-
Notifications
You must be signed in to change notification settings - Fork 201
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Adds implementation for supporting columnar batch reads from Spark. (#…
…198) This bypasses most of the existing translation code for the following reasons: 1. I think there might be a memory leak because the existing code doesn't close the allocator. 2. This avoids continuously recopying the schema. I didn't delete the old code because it appears the BigQueryRDD still relies on it partially. I also couldn't find instructions on formatting/testing (I couldn't find explicit unit tests for existing arrow code, I'll update accordingly if pointers can be provided).
- Loading branch information
1 parent
a209789
commit 22b41d3
Showing
4 changed files
with
265 additions
and
9 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
143 changes: 143 additions & 0 deletions
143
...tor/src/main/java/com/google/cloud/spark/bigquery/v2/ArrowColumnBatchPartitionReader.java
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,143 @@ | ||
| /* | ||
| * Copyright 2018 Google Inc. All Rights Reserved. | ||
| * | ||
| * Licensed 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 com.google.cloud.spark.bigquery.v2; | ||
|
|
||
| import com.google.cloud.bigquery.connector.common.ReadRowsHelper; | ||
| import com.google.cloud.bigquery.storage.v1.ReadRowsResponse; | ||
| import com.google.cloud.spark.bigquery.ArrowSchemaConverter; | ||
| import com.google.protobuf.ByteString; | ||
| import java.io.IOException; | ||
| import java.io.InputStream; | ||
| import java.io.SequenceInputStream; | ||
| import java.util.Iterator; | ||
| import java.util.List; | ||
| import java.util.NoSuchElementException; | ||
| import org.apache.arrow.memory.BufferAllocator; | ||
| import org.apache.arrow.memory.RootAllocator; | ||
| import org.apache.arrow.util.AutoCloseables; | ||
| import org.apache.arrow.vector.VectorSchemaRoot; | ||
| import org.apache.arrow.vector.ipc.ArrowStreamReader; | ||
| import org.apache.spark.sql.vectorized.ColumnVector; | ||
| import org.apache.spark.sql.vectorized.ColumnarBatch; | ||
| import org.apache.spark.sql.sources.v2.reader.InputPartitionReader; | ||
|
|
||
| class ArrowColumnBatchPartitionColumnBatchReader implements InputPartitionReader<ColumnarBatch> { | ||
| private static final long maxAllocation = 500 * 1024 * 1024; | ||
|
|
||
| private final ReadRowsHelper readRowsHelper; | ||
| private final ArrowStreamReader reader; | ||
| private final BufferAllocator allocator; | ||
| private final List<String> namesInOrder; | ||
| private ColumnarBatch currentBatch; | ||
| private boolean closed = false; | ||
|
|
||
| static class ReadRowsResponseInputStreamEnumeration | ||
| implements java.util.Enumeration<InputStream> { | ||
| private Iterator<ReadRowsResponse> responses; | ||
| private ReadRowsResponse currentResponse; | ||
|
|
||
| ReadRowsResponseInputStreamEnumeration(Iterator<ReadRowsResponse> responses) { | ||
| this.responses = responses; | ||
| loadNextResponse(); | ||
| } | ||
|
|
||
| public boolean hasMoreElements() { | ||
| return currentResponse != null; | ||
| } | ||
|
|
||
| public InputStream nextElement() { | ||
| if (!hasMoreElements()) { | ||
| throw new NoSuchElementException("No more responses"); | ||
| } | ||
| ReadRowsResponse ret = currentResponse; | ||
| loadNextResponse(); | ||
| return ret.getArrowRecordBatch().getSerializedRecordBatch().newInput(); | ||
| } | ||
|
|
||
| void loadNextResponse() { | ||
| if (responses.hasNext()) { | ||
| currentResponse = responses.next(); | ||
| } else { | ||
| currentResponse = null; | ||
| } | ||
| } | ||
| } | ||
|
|
||
| ArrowColumnBatchPartitionColumnBatchReader( | ||
| Iterator<ReadRowsResponse> readRowsResponses, | ||
| ByteString schema, | ||
| ReadRowsHelper readRowsHelper, | ||
| List<String> namesInOrder) { | ||
| this.allocator = | ||
| (new RootAllocator(maxAllocation)) | ||
| .newChildAllocator("ArrowBinaryIterator", 0, maxAllocation); | ||
| this.readRowsHelper = readRowsHelper; | ||
| this.namesInOrder = namesInOrder; | ||
|
|
||
| InputStream batchStream = | ||
| new SequenceInputStream(new ReadRowsResponseInputStreamEnumeration(readRowsResponses)); | ||
| InputStream fullStream = new SequenceInputStream(schema.newInput(), batchStream); | ||
|
|
||
| reader = new ArrowStreamReader(fullStream, allocator); | ||
| } | ||
|
|
||
| @Override | ||
| public boolean next() throws IOException { | ||
| if (closed) { | ||
| return false; | ||
| } | ||
| closed = !reader.loadNextBatch(); | ||
| if (closed) { | ||
| return false; | ||
| } | ||
| VectorSchemaRoot root = reader.getVectorSchemaRoot(); | ||
| if (currentBatch == null) { | ||
| // trying to verify from dev@spark but this object | ||
| // should only need to get created once. The underlying | ||
| // vectors should stay the same. | ||
| ColumnVector[] columns = | ||
| namesInOrder.stream() | ||
| .map(root::getVector) | ||
| .map(ArrowSchemaConverter::new) | ||
| .toArray(ColumnVector[]::new); | ||
|
|
||
| currentBatch = new ColumnarBatch(columns); | ||
| } | ||
| currentBatch.setNumRows(root.getRowCount()); | ||
| return true; | ||
| } | ||
|
|
||
| @Override | ||
| public ColumnarBatch get() { | ||
| return currentBatch; | ||
| } | ||
|
|
||
| @Override | ||
| public void close() throws IOException { | ||
| closed = true; | ||
| try { | ||
| readRowsHelper.close(); | ||
| } catch (Exception e) { | ||
| throw new IOException("Failure closing stream: " + readRowsHelper, e); | ||
| } finally { | ||
| try { | ||
| AutoCloseables.close(reader, allocator); | ||
| } catch (Exception e) { | ||
| throw new IOException("Failure closing arrow components. stream: " + readRowsHelper, e); | ||
| } | ||
| } | ||
| } | ||
| } |
62 changes: 62 additions & 0 deletions
62
connector/src/main/java/com/google/cloud/spark/bigquery/v2/ArrowInputPartition.java
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,62 @@ | ||
| /* | ||
| * Copyright 2018 Google Inc. All Rights Reserved. | ||
| * | ||
| * Licensed 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 com.google.cloud.spark.bigquery.v2; | ||
|
|
||
| import com.google.cloud.bigquery.connector.common.BigQueryReadClientFactory; | ||
| import com.google.cloud.bigquery.connector.common.ReadRowsHelper; | ||
| import com.google.cloud.bigquery.connector.common.ReadSessionResponse; | ||
| import com.google.cloud.bigquery.storage.v1.ReadRowsRequest; | ||
| import com.google.cloud.bigquery.storage.v1.ReadRowsResponse; | ||
| import com.google.common.collect.ImmutableList; | ||
| import com.google.protobuf.ByteString; | ||
| import java.util.Iterator; | ||
| import org.apache.spark.sql.sources.v2.reader.InputPartition; | ||
| import org.apache.spark.sql.sources.v2.reader.InputPartitionReader; | ||
| import org.apache.spark.sql.vectorized.ColumnarBatch; | ||
|
|
||
| public class ArrowInputPartition implements InputPartition<ColumnarBatch> { | ||
|
|
||
| private final BigQueryReadClientFactory bigQueryReadClientFactory; | ||
| private final String streamName; | ||
| private final int maxReadRowsRetries; | ||
| private final ImmutableList<String> selectedFields; | ||
| private final ByteString serializedArrowSchema; | ||
|
|
||
| public ArrowInputPartition( | ||
| BigQueryReadClientFactory bigQueryReadClientFactory, | ||
| String name, | ||
| int maxReadRowsRetries, | ||
| ImmutableList<String> selectedFields, | ||
| ReadSessionResponse readSessionResponse) { | ||
| this.bigQueryReadClientFactory = bigQueryReadClientFactory; | ||
| this.streamName = name; | ||
| this.maxReadRowsRetries = maxReadRowsRetries; | ||
| this.selectedFields = selectedFields; | ||
| this.serializedArrowSchema = | ||
| readSessionResponse.getReadSession().getArrowSchema().getSerializedSchema(); | ||
| } | ||
|
|
||
| @Override | ||
| public InputPartitionReader<ColumnarBatch> createPartitionReader() { | ||
| ReadRowsRequest.Builder readRowsRequest = | ||
| ReadRowsRequest.newBuilder().setReadStream(streamName); | ||
| ReadRowsHelper readRowsHelper = | ||
| new ReadRowsHelper(bigQueryReadClientFactory, readRowsRequest, maxReadRowsRetries); | ||
| Iterator<ReadRowsResponse> readRowsResponses = readRowsHelper.readRows(); | ||
| return new ArrowColumnBatchPartitionColumnBatchReader( | ||
| readRowsResponses, serializedArrowSchema, readRowsHelper, selectedFields); | ||
| } | ||
| } |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters