Skip to content

Arrow: Fix vectorized read of all-null DELTA-encoded Parquet pages #17017

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
raunaqmorarka wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Arrow: Fix vectorized read of all-null DELTA-encoded Parquet pages #17017

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .../java/org/apache/iceberg/arrow/vectorized/parquet/VectorizedDeltaEncodedValuesReader.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,7 +149,7 @@ private void readValues(
int remaining = total;
int currentRowId = rowId;
// First value
if (valuesRead == 0) {
if (valuesRead == 0 && total > 0) {
outputWriter.write(vec, ((long) (currentRowId + valuesRead) * typeWidth), firstValue);
lastValueRead = firstValue;
currentRowId++;
Expand Down
43 changes: 43 additions & 0 deletions ...est/java/org/apache/iceberg/spark/data/vectorized/parquet/TestParquetVectorizedReads.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,6 +65,7 @@
import org.apache.iceberg.types.TypeUtil;
import org.apache.iceberg.types.Types;
import org.apache.parquet.column.ParquetProperties;
import org.apache.parquet.hadoop.ParquetOutputFormat;
import org.apache.parquet.schema.GroupType;
import org.apache.parquet.schema.MessageType;
import org.apache.parquet.schema.Type;
Expand Down Expand Up @@ -254,6 +255,17 @@ FileAppender<Record> getParquetV2Writer(Schema schema, File testFile) throws IOE
.build();
}

FileAppender<Record> parquetV2WriterWithoutDictionary(Schema schema, File testFile)
throws IOException {
return Parquet.write(Files.localOutput(testFile))
.schema(schema)
.createWriterFunc(GenericParquetWriter::create)
.named("test")
.writerVersion(ParquetProperties.WriterVersion.PARQUET_2_0)
.set(ParquetOutputFormat.ENABLE_DICTIONARY, "false")
.build();
}

void assertRecordsMatch(
Schema schema,
int expectedSize,
Expand Down Expand Up @@ -430,6 +442,37 @@ public void testSupportedReadsForParquetV2() throws Exception {
assertRecordsMatch(schema, 30000, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testAllNullStringAndBinaryColumnsParquetV2() throws Exception {
// V2 string/binary uses DELTA_BYTE_ARRAY; an all-null page decodes to a delta length stream
// with totalValueCount == 0.
Schema schema =
new Schema(
required(100, "id", Types.LongType.get()),
optional(101, "string_data", Types.StringType.get()),
optional(102, "binary_data", Types.BinaryType.get()));

int stringOrdinal = 1;
int binaryOrdinal = 2;
File dataFile = temp.resolve("all-null-v2.parquet").toFile();
Iterable<Record> data =
generateData(
schema,
100,
0L,
RandomData.DEFAULT_NULL_PERCENTAGE,
record -> {
record.set(stringOrdinal, null);
record.set(binaryOrdinal, null);
return record;
});
try (FileAppender<Record> writer = parquetV2WriterWithoutDictionary(schema, dataFile)) {
writer.addAll(data);
}

assertRecordsMatch(schema, 100, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testUnsupportedReadsForParquetV2() throws Exception {
// Some types use delta encoding and which are not supported for vectorized reads
Expand Down
43 changes: 43 additions & 0 deletions ...est/java/org/apache/iceberg/spark/data/vectorized/parquet/TestParquetVectorizedReads.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,6 +65,7 @@
import org.apache.iceberg.types.TypeUtil;
import org.apache.iceberg.types.Types;
import org.apache.parquet.column.ParquetProperties;
import org.apache.parquet.hadoop.ParquetOutputFormat;
import org.apache.parquet.schema.GroupType;
import org.apache.parquet.schema.MessageType;
import org.apache.parquet.schema.Type;
Expand Down Expand Up @@ -254,6 +255,17 @@ FileAppender<Record> getParquetV2Writer(Schema schema, File testFile) throws IOE
.build();
}

