[native] Fix varchar cast for json

presto-native-execution/presto_cpp/main/types/PrestoToVeloxExpr.cpp

@@ -181,8 +181,13 @@ std::optional<TypedExprPtr> convertCastToVarcharWithMaxLength(
   VELOX_DCHECK(end == returnType.data() + returnType.size() - 1);
 
   VELOX_DCHECK_EQ(args.size(), 1);
-  const auto arg = args[0];
 
+  auto arg = args[0];
+  // If the argument is of JSON type, convert it to VARCHAR before applying
+  // substr.
+  if (velox::isJsonType(arg->type())) {
+    arg = std::make_shared<CastTypedExpr>(velox::VARCHAR(), arg, false);
+  }
   return std::make_shared<CallTypedExpr>(
       arg->type(),
       std::vector<TypedExprPtr>{
@@ -256,8 +261,8 @@ std::optional<TypedExprPtr> tryConvertCast(
   }
 
   // When the return type is varchar with max length, truncate if only the
-  // argument type is varchar (or varchar with max length). Non-varchar argument
-  // types are not truncated.
+  // argument type is varchar, or varchar with max length or json. Non-varchar
+  // argument types are not truncated.
   if (returnType.find(kVarchar) == 0 &&
       args[0]->type()->kind() == TypeKind::VARCHAR &&
       returnType.size() > strlen(kVarchar)) {

presto-native-execution/presto_cpp/main/types/tests/RowExpressionTest.cpp

@@ -18,6 +18,7 @@
 #include "presto_cpp/presto_protocol/core/presto_protocol_core.h"
 #include "velox/core/Expressions.h"
 #include "velox/type/Type.h"
+#include "velox/functions/prestosql/types/JsonType.h"
 
 using namespace facebook::presto;
 using namespace facebook::velox;
@@ -30,6 +31,7 @@ class RowExpressionTest : public ::testing::Test {
   }
 
   void SetUp() override {
+    registerJsonType();
     pool_ = memory::MemoryManager::getInstance()->addLeafPool();
     converter_ =
         std::make_unique<VeloxExprConverter>(pool_.get(), &typeParser_);
@@ -626,6 +628,29 @@ TEST_F(RowExpressionTest, castToVarchar) {
     ASSERT_TRUE(returnExpr->nullOnFailure());
     ASSERT_EQ(returnExpr->type()->toString(), "VARCHAR");
   }
+  // CAST(json AS varchar(3))
+  {
+    std::shared_ptr<protocol::CallExpression> p =
+        json::parse(makeCastToVarchar(false, "json", "varchar(3)"));
+    auto expr = converter_->toVeloxExpr(p);
+    auto returnExpr = std::dynamic_pointer_cast<const CallTypedExpr>(expr);
+
+    ASSERT_NE(returnExpr, nullptr);
+    ASSERT_EQ(returnExpr->name(), "presto.default.substr");
+
+    auto returnArg1 = std::dynamic_pointer_cast<const CastTypedExpr>(
+        returnExpr->inputs()[0]);
+    auto returnArg2 = std::dynamic_pointer_cast<const ConstantTypedExpr>(
+        returnExpr->inputs()[1]);
+    auto returnArg3 = std::dynamic_pointer_cast<const ConstantTypedExpr>(
+        returnExpr->inputs()[2]);
+
+    ASSERT_EQ(returnArg1->type()->toString(), "VARCHAR");
+    ASSERT_EQ(returnArg2->type()->toString(), "BIGINT");
+    ASSERT_EQ(returnArg2->value().toJson(returnArg2->type()), "1");
+    ASSERT_EQ(returnArg3->type()->toString(), "BIGINT");
+    ASSERT_EQ(returnArg3->value().toJson(returnArg3->type()), "3");
+  }
 }
 
 TEST_F(RowExpressionTest, special) {

presto-native-execution/src/test/java/com/facebook/presto/nativeworker/AbstractTestNativeGeneralQueries.java

@@ -510,6 +510,7 @@ public void testCast()
         // Round-trip tests of casts for Json.
         assertQuery("SELECT cast(cast(name as JSON) as VARCHAR), cast(cast(size as JSON) as INTEGER), cast(cast(size + 0.01 as JSON) as DOUBLE), cast(cast(size > 5 as JSON) as BOOLEAN) FROM part");
         assertQuery("SELECT cast(cast(array[suppkey, nationkey] as JSON) as ARRAY(INTEGER)), cast(cast(map(array[name, address, phone], array[1.1, 2.2, 3.3]) as JSON) as MAP(VARCHAR(40), DOUBLE)), cast(cast(map(array[name], array[phone]) as JSON) as MAP(VARCHAR(25), JSON)), cast(cast(array[array[suppkey], array[nationkey]] as JSON) as ARRAY(JSON)) from supplier");
+        assertQuery("SELECT cast(json_extract(x, '$.a') AS varchar(255)) AS extracted_value FROM (VALUES ('{\"a\": \"Some long string\"}')) AS t(x)");
 
         // Cast from date to timestamp
         assertQuery("SELECT CAST(date(shipdate) AS timestamp) FROM lineitem");