[NSE-653] Add validity checking for get_json_object in WSCG

.github/workflows/unittests.yml

@@ -159,7 +159,7 @@ jobs:
     steps:
     - uses: actions/checkout@v2
     - name: Run clang-format style check for C/C++ programs.
-      uses: jidicula/clang-format-action@v3.2.0
+      uses: jidicula/clang-format-action@v3.5.1
       with:
         clang-format-version: '10'
         check-path: 'native-sql-engine/cpp/src'

native-sql-engine/cpp/src/codegen/arrow_compute/ext/expression_codegen_visitor.cc

@@ -255,11 +255,18 @@ arrow::Status ExpressionCodegenVisitor::Visit(const gandiva::FunctionNode& node)
     real_codes_str_ = codes_str_;
     real_validity_str_ = check_str_;
     std::stringstream prepare_ss;
-    prepare_ss << "bool " << check_str_ << " = true;" << std::endl;
-    prepare_ss << "std::string " << codes_str_ << " = get_json_object("
+    auto validity = codes_str_ + "_validity";
+    prepare_ss << "std::string " << codes_str_ << ";" << std::endl;
+    prepare_ss << "bool " << validity << " = " << child_visitor_list[0]->GetPreCheck()
+               << ";" << std::endl;
+    prepare_ss << "if (" << validity << ") {" << std::endl;
+    prepare_ss << codes_str_ << " = get_json_object("
                << child_visitor_list[0]->GetResult() << ", "
-               << child_visitor_list[1]->GetResult() << ");\n";
+               << child_visitor_list[1]->GetResult() << ", "
+               << "&" << validity << ");\n";
+    prepare_ss << "}" << std::endl;
     prepare_str_ += prepare_ss.str();
+    check_str_ = validity;
     header_list_.push_back(R"(#include "precompile/gandiva.h")");
   } else if (func_name.compare("substr") == 0) {
     ss << child_visitor_list[0]->GetResult() << ".substr("

native-sql-engine/cpp/src/precompile/gandiva.h

@@ -230,7 +230,8 @@ arrow::Decimal128 round(arrow::Decimal128 in, int32_t original_precision,
   return arrow::Decimal128(out);
 }
 
-std::string get_json_object(const std::string& json_str, const std::string& json_path) {
+std::string get_json_object(const std::string& json_str, const std::string& json_path,
+                            bool* validity) {
   std::unique_ptr<arrow::json::BlockParser> parser;
   (arrow::json::BlockParser::Make(arrow::json::ParseOptions::Defaults(), &parser));
   (parser->Parse(std::make_shared<arrow::Buffer>(json_str)));
@@ -239,27 +240,29 @@ std::string get_json_object(const std::string& json_str, const std::string& json
   auto struct_parsed = std::dynamic_pointer_cast<arrow::StructArray>(parsed);
   // json_path example: $.col_14, will extract col_14 here
   if (json_path.length() < 3) {
-    return nullptr;
+    *validity = false;
+    return "";
   }
   auto col_name = json_path.substr(2);
   // illegal json string.
   if (struct_parsed == nullptr) {
-    return nullptr;
+    *validity = false;
+    return "";
   }
   auto dict_parsed = std::dynamic_pointer_cast<arrow::DictionaryArray>(
       struct_parsed->GetFieldByName(col_name));
   // no data contained for given field.
   if (dict_parsed == nullptr) {
-    return nullptr;
+    *validity = false;
+    return "";
   }
 
   auto dict_array = dict_parsed->dictionary();
   // needs to see whether there is a case that has more than one indices.
   auto res_index = dict_parsed->GetValueIndex(0);
   // TODO(): check null results
   auto utf8_array = std::dynamic_pointer_cast<arrow::BinaryArray>(dict_array);
-
   auto res = utf8_array->GetString(res_index);
-
+  *validity = true;
   return res;
 }

native-sql-engine/cpp/src/tests/arrow_compute_test_precompile.cc

@@ -112,8 +112,17 @@ TEST(TestArrowCompute, ArithmeticComparisonTest) {
 }
 
 TEST(TestArrowCompute, JsonTest) {
-  std::string data = get_json_object(R"({"hello": "3.5"})", "$.hello");
+  bool validity;
+  std::string data = get_json_object(R"({"hello": "3.5"})", "$.hello", &validity);
   EXPECT_EQ(data, "3.5");
+  EXPECT_EQ(validity, true);
+
+  data = get_json_object(R"({"hello": ""})", "$.hello", &validity);
+  EXPECT_EQ(data, "");
+  EXPECT_EQ(validity, true);
+
+  data = get_json_object(R"({"hello": "3.5"})", "$.hi", &validity);
+  EXPECT_EQ(validity, false);
 }
 
 }  // namespace codegen