Provision for different mvt_values to share a string pool

evaluator.cpp

@@ -11,7 +11,7 @@
 static std::string mvt_value_to_string(mvt_value const &one, bool &fail) {
 	switch (one.type) {
 	case mvt_string:
-		return *(one.string_value);
+		return one.get_string_value();
 	case mvt_float:
 		return milo::dtoa_milo(one.numeric_value.float_value);
 	case mvt_double:
@@ -53,7 +53,7 @@ int compare_fsl(mvt_value const &one, json_object *two, bool &fail) {
 		switch (one.type) {
 		case mvt_string: {
 			char *endptr = NULL;
-			const char *s = one.string_value->c_str();
+			const char *s = one.c_str();
 			lhs = strtod(s, &endptr);
 			if (endptr == s) {
 				fail = true;  // non-numeric-string op number => null
@@ -106,7 +106,7 @@ int compare_fsl(mvt_value const &one, json_object *two, bool &fail) {
 
 		switch (one.type) {
 		case mvt_string:
-			lhs = one.string_value->size() > 0;
+			lhs = one.get_string_view().size() > 0;
 			break;
 		case mvt_float:
 			lhs = one.numeric_value.float_value != 0;
@@ -152,7 +152,7 @@ int compare(mvt_value const &one, json_object *two, bool &fail) {
 			return false;  // string vs non-string
 		}
 
-		return strcmp(one.string_value->c_str(), two->value.string.string);
+		return strcmp(one.c_str(), two->value.string.string);
 
 	case mvt_double:
 	case mvt_float:
@@ -706,13 +706,13 @@ bool evaluate(mvt_feature const &feat, mvt_layer const &layer, json_object *filt
 
 			if (feat.type == mvt_point) {
 				const static std::string point = "Point";
-				v.string_value = point;
+				v.set_string_value(point);
 			} else if (feat.type == mvt_linestring) {
 				const static std::string linestring = "LineString";
-				v.string_value = linestring;
+				v.set_string_value(linestring);
 			} else if (feat.type == mvt_polygon) {
 				const static std::string polygon = "Polygon";
-				v.string_value = polygon;
+				v.set_string_value(polygon);
 			}
 			return v;
 		}

mvt.cpp

@@ -157,7 +157,7 @@ bool mvt_tile::decode(const std::string &message, bool &was_compressed) {
 						switch (value_reader.tag()) {
 						case 1: /* string */
 							value.type = mvt_string;
-							value.string_value = value_reader.get_string();
+							value.set_string_value(value_reader.get_string());
 							break;
 
 						case 2: /* float */
@@ -332,7 +332,7 @@ std::string mvt_tile::encode() {
 
 			switch (pbv.type) {
 			case mvt_string:
-				value_writer.add_string(1, *(pbv.string_value));
+				value_writer.add_string(1, pbv.get_string_value());
 				break;
 			case mvt_float:
 				value_writer.add_float(2, pbv.numeric_value.float_value);
@@ -461,7 +461,7 @@ bool mvt_value::operator<(const mvt_value &o) const {
 	if (type == o.type) {
 		switch (type) {
 		case mvt_string:
-			return *string_value < *(o.string_value);
+			return get_string_view() < o.get_string_view();
 
 		case mvt_float:
 			return numeric_value.float_value < o.numeric_value.float_value;
@@ -497,7 +497,7 @@ bool mvt_value::operator==(const mvt_value &o) const {
 	if (type == o.type) {
 		switch (type) {
 		case mvt_string:
-			return *string_value == *(o.string_value);
+			return get_string_view() == o.get_string_view();
 
 		case mvt_float:
 			return numeric_value.float_value == o.numeric_value.float_value;
@@ -553,7 +553,7 @@ static std::string quote(std::string const &s) {
 std::string mvt_value::toString() const {
 	switch (type) {
 	case mvt_string:
-		return quote(*string_value);
+		return quote(get_string_value());
 	case mvt_int:
 		return std::to_string(numeric_value.int_value);
 	case mvt_double: {
@@ -753,7 +753,7 @@ mvt_value stringified_to_mvt_value(int type, const char *s) {
 		break;
 	default:
 		tv.type = mvt_string;
-		tv.string_value = s;
+		tv.set_string_value(s);
 	}
 
 	return tv;
@@ -770,7 +770,7 @@ serial_val mvt_value_to_serial_val(mvt_value const &v) {
 	switch (v.type) {
 	case mvt_string:
 		sv.type = mvt_string;
-		sv.s = *(v.string_value);
+		sv.s = v.get_string_value();
 		break;
 	case mvt_float:
 		sv.type = mvt_double;
@@ -812,7 +812,7 @@ serial_val mvt_value_to_serial_val(mvt_value const &v) {
 long long mvt_value_to_long_long(mvt_value const &v) {
 	switch (v.type) {
 	case mvt_string:
-		return atoll(v.string_value->c_str());
+		return atoll(v.c_str());
 	case mvt_float:
 		return v.numeric_value.float_value;
 	case mvt_double:

mvt.hpp

@@ -3,10 +3,12 @@
 
 #include <sqlite3.h>
 #include <string>
+#include <string_view>
 #include <unordered_map>
 #include <set>
 #include <vector>
 #include <optional>
+#include <memory>
 
 #include "errors.hpp"
 
