Use JSON visitor for serialization

power_grid_model_c/power_grid_model/include/power_grid_model/auxiliary/serialization/deserializer.hpp

@@ -19,6 +19,7 @@
 #include <set>
 #include <span>
 #include <sstream>
+#include <stack>
 #include <string_view>
 #include <utility>
 
@@ -38,6 +39,104 @@ constexpr from_json_t from_json;
 
 namespace detail {
 
+using nlohmann::json;
+
+// visitor for json conversion
+struct JsonMapArrayData {
+    size_t size{};
+    std::shared_ptr<msgpack::sbuffer> buffer{};
+
+    JsonMapArrayData() : size{}, buffer{std::make_shared<msgpack::sbuffer>()} {}
+    JsonMapArrayData(size_t size_, std::shared_ptr<msgpack::sbuffer> buffer_) : size{size_}, buffer{buffer_} {}
+};
+
+struct JsonSAXVisitor {
+    msgpack::packer<msgpack::sbuffer> top_packer() {
+        if (data_buffers.empty()) {
+            throw SerializationError{"Json root should be a map!\n"};
+        }
+        return {*data_buffers.top().buffer};
+    }
+
+    template <class T> bool pack_data(T const& val) {
+        top_packer().pack(val);
+        ++data_buffers.top().size;
+        return true;
+    }
+
+    bool null() {
+        top_packer().pack_nil();
+        ++data_buffers.top().size;
+        return true;
+    }
+    bool boolean(bool val) { return pack_data(val); }
+    bool number_integer(json::number_integer_t val) { return pack_data(val); }
+    bool number_unsigned(json::number_unsigned_t val) { return pack_data(val); }
+    bool number_float(json::number_float_t val, json::string_t const& /* s */) { return pack_data(val); }
+    bool string(json::string_t const& val) {
+        if (val == "inf" || val == "+inf") {
+            return pack_data(std::numeric_limits<double>::infinity());
+        }
+        if (val == "-inf") {
+            return pack_data(-std::numeric_limits<double>::infinity());
+        }
+        return pack_data(val);
+    }
+    bool key(json::string_t const& val) {
+        top_packer().pack(val);
+        return true;
+    }
+    static bool binary(json::binary_t const& /* val */) { return true; }
+
+    bool start_object(size_t /* elements */) {
+        data_buffers.emplace();
+        return true;
+    }
+    bool end_object() {
+        JsonMapArrayData object_data{std::move(data_buffers.top())};
+        data_buffers.pop();
+        if (!std::in_range<uint32_t>(object_data.size)) {
+            throw SerializationError{"Json map/array size exceeds the msgpack limit (2^32)!\n"};
+        }
+        if (data_buffers.empty()) {
+            msgpack::packer<msgpack::sbuffer> root_packer{root_buffer};
+            root_packer.pack_map(static_cast<uint32_t>(object_data.size));
+            root_buffer.write(object_data.buffer->data(), object_data.buffer->size());
+        } else {
+            top_packer().pack_map(static_cast<uint32_t>(object_data.size));
+            data_buffers.top().buffer->write(object_data.buffer->data(), object_data.buffer->size());
+            ++data_buffers.top().size;
+        }
+        return true;
+    }
+    bool start_array(size_t /* elements */) {
+        data_buffers.emplace();
+        return true;
+    }
+    bool end_array() {
+        JsonMapArrayData array_data{std::move(data_buffers.top())};
+        data_buffers.pop();
+        if (!std::in_range<uint32_t>(array_data.size)) {
+            throw SerializationError{"Json map/array size exceeds the msgpack limit (2^32)!\n"};
+        }
+        top_packer().pack_array(static_cast<uint32_t>(array_data.size));
+        data_buffers.top().buffer->write(array_data.buffer->data(), array_data.buffer->size());
+        ++data_buffers.top().size;
+        return true;
+    }
+
+    [[noreturn]] static bool parse_error(std::size_t position, std::string const& last_token,
+                                         json::exception const& ex) {
+        std::stringstream ss;
+        ss << "Parse error in JSON. Position: " << position << ", last token: " << last_token
+           << ". Exception message: " << ex.what() << '\n';
+        throw SerializationError{ss.str()};
+    }
+
+    std::stack<JsonMapArrayData> data_buffers{};
+    msgpack::sbuffer root_buffer{};
+};
+
 // visitors for parsing
 struct DefaultNullVisitor : msgpack::null_visitor {
     static std::string msg_for_parse_error(size_t parsed_offset, size_t error_offset, std::string_view msg) {
@@ -255,6 +354,7 @@ class Deserializer {
     using visit_map_t = detail::visit_map_t;
     using visit_array_t = detail::visit_array_t;
     using visit_map_array_t = detail::visit_map_array_t;
+    using JsonSAXVisitor = detail::JsonSAXVisitor;
 
     struct ComponentByteMeta {
         std::string_view component;
@@ -308,7 +408,7 @@ class Deserializer {
     // data members are order dependent
     // DO NOT modify the order!
     // own buffer if from json
-    std::vector<char> buffer_from_json_;
+    msgpack::sbuffer buffer_from_json_;
     // pointer to buffers
     char const* data_;
     size_t size_;
@@ -332,48 +432,13 @@ class Deserializer {
     std::vector<std::vector<ComponentByteMeta>> msg_data_offsets_;
     WritableDatasetHandler dataset_handler_;
 
