPOS].value(tag.c_str() + (index ? 1 : 0), index - (index ? 1 : 0));
	0	0	per_tag[i][j].values[TAG_UPOS] = maps[MAP_TAG_UPOS].value(tag.c_str() + (index ? 1 : 0), index - (index ? 1 : 0));
9128	0	0	if (index < tag.size()) index++;
9129	0	0	if (index < tag.size()) index = tag.find(separator, index);
9130	0	0	if (index < tag.size()) index++;
9131	0	0	for (size_t length; index < tag.size(); index += length + 1) {
9133	0	0	length = (length == string::npos ? tag.size() : length) - index;
9135	0	0	for (size_t equal_sign = 0; equal_sign + 1 < length; equal_sign++)
9136	0	0	if (tag[index + equal_sign] == '=') {
9140	0	0	if (tag.compare(index, equal_sign, "Case") == 0) value = TAG_CASE, map = MAP_TAG_CASE;
9143	0	0	if (tag.compare(index, equal_sign, "Gender") == 0) value = TAG_GENDER, map = MAP_TAG_GENDER;
9144	0	0	if (tag.compare(index, equal_sign, "Number") == 0) value = TAG_NUMBER, map = MAP_TAG_NUMBER;
9145	0	0	if (tag.compare(index, equal_sign, "Person") == 0) value = TAG_PERSON, map = MAP_TAG_PERSON;
9148	0	0	if (tag.compare(index, equal_sign, "Negative") == 0) value = TAG_NEGATIVE, map = MAP_TAG_NEGATIVE;
9152	0	0	if (value >= 0)
9158	0	0	if (tag.size() >= 2 && tag[1] == 'V') {
	0	0	if (tag.size() >= 2 && tag[1] == 'V') {
	0	0	if (tag.size() >= 2 && tag[1] == 'V') {
9160	0	0	verb_candidate = verb_candidate < 0 || (tag_compare = tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
	0	0	verb_candidate = verb_candidate < 0 || (tag_compare = tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
9170	0	0	if (verb_candidate >= 0) {
9176	0	0	if (analyses[i].size() == 1) {
9184	0	0	} else if (forms[i].len <= 0) {
9199	0	0	while (form.len) {
9203	0	0	num = num || cat & unicode::N;
	0	0	num = num || cat & unicode::N;
9204	0	0	cap = cap || cat & unicode::Lut;
	0	0	cap = cap || cat & unicode::Lut;
9205	0	0	dash = dash || cat & unicode::Pd;
	0	0	dash = dash || cat & unicode::Pd;
9207	0	0	if (index == 10 || (!form.len && index < 10)) {
	0	0	if (index == 10 || (!form.len && index < 10)) {
	0	0	if (index == 10 || (!form.len && index < 10)) {
9237	0	0	if (prev_dynamic) {
9245	0	0	if (tag.tag.size() >= 2 && tag.tag[1] == 'V') {
	0	0	if (tag.tag.size() >= 2 && tag.tag[1] == 'V') {
	0	0	if (tag.tag.size() >= 2 && tag.tag[1] == 'V') {
9299	0	0	maps.resize(MAP_TOTAL);
9335	0	0	for (unsigned i = forms.size(); i--;) {
9339	0	0	for (unsigned j = 0; j < analyses[i].size(); j++) {
9342	0	0	per_tag[i][j].values[TAG3] = analyses[i][j].tag.size() >= 3 ? maps[MAP_TAG3].value(analyses[i][j].tag.c_str() + 2, 1) : elementary_feature_empty;
9343	0	0	per_tag[i][j].values[TAG5] = analyses[i][j].tag.size() >= 5 ? maps[MAP_TAG5].value(analyses[i][j].tag.c_str() + 4, 1) : elementary_feature_empty;
9344	0	0	per_tag[i][j].values[TAG25] = analyses[i][j].tag.size() >= 5 ? maps[MAP_TAG25].value((tag25[0] = analyses[i][j].tag[1], tag25[1] = analyses[i][j].tag[4], tag25), 2) : elementary_feature_empty;
9345	0	0	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == analyses[i][j].lemma ? per_tag[i][j-1].values[LEMMA] :
	0	0	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == analyses[i][j].lemma ? per_tag[i][j-1].values[LEMMA] :
9348	0	0	if (analyses[i][j].tag[0] == 'V') {
9350	0	0	verb_candidate = verb_candidate < 0 || (tag_compare = analyses[i][j].tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && analyses[i][j].lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
	0	0	verb_candidate = verb_candidate < 0 || (tag_compare = analyses[i][j].tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && analyses[i][j].lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
9360	0	0	if (verb_candidate >= 0) {
9366	0	0	if (analyses[i].size() == 1) {
9370	0	0	} else if (forms[i].len <= 0) {
9381	0	0	while (form.len) {
9385	0	0	num = num || cat & unicode::N;
	0	0	num = num || cat & unicode::N;
9386	0	0	cap = cap || cat & unicode::Lut;
	0	0	cap = cap || cat & unicode::Lut;
9387	0	0	dash = dash || cat & unicode::Pd;
	0	0	dash = dash || cat & unicode::Pd;
9389	0	0	if (index == 5 || (!form.len && index < 5)) {
	0	0	if (index == 5 || (!form.len && index < 5)) {
	0	0	if (index == 5 || (!form.len && index < 5)) {
9409	0	0	if (prev_dynamic) {
9417	0	0	if (tag.tag[0] == 'V') {
9471	1	0	maps.resize(MAP_TOTAL);
9519	7	2	for (unsigned i = forms.size(); i--;) {
9523	11	7	for (unsigned j = 0; j < analyses[i].size(); j++) {
9525	11	0	per_tag[i][j].values[TAG1] = analyses[i][j].tag.size() >= 1 ? maps[MAP_TAG1].value(analyses[i][j].tag.c_str() + 0, 1) : elementary_feature_empty;
9526	11	0	per_tag[i][j].values[TAG2] = analyses[i][j].tag.size() >= 2 ? maps[MAP_TAG2].value(analyses[i][j].tag.c_str() + 1, 1) : elementary_feature_empty;
9527	11	0	per_tag[i][j].values[TAG3] = analyses[i][j].tag.size() >= 3 ? maps[MAP_TAG3].value(analyses[i][j].tag.c_str() + 2, 1) : elementary_feature_empty;
9528	2	9	per_tag[i][j].values[TAG4] = analyses[i][j].tag.size() >= 4 ? maps[MAP_TAG4].value(analyses[i][j].tag.c_str() + 3, 1) : elementary_feature_empty;
9529	0	11	per_tag[i][j].values[TAG5] = analyses[i][j].tag.size() >= 5 ? maps[MAP_TAG5].value(analyses[i][j].tag.c_str() + 4, 1) : elementary_feature_empty;
9530	4	7	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == analyses[i][j].lemma ? per_tag[i][j-1].values[LEMMA] :
	4	0	per_tag[i][j].values[LEMMA] = j && analyses[i][j-1].lemma == analyses[i][j].lemma ? per_tag[i][j-1].values[LEMMA] :
9533	3	8	if (analyses[i][j].tag[0] == 'V') {
9535	1	2	verb_candidate = verb_candidate < 0 || (tag_compare = analyses[i][j].tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && analyses[i][j].lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
	1	0	verb_candidate = verb_candidate < 0 || (tag_compare = analyses[i][j].tag.compare(analyses[i][verb_candidate].tag), tag_compare < 0) || (tag_compare == 0 && analyses[i][j].lemma < analyses[i][verb_candidate].lemma) ? j : verb_candidate;
9545	2	5	if (verb_candidate >= 0) {
9551	5	2	if (analyses[i].size() == 1) {
9559	0	2	} else if (forms[i].len <= 0) {
9574	9	2	while (form.len) {
9578	9	0	num = num || cat & unicode::N;
	9	0	num = num || cat & unicode::N;
9579	9	0	cap = cap || cat & unicode::Lut;
	9	0	cap = cap || cat & unicode::Lut;
9580	9	0	dash = dash || cat & unicode::Pd;
	9	0	dash = dash || cat & unicode::Pd;
9582	9	0	if (index == 10 || (!form.len && index < 10)) {
	7	2	if (index == 10 || (!form.len && index < 10)) {
	0	2	if (index == 10 || (!form.len && index < 10)) {
9612	20	2	if (prev_dynamic) {
9620	3	19	if (tag.tag[0] == 'V') {
9667	0	0	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
	0	0	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
9676	0	0	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
	0	0	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
	0	2	while (lock.test_and_set(memory_order_acquire)) {}
9677	0	0	if (!stack.empty()) {
	1	1	if (!stack.empty()) {
	0	0	if (!stack.empty()) {
	1	1	if (!stack.empty()) {
	1	1	if (!stack.empty()) {
9728	0	0	cache(const perceptron_tagger& self) : decoder_cache(self.decoder) {}
	1	0	cache(const perceptron_tagger& self) : decoder_cache(self.decoder) {}
	0	0	cache(const perceptron_tagger& self) : decoder_cache(self.decoder) {}
9742	0	0	if (dict.reset(morpho::load(is)), !dict) return false;
	1	0	if (dict.reset(morpho::load(is)), !dict) return false;
	0	0	if (dict.reset(morpho::load(is)), !dict) return false;
9744	0	0	if (!features.load(is)) return false;
	1	0	if (!features.load(is)) return false;
	0	0	if (!features.load(is)) return false;
9756	0	0	if (!dict) return;
	2	0	if (!dict) return;
	0	0	if (!dict) return;
9759	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	1	1	if (!c) c = new cache(*this);
	1	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
9762	0	0	if (c->analyses.size() < forms.size()) c->analyses.resize(forms.size());
	2	0	if (c->analyses.size() < forms.size()) c->analyses.resize(forms.size());
	0	0	if (c->analyses.size() < forms.size()) c->analyses.resize(forms.size());
9763	0	0	for (unsigned i = 0; i < forms.size(); i++) {
	7	2	for (unsigned i = 0; i < forms.size(); i++) {
	0	0	for (unsigned i = 0; i < forms.size(); i++) {
9766	0	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
	0	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
	7	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
	7	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
	0	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
	0	0	dict->analyze(forms[i], guesser >= 0 ? guesser : use_guesser ? morpho::GUESSER : morpho::NO_GUESSER, c->analyses[i]);
9769	0	0	if (c->tags.size() < forms.size()) c->tags.resize(forms.size() * 2);
	1	1	if (c->tags.size() < forms.size()) c->tags.resize(forms.size() * 2);
	0	0	if (c->tags.size() < forms.size()) c->tags.resize(forms.size() * 2);
9772	0	0	for (unsigned i = 0; i < forms.size(); i++)
	7	2	for (unsigned i = 0; i < forms.size(); i++)
	0	0	for (unsigned i = 0; i < forms.size(); i++)
9783	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
	0	0	if (!c) c = new cache(*this);
9878	1	0	tagger_id id = tagger_id(is.get());
	0	1	tagger_id id = tagger_id(is.get());
	0	0	tagger_id id = tagger_id(is.get());
	0	0	tagger_id id = tagger_id(is.get());
	0	0	tagger_id id = tagger_id(is.get());
	0	0	tagger_id id = tagger_id(is.get());
	0	0	tagger_id id = tagger_id(is.get());
9884	0	0	auto res = new_unique_ptr>>(tagger_ids::decoding_order(id), tagger_ids::window_size(id));
9885	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
9893	1	0	auto res = new_unique_ptr>>(tagger_ids::decoding_order(id), tagger_ids::window_size(id));
9894	1	0	if (res->load(is)) return res.release();
	1	0	if (res->load(is)) return res.release();
9901	0	0	auto res = new_unique_ptr>>(tagger_ids::decoding_order(id), tagger_ids::window_size(id));
9902	0	0	if (res->load(is)) return res.release();
	0	0	if (res->load(is)) return res.release();
9911	0	0	ifstream f(path_from_utf8(fname).c_str(), ifstream::binary);
9912	0	0	if (!f) return nullptr;
9914	0	0	return load(f);
9919	0	0	return morpho ? morpho->new_tokenizer() : nullptr;
10024	0	0	for (int i = 0; i < 15 && pdt_tag[i]; i++)
	0	0	for (int i = 0; i < 15 && pdt_tag[i]; i++)
10025	0	0	if (pdt_tag[i] != '-') {
10026	0	0	if (!tag.empty()) tag.push_back('|');
10033	0	0	for (unsigned i = 0; i + 2 < lemma.size(); i++)
10034	0	0	if (lemma[i] == '_' && lemma[i + 1] == ';') {
	0	0	if (lemma[i] == '_' && lemma[i + 1] == ';') {
	0	0	if (lemma[i] == '_' && lemma[i + 1] == ';') {
10035	0	0	if (!tag.empty()) tag.push_back('|');
10044	0	0	return raw_lemma < lemma.size() ? (lemma.resize(raw_lemma), true) : false;
10055	0	0	for (auto&& tagged_lemma : tagged_lemmas) {
10061	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
10069	0	0	for (auto&& tagged_lemma_forms : forms) {
10070	0	0	for (auto&& tagged_form : tagged_lemma_forms.forms)
10076	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
10130	0	0	return lemma_id_len < lemma.size() ? (lemma.resize(lemma_id_len), true) : false;
10140	0	0	for (auto&& tagged_lemma : tagged_lemmas)
10144	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
10152	0	0	for (auto&& tagged_lemma_forms : forms)
10156	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
10210	0	0	return raw_lemma_len < lemma.size() ? (lemma.resize(raw_lemma_len), true) : false;
10220	0	0	for (auto&& tagged_lemma : tagged_lemmas)
10224	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
	0	0	if (!lemma_changed || tagged_lemmas.size() < 2) return;
