Deserialize StringColumnStatisticsData, handle default value when averageLength is null

Author: avibhstarburst

trino-aws-proxy-glue/src/main/java/io/trino/aws/proxy/glue/rest/GlueDeserializer.java

@@ -21,6 +21,7 @@
 import com.fasterxml.jackson.databind.ObjectMapper;
 import software.amazon.awssdk.core.SdkBytes;
 import software.amazon.awssdk.core.SdkField;
+import software.amazon.awssdk.services.glue.model.StringColumnStatisticsData;
 
 import java.io.IOException;
 import java.lang.reflect.Field;
@@ -70,9 +71,17 @@ public T deserialize(JsonParser parser, DeserializationContext context)
                 if (type == null) {
                     return (T) context.handleUnexpectedToken(context.getContextualType(), parser);
                 }
+
                 if (type == SdkBytes.class) {
                     sdkField.set(builder, SdkBytes.fromByteArray(fieldValue.binaryValue()));
                 }
+                else if (type == StringColumnStatisticsData.class) {
+                    StringColumnStatisticsData stringColumnStatisticsData = mapper.convertValue(fieldValue, StringColumnStatisticsData.class);
+                    if (stringColumnStatisticsData.averageLength() == null) {
+                        stringColumnStatisticsData = stringColumnStatisticsData.toBuilder().averageLength(0D).build();
+                    }
+                    sdkField.set(builder, stringColumnStatisticsData);
+                }
                 else {
                     JavaType javaType = context.getTypeFactory().constructType(type);
                     sdkField.set(builder, mapper.convertValue(fieldValue, javaType));

trino-aws-proxy-glue/src/test/java/io/trino/aws/proxy/glue/TestGlueBase.java

@@ -48,13 +48,13 @@
 import software.amazon.awssdk.services.glue.model.OpenTableFormatInput;
 import software.amazon.awssdk.services.glue.model.PartitionIndex;
 import software.amazon.awssdk.services.glue.model.StorageDescriptor;
+import software.amazon.awssdk.services.glue.model.StringColumnStatisticsData;
 import software.amazon.awssdk.services.glue.model.TableIdentifier;
 import software.amazon.awssdk.services.glue.model.TableInput;
 import software.amazon.awssdk.services.glue.model.UpdateColumnStatisticsForTableRequest;
 
 import java.net.URI;
 import java.time.temporal.ChronoUnit;
-import java.util.List;
 import java.util.Map;
 
 import static io.trino.aws.proxy.glue.TestingGlueRequestHandler.DATABASE_1;
@@ -174,7 +174,7 @@ public void testComplexSerialization()
                         .databaseName(databaseName)
                         .tableName("table1")
                         .columnStatisticsList(
-                                List.of(ColumnStatistics.builder()
+                                ColumnStatistics.builder()
                                         .columnName("a")
                                         .columnType("b")
                                         .statisticsData(ColumnStatisticsData.builder()
@@ -186,7 +186,20 @@ public void testComplexSerialization()
                                                                 .build())
                                                         .build())
                                                 .build())
-                                        .build()))
+                                        .build(),
+                                ColumnStatistics.builder()
+                                        .columnName("a")
+                                        .columnType("b")
+                                        .statisticsData(ColumnStatisticsData.builder()
+                                                .type(ColumnStatisticsType.STRING)
+                                                .stringColumnStatisticsData(StringColumnStatisticsData.builder()
+                                                        .maximumLength(1L)
+//                                                        .averageLength(0D)
+                                                        .numberOfNulls(0L)
+                                                        .numberOfDistinctValues(1L)
+                                                        .build())
+                                                .build())
+                                        .build())
                         .build());
     }
 

trino-aws-proxy-glue/src/test/java/io/trino/aws/proxy/glue/TestingGlueRequestHandler.java

@@ -122,9 +122,18 @@ public GlueResponse handleRequest(ParsedGlueRequest request, SigningMetadata sig
             case UpdateColumnStatisticsForTableRequest columnStatisticsForTableRequest -> {
                 assertThat(columnStatisticsForTableRequest.catalogId()).isEqualTo("1");
                 assertThat(columnStatisticsForTableRequest.databaseName()).isEqualTo("database1");
-                assertThat(columnStatisticsForTableRequest.columnStatisticsList()).hasSize(1);
+                assertThat(columnStatisticsForTableRequest.columnStatisticsList()).hasSize(2);
                 assertThat(columnStatisticsForTableRequest.columnStatisticsList().getFirst().statisticsData()).isNotNull();
-                yield  UpdateColumnStatisticsForTableResponse.builder().build();
+                assertThat(columnStatisticsForTableRequest.columnStatisticsList()
+                        .stream()
+                        .filter(columnStatistics -> columnStatistics.statisticsData().stringColumnStatisticsData() != null)
+                        .findFirst()
+                        .orElseThrow()
+                        .statisticsData()
+                        .stringColumnStatisticsData()
+                        .averageLength())
+                        .isNotNull();
+                yield UpdateColumnStatisticsForTableResponse.builder().build();
             }
 
             default -> throw new WebApplicationException(NOT_FOUND);