-    static std::vector<char> json_to_msgpack(std::string_view json_string) {
-        nlohmann::json json_document = nlohmann::json::parse(json_string);
-        json_convert_inf(json_document);
-        std::vector<char> msgpack_data;
-        nlohmann::json::to_msgpack(json_document, msgpack_data);
+    static msgpack::sbuffer json_to_msgpack(std::string_view json_string) {
+        JsonSAXVisitor visitor{};
+        nlohmann::json::sax_parse(json_string, &visitor);
+        msgpack::sbuffer msgpack_data{std::move(visitor.root_buffer)};
         return msgpack_data;
     }
 
-    static void json_convert_inf(nlohmann::json& json_document) {
-        switch (json_document.type()) {
-        case nlohmann::json::value_t::object:
-        case nlohmann::json::value_t::array:
-            for (auto& value : json_document) {
-                json_convert_inf(value);
-            }
-            break;
-        case nlohmann::json::value_t::string:
-            json_string_to_inf(json_document);
-            break;
-        default:
-            break;
-        }
-    }
-
-    static void json_string_to_inf(nlohmann::json& value) {
-        std::string const str = value.get<std::string>();
-        if (str == "inf" || str == "+inf") {
-            value = std::numeric_limits<double>::infinity();
-        }
-        if (str == "-inf") {
-            value = -std::numeric_limits<double>::infinity();
-        }
-    }
-
-    static std::string_view key_to_string(msgpack::object_kv const& kv) {
-        try {
-            return kv.key.as<std::string_view>();
-        } catch (msgpack::type_error&) {
-            throw SerializationError{"Keys in the dictionary should always be a string!\n"};
-        }
-    }
-
     template <class map_array, bool move_forward> MapArrayVisitor<map_array> parse_map_array() {
         MapArrayVisitor<map_array> visitor{};
         if constexpr (move_forward) {
@@ -747,7 +812,7 @@ class Deserializer {
         std::stringstream ss;
         ss << e.what();
         if (!root_key_.empty()) {
-            ss << "Position of error: " << root_key_;
+            ss << " Position of error: " << root_key_;
             root_key_ = "";
         }
         if (is_batch_ && scenario_number_ >= 0) {

power_grid_model_c/power_grid_model/include/power_grid_model/auxiliary/serialization/serializer.hpp

@@ -16,7 +16,11 @@
 
 #include <msgpack.hpp>
 
+#include <iomanip>
+#include <limits>
 #include <span>
+#include <sstream>
+#include <stack>
 #include <string_view>
 
 // custom packers
@@ -59,6 +63,141 @@ MSGPACK_API_VERSION_NAMESPACE(MSGPACK_DEFAULT_API_NS) {
 
