Skip to content

Commit

Permalink
Add warning about using Schema.FieldType.BYTES keys in a Map (#31600)
Browse files Browse the repository at this point in the history
  • Loading branch information
Amar3tto authored Jun 14, 2024
1 parent ffaa20c commit c02a5e6
Show file tree
Hide file tree
Showing 2 changed files with 16 additions and 8 deletions.
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,8 @@
import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.collect.Maps;
import org.checkerframework.checker.nullness.qual.NonNull;
import org.checkerframework.checker.nullness.qual.Nullable;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

/** {@link Schema} describes the fields in {@link Row}. */
@SuppressWarnings({
Expand Down Expand Up @@ -681,6 +683,9 @@ public interface LogicalType<InputT, BaseT> extends Serializable {
@AutoValue
@Immutable
public abstract static class FieldType implements Serializable {

private static final Logger LOG = LoggerFactory.getLogger(FieldType.class);

// Returns the type of this field.
public abstract TypeName getTypeName();

Expand Down Expand Up @@ -816,6 +821,13 @@ public static FieldType iterable(FieldType elementType) {

/** Create a map type for the given key and value types. */
public static FieldType map(FieldType keyType, FieldType valueType) {
if (FieldType.BYTES.equals(keyType)) {
LOG.warn(
"Using byte arrays as keys in a Map may lead to unexpected behavior and may not work as intended. "
+ "Since arrays do not override equals() or hashCode, comparisons will be done on reference equality only. "
+ "ByteBuffers, when used as keys, present similar challenges because Row stores ByteBuffer as a byte array. "
+ "Consider using a different type of key for more consistent and predictable behavior.");
}
return FieldType.forTypeName(TypeName.MAP)
.setMapKeyType(keyType)
.setMapValueType(valueType)
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,6 @@
import org.joda.time.DateTime;
import org.joda.time.DateTimeZone;
import org.junit.Assume;
import org.junit.Ignore;
import org.junit.Test;

/** Unit tests for {@link RowCoder}. */
Expand Down Expand Up @@ -285,17 +284,14 @@ public void testConsistentWithEqualsBytesField() throws Exception {
}

@Test
@Ignore
public void testConsistentWithEqualsMapWithBytesKeyField() throws Exception {
public void testEqualsMapWithBytesKeyFieldWorksOnReferenceEquality() throws Exception {
FieldType fieldType = FieldType.map(FieldType.BYTES, FieldType.INT32);
Schema schema = Schema.of(Schema.Field.of("f1", fieldType));
RowCoder coder = RowCoder.of(schema);

Map<byte[], Integer> map1 = Collections.singletonMap(new byte[] {1, 2, 3, 4}, 1);
Row row1 = Row.withSchema(schema).addValue(map1).build();

Map<byte[], Integer> map2 = Collections.singletonMap(new byte[] {1, 2, 3, 4}, 1);
Row row2 = Row.withSchema(schema).addValue(map2).build();
Map<byte[], Integer> map = Collections.singletonMap(new byte[] {1, 2, 3, 4}, 1);
Row row1 = Row.withSchema(schema).addValue(map).build();
Row row2 = Row.withSchema(schema).addValue(map).build();

Assume.assumeTrue(coder.consistentWithEquals());

Expand Down

0 comments on commit c02a5e6

Please sign in to comment.