Fix issues in multi-key json case and empty value case (apache#54)

rui-mo · Mar 15, 2022 · 66c3ce4 · 66c3ce4
1 parent 7324c28
commit 66c3ce4
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 9 deletions.
diff --git a/cpp/src/gandiva/json_holder.cc b/cpp/src/gandiva/json_holder.cc
@@ -17,7 +17,6 @@
 
 #include "gandiva/json_holder.h"
 
-#include <iostream>
 #include <regex>
 
 #include "gandiva/node.h"
@@ -42,7 +41,6 @@ const uint8_t* JsonHolder::operator()(gandiva::ExecutionContext* ctx, const std:
   (parser->Finish(&parsed));
   auto struct_parsed = std::dynamic_pointer_cast<arrow::StructArray>(parsed);
   //json_path example: $.col_14, will extract col_14 here
-  // needs to gurad failure here
   if (json_path.length() < 3) {
     return nullptr;
   }
@@ -58,9 +56,14 @@ const uint8_t* JsonHolder::operator()(gandiva::ExecutionContext* ctx, const std:
     return nullptr;
   }
   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);
   auto utf8_array = std::dynamic_pointer_cast<arrow::BinaryArray>(dict_array);
-  auto res = utf8_array->GetValue(0, out_len);
-
+  auto res = utf8_array->GetValue(res_index, out_len);
+  // empty string case.
+  if (*out_len == 0) {
+    return reinterpret_cast<const uint8_t*>("");
+  }
   uint8_t* result_buffer = reinterpret_cast<uint8_t*>(ctx->arena()->Allocate(*out_len));
   memcpy(result_buffer, std::string((char*)res, *out_len).data(), *out_len);
   return result_buffer;

diff --git a/cpp/src/gandiva/json_holder_test.cc b/cpp/src/gandiva/json_holder_test.cc
@@ -41,24 +41,42 @@ TEST_F(TestJsonHolder, TestJson) {
 
   int32_t out_len;
 
-  const uint8_t* data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$.hello", &out_len);
+  const uint8_t* data = get_json_object(&execution_context_, R"({"hello": "3.5"})", "$.hello", &out_len);
+  EXPECT_EQ(std::string((char*)data, out_len), "3.5");
+
+  // test the case that value is not surrended by double quotes.
+  data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$.hello", &out_len);
+  EXPECT_EQ(out_len, 3);
   EXPECT_EQ(std::string((char*)data, out_len), "3.5");
 
   // no data contained for given field.
   data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$.hi", &out_len);
   EXPECT_EQ(data, nullptr);
 
+  // empty string.
+  data = get_json_object(&execution_context_, R"({"hello": ""})", "$.hello", &out_len);
+  EXPECT_EQ(out_len, 0);
+  EXPECT_EQ(std::string((char*)data, out_len), "");
+
   // illegal json string.
   data = get_json_object(&execution_context_, R"({"hello"-3.5})", "$.hello", &out_len);
   EXPECT_EQ(data, nullptr);
 
-  // illegal field is given.
-  data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$xx", &out_len);
+  // field name is incorrectly given.
+  data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$hello", &out_len);
   EXPECT_EQ(data, nullptr);
 
-  // illegal field is given and a short string field.
-  data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$", &out_len);
+  // field name is not given.
+  data = get_json_object(&execution_context_, R"({"hello": 3.5})", "$.", &out_len);
   EXPECT_EQ(data, nullptr);
+
+  data = get_json_object(&execution_context_, R"({"name": "fang", "age": 5, "id": "001"})", "$.age", &out_len);
+  EXPECT_EQ(out_len, 1);
+  EXPECT_EQ(std::string((char*)data, out_len), "5");
+
+  data = get_json_object(&execution_context_, R"({"name": "fang", "age": "5", "id": "001"})", "$.id", &out_len);
+  EXPECT_EQ(out_len, 3);
+  EXPECT_EQ(std::string((char*)data, out_len), "001");
 }
 
 }  // namespace gandiva