feat: add ut for avro

Author: zjw1111

src/paimon/format/avro/CMakeLists.txt

@@ -50,6 +50,7 @@ if(PAIMON_ENABLE_AVRO)
     if(PAIMON_BUILD_TESTS)
         add_paimon_test(avro_format_test
                         SOURCES
+                        avro_direct_encoder_decoder_test.cpp
                         avro_file_batch_reader_test.cpp
                         avro_file_format_test.cpp
                         avro_format_writer_test.cpp

src/paimon/format/avro/avro_direct_decoder.cpp

@@ -418,6 +418,7 @@ Status DecodeFieldToBuilder(const ::avro::NodePtr& avro_node,
 
         const auto& branch_node = avro_node->leafAt(branch_index);
         if (branch_node->type() == ::avro::AVRO_NULL) {
+            decoder->decodeNull();
             PAIMON_RETURN_NOT_OK_FROM_ARROW(array_builder->AppendNull());
             return Status::OK();
         } else {

src/paimon/format/avro/avro_direct_encoder.cpp

@@ -57,7 +57,8 @@ Result<UnionBranches> ValidateUnion(const ::avro::NodePtr& union_node) {
         return UnionBranches{.null_index = 0, .value_index = 1, .value_node = branch_1};
     }
     if (branch_1->type() == ::avro::AVRO_NULL && branch_0->type() != ::avro::AVRO_NULL) {
-        return UnionBranches{.null_index = 1, .value_index = 0, .value_node = branch_0};
+        return Status::Invalid(
+            "Unexpected: In paimon, we expect the null branch to be the first branch in a union.");
     }
     return Status::Invalid("Union must have exactly one null branch");
 }
@@ -92,10 +93,6 @@ Status AvroDirectEncoder::EncodeArrowToAvro(const ::avro::NodePtr& avro_node,
     }
 
     switch (avro_node->type()) {
-        case ::avro::AVRO_NULL:
-            encoder->encodeNull();
-            return Status::OK();
-
         case ::avro::AVRO_BOOL: {
             const auto& bool_array =
                 arrow::internal::checked_cast<const arrow::BooleanArray&>(array);
@@ -294,7 +291,7 @@ Status AvroDirectEncoder::EncodeArrowToAvro(const ::avro::NodePtr& avro_node,
                                     element_node->leaves() != 2)) {
                     return Status::Invalid(
                         fmt::format("Expected AVRO_RECORD for map key-value pair, got {}",
-                                    ::avro::toString(element_node->type())));
+                                    AvroUtils::ToString(avro_node)));
                 }
 
                 const auto& map_array =
@@ -366,9 +363,11 @@ Status AvroDirectEncoder::EncodeArrowToAvro(const ::avro::NodePtr& avro_node,
             return Status::OK();
         }
 
+        case ::avro::AVRO_NULL:
         case ::avro::AVRO_UNION:
             // Already handled above
-            return Status::Invalid("Unexpected union handling");
+            return Status::Invalid(fmt::format("Unexpected Avro type handling: {}",
+                                               ::avro::toString(avro_node->type())));
         default:
             return Status::Invalid(
                 fmt::format("Unsupported Avro type: {}", ::avro::toString(avro_node->type())));