10232	0	0	for (auto&& tagged_lemma_forms : forms)
10236	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
	0	0	if (!lemma_changed || forms.size() < 2) return;
10275	0	0	if (name == "pdt_to_conll2009") return tagset_converter::new_pdt_to_conll2009_converter();
10276	0	0	if (name == "strip_lemma_comment") return tagset_converter::new_strip_lemma_comment_converter(dictionary);
10277	0	0	if (name == "strip_lemma_id") return tagset_converter::new_strip_lemma_id_converter(dictionary);
10284	0	0	inline static bool eq(const tagged_lemma& a, const tagged_lemma& b) { return a.lemma == b.lemma && a.tag == b.tag; }
	0	0	inline static bool eq(const tagged_lemma& a, const tagged_lemma& b) { return a.lemma == b.lemma && a.tag == b.tag; }
10285	0	0	inline static bool lt(const tagged_lemma& a, const tagged_lemma& b) { int lemma_compare = a.lemma.compare(b.lemma); return lemma_compare < 0 || (lemma_compare == 0 && a.tag < b.tag); }
10294	0	0	for (unsigned i = 0; i < forms.size(); i++) {
10296	0	0	for (unsigned j = forms.size() - 1; j > i; j--)
10297	0	0	if (forms[j].lemma == forms[i].lemma) {
10299	0	0	for (auto&& tagged_form : forms[j].forms)
10303	0	0	if (j < forms.size() - 1) {
10311	0	0	if (any_merged && forms[i].forms.size() > 1) {
	0	0	if (any_merged && forms[i].forms.size() > 1) {
	0	0	if (any_merged && forms[i].forms.size() > 1) {
10314	0	0	inline static bool eq(const tagged_form& a, const tagged_form& b) { return a.tag == b.tag && a.form == b.form; }
	0	0	inline static bool eq(const tagged_form& a, const tagged_form& b) { return a.tag == b.tag && a.form == b.form; }
10315	0	0	inline static bool lt(const tagged_form& a, const tagged_form& b) { int tag_compare = a.tag.compare(b.tag); return tag_compare < 0 || (tag_compare == 0 && a.form < b.form); }
10473	321	3	const unordered_set czech_tokenizer::abbreviations_czech = {
	0	0	const unordered_set czech_tokenizer::abbreviations_czech = {
10489	309	3	const unordered_set czech_tokenizer::abbreviations_slovak = {
	0	0	const unordered_set czech_tokenizer::abbreviations_slovak = {
10506	0	0	: ragel_tokenizer(version <= 1 ? 1 : 2), m(m) {
	0	0	: ragel_tokenizer(version <= 1 ? 1 : 2), m(m) {
10520	0	0	if (!m) return;
10521	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unicode::L) return;
	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unicode::L) return;
	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unicode::L) return;
10524	0	0	for (unsigned hyphens = 1; hyphens <= 2; hyphens++) {
10526	0	0	if (tokens.size() < 2*hyphens + 1) break;
10528	0	0	if (tokens[first_hyphen].length != 1 || chars[tokens[first_hyphen].start].cat & ~unicode::P ||
	0	0	if (tokens[first_hyphen].length != 1 || chars[tokens[first_hyphen].start].cat & ~unicode::P ||
	0	0	if (tokens[first_hyphen].length != 1 || chars[tokens[first_hyphen].start].cat & ~unicode::P ||
10529	0	0	tokens[first_hyphen].start + tokens[first_hyphen].length != tokens[first_hyphen + 1].start ||
10530	0	0	tokens[first_hyphen-1].start + tokens[first_hyphen-1].length != tokens[first_hyphen].start ||
	0	0	tokens[first_hyphen-1].start + tokens[first_hyphen-1].length != tokens[first_hyphen].start ||
10534	0	0	if (m->analyze(string_piece(chars[tokens[first_hyphen-1].start].str, chars[tokens.back().start + tokens.back().length].str - chars[tokens[first_hyphen-1].start].str), morpho::NO_GUESSER, lemmas) >= 0)
10538	0	0	if (matched_hyphens) {
10552	0	0	while (tokenize_url_email(tokens))
10553	0	0	if (emergency_sentence_split(tokens))
10569	0	0	if ( ( current) == ( (chars.size() - 1)) )
10574	0	0	switch ( _czech_tokenizer_from_state_actions[cs] ) {
10583	0	0	if ( _klen > 0 ) {
10588	0	0	if ( _upper < _lower )
10592	0	0	if ( _widec < _mid[0] )
10594	0	0	else if ( _widec > _mid[1] )
10600	0	0	if (
10601	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
10606	0	0	if (
10607	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
10620	0	0	if ( _klen > 0 ) {
10625	0	0	if ( _upper < _lower )
10629	0	0	if ( _widec < *_mid )
10631	0	0	else if ( _widec > *_mid )
10643	0	0	if ( _klen > 0 ) {
10648	0	0	if ( _upper < _lower )
10652	0	0	if ( _widec < _mid[0] )
10654	0	0	else if ( _widec > _mid[1] )
10669	0	0	if ( _czech_tokenizer_trans_actions[_trans] == 0 )
10683	0	0	do
10684	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10692	0	0	for (current = ts; current < whitespace; current++)
10695	0	0	if (eos) {( current)++; goto _out; }
10700	0	0	if (!tokens.empty()) {( current)++; goto _out; }
10702	0	0	do
10703	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10712	0	0	do
10713	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10721	0	0	do
10722	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10729	0	0	if (!tokens.empty()) {( current)++; goto _out; }
10731	0	0	do
10732	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10741	0	0	do
10742	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
10750	0	0	switch ( _czech_tokenizer_to_state_actions[cs] ) {
10756	0	0	if ( cs == 0 )
10758	0	0	if ( ++( current) != ( (chars.size() - 1)) )
10761	0	0	if ( ( current) == ( (chars.size() - 1)) )
10763	0	0	if ( _czech_tokenizer_eof_trans[cs] > 0 ) {
10795	342	3	const unordered_set english_tokenizer::abbreviations = {
	0	0	const unordered_set english_tokenizer::abbreviations = {
10894	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unilib::unicode::L) return;
	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unilib::unicode::L) return;
	0	0	if (tokens.empty() || chars[tokens.back().start].cat & ~unilib::unicode::L) return;
10909	0	0	if ( ( index) == ( end) )
10918	0	0	if ( _klen > 0 ) {
10923	0	0	if ( _upper < _lower )
10927	0	0	if ( ( ragel_char(chars[tokens.back().start + end - index - 1])) < *_mid )
