Add SimdJsonFacade and integrate simdjson for JSON query evaluation

Author: jkurdek

CMakeLists.txt

@@ -20,6 +20,7 @@ endif()
 
 find_package(GTest CONFIG REQUIRED)
 find_package(RapidJSON CONFIG REQUIRED)
+find_package(simdjson CONFIG REQUIRED)
 
 enable_testing()
 

include/json_facade.h

@@ -43,6 +43,11 @@ class RapidJsonFacade : public JsonFacade {
     [[nodiscard]] virtual bool EvaluateQuery(std::string_view jsonStr, const JsonQuery& query) override;
 };
 
+class SimdJsonFacade : public JsonFacade {
+   public:
+    [[nodiscard]] virtual bool EvaluateQuery(std::string_view jsonStr, const JsonQuery& query) override;
+};
+
 class JsonQueryDriver {
    public:
     explicit JsonQueryDriver(std::unique_ptr<JsonFacade>&& json_facade = {}) : json_facade_(std::move(json_facade)) {}

src/main.cpp

@@ -22,7 +22,11 @@ int main(int argc, char* argv[]) {
 
         auto facade = std::make_unique<RapidJsonFacade>();
         auto json_query_driver = std::make_unique<JsonQueryDriver>(std::move(facade));
-        auto sparser = Sparser(std::move(json_query_driver));
+
+        auto simdjsonfacade = std::make_unique<SimdJsonFacade>();
+        auto simdjson_query_driver = std::make_unique<JsonQueryDriver>(std::move(simdjsonfacade));
+
+        auto sparser = Sparser(std::move(simdjson_query_driver));
         sparser.Run(filename, JsonQuery(disj));
 
     } catch (const std::exception& e) {

src/sparser/CMakeLists.txt

@@ -1,4 +1,7 @@
 file(GLOB SRC_FILES "${CMAKE_CURRENT_SOURCE_DIR}/*.cpp")
 add_library(SparserCpp ${SRC_FILES})
 target_include_directories(SparserCpp PUBLIC ${CMAKE_SOURCE_DIR}/include)
-target_link_libraries(SparserCpp PUBLIC rapidjson)
+target_link_libraries(SparserCpp PUBLIC 
+    rapidjson
+    simdjson::simdjson
+)

src/sparser/json_facade.cpp

@@ -1,8 +1,9 @@
 #include "json_facade.h"
 
+#include <simdjson.h>
+
 #include <cstring>
 #include <sstream>
-#include <stdexcept>
 #include <string>
 #include <string_view>
 
@@ -77,6 +78,50 @@ bool RapidJsonFacade::EvaluateQuery(std::string_view jsonStr, const JsonQuery& q
     return ans;
 }
 
+bool SimdJsonFacade::EvaluateQuery(std::string_view jsonStr, const JsonQuery& query) {
+    simdjson::ondemand::parser parser;
+    auto json = simdjson::padded_string(jsonStr);
+
+    try {
+        auto doc = parser.iterate(json);
+        auto obj = doc.get_object();
+
+        bool ans = false;
+        for (const auto& conjunction : query.GetDisjunction().conjunctions) {
+            bool all_predicates_satisfied = true;
+
+            for (const auto& predicate : conjunction.predicates) {
+                simdjson::ondemand::value field;
+                auto error = obj[predicate.key].get(field);
+
+                if (error || !field.is_string()) {
+                    all_predicates_satisfied = false;
+                    break;
+                }
+
+                std::string_view field_str;
+                error = field.get_string().get(field_str);
+                if (error || field_str.find(predicate.value) == std::string_view::npos) {
+                    all_predicates_satisfied = false;
+                    break;
+                }
+            }
+
+            if (all_predicates_satisfied) {
+                ans = true;
+                break;
+            }
+        }
+        return ans;
+
+    } catch (const simdjson::simdjson_error& e) {
+#ifndef NDEBUG
+        throw std::runtime_error(e.what());
+#endif
+        return false;
+    }
+}
+
 bool JsonQueryDriver::RunQuery(std::string_view buffer, const JsonQuery& query) {
     return json_facade_->EvaluateQuery(buffer, query);
 }

vcpkg.json

@@ -1,6 +1,7 @@
 {
-  "dependencies": [
-    "gtest",
-    "rapidjson"
-  ]
+    "dependencies": [
+        "gtest",
+        "rapidjson",
+        "simdjson"
+    ]
 }