Github user nongli commented on a diff in the pull request:
https://github.com/apache/spark/pull/9774#discussion_r45161743
--- Diff:
sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/UnsafeRowParquetRecordReader.java
---
@@ -0,0 +1,428 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql.execution.datasources.parquet;
+
+import static org.apache.parquet.column.ValuesType.DEFINITION_LEVEL;
+import static org.apache.parquet.column.ValuesType.REPETITION_LEVEL;
+import static org.apache.parquet.column.ValuesType.VALUES;
+
+import org.apache.hadoop.mapreduce.InputSplit;
+import org.apache.hadoop.mapreduce.TaskAttemptContext;
+import org.apache.parquet.column.ColumnDescriptor;
+import org.apache.parquet.column.Dictionary;
+import org.apache.parquet.column.page.*;
+import org.apache.parquet.column.Encoding;
+import org.apache.parquet.column.values.ValuesReader;
+import org.apache.parquet.io.api.Binary;
+import org.apache.parquet.schema.Type;
+import org.apache.spark.sql.catalyst.expressions.UnsafeRow;
+import org.apache.spark.unsafe.Platform;
+
+import java.io.IOException;
+import java.util.*;
+
+/**
+ * A specialized RecordReader that reads into UnsafeRows directly using
the Parquet column APIs.
+ *
+ * TODO: handle complex types.
+ */
+public class UnsafeRowParquetRecordReader extends
SpecificParquetRecordReaderBase<UnsafeRow> {
+ /**
+ * Batch of unsafe rows that we assemble and the current index we've
returned. Everytime this
+ * batch is used up (batchIdx == numBatched), we populated the batch.
+ */
+ private UnsafeRow[] rows = new UnsafeRow[64];
+ private int batchIdx = 0;
+ private int numBatched = 0;
+
+ /**
+ * For each request column, the reader to read this column.
+ * columnsReaders[i] populated the UnsafeRow's attribute at i.
+ */
+ private ColumnReader[] columnReaders;
+
+ /**
+ * The number of rows that have been returned.
+ */
+ private long rowsReturned;
+
+ /**
+ * The number of rows that have been reading, including the current in
flight row group.
+ */
+ private long totalCountLoadedSoFar = 0;
+
+ /**
+ * Implementation of RecordReader API.
+ */
+ @Override
+ public void initialize(InputSplit inputSplit, TaskAttemptContext
taskAttemptContext)
+ throws IOException, InterruptedException {
+ super.initialize(inputSplit, taskAttemptContext);
+
+ int rowByteSize =
UnsafeRow.calculateBitSetWidthInBytes(requestedSchema.getFieldCount());
+ rowByteSize += 8 * requestedSchema.getFieldCount();
+
+ byte[] buffer = new byte[rowByteSize * rows.length];
+ for (int i = 0; i < rows.length; ++i) {
+ rows[i] = new UnsafeRow();
+ rows[i].pointTo(buffer, Platform.BYTE_ARRAY_OFFSET + i * rowByteSize,
+ requestedSchema.getFieldCount(), rowByteSize);
+ }
+
+ /**
+ * Check that the requested schema is supported.
+ */
+ for (Type t: requestedSchema.getFields()) {
+ if (!t.isPrimitive()) throw new IOException("Unsupported schema");
+ }
+
+ // TODO: this needs to handle the schema resolution rules
+ }
+
+ @Override
+ public boolean nextKeyValue() throws IOException, InterruptedException {
+ if (batchIdx >= numBatched) {
+ if (!loadBatch()) return false;
+ }
+ ++batchIdx;
+ return true;
+ }
+
+ @Override
+ public UnsafeRow getCurrentValue() throws IOException,
InterruptedException {
+ return rows[batchIdx - 1];
+ }
+
+ @Override
+ public float getProgress() throws IOException, InterruptedException {
+ return (float) rowsReturned / totalRowCount;
+ }
+
+ /**
+ * Decodes a batch of values into `rows`. This function is the hot path.
+ */
+ private boolean loadBatch() throws IOException {
+ // no more records left
+ if (rowsReturned >= totalRowCount) { return false; }
+ checkEndOfRowGroup();
+
+ int num = (int)Math.min(rows.length, totalCountLoadedSoFar -
rowsReturned);
+ rowsReturned += num;
+
+ for (int i = 0; i < columnReaders.length; ++i) {
+ switch (columnReaders[i].descriptor.getType()) {
+ case BOOLEAN:
+ decodeBooleanBatch(i, num);
+ break;
+ case INT32:
+ decodeIntBatch(i, num);
+ break;
+ case INT64:
+ decodeLongBatch(i, num);
+ break;
+ case FLOAT:
+ decodeFloatBatch(i, num);
+ break;
+ case DOUBLE:
+ decodeDoubleBatch(i, num);
+ break;
--- End diff --
It handles NULLs. The batch decode reads the def levels.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
