[GitHub] [iceberg] SreeramGarlapati commented on a change in pull request #2272: Support structured streaming read for Iceberg

[GitBox <gi...@apache.org>]

SreeramGarlapati commented on a change in pull request #2272:
URL: https://github.com/apache/iceberg/pull/2272#discussion_r632820864



##########
File path: spark2/src/main/java/org/apache/iceberg/spark/source/StreamingReader.java
##########
@@ -0,0 +1,370 @@
+/*
+ * 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.iceberg.spark.source;
+
+import java.util.Collections;
+import java.util.List;
+import java.util.Objects;
+import java.util.Optional;
+import java.util.stream.Collectors;
+import org.apache.commons.collections.CollectionUtils;
+import org.apache.iceberg.CombinedScanTask;
+import org.apache.iceberg.DataOperations;
+import org.apache.iceberg.FileScanTask;
+import org.apache.iceberg.MicroBatches;
+import org.apache.iceberg.MicroBatches.MicroBatch;
+import org.apache.iceberg.Snapshot;
+import org.apache.iceberg.Table;
+import org.apache.iceberg.encryption.EncryptionManager;
+import org.apache.iceberg.io.CloseableIterable;
+import org.apache.iceberg.io.FileIO;
+import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting;
+import org.apache.iceberg.relocated.com.google.common.base.Preconditions;
+import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList;
+import org.apache.iceberg.relocated.com.google.common.collect.Iterables;
+import org.apache.iceberg.relocated.com.google.common.collect.Lists;
+import org.apache.iceberg.spark.SparkReadOptions;
+import org.apache.iceberg.util.Pair;
+import org.apache.iceberg.util.PropertyUtil;
+import org.apache.iceberg.util.SnapshotUtil;
+import org.apache.iceberg.util.TableScanUtil;
+import org.apache.spark.broadcast.Broadcast;
+import org.apache.spark.sql.sources.v2.DataSourceOptions;
+import org.apache.spark.sql.sources.v2.reader.streaming.MicroBatchReader;
+import org.apache.spark.sql.sources.v2.reader.streaming.Offset;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import static org.apache.iceberg.TableProperties.SPLIT_LOOKBACK;
+import static org.apache.iceberg.TableProperties.SPLIT_LOOKBACK_DEFAULT;
+import static org.apache.iceberg.TableProperties.SPLIT_OPEN_FILE_COST;
+import static org.apache.iceberg.TableProperties.SPLIT_OPEN_FILE_COST_DEFAULT;
+import static org.apache.iceberg.TableProperties.SPLIT_SIZE;
+import static org.apache.iceberg.TableProperties.SPLIT_SIZE_DEFAULT;
+
+/**
+ * A micro-batch based Spark Structured Streaming reader for Iceberg table. It will track the added
+ * files and generate tasks per batch to process newly added files. By default it will process
+ * all the newly added files to the current snapshot in each batch, user could also set this
+ * configuration "max-files-per-trigger" to control the number of files processed per batch.
+ */
+class StreamingReader extends Reader implements MicroBatchReader {
+  private static final Logger LOG = LoggerFactory.getLogger(StreamingReader.class);
+  private static final String MAX_SIZE_PER_BATCH = "max-size-per-batch";
+  private static final String START_SNAPSHOT_ID = "start-snapshot-id";
+
+  private StreamingOffset startOffset;
+  private StreamingOffset endOffset;
+
+  private final Table table;
+  private final long maxSizePerBatch;
+  private final Long startSnapshotId;
+  private final long splitSize;
+  private final int splitLookback;
+  private final long splitOpenFileCost;
+  private Boolean readUsingBatch = null;
+
+  /**
+   * Used to cache the pending batches for this streaming batch interval.