FileAppender<Record> parquetV2WriterWithoutDictionary(Schema schema, File testFile)
throws IOException {
return Parquet.write(Files.localOutput(testFile))
.schema(schema)
.createWriterFunc(GenericParquetWriter::create)
.named("test")
.writerVersion(ParquetProperties.WriterVersion.PARQUET_2_0)
.set(ParquetOutputFormat.ENABLE_DICTIONARY, "false")
.build();
}

void assertRecordsMatch(
Schema schema,
int expectedSize,
Expand Down Expand Up @@ -430,6 +442,37 @@ public void testSupportedReadsForParquetV2() throws Exception {
assertRecordsMatch(schema, 30000, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testAllNullStringAndBinaryColumnsParquetV2() throws Exception {
// V2 string/binary uses DELTA_BYTE_ARRAY; an all-null page decodes to a delta length stream
// with totalValueCount == 0.
Schema schema =
new Schema(
required(100, "id", Types.LongType.get()),
optional(101, "string_data", Types.StringType.get()),
optional(102, "binary_data", Types.BinaryType.get()));

int stringOrdinal = 1;
int binaryOrdinal = 2;
File dataFile = temp.resolve("all-null-v2.parquet").toFile();
Iterable<Record> data =
generateData(
schema,
100,
0L,
RandomData.DEFAULT_NULL_PERCENTAGE,
record -> {
record.set(stringOrdinal, null);
record.set(binaryOrdinal, null);
return record;
});
try (FileAppender<Record> writer = parquetV2WriterWithoutDictionary(schema, dataFile)) {
writer.addAll(data);
}

assertRecordsMatch(schema, 100, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testUnsupportedReadsForParquetV2() throws Exception {
// Some types use delta encoding and which are not supported for vectorized reads
Expand Down
42 changes: 42 additions & 0 deletions ...est/java/org/apache/iceberg/spark/data/vectorized/parquet/TestParquetVectorizedReads.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -267,6 +267,17 @@ FileAppender<Record> parquetWriterWithoutDictionary(Schema schema, File testFile
.build();
}

FileAppender<Record> parquetV2WriterWithoutDictionary(Schema schema, File testFile)
Comment thread
raunaqmorarka marked this conversation as resolved.
throws IOException {
return Parquet.write(Files.localOutput(testFile))
.schema(schema)
.createWriterFunc(GenericParquetWriter::create)
.named("test")
.writerVersion(ParquetProperties.WriterVersion.PARQUET_2_0)
.set(ParquetOutputFormat.ENABLE_DICTIONARY, "false")
.build();
}

void assertRecordsMatch(
Schema schema,
int expectedSize,
Expand Down Expand Up @@ -443,6 +454,37 @@ public void testSupportedReadsForParquetV2() throws Exception {
assertRecordsMatch(schema, 30000, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testAllNullStringAndBinaryColumnsParquetV2() throws Exception {
// V2 string/binary uses DELTA_BYTE_ARRAY; an all-null page decodes to a delta length stream
// with totalValueCount == 0.
Schema schema =
new Schema(
required(100, "id", Types.LongType.get()),
optional(101, "string_data", Types.StringType.get()),
optional(102, "binary_data", Types.BinaryType.get()));

int stringOrdinal = 1;
int binaryOrdinal = 2;
File dataFile = temp.resolve("all-null-v2.parquet").toFile();
Iterable<Record> data =
generateData(
schema,
100,
0L,
RandomData.DEFAULT_NULL_PERCENTAGE,
record -> {
record.set(stringOrdinal, null);
record.set(binaryOrdinal, null);
return record;
});
try (FileAppender<Record> writer = parquetV2WriterWithoutDictionary(schema, dataFile)) {
writer.addAll(data);
}

assertRecordsMatch(schema, 100, data, dataFile, false, BATCH_SIZE);
}

@Test
public void testUnsupportedReadsForParquetV2() throws Exception {
// Some types use delta encoding and which are not supported for vectorized reads
Expand Down
Loading