10929	0	0	else if ( ( ragel_char(chars[tokens.back().start + end - index - 1])) > *_mid )
10941	0	0	if ( _klen > 0 ) {
10946	0	0	if ( _upper < _lower )
10950	0	0	if ( ( ragel_char(chars[tokens.back().start + end - index - 1])) < _mid[0] )
10952	0	0	else if ( ( ragel_char(chars[tokens.back().start + end - index - 1])) > _mid[1] )
10966	0	0	if ( _english_tokenizer_split_token_trans_actions[_trans] == 0 )
10980	0	0	if ( cs == 0 )
10982	0	0	if ( ++( index) != ( end) )
10985	0	0	if ( ( index) == ( end) )
10987	0	0	switch ( _english_tokenizer_split_token_eof_actions[cs] ) {
10997	0	0	if (split_len && split_len < end) {
11151	0	0	english_tokenizer::english_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	english_tokenizer::english_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	english_tokenizer::english_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	english_tokenizer::english_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
11160	0	0	while (tokenize_url_email(tokens))
11161	0	0	if (emergency_sentence_split(tokens))
11177	0	0	if ( ( current) == ( (chars.size() - 1)) )
11182	0	0	switch ( _english_tokenizer_from_state_actions[cs] ) {
11191	0	0	if ( _klen > 0 ) {
11196	0	0	if ( _upper < _lower )
11200	0	0	if ( _widec < _mid[0] )
11202	0	0	else if ( _widec > _mid[1] )
11208	0	0	if (
11209	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
11214	0	0	if (
11215	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
11228	0	0	if ( _klen > 0 ) {
11233	0	0	if ( _upper < _lower )
11237	0	0	if ( _widec < *_mid )
11239	0	0	else if ( _widec > *_mid )
11251	0	0	if ( _klen > 0 ) {
11256	0	0	if ( _upper < _lower )
11260	0	0	if ( _widec < _mid[0] )
11262	0	0	else if ( _widec > _mid[1] )
11277	0	0	if ( _english_tokenizer_trans_actions[_trans] == 0 )
11291	0	0	do
11292	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11300	0	0	for (current = ts; current < whitespace; current++)
11303	0	0	if (eos) {( current)++; goto _out; }
11308	0	0	if (!tokens.empty()) {( current)++; goto _out; }
11310	0	0	do
11311	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11320	0	0	do
11321	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11329	0	0	do
11330	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11337	0	0	if (!tokens.empty()) {( current)++; goto _out; }
11339	0	0	do
11340	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11349	0	0	do
11350	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11358	0	0	switch ( _english_tokenizer_to_state_actions[cs] ) {
11364	0	0	if ( cs == 0 )
11366	0	0	if ( ++( current) != ( (chars.size() - 1)) )
11369	0	0	if ( ( current) == ( (chars.size() - 1)) )
11371	0	0	if ( _english_tokenizer_eof_trans[cs] > 0 ) {
11528	3	0	generic_tokenizer::generic_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	generic_tokenizer::generic_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	generic_tokenizer::generic_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
	0	0	generic_tokenizer::generic_tokenizer(unsigned version) : ragel_tokenizer(version <= 1 ? 1 : 2) {}
11537	0	2	while (tokenize_url_email(tokens))
11538	0	0	if (emergency_sentence_split(tokens))
11554	2	0	if ( ( current) == ( (chars.size() - 1)) )
11559	10	27	switch ( _generic_tokenizer_from_state_actions[cs] ) {
11568	10	27	if ( _klen > 0 ) {
11573	17	10	if ( _upper < _lower )
11577	3	14	if ( _widec < _mid[0] )
11579	14	0	else if ( _widec > _mid[1] )
11585	0	0	if (
11586	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
	0	0	!current || (chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N | unicode::Pd)) ) _widec += 256;
11591	0	0	if (
11592	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
	0	0	!current || ((chars[current-1].cat & ~(unicode::L | unicode::M | unicode::N)) && chars[current-1].chr != '+') ) _widec += 256;
11605	37	0	if ( _klen > 0 ) {
11610	112	27	if ( _upper < _lower )
11614	66	46	if ( _widec < *_mid )
11616	36	10	else if ( _widec > *_mid )
11628	24	3	if ( _klen > 0 ) {
11633	32	5	if ( _upper < _lower )
11637	10	22	if ( _widec < _mid[0] )
11639	3	19	else if ( _widec > _mid[1] )
11654	12	27	if ( _generic_tokenizer_trans_actions[_trans] == 0 )
11667	0	0	do
11668	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11676	0	0	for (current = ts; current < whitespace; current++)
11679	0	0	if (eos) {( current)++; goto _out; }
11684	0	0	if (!tokens.empty()) {( current)++; goto _out; }
11686	0	0	do
11687	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11695	0	7	do
11696	0	7	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11704	0	3	do
11705	0	3	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11712	0	0	if (!tokens.empty()) {( current)++; goto _out; }
11714	0	0	do
11715	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11723	0	0	do
11724	0	0	if (emergency_sentence_split(tokens)) { ( current)--; {( current)++; goto _out; } }
11732	10	29	switch ( _generic_tokenizer_to_state_actions[cs] ) {
11738	39	0	if ( cs == 0 )
11740	35	4	if ( ++( current) != ( (chars.size() - 1)) )
11743	4	0	if ( ( current) == ( (chars.size() - 1)) )
11745	2	2	if ( _generic_tokenizer_eof_trans[cs] > 0 ) {
12127	3	0	initialize_ragel_map();
12131	0	7	while (ragel_map_flag.test_and_set()) {}
12132	2	5	if (ragel_map.empty()) {
12133	256	2	for (uint8_t ascii = 0; ascii < 128; ascii++)
12145	2	6	if (chr >= ragel_map.size())
12165	10	0	if ( ( current) == ( (chars.size() - 1)) )
12173	0	22	if ( _klen > 0 ) {
12178	0	0	if ( _upper < _lower )
12182	0	0	if ( _widec < _mid[0] )
12184	0	0	else if ( _widec > _mid[1] )
12190	0	0	if (
12196	0	0	if (
12210	22	0	if ( _klen > 0 ) {
12215	63	22	if ( _upper < _lower )
12219	12	51	if ( _widec < *_mid )
12221	51	0	else if ( _widec > *_mid )
12233	22	0	if ( _klen > 0 ) {
12238	61	8	if ( _upper < _lower )
12242	8	53	if ( _widec < _mid[0] )
12244	39	14	else if ( _widec > _mid[1] )
12258	0	22	if ( _ragel_url_email_trans_actions[_trans] == 0 )
12279	14	8	if ( cs == 0 )
12281	12	2	if ( ++( current) != ( (chars.size() - 1)) )
12287	0	10	if (end > start) {
12316	1	0	vertical_tokenizer() : unicode_tokenizer(0) {}
12372	4	0	ragel_tokenizer::initialize_ragel_map();
12374	4	0	set_text(string_piece(nullptr, 0));
12380	3	4	if (make_copy && text.str) {
	3	0	if (make_copy && text.str) {
12387	145	7	for (const char* curr_str = text.str; text.len; curr_str = text.str)
12393	7	0	vector& tokens = tokens_ptr ? *tokens_ptr : tokens_buffer;
12395	7	0	if (forms) forms->clear();
12396	4	3	if (current >= chars.size() - 1) return false;
12399	4	0	if (forms)
12400	33	4	for (auto&& token : tokens)
12407	10	2	if (current >= chars.size() - 1) return false;
12409	10	0	return url_email_tokenizer ? ragel_tokenizer::ragel_url_email(url_email_tokenizer, chars, current, tokens) : false;
12416	0	10	return tokens.size() >= 500 ||
12417	10	0	(tokens.size() >= 450 && chars[tokens.back().start].cat & unicode::P) ||
	0	0	(tokens.size() >= 450 && chars[tokens.back().start].cat & unicode::P) ||
	0	10	(tokens.size() >= 450 && chars[tokens.back().start].cat & unicode::P) ||
12418	0	0	(tokens.size() >= 400 && chars[tokens.back().start].cat & unicode::Po);
12424	0	0	if (eos_chr == '.' && !tokens.empty()) {
	0	0	if (eos_chr == '.' && !tokens.empty()) {
	0	0	if (eos_chr == '.' && !tokens.empty()) {
12426	0	0	if (tokens.back().length == 1 && chars[tokens.back().start].cat & unicode::Lut)
	0	0	if (tokens.back().length == 1 && chars[tokens.back().start].cat & unicode::Lut)
	0	0	if (tokens.back().length == 1 && chars[tokens.back().start].cat & unicode::Lut)
12430	0	0	if (abbreviations) {
12432	0	0	for (size_t i = 0; i < tokens.back().length; i++)
12434	0	0	if (abbreviations->count(eos_buffer))
12459	2	0	if (current >= chars.size() - 1) return false;
12463	116	2	while (current < chars.size() - 1 && chars[current].chr != '\r' && chars[current].chr != '\n') current++;
	116	0	while (current < chars.size() - 1 && chars[current].chr != '\r' && chars[current].chr != '\n') current++;
	26	90	while (current < chars.size() - 1 && chars[current].chr != '\r' && chars[current].chr != '\n') current++;
	90	28	while (current < chars.size() - 1 && chars[current].chr != '\r' && chars[current].chr != '\n') current++;
12466	26	2	if (current < chars.size() - 1) {
12468	26	0	if (current < chars.size() - 1 &&
	0	26	if (current < chars.size() - 1 &&
	0	26	if (current < chars.size() - 1 &&
12469	0	0	((chars[current-1].chr == '\r' && chars[current].chr == '\n') ||
	26	0	((chars[current-1].chr == '\r' && chars[current].chr == '\n') ||
12470	0	26	(chars[current-1].chr == '\n' && chars[current].chr == '\r')))
12474	26	2	if (line_start < line_end)
12562	0	0	return {1, 11, 1, ""};
	0	0	return {1, 11, 1, ""};
12573	0	0	<< (morphodita.prerelease.empty() ? "" : "-") << morphodita.prerelease
	0	0	<< (morphodita.prerelease.empty() ? "" : "-") << morphodita.prerelease
12575	0	0	<< (other_libraries.empty() ? "" : " and ") << other_libraries << ")\n"
	0	0	<< (other_libraries.empty() ? "" : " and ") << other_libraries << ")\n"
12577	0	0	"Mathematics and Physics, Charles University in Prague, Czech Republic.";
12611	42	42	default: return (bilou_entity - B_first) & 1 ? bilou_type_U : bilou_type_B;
12615	28	42	switch (bilou_entity) {
12750	1	0	if (tagger.reset(tagger::load_instance(is)), !tagger) return false;
12751	1	0	if (!named_entities.load(is)) return false;
12754	1	0	if (!templates.load(is, nlp_pipeline(tokenizer.get(), tagger.get()))) return false;
	1	0	if (!templates.load(is, nlp_pipeline(tokenizer.get(), tagger.get()))) return false;
12756	1	0	int stages = is.get();
12757	1	0	if (stages == EOF) return false;
12758	1	0	networks.resize(stages);
12759	2	1	for (auto&& network : networks)
12760	2	0	if (!network.load(is)) return false;
	2	0	if (!network.load(is)) return false;
12767	2	0	if (forms.empty() || !tagger || !named_entities.size() || !networks.size()) return;
	2	0	if (forms.empty() || !tagger || !named_entities.size() || !networks.size()) return;
	2	0	if (forms.empty() || !tagger || !named_entities.size() || !networks.size()) return;
	0	2	if (forms.empty() || !tagger || !named_entities.size() || !networks.size()) return;
	2	0	if (forms.empty() || !tagger || !named_entities.size() || !networks.size()) return;
12771	1	1	if (!c) c = new cache();
12777	2	0	if (sentence.size) {
12781	4	2	for (auto&& network : networks) {
12789	14	4	for (unsigned i = 0; i < sentence.size; i++) {
12790	14	0	if (!sentence.probabilities[i].local_filled) {
12796	4	10	if (i == 0) {
12808	7	2	for (unsigned i = 0; i < sentence.size; i++)
12809	3	4	if (sentence.probabilities[i].global.best == bilou_type_U) {
12811	0	4	} else if (sentence.probabilities[i].global.best == bilou_type_B) {
12813	0	0	while (i < sentence.size && sentence.probabilities[i].global.best != bilou_type_L) i++;
	0	0	while (i < sentence.size && sentence.probabilities[i].global.best != bilou_type_L) i++;
	0	0	while (i < sentence.size && sentence.probabilities[i].global.best != bilou_type_L) i++;
12830	0	0	for (unsigned i = 0; i < types.size(); i++)
12836	0	0	if (gazetteer_types) gazetteer_types->clear();
12842	14	70	for (auto&& prob_bilou : prob.bilou)
12845	126	14	for (bilou_entity::value i = 0; i < outcomes.size(); i++) {
12847	70	56	if (outcomes[i] > prob.bilou[bilou].probability) {
12882	0	0	if (it == str2id.end() && add_entity) {
	0	0	if (it == str2id.end() && add_entity) {
	0	0	if (it == str2id.end() && add_entity) {
12886	0	0	return it == str2id.end() ? entity_type_unknown : it->second;
12890	0	0	return entity < id2str.size() ? id2str[entity] : empty;
	0	0	return entity < id2str.size() ? id2str[entity] : empty;
	3	0	return entity < id2str.size() ? id2str[entity] : empty;
	0	0	return entity < id2str.size() ? id2str[entity] : empty;
	0	0	return entity < id2str.size() ? id2str[entity] : empty;
12895	1	0	if (!compressor::load(is, data)) return false;
	1	0	if (!compressor::load(is, data)) return false;
12899	1	0	id2str.resize(data.next_4B());
	1	0	id2str.resize(data.next_4B());
12900	3	1	for (unsigned i = 0; i < id2str.size(); i++) {
12901	3	0	data.next_str(id2str[i]);
12903	0	0	}
12930	1	0	switch (id) {
12936	1	0	if (res->load(is)) return res.release();
	0	1	if (res->load(is)) return res.release();
12945	1	0	ifstream in(path_from_utf8(fname).c_str(), ifstream::in | ifstream::binary);
12946	1	0	if (!in.is_open()) return nullptr;
12948	1	0	return load(in);
12986	0	0	for (; len--; str++, pos++)
	0	0	for (; len--; str++, pos++)
	0	0	for (; len--; str++, pos++)
12987	0	0	if (*str == c)
	0	0	if (*str == c)
	0	0	if (*str == c)
13003	0	0	for (unsigned i = 0; i < forms.size(); i++) {
13007	0	0	if (space < form.len) {
13013	0	0	if (space < form.len) {
13081	1	0	morpho = tagger ? tagger->get_morpho() : nullptr;
13082	1	0	return tagger && morpho;
	0	1	return tagger && morpho;
13086	0	0	if (params.empty()) return cerr << "Missing tagger_file argument to morphodita_tagger!" << endl, false;
13089	0	0	if (!in.is_open()) return cerr << "Cannot open morphodita tagger file '" << params << "'!" << endl, false;
13090	0	0	if (!load(in)) return cerr << "Cannot load morphodita tagger from file '" << params << "'!" << endl, false;
	0	0	if (!load(in)) return cerr << "Cannot load morphodita tagger from file '" << params << "'!" << endl, false;
13092	0	0	if (!in.seekg(0, ifstream::beg)) return cerr << "Cannot seek in morphodita tagger file '" << params << "'!" << endl, false;
	0	0	if (!in.seekg(0, ifstream::beg)) return cerr << "Cannot seek in morphodita tagger file '" << params << "'!" << endl, false;
13093	0	0	os << in.rdbuf();
13100	2	0	if (!tagger || !morpho) return;
	2	0	if (!tagger || !morpho) return;
	2	0	if (!tagger || !morpho) return;
13104	1	1	if (!c) c = new cache();
13110	2	0	if (c->tags.size() >= forms.size()) {
13112	7	2	for (unsigned i = 0; i < forms.size(); i++) {
13122	11	7	for (auto&& analysis : c->analyses)
13123	11	0	sentence.words[i].raw_lemmas_all.emplace_back(analysis.lemma, 0, morpho->raw_lemma_len(analysis.lemma));
	11	0	sentence.words[i].raw_lemmas_all.emplace_back(analysis.lemma, 0, morpho->raw_lemma_len(analysis.lemma));
13174	1	0	if (!res) return nullptr;
13175	1	0	if (!res->load(is)) return nullptr;
	1	0	if (!res->load(is)) return nullptr;
13185	0	0	if (colon == string::npos) {
13188	0	0	tagger_id = tagger_id_and_params.substr(0, colon);
13189	0	0	params = tagger_id_and_params.substr(colon + 1);
13194	0	0	if (!tagger_ids::parse(tagger_id, id)) return cerr << "Unknown tagger_id '" << tagger_id << "'!" << endl, nullptr;
	0	0	if (!tagger_ids::parse(tagger_id, id)) return cerr << "Unknown tagger_id '" << tagger_id << "'!" << endl, nullptr;
13197	0	0	unique_ptr res(create(id));
13198	0	0	if (!res) return cerr << "Cannot create instance for tagger_id '" << tagger_id << "'!" << endl, nullptr;
13201	0	0	os.put(id);
13202	0	0	if (!res->create_and_encode(params, os)) return cerr << "Cannot encode instance of tagger_id '" << tagger_id << "'!" << endl, nullptr;
	0	0	if (!res->create_and_encode(params, os)) return cerr << "Cannot encode instance of tagger_id '" << tagger_id << "'!" << endl, nullptr;
13243	0	0	for (unsigned i = 0; i < forms.size(); i++) {
13544	0	0	for (; *str; str++)
13545	0	0	if (((unsigned char)*str) >= 0x80) {
13546	0	0	if (((unsigned char)*str) < 0xC0) return false;
13547	0	0	else if (((unsigned char)*str) < 0xE0) {
13548	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13549	0	0	} else if (((unsigned char)*str) < 0xF0) {
13550	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13551	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13552	0	0	} else if (((unsigned char)*str) < 0xF8) {
13553	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13554	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13555	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13562	0	0	for (; len > 0; str++, len--)
13563	0	0	if (((unsigned char)*str) >= 0x80) {
13564	0	0	if (((unsigned char)*str) < 0xC0) return false;
13565	0	0	else if (((unsigned char)*str) < 0xE0) {
13566	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13567	0	0	} else if (((unsigned char)*str) < 0xF0) {
13568	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13569	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13570	0	0	} else if (((unsigned char)*str) < 0xF8) {
13571	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13572	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13573	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
	0	0	str++; if (!--len || ((unsigned char)*str) < 0x80 || ((unsigned char)*str) >= 0xC0) return false;
13582	0	0	for (char32_t chr; (chr = decode(str)); )
13589	0	0	while (len)
13596	0	0	for (auto&& chr : str)
13624	0	0	return {3, 3, 0, ""};
	0	0	return {3, 3, 0, ""};
14154	244	31893	IF_BIT_0(prob)
	784	31353	IF_BIT_0(prob)
14159	777	7	if (checkDicSize != 0 || processedPos != 0)
14161	0	777	(dic[(dicPos == 0 ? dicBufSize : dicPos) - 1] >> (8 - lc))));
14163	440	344	if (state < kNumLitStates)
14167	385	3135	do { GET_BIT(prob + symbol, symbol) } while (symbol < 0x100);
	2054	1466	do { GET_BIT(prob + symbol, symbol) } while (symbol < 0x100);
	3080	440	do { GET_BIT(prob + symbol, symbol) } while (symbol < 0x100);
14171	0	344	unsigned matchByte = p->dic[(dicPos - rep0) + ((dicPos < rep0) ? dicBufSize : 0)];
14173	95	249	state -= (state < 10) ? 3 : 6;
14182	310	2442	GET_BIT2(probLit, symbol, offs &= ~bit, offs &= bit)
	1825	927	GET_BIT2(probLit, symbol, offs &= ~bit, offs &= bit)
14184	2408	344	while (symbol < 0x100);
14194	198	31155	IF_BIT_0(prob)
	267	31086	IF_BIT_0(prob)
14203	31086	0	if (checkDicSize == 0 && processedPos == 0)
14206	120	30966	IF_BIT_0(prob)
	30975	111	IF_BIT_0(prob)
14210	123	30852	IF_BIT_0(prob)
	54	30921	IF_BIT_0(prob)
14213	0	54	dic[dicPos] = dic[(dicPos - rep0) + ((dicPos < rep0) ? dicBufSize : 0)];
14216	3	51	state = state < kNumLitStates ? 9 : 11;
14226	22	89	IF_BIT_0(prob)
	61	50	IF_BIT_0(prob)
14235	4	46	IF_BIT_0(prob)
	31	19	IF_BIT_0(prob)
14251	30873	159	state = state < kNumLitStates ? 8 : 11;
14257	206	31093	IF_BIT_0(probLen)
	276	31023	IF_BIT_0(probLen)
14268	111	30912	IF_BIT_0(probLen)
	54	30969	IF_BIT_0(probLen)
14283	1134	247608	TREE_DECODE(probLen, limit, len);
	1269	247473	TREE_DECODE(probLen, limit, len);
	217443	31299	TREE_DECODE(probLen, limit, len);
14287	267	31032	if (state >= kNumStates)
14292	34	233	TREE_6_DECODE(prob, distance);
	194	73	TREE_6_DECODE(prob, distance);
	39	228	TREE_6_DECODE(prob, distance);
	244	23	TREE_6_DECODE(prob, distance);
	22	245	TREE_6_DECODE(prob, distance);
	169	98	TREE_6_DECODE(prob, distance);
	29	238	TREE_6_DECODE(prob, distance);
	112	155	TREE_6_DECODE(prob, distance);
	25	242	TREE_6_DECODE(prob, distance);
	154	113	TREE_6_DECODE(prob, distance);
	30	237	TREE_6_DECODE(prob, distance);
	178	89	TREE_6_DECODE(prob, distance);
14293	229	38	if (distance >= kStartPosModelIndex)
14298	119	110	if (posSlot < kEndPosModelIndex)
14305	218	119	do
14307	41	296	GET_BIT2(prob + i, i, ; , distance |= mask);
	147	190	GET_BIT2(prob + i, i, ; , distance |= mask);
14316	1025	110	do
14318	137	998	NORMALIZE
14342	14	96	GET_BIT2(prob + i, i, ; , distance |= 1);
	53	57	GET_BIT2(prob + i, i, ; , distance |= 1);
14343	20	90	GET_BIT2(prob + i, i, ; , distance |= 2);
	51	59	GET_BIT2(prob + i, i, ; , distance |= 2);
14344	9	101	GET_BIT2(prob + i, i, ; , distance |= 4);
	66	44	GET_BIT2(prob + i, i, ; , distance |= 4);
14345	18	92	GET_BIT2(prob + i, i, ; , distance |= 8);
	70	40	GET_BIT2(prob + i, i, ; , distance |= 8);
14347	0	110	if (distance == (uint32_t)0xFFFFFFFF)
14359	267	0	if (checkDicSize == 0)
14361	267	0	if (distance >= processedPos)
14364	0	0	else if (distance >= checkDicSize)
14366	130	137	state = (state < kNumStates + kNumLitStates) ? kNumLitStates : kNumLitStates + 3;
14371	31299	0	if (limit == dicPos)
14375	0	31299	unsigned curLen = ((rem < len) ? (unsigned)rem : len);
14376	0	31299	size_t pos = (dicPos - rep0) + ((dicPos < rep0) ? dicBufSize : 0);
14381	31299	0	if (pos + curLen <= dicBufSize)
14387	8397330	31299	do
14393	0	0	do
14396	0	0	if (++pos == dicBufSize)
14404	32017	120	while (dicPos < limit && buf < bufLimit);
14405	26	94	NORMALIZE;
14423	0	127	if (p->remainLen != 0 && p->remainLen < kMatchSpecLenStart)
14430	0	0	if (limit - dicPos < len)
14433	0	0	if (p->checkDicSize == 0 && p->prop.dicSize - p->processedPos <= len)
	0	0	if (p->checkDicSize == 0 && p->prop.dicSize - p->processedPos <= len)
14438	0	0	while (len-- != 0)
14440	0	0	dic[dicPos] = dic[(dicPos - rep0) + ((dicPos < rep0) ? dicBufSize : 0)];
14452	120	0	if (p->checkDicSize == 0)
14455	0	120	if (limit - p->dicPos > rem)
14458	120	0	RINOK(LzmaDec_DecodeReal(p, limit2, bufLimit));
14459	0	120	if (p->processedPos >= p->prop.dicSize)
14463	113	7	while (p->dicPos < limit && p->buf < bufLimit && p->remainLen < kMatchSpecLenStart);
	0	113	while (p->dicPos < limit && p->buf < bufLimit && p->remainLen < kMatchSpecLenStart);
	0	0	while (p->dicPos < limit && p->buf < bufLimit && p->remainLen < kMatchSpecLenStart);
14465	0	120	if (p->remainLen > kMatchSpecLenStart)
14496	0	110	IF_BIT_0_CHECK(prob)
	0	0	IF_BIT_0_CHECK(prob)
	69	41	IF_BIT_0_CHECK(prob)
14503	68	1	if (p->checkDicSize != 0 || p->processedPos != 0)
14506	0	68	(p->dic[(p->dicPos == 0 ? p->dicBufSize : p->dicPos) - 1] >> (8 - p->prop.lc))));
14508	44	25	if (state < kNumLitStates)
14511	28	324	do { GET_BIT_CHECK(prob + symbol, symbol) } while (symbol < 0x100);
	28	0	do { GET_BIT_CHECK(prob + symbol, symbol) } while (symbol < 0x100);
	218	134	do { GET_BIT_CHECK(prob + symbol, symbol) } while (symbol < 0x100);
	308	44	do { GET_BIT_CHECK(prob + symbol, symbol) } while (symbol < 0x100);
14516	0	25	((p->dicPos < p->reps[0]) ? p->dicBufSize : 0)];
14526	20	180	GET_BIT2_CHECK(probLit, symbol, offs &= ~bit, offs &= bit)
	20	0	GET_BIT2_CHECK(probLit, symbol, offs &= ~bit, offs &= bit)
	146	54	GET_BIT2_CHECK(probLit, symbol, offs &= ~bit, offs &= bit)
14528	175	25	while (symbol < 0x100);
14538	3	38	IF_BIT_0_CHECK(prob)
	3	0	IF_BIT_0_CHECK(prob)
	20	21	IF_BIT_0_CHECK(prob)
14550	1	20	IF_BIT_0_CHECK(prob)
	1	0	IF_BIT_0_CHECK(prob)
	15	6	IF_BIT_0_CHECK(prob)
14554	2	13	IF_BIT_0_CHECK(prob)
	2	0	IF_BIT_0_CHECK(prob)
	9	6	IF_BIT_0_CHECK(prob)
14557	3	6	NORMALIZE_CHECK;
	3	0	NORMALIZE_CHECK;
14569	2	4	IF_BIT_0_CHECK(prob)
	2	0	IF_BIT_0_CHECK(prob)
	4	2	IF_BIT_0_CHECK(prob)
14577	1	3	IF_BIT_0_CHECK(prob)
	1	0	IF_BIT_0_CHECK(prob)
	1	3	IF_BIT_0_CHECK(prob)
14593	2	30	IF_BIT_0_CHECK(probLen)
	2	0	IF_BIT_0_CHECK(probLen)
	23	9	IF_BIT_0_CHECK(probLen)
14604	0	9	IF_BIT_0_CHECK(probLen)
	0	0	IF_BIT_0_CHECK(probLen)
	4	5	IF_BIT_0_CHECK(probLen)
14619	15	106	TREE_DECODE_CHECK(probLen, limit, len);
	15	0	TREE_DECODE_CHECK(probLen, limit, len);
	75	46	TREE_DECODE_CHECK(probLen, limit, len);
	89	32	TREE_DECODE_CHECK(probLen, limit, len);
14623	20	12	if (state < 4)
14629	15	105	TREE_DECODE_CHECK(prob, 1 << kNumPosSlotBits, posSlot);
	15	0	TREE_DECODE_CHECK(prob, 1 << kNumPosSlotBits, posSlot);
	80	40	TREE_DECODE_CHECK(prob, 1 << kNumPosSlotBits, posSlot);
	100	20	TREE_DECODE_CHECK(prob, 1 << kNumPosSlotBits, posSlot);
14630	18	2	if (posSlot >= kStartPosModelIndex)
14636	10	8	if (posSlot < kEndPosModelIndex)
14643	47	8	do
14645	6	49	NORMALIZE_CHECK
	6	0	NORMALIZE_CHECK
14656	45	18	do
14658	8	55	GET_BIT_CHECK(prob + i, i);
	8	0	GET_BIT_CHECK(prob + i, i);
	18	45	GET_BIT_CHECK(prob + i, i);
14666	22	79	NORMALIZE_CHECK;
	22	0	NORMALIZE_CHECK;
14683	0	0	if (initDic)
14689	0	0	if (initState)
14704	55930	7	for (i = 0; i < numProbs; i++)
14720	127	0	while (p->remainLen != kMatchSpecLenStart)
14724	7	120	if (p->needFlush != 0)
14726	42	0	for (; inSize > 0 && p->tempBufSize < RC_INIT_SIZE; (*srcLen)++, inSize--)
	35	7	for (; inSize > 0 && p->tempBufSize < RC_INIT_SIZE; (*srcLen)++, inSize--)
14728	0	7	if (p->tempBufSize < RC_INIT_SIZE)
14733	7	0	if (p->tempBuf[0] != 0)
14741	7	120	if (p->dicPos >= dicLimit)
14743	7	0	if (p->remainLen == 0 && p->code == 0)
	7	0	if (p->remainLen == 0 && p->code == 0)
14748	0	0	if (finishMode == LZMA_FINISH_ANY)
14753	0	0	if (p->remainLen != 0)
14761	7	113	if (p->needInitState)
14764	0	120	if (p->tempBufSize == 0)
14768	110	10	if (inSize < LZMA_REQUIRED_INPUT_MAX || checkEndMarkNow)
14771	0	110	if (dummyRes == DUMMY_ERROR)
14779	0	110	if (checkEndMarkNow && dummyRes != DUMMY_MATCH)
14789	120	0	if (LzmaDec_DecodeReal2(p, dicLimit, bufLimit) != 0)
14799	0	0	while (rem < LZMA_REQUIRED_INPUT_MAX && lookAhead < inSize)
	0	0	while (rem < LZMA_REQUIRED_INPUT_MAX && lookAhead < inSize)
14802	0	0	if (rem < LZMA_REQUIRED_INPUT_MAX || checkEndMarkNow)
14805	0	0	if (dummyRes == DUMMY_ERROR)
14811	0	0	if (checkEndMarkNow && dummyRes != DUMMY_MATCH)
14818	0	0	if (LzmaDec_DecodeReal2(p, dicLimit, p->buf) != 0)
14827	0	0	if (p->code == 0)
14842	0	0	if (p->dicPos == p->dicBufSize)
14845	0	0	if (outSize > p->dicBufSize - dicPos)
14865	0	0	if (res != 0)
14867	0	0	if (outSizeCur == 0 || outSize == 0)
14895	7	0	if (size < LZMA_PROPS_SIZE)
14900	0	7	if (dicSize < LZMA_DIC_MIN)
14905	7	0	if (d >= (9 * 5 * 5))
14919	0	7	if (p->probs == 0 || numProbs != p->numProbs)
	0	0	if (p->probs == 0 || numProbs != p->numProbs)
14924	7	0	if (p->probs == 0)
14933	7	0	RINOK(LzmaProps_Decode(&propNew, props, propsSize));
14934	7	0	RINOK(LzmaDec_AllocateProbs2(p, &propNew, alloc));
14943	0	0	RINOK(LzmaProps_Decode(&propNew, props, propsSize));
14944	0	0	RINOK(LzmaDec_AllocateProbs2(p, &propNew, alloc));
14946	0	0	if (p->dic == 0 || dicBufSize != p->dicBufSize)
	0	0	if (p->dic == 0 || dicBufSize != p->dicBufSize)
14950	0	0	if (p->dic == 0)
14970	7	0	if (inSize < RC_INIT_SIZE)
14975	7	0	if (res != 0)
14985	7	0	if (res == SZ_OK && *status == LZMA_STATUS_NEEDS_MORE_INPUT)
	0	7	if (res == SZ_OK && *status == LZMA_STATUS_NEEDS_MORE_INPUT)
14999	7	7	static void LzmaFree(void* /*p*/, void *address) { delete[] (char*) address; }
15007	7	0	if (!is.read((char *) &uncompressed_len, sizeof(uncompressed_len))) return false;
15008	7	0	if (!is.read((char *) &compressed_len, sizeof(compressed_len))) return false;
15009	7	0	if (!is.read((char *) &poor_crc, sizeof(poor_crc))) return false;
15010	7	0	if (poor_crc != uncompressed_len * 19991 + compressed_len * 199999991 + 1234567890) return false;
15011	7	0	if (!is.read((char *) props_encoded, sizeof(props_encoded))) return false;
15014	7	0	if (!is.read((char *) compressed.data(), compressed_len)) return false;
	7	0	if (!is.read((char *) compressed.data(), compressed_len)) return false;
15018	7	0	auto res = lzma::LzmaDecode(data.fill(uncompressed_len), &uncompressed_size, compressed.data(), &compressed_size, props_encoded, LZMA_PROPS_SIZE, lzma::LZMA_FINISH_ANY, &status, &lzmaAllocator);
15019	7	0	if (res != SZ_OK || uncompressed_size != uncompressed_len || compressed_size != compressed_len) return false;
	7	0	if (res != SZ_OK || uncompressed_size != uncompressed_len || compressed_size != compressed_len) return false;
	7	0	if (res != SZ_OK || uncompressed_size != uncompressed_len || compressed_size != compressed_len) return false;
16264	14	0	if ( p == ( (str.str + str.len)) )
16272	0	38	if ( _klen > 0 ) {
16277	0	0	if ( _upper < _lower )
16281	0	0	if ( _widec < _mid[0] )
16283	0	0	else if ( _widec > _mid[1] )
16286	0	0	switch ( _url_detector_cond_spaces[_url_detector_cond_offsets[cs] + ((_mid - _keys)>>1)] ) {
16289	0	0	if (
16303	38	0	if ( _klen > 0 ) {
16308	104	38	if ( _upper < _lower )
16312	32	72	if ( _widec < *_mid )
16314	72	0	else if ( _widec > *_mid )
16326	38	0	if ( _klen > 0 ) {
16331	100	10	if ( _upper < _lower )
16335	24	76	if ( _widec < _mid[0] )
16337	48	28	else if ( _widec > _mid[1] )
16351	0	38	if ( _url_detector_trans_actions[_trans] == 0 )
16356	0	0	while ( _nacts-- > 0 )
16376	28	10	if ( cs == 0 )
16378	24	4	if ( ++p != ( (str.str + str.len)) )
16407	0	14	if (length) *length = result_length;
16408	14	0	return length || result_length == str.len ? result : NO_URL;
	0	14	return length || result_length == str.len ? result : NO_URL;
16454	0	0	return {1, 2, 0, ""};
16466	0	0	<< (nametag.prerelease.empty() ? "" : "-") << nametag.prerelease
	0	0	<< (nametag.prerelease.empty() ? "" : "-") << nametag.prerelease
16468	0	0	<< (unilib.prerelease.empty() ? "" : "-") << unilib.prerelease
	0	0	<< (unilib.prerelease.empty() ? "" : "-") << unilib.prerelease
16470	0	0	<< (morphodita.prerelease.empty() ? "" : "-") << morphodita.prerelease
	0	0	<< (morphodita.prerelease.empty() ? "" : "-") << morphodita.prerelease
16471	0	0	<< (other_libraries.empty() ? "" : "\nand ") << other_libraries << ")\n"
	0	0	<< (other_libraries.empty() ? "" : "\nand ") << other_libraries << ")\n"
16473	0	0	"Mathematics and Physics, Charles University in Prague, Czech Republic.";
16479	3	0	} // namespace ufal
	3	0	} // namespace ufal