+   */
+  private Pair<StreamingOffset, List<MicroBatch>> cachedPendingBatches = null;
+
+  StreamingReader(Table table, Broadcast<FileIO> io, Broadcast<EncryptionManager> encryptionManager,
+      boolean caseSensitive, DataSourceOptions options) {
+    super(table, io, encryptionManager, caseSensitive, options);
+
+    this.table = table;
+    this.maxSizePerBatch = options.get(MAX_SIZE_PER_BATCH).map(Long::parseLong).orElse(Long.MAX_VALUE);
+    Preconditions.checkArgument(maxSizePerBatch > 0L,
+        "Option max-size-per-batch '%s' should > 0", maxSizePerBatch);
+
+    this.startSnapshotId = options.get(START_SNAPSHOT_ID).map(Long::parseLong).orElse(null);
+    if (startSnapshotId != null) {
+      if (!SnapshotUtil.ancestorOf(table, table.currentSnapshot().snapshotId(), startSnapshotId)) {
+        throw new IllegalArgumentException("The option start-snapshot-id " + startSnapshotId +
+            " is not an ancestor of the current snapshot");
+      }
+    }
+
+    long tableSplitSize = Optional.ofNullable(splitSize())
+        .orElseGet(
+            () -> PropertyUtil.propertyAsLong(table.properties(), SPLIT_SIZE, SPLIT_SIZE_DEFAULT));
+    int tableSplitLookback = Optional.ofNullable(splitLookback())
+        .orElseGet(() -> PropertyUtil
+            .propertyAsInt(table.properties(), SPLIT_LOOKBACK, SPLIT_LOOKBACK_DEFAULT));
+    long tableSplitOpenFileCost = Optional.ofNullable(splitOpenFileCost())
+        .orElseGet(() -> PropertyUtil.propertyAsLong(table.properties(), SPLIT_OPEN_FILE_COST,
+            SPLIT_OPEN_FILE_COST_DEFAULT));
+
+    this.splitSize = options.getLong(SparkReadOptions.SPLIT_SIZE, tableSplitSize);
+    this.splitLookback = options.getInt(SparkReadOptions.LOOKBACK, tableSplitLookback);
+    this.splitOpenFileCost = options
+        .getLong(SparkReadOptions.FILE_OPEN_COST, tableSplitOpenFileCost);
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public void setOffsetRange(Optional<Offset> start, Optional<Offset> end) {
+    table.refresh();
+
+    if (start.isPresent() && !StreamingOffset.START_OFFSET.equals(start.get())) {
+      this.startOffset = (StreamingOffset) start.get();
+      this.endOffset = (StreamingOffset) end.orElseGet(() -> calculateEndOffset(startOffset));
+    } else {
+      // If starting offset is "START_OFFSET" (there's no snapshot in the last batch), or starting
+      // offset is not set, then we need to calculate the starting offset again.
+      this.startOffset = calculateStartingOffset();
+      this.endOffset = calculateEndOffset(startOffset);
+    }
+  }
+
+  @Override
+  public Offset getStartOffset() {
+    if (startOffset == null) {
+      throw new IllegalStateException("Start offset is not set");
+    }
+
+    return startOffset;
+  }
+
+  @Override
+  public Offset getEndOffset() {
+    if (endOffset == null) {
+      throw new IllegalStateException("End offset is not set");
+    }
+
+    return endOffset;
+  }
+
+  @Override
+  public Offset deserializeOffset(String json) {
+    return StreamingOffset.fromJson(json);
+  }
+
+  @Override
+  public void commit(Offset end) {
+    // Since all the data and metadata of Iceberg is as it is, nothing needs to commit when
+    // offset is processed, so no need to implement this method.
+  }
+
+  @Override
+  public void stop() {
+  }
+
+  @Override
+  public boolean enableBatchRead() {
+    return readUsingBatch != null && readUsingBatch;
+  }
+
+  @Override
+  public String toString() {
+    return String.format(
+        "IcebergStreamScan(table=%s, type=%s)", table, table.schema().asStruct());
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  protected List<CombinedScanTask> tasks() {
+    if (startOffset.equals(endOffset)) {
+      LOG.info("Start offset {} equals to end offset {}, no data to process", startOffset, endOffset);
+      return Collections.emptyList();
+    }
+
+    Preconditions.checkState(cachedPendingBatches != null,
+        "pendingBatches is null, which is unexpected as it will be set when calculating end offset");
+
+    List<MicroBatch> pendingBatches = cachedPendingBatches.second();
+    if (pendingBatches.isEmpty()) {
+      LOG.info("Current start offset {} and end offset {}, there's no task to process in this batch",
+          startOffset, endOffset);
+      return Collections.emptyList();
+    }
+
+    MicroBatch lastBatch = pendingBatches.get(pendingBatches.size() - 1);
+    Preconditions.checkState(
+        lastBatch.snapshotId() == endOffset.snapshotId() && lastBatch.endFileIndex() == endOffset.position(),
+        "The cached pendingBatches doesn't match the current end offset " + endOffset);
+
+    LOG.info("Processing data from {} to {}", startOffset, endOffset);
+    List<FileScanTask> tasks = pendingBatches.stream()
+        .flatMap(batch -> batch.tasks().stream())
+        .collect(Collectors.toList());
+    CloseableIterable<FileScanTask> splitTasks = TableScanUtil.splitFiles(CloseableIterable.withNoopClose(tasks),
+        splitSize);
+    List<CombinedScanTask> combinedScanTasks = Lists.newArrayList(
+        TableScanUtil.planTasks(splitTasks, splitSize, splitLookback, splitOpenFileCost));
+
+    if (readUsingBatch == null) {
+      this.readUsingBatch = checkEnableBatchRead(combinedScanTasks);
+    }
+
+    return combinedScanTasks;
+  }
+
+  /**
+   * Used to calculate start offset. If the startSnapshotId has a value, start the construction
+   * from the specified snapshot, otherwise, start the construction from the beginning.
+   *
+   * @return The start offset to scan from.
+   */
+  private StreamingOffset calculateStartingOffset() {
+    StreamingOffset startingOffset;
+    if (startSnapshotId != null) {
+      startingOffset = new StreamingOffset(startSnapshotId, 0, false);
+    } else {
+      List<Long> snapshotIds = SnapshotUtil.currentAncestors(table);
+      if (snapshotIds.isEmpty()) {
+        // there's no snapshot currently.
+        startingOffset = StreamingOffset.START_OFFSET;
+      } else {
+        startingOffset = new StreamingOffset(Iterables.getLast(snapshotIds), 0, true);
+      }
+    }
+
+    return startingOffset;
+  }
+
+  /**
+   * Used to calculate end offset.
+   *
+   * @param start The start offset to scan from
+   * @return The end offset to scan to
+   */
+  private StreamingOffset calculateEndOffset(StreamingOffset start) {
+    if (start.equals(StreamingOffset.START_OFFSET)) {
+      return StreamingOffset.START_OFFSET;
+    }
+
+    // Spark will invoke setOffsetRange more than once. If this is already calculated, use the cached one to avoid
+    // calculating again.
+    if (cachedPendingBatches == null || CollectionUtils.isEmpty(cachedPendingBatches.second()) ||
+        !cachedPendingBatches.first().equals(start)) {
+      this.cachedPendingBatches = Pair.of(start, getChangesWithRateLimit(start, maxSizePerBatch));
+    }
+
+    List<MicroBatch> batches = cachedPendingBatches.second();
+    MicroBatch lastBatch = Iterables.getLast(batches, null);
+
+    if (lastBatch == null) {
+      return start;
+    } else {
+      boolean isStarting = lastBatch.snapshotId() == start.snapshotId() && start.shouldScanAllFiles();
+      return new StreamingOffset(lastBatch.snapshotId(), lastBatch.endFileIndex(), isStarting);
+    }
+  }
+
+  /**
+   * Streaming Read control is performed by changing the offset and maxSize.
+   *
+   * @param offset The start offset to scan from
+   * @param maxSize     The maximum size of Bytes can calculate how many batches
+   * @return MicroBatch of list
+   */
+  @VisibleForTesting
+  List<MicroBatch> getChangesWithRateLimit(StreamingOffset offset, long maxSize) {
+    List<MicroBatch> batches = Lists.newArrayList();
+    long currentLeftSize = maxSize;
+    MicroBatch lastBatch = null;
+
+    assertNoOverwrite(table.snapshot(offset.snapshotId()));
+    if (shouldGenerateFromStartOffset(offset)) {
+      MicroBatch batch = generateMicroBatch(offset.snapshotId(), offset.position(),
+          offset.shouldScanAllFiles(), currentLeftSize);
+      if (!batch.tasks().isEmpty()) {
+        batches.add(batch);
+        currentLeftSize -= batch.sizeInBytes();
+        lastBatch = Iterables.getLast(batches);
+      }
+    }
+
+    // Current snapshot can already satisfy the size needs.
+    if (currentLeftSize <= 0L || (lastBatch != null && !lastBatch.lastIndexOfSnapshot())) {
+      return batches;
+    }
+
+    long currentSnapshotId = table.currentSnapshot().snapshotId();
+    if (currentSnapshotId == offset.snapshotId()) {
+      // the snapshot of current offset is already the latest snapshot of this table.
+      return batches;
+    }
+
+    ImmutableList<Long> snapshotIds = ImmutableList.<Long>builder()
+        .addAll(SnapshotUtil.snapshotIdsBetween(table, offset.snapshotId(), currentSnapshotId))
+        .build()
+        .reverse();
+
+    for (Long id : snapshotIds) {
+      Snapshot snapshot = table.snapshot(id);
+      assertNoOverwrite(snapshot);
+      if (!isAppend(snapshot)) {
+        continue;
+      }
+
+      MicroBatch batch = generateMicroBatch(id, 0, false, currentLeftSize);
+      if (!batch.tasks().isEmpty()) {
+        batches.add(batch);
+        currentLeftSize -= batch.sizeInBytes();
+        lastBatch = Iterables.getLast(batches);
+      }
+
+      // If the current request size is already satisfied, or none of the DataFile size can satisfy the current left
+      // size, break the current loop.
+      if (currentLeftSize <= 0L || !Objects.requireNonNull(lastBatch).lastIndexOfSnapshot()) {
+        break;
+      }
+    }
+
+    return batches;
+  }
+
+  private MicroBatch generateMicroBatch(long snapshotId, long startIndex, boolean isStart, long currentLeftSize) {
+    return MicroBatches.from(table.snapshot(snapshotId), table.io())
+        .caseSensitive(caseSensitive())
+        .specsById(table.specs())
+        .generate(startIndex, currentLeftSize, isStart);
+  }
+
+  @SuppressWarnings("checkstyle:HiddenField")
+  private boolean shouldGenerateFromStartOffset(StreamingOffset startOffset) {
+    boolean isSnapshotFullyProcessed;
+    if (cachedPendingBatches != null && !cachedPendingBatches.second().isEmpty()) {
+      List<MicroBatch> batches = cachedPendingBatches.second();
+      MicroBatch lastBatch = Iterables.getLast(batches, null);
+      isSnapshotFullyProcessed = lastBatch.lastIndexOfSnapshot();
+    } else {
+      isSnapshotFullyProcessed = false;
+    }
+    return !isSnapshotFullyProcessed &&
+        (startOffset.shouldScanAllFiles() || isAppend(table.snapshot(startOffset.snapshotId())));
+  }
+
+  private static void assertNoOverwrite(Snapshot snapshot) {

Review comment:
       can we pl. take a configuration whether or not to assert. essentially - we want to factor in cases like compaction - which will perform harmless `REPLACE` operations.
   
   https://docs.microsoft.com/en-us/azure/databricks/delta/delta-streaming#ignore-updates-and-deletes




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org