@@ -75,7 +77,8 @@ enum mvt_value_type {
 
 struct mvt_value {
 	mvt_value_type type;
-	std::optional<std::string> string_value;
+	std::shared_ptr<std::string> s;
+
 	union {
 		float float_value;
 		double double_value;
@@ -84,8 +87,36 @@ struct mvt_value {
 		long long sint_value;
 		bool bool_value;
 		int null_value;
+		struct {
+			size_t off;
+			size_t len;
+		} string_value;
 	} numeric_value;
 
+	std::string get_string_value() const {
+		return std::string(*s, numeric_value.string_value.off, numeric_value.string_value.len);
+	}
+
+	std::string_view get_string_view() const {
+		return std::string_view(s->c_str() + numeric_value.string_value.off, numeric_value.string_value.len);
+	}
+
+	const char *c_str() const {
+		return s->c_str() + numeric_value.string_value.off;
+	}
+
+	void set_string_value(const std::string &val) {
+		if (s == nullptr) {
+			s = std::make_shared<std::string>();
+		}
+
+		type = mvt_string;
+		numeric_value.string_value.off = s->size();
+		numeric_value.string_value.len = val.size();
+		s->append(val);
+		s->push_back('\0');
+	}
+
 	bool operator<(const mvt_value &o) const;
 	bool operator==(const mvt_value &o) const;
 	std::string toString() const;
@@ -99,13 +130,9 @@ struct mvt_value {
 template <>
 struct std::hash<mvt_value> {
 	std::size_t operator()(const mvt_value &k) const {
-		using std::hash;
-		using std::size_t;
-		using std::string;
-
 		switch (k.type) {
 		case mvt_string:
-			return std::hash<string>()(*(k.string_value));
+			return std::hash<std::string_view>()(k.get_string_view());
 
 		case mvt_float:
 			return std::hash<float>()(k.numeric_value.float_value);

tile-join.cpp

@@ -197,7 +197,7 @@ void append_tile(std::string message, int z, unsigned x, unsigned y, std::map<st
 								mvt_value outval;
 								if (attr_type == mvt_string) {
 									outval.type = mvt_string;
-									outval.string_value = joinval;
+									outval.set_string_value(joinval);
 								} else {
 									outval.type = mvt_double;
 									outval.numeric_value.double_value = atof(joinval.c_str());

tile.cpp

@@ -217,7 +217,7 @@ void decode_meta(std::vector<long long> const &metakeys, std::vector<long long>
 		mvt_value key = retrieve_string(metakeys[i], stringpool, NULL);
 		mvt_value value = retrieve_string(metavals[i], stringpool, &otype);
 
-		layer.tag(feature, *(key.string_value), value);
+		layer.tag(feature, key.get_string_value(), value);
 	}
 }
 
@@ -227,9 +227,9 @@ static int metacmp(const std::vector<long long> &keys1, const std::vector<long l
 		mvt_value key1 = retrieve_string(keys1[i], stringpool1, NULL);
 		mvt_value key2 = retrieve_string(keys2[i], stringpool2, NULL);
 
-		if (*(key1.string_value) < *(key2.string_value)) {
+		if (key1.get_string_view() < key2.get_string_view()) {
 			return -1;
-		} else if (*(key1.string_value) > *(key2.string_value)) {
+		} else if (key1.get_string_view() > key2.get_string_view()) {
 			return 1;
 		}
 
@@ -259,7 +259,7 @@ static int metacmp(const std::vector<long long> &keys1, const std::vector<long l
 	}
 }
 
-static mvt_value find_attribute_value(const struct coalesce *c1, std::string key) {
+static mvt_value find_attribute_value(const struct coalesce *c1, std::string const &key) {
 	if (key == ORDER_BY_SIZE) {
 		mvt_value v;
 		v.type = mvt_double;
@@ -273,7 +273,7 @@ static mvt_value find_attribute_value(const struct coalesce *c1, std::string key
 
 	for (size_t i = 0; i < keys1.size(); i++) {
 		mvt_value key1 = retrieve_string(keys1[i], stringpool1, NULL);
-		if (key == *(key1.string_value)) {
+		if (key == key1.get_string_value()) {
 			return retrieve_string(values1[i], stringpool1, NULL);
 		}
 	}
@@ -1628,11 +1628,11 @@ serial_feature next_feature(decompressor *geoms, std::atomic<long long> *geompos
 			v.type = mvt_string;
 
 			if (sf.t == mvt_point) {
-				v.string_value = "Point";
+				v.set_string_value("Point");
 			} else if (sf.t == mvt_linestring) {
-				v.string_value = "LineString";
+				v.set_string_value("LineString");
 			} else if (sf.t == mvt_polygon) {
-				v.string_value = "Polygon";
+				v.set_string_value("Polygon");
 			}
 
 			attributes.insert(std::pair<std::string, mvt_value>("$type", v));

write_json.cpp

@@ -339,7 +339,7 @@ void layer_to_geojson(mvt_layer const &layer, unsigned z, unsigned x, unsigned y
 			switch (val.type) {
 			case mvt_string:
 				state.json_write_string(key);
-				state.json_write_string(*(val.string_value));
+				state.json_write_string(val.get_string_value());
 				break;
 			case mvt_int:
 				state.json_write_string(key);