import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Method returns the reader given the specified arguments.
 * TODO This is a bad abstraction.  See HBASE-6635.
 *
 * @param path hfile's path
 * @param fsdis stream of path's file
 * @param size max size of the trailer.
 * @param cacheConf Cache configuation values, cannot be null.
 * @param hfs
 * @return an appropriate instance of HFileReader
 * @throws IOException If file is invalid, will throw CorruptHFileException flavored IOException
 */
@edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "SF_SWITCH_FALLTHROUGH", justification = "Intentional")
private static Reader pickReaderVersion(Path path, FSDataInputStreamWrapper fsdis, long size,
    CacheConfig cacheConf, HFileSystem hfs, Configuration conf) throws IOException {
  FixedFileTrailer trailer = null;
  try {
    boolean isHBaseChecksum = fsdis.shouldUseHBaseChecksum();
    assert !isHBaseChecksum; // Initially we must read with FS checksum.
    trailer = FixedFileTrailer.readFromStream(fsdis.getStream(isHBaseChecksum), size);
    switch (trailer.getMajorVersion()) {
    case 2:
      return new HFileReaderV2(path, trailer, fsdis, size, cacheConf, hfs, conf);
    case 3:
      return new HFileReaderV3(path, trailer, fsdis, size, cacheConf, hfs, conf);
    default:
      throw new IllegalArgumentException("Invalid HFile version " + trailer.getMajorVersion());
    }
  } catch (Throwable t) {
    try {
      fsdis.close();
    } catch (Throwable t2) {
      LOG.warn("Error closing fsdis FSDataInputStreamWrapper", t2);
    }
    throw new CorruptHFileException("Problem reading HFile Trailer from file " + path, t);
  }
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
@Test
public void testNewBlocksHaveDefaultChecksum() throws IOException {
  Path path = new Path(TEST_UTIL.getDataTestDir(), "default_checksum");
  FSDataOutputStream os = fs.create(path);
  HFileContext meta = new HFileContextBuilder().build();
  HFileBlock.Writer hbw = new HFileBlock.Writer(null, meta);
  DataOutputStream dos = hbw.startWriting(BlockType.DATA);
  for (int i = 0; i < 1000; ++i)
    dos.writeInt(i);
  hbw.writeHeaderAndData(os);
  int totalSize = hbw.getOnDiskSizeWithHeader();
  os.close();

  // Use hbase checksums.
  assertEquals(true, hfs.useHBaseChecksum());

  FSDataInputStreamWrapper is = new FSDataInputStreamWrapper(fs, path);
  meta = new HFileContextBuilder().withHBaseCheckSum(true).build();
  HFileBlock.FSReader hbr = new HFileBlock.FSReaderImpl(
      is, totalSize, (HFileSystem) fs, path, meta);
  HFileBlock b = hbr.readBlockData(0, -1, -1, false);
  assertEquals(b.getChecksumType(), ChecksumType.getDefaultChecksumType().getCode());
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Read all blocks from {@code path} to populate {@code blockCache}.
 */
private static void cacheBlocks(Configuration conf, CacheConfig cacheConfig, FileSystem fs,
    Path path, HFileContext cxt) throws IOException {
  FSDataInputStreamWrapper fsdis = new FSDataInputStreamWrapper(fs, path);
  long fileSize = fs.getFileStatus(path).getLen();
  FixedFileTrailer trailer =
    FixedFileTrailer.readFromStream(fsdis.getStream(false), fileSize);
  HFileReaderV2 reader = new HFileReaderV2(path, trailer, fsdis, fileSize, cacheConfig,
    fsdis.getHfs(), conf);
  reader.loadFileInfo();
  long offset = trailer.getFirstDataBlockOffset(),
    max = trailer.getLastDataBlockOffset();
  List<HFileBlock> blocks = new ArrayList<HFileBlock>(4);
  HFileBlock block;
  while (offset <= max) {
    block = reader.readBlock(offset, -1, /* cacheBlock */ true, /* pread */ false,
    /* isCompaction */ false, /* updateCacheMetrics */ true, null, null);
    offset += block.getOnDiskSizeWithHeader();
    blocks.add(block);
  }
  LOG.info("read " + Iterables.toString(blocks));
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Method returns the reader given the specified arguments.
 * TODO This is a bad abstraction.  See HBASE-6635.
 *
 * @param path hfile's path
 * @param fsdis stream of path's file
 * @param size max size of the trailer.
 * @param cacheConf Cache configuation values, cannot be null.
 * @param hfs
 * @return an appropriate instance of HFileReader
 * @throws IOException If file is invalid, will throw CorruptHFileException flavored IOException
 */
private static Reader pickReaderVersion(Path path, FSDataInputStreamWrapper fsdis,
    long size, CacheConfig cacheConf, HFileSystem hfs, Configuration conf) throws IOException {
  FixedFileTrailer trailer = null;
  try {
    boolean isHBaseChecksum = fsdis.shouldUseHBaseChecksum();
    assert !isHBaseChecksum; // Initially we must read with FS checksum.
    trailer = FixedFileTrailer.readFromStream(fsdis.getStream(isHBaseChecksum), size);
    switch (trailer.getMajorVersion()) {
    case 2:
      return new HFileReaderV2(path, trailer, fsdis, size, cacheConf, hfs, conf);
    case 3 :
      return new HFileReaderV3(path, trailer, fsdis, size, cacheConf, hfs, conf);
    default:
      throw new IllegalArgumentException("Invalid HFile version " + trailer.getMajorVersion());
    }
  } catch (Throwable t) {
    try {
      fsdis.close();
    } catch (Throwable t2) {
      LOG.warn("Error closing fsdis FSDataInputStreamWrapper", t2);
    }
    throw new CorruptHFileException("Problem reading HFile Trailer from file " + path, t);
  }
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * @param fs fileystem to read from
 * @param p path to the file
 * @param in {@link FSDataInputStreamWrapper}
 * @param size Full size of the file
 * @param cacheConf
 * @param r original reference file. This will be not null only when reading a split file.
 * @param reader the base reader instance
 * @return The reader to use
 * @throws IOException
 */
public StoreFile.Reader postStoreFileReaderOpen(final FileSystem fs, final Path p,
    final FSDataInputStreamWrapper in, final long size, final CacheConfig cacheConf,
    final Reference r, StoreFile.Reader reader) throws IOException {
  ObserverContext<RegionCoprocessorEnvironment> ctx = null;
  for (RegionEnvironment env : coprocessors) {
    if (env.getInstance() instanceof RegionObserver) {
      ctx = ObserverContext.createAndPrepare(env, ctx);
      Thread currentThread = Thread.currentThread();
      ClassLoader cl = currentThread.getContextClassLoader();
      try {
        currentThread.setContextClassLoader(env.getClassLoader());
        reader = ((RegionObserver) env.getInstance()).postStoreFileReaderOpen(ctx, fs, p, in,
          size, cacheConf, r, reader);
      } catch (Throwable e) {
        handleCoprocessorThrowable(env, e);
      } finally {
        currentThread.setContextClassLoader(cl);
      }
      if (ctx.shouldComplete()) {
        break;
      }
    }
  }
  return reader;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
@Override
public StoreFile.Reader preStoreFileReaderOpen(
  ObserverContext<RegionCoprocessorEnvironment> ctx, FileSystem fs, Path p,
  FSDataInputStreamWrapper in, long size, CacheConfig cacheConf, Reference r,
  StoreFile.Reader reader) throws IOException {
  try {
    long delay = rnd.nextInt(3);
    LOG.info("@@@ Delaying region " +
      ctx.getEnvironment().getRegion().getRegionInfo().
        getRegionNameAsString() + " for " + delay + " seconds...");
    Thread.sleep(delay * 1000);
  } catch (InterruptedException ie) {
    LOG.error(ie);
  }
  return reader;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * @param fs fileystem to read from
 * @param p path to the file
 * @param in {@link FSDataInputStreamWrapper}
 * @param size Full size of the file
 * @param cacheConf
 * @param r original reference file. This will be not null only when reading a split file.
 * @return a Reader instance to use instead of the base reader if overriding
 * default behavior, null otherwise
 * @throws IOException
 */
public StoreFileReader preStoreFileReaderOpen(final FileSystem fs, final Path p,
    final FSDataInputStreamWrapper in, final long size, final CacheConfig cacheConf,
    final Reference r) throws IOException {
  if (coprocEnvironments.isEmpty()) {
    return null;
  }
  return execOperationWithResult(
      new ObserverOperationWithResult<RegionObserver, StoreFileReader>(regionObserverGetter, null) {
        @Override
        public StoreFileReader call(RegionObserver observer) throws IOException {
          return observer.preStoreFileReaderOpen(this, fs, p, in, size, cacheConf, r,
              getResult());
        }
      });
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * @param fs fileystem to read from
 * @param p path to the file
 * @param in {@link FSDataInputStreamWrapper}
 * @param size Full size of the file
 * @param cacheConf
 * @param r original reference file. This will be not null only when reading a split file.
 * @param reader the base reader instance
 * @return The reader to use
 * @throws IOException
 */
public StoreFileReader postStoreFileReaderOpen(final FileSystem fs, final Path p,
    final FSDataInputStreamWrapper in, final long size, final CacheConfig cacheConf,
    final Reference r, final StoreFileReader reader) throws IOException {
  if (this.coprocEnvironments.isEmpty()) {
    return reader;
  }
  return execOperationWithResult(
      new ObserverOperationWithResult<RegionObserver, StoreFileReader>(regionObserverGetter, reader) {
        @Override
        public StoreFileReader call(RegionObserver observer) throws IOException {
          return observer.postStoreFileReaderOpen(this, fs, p, in, size, cacheConf, r,
              getResult());
        }
      });
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
FSReaderImpl(FSDataInputStreamWrapper stream, long fileSize, HFileSystem hfs, