yihua commented on code in PR #12866: URL: https://github.com/apache/hudi/pull/12866#discussion_r2105661387
########## hudi-io/src/main/java/org/apache/hudi/io/hfile/HFileWriterImpl.java: ########## @@ -0,0 +1,264 @@ +/* + * 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.hudi.io.hfile; + +import org.apache.hudi.common.util.StringUtils; +import org.apache.hudi.common.util.io.ByteBufferBackedInputStream; +import org.apache.hudi.io.ByteArraySeekableDataInputStream; +import org.apache.hudi.io.compress.CompressionCodec; +import org.apache.hudi.io.hfile.protobuf.generated.HFileProtos; + +import com.google.protobuf.ByteString; +import com.google.protobuf.CodedOutputStream; + +import java.io.ByteArrayOutputStream; +import java.io.DataOutputStream; +import java.io.IOException; +import java.io.OutputStream; +import java.nio.ByteBuffer; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import java.util.Arrays; +import java.util.HashMap; +import java.util.Map; + +import static org.apache.hudi.io.hfile.DataSize.SIZEOF_INT16; + +/** + * Pure Java implementation of HFile writer (HFile v3 format) for Hudi. + */ +public class HFileWriterImpl implements HFileWriter { + private final OutputStream outputStream; + private final HFileContext context; + // Meta Info map. + private final Map<String, byte[]> metaInfo = new HashMap<>(); + // Data block under construction. + private HFileDataBlock currentDataBlock; + // Meta block under construction. + private final HFileRootIndexBlock rootIndexBlock; + private final HFileMetaIndexBlock metaIndexBlock; + private final HFileFileInfoBlock fileInfoBlock; + private long uncompressedBytes; + private long totalUncompressedBytes; + private long currentOffset; + private long loadOnOpenSectionOffset; + private final int blockSize; + + // Variables used to record necessary information to reduce + // the memory usage. + private byte[] lastKey = new byte[0]; + private long firstDataBlockOffset = -1; + private long lastDataBlockOffset; + private long totalNumberOfRecords = 0; + + public HFileWriterImpl(HFileContext context, OutputStream outputStream) { + this.outputStream = outputStream; + this.context = context; + this.blockSize = this.context.getBlockSize(); + this.uncompressedBytes = 0L; + this.totalUncompressedBytes = 0L; + this.currentOffset = 0L; + this.currentDataBlock = new HFileDataBlock(context); + this.rootIndexBlock = new HFileRootIndexBlock(context); + this.metaIndexBlock = new HFileMetaIndexBlock(context); + this.fileInfoBlock = new HFileFileInfoBlock(context); + initFileInfo(); + } + + // Append a data kv pair. + public void append(String key, byte[] value) throws IOException { + byte[] keyBytes = StringUtils.getUTF8Bytes(key); + lastKey = keyBytes; + // Records with the same key must be put into the same block. + if (!Arrays.equals(currentDataBlock.getLastKeyContent(), keyBytes) + && uncompressedBytes + keyBytes.length + value.length + 9 > blockSize) { + flushCurrentDataBlock(); + uncompressedBytes = 0; + } + currentDataBlock.add(keyBytes, value); + int uncompressedKeyValueSize = keyBytes.length + value.length; + uncompressedBytes += uncompressedKeyValueSize + 9; + totalUncompressedBytes += uncompressedKeyValueSize + 9; + } + + // Append a metadata kv pair. + public void appendMetaInfo(String name, byte[] value) { + metaInfo.put(name, value); + } + + // Append a file info kv pair. + public void appendFileInfo(String name, byte[] value) { + fileInfoBlock.add(name, value); + } + + @Override + public void close() throws IOException { + flushCurrentDataBlock(); + flushMetaBlocks(); + writeLoadOnOpenSection(); + writeTrailer(); + outputStream.flush(); + outputStream.close(); + } + + private void flushCurrentDataBlock() throws IOException { + // 0. Skip flush if no data. + if (currentDataBlock.isEmpty()) { + return; + } + // 1. Update metrics. + if (firstDataBlockOffset < 0) { + firstDataBlockOffset = currentOffset; + } + lastDataBlockOffset = currentOffset; + totalNumberOfRecords += currentDataBlock.getNumOfEntries(); + // 2. Flush data block. + ByteBuffer blockBuffer = currentDataBlock.serialize(); + writeBuffer(blockBuffer); + // 3. Create an index entry. + rootIndexBlock.add( + currentDataBlock.getFirstKey(), lastDataBlockOffset, blockBuffer.limit()); Review Comment: HBase-based HFile writer supports fake first key in the block index (see #13179) to reduce the block index size. We can consider this as a follow-up optimization (make sure to file a JIRA ticket to track that). -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