 namespace power_grid_model::meta_data {
 
+namespace json_converter {
+
+struct MapArray {
+    MapArray(uint32_t size_input) : size{size_input}, empty{size_input == 0} {}
+
+    uint32_t size;
+    bool empty;
+    bool begin{true};
+};
+
+struct JsonConverter : msgpack::null_visitor {
+    static constexpr char sep_char = ' ';
+
+    Idx indent;
+    Idx max_indent_level;
+    std::stringstream ss{};
+    std::stack<MapArray> map_array{};
+
+    void print_indent() {
+        if (indent < 0) {
+            return;
+        }
+        Idx const indent_level = static_cast<Idx>(map_array.size());
+        if (indent_level > max_indent_level) {
+            if (map_array.top().begin) {
+                map_array.top().begin = false;
+                return;
+            }
+            ss << sep_char;
+            return;
+        }
+        ss << '\n';
+        ss << std::string(indent_level * indent, sep_char);
+    }
+
+    void print_key_val_sep() {
+        if (indent < 0) {
+            return;
+        }
+        ss << sep_char;
+    }
+
+    bool visit_nil() {
+        ss << "null";
+        return true;
+    }
+    bool visit_boolean(bool v) {
+        if (v) {
+            ss << "true";
+        } else {
+            ss << "false";
+        }
+        return true;
+    }
+    bool visit_positive_integer(uint64_t v) {
+        ss << v;
+        return true;
+    }
+    bool visit_negative_integer(int64_t v) {
+        ss << v;
+        return true;
+    }
+    bool visit_float32(float v) { return visit_float64(v); }
+    bool visit_float64(double v) {
+        if (std::isinf(v)) {
+            using namespace std::string_view_literals;
+            ss << '"' << (v > 0.0 ? "inf"sv : "-inf"sv) << '"';
+        } else {
+            ss << std::setprecision(std::numeric_limits<double>::digits10 + 1) << v;
+        }
+        return true;
+    }
+    bool visit_str(const char* v, uint32_t size) {
+        ss << '"' << std::string_view{v, size} << '"';
+        return true;
+    }
+    bool start_array(uint32_t num_elements) {
+        map_array.emplace(num_elements);
+        ss << '[';
+        return true;
+    }
+    bool start_array_item() {
+        print_indent();
+        return true;
+    }
+    bool end_array_item() {
+        --map_array.top().size;
+        if (map_array.top().size > 0) {
+            ss << ',';
+        }
+        return true;
+    }
+    bool end_array() {
+        bool const empty = map_array.top().empty;
+        map_array.pop();
+        if (static_cast<Idx>(map_array.size()) < max_indent_level && !empty) {
+            print_indent();
+        }
+        ss << ']';
+        return true;
+    }
+    bool start_map(uint32_t num_kv_pairs) {
+        map_array.emplace(num_kv_pairs);
+        ss << '{';
+        return true;
+    }
+    bool start_map_key() {
+        print_indent();
+        return true;
+    }
+    bool end_map_key() {
+        ss << ':';
+        print_key_val_sep();
+        return true;
+    }
+    bool end_map_value() {
+        --map_array.top().size;
+        if (map_array.top().size > 0) {
+            ss << ',';
+        }
+        return true;
+    }
+    bool end_map() {
+        bool const empty = map_array.top().empty;
+        map_array.pop();
+        if (static_cast<Idx>(map_array.size()) < max_indent_level && !empty) {
+            print_indent();
+        }
+        ss << '}';
+        return true;
+    }
+};
+
+} // namespace json_converter
+
 class Serializer {
     using RawElementPtr = void const*;
 
@@ -210,10 +349,11 @@ class Serializer {
 
     std::string const& get_json(bool use_compact_list, Idx indent) {
         if (json_buffer_.empty() || (use_compact_list_ != use_compact_list) || (json_indent_ != indent)) {
-            auto json_document = nlohmann::json::from_msgpack(get_msgpack(use_compact_list));
-            json_convert_inf(json_document);
-            json_indent_ = indent;
-            json_buffer_ = json_document.dump(static_cast<int>(indent));
+            Idx const max_indent_level = dataset_handler_.is_batch() ? 4 : 3;
+            json_converter::JsonConverter visitor{{}, indent, max_indent_level};
+            auto const msgpack_data = get_msgpack(use_compact_list);
+            msgpack::parse(msgpack_data.data(), msgpack_data.size(), visitor);
+            json_buffer_ = visitor.ss.str();
         }
         return json_buffer_;
     }

tests/c_api_tests/test_serialization.cpp

@@ -19,7 +19,7 @@ namespace {
 using namespace std::string_literals;
 
 constexpr char const* json_data =
-    R"({"attributes":{},"data":{"node":[{"id":5}],"source":[{"id":6},{"id":7}]},"is_batch":false,"type":"input","version":"1.0"})";
+    R"({"version":"1.0","type":"input","is_batch":false,"attributes":{},"data":{"node":[{"id":5}],"source":[{"id":6},{"id":7}]}})";
 } // namespace
 
 TEST_CASE("Serialization") {
@@ -75,7 +75,7 @@ TEST_CASE("Serialization") {
             Idx msgpack_size{};
             PGM_serializer_get_to_binary_buffer(hl, serializer, 0, &msgpack_data, &msgpack_size);
             CHECK(PGM_error_code(hl) == PGM_no_error);
-            auto const json_document = nlohmann::json::from_msgpack(msgpack_data, msgpack_data + msgpack_size);
+            auto const json_document = nlohmann::ordered_json::from_msgpack(msgpack_data, msgpack_data + msgpack_size);
             auto const json_result = json_document.dump(-1);
             CHECK(json_result == json_data);
         }

tests/cpp_unit_tests/test_deserializer.cpp

@@ -244,10 +244,6 @@ constexpr std::string_view json_batch = R"(
 }
 )";
 
-} // namespace
-
-namespace {
-
 void check_error(std::string_view json, char const* err_msg) {
     std::vector<NodeInput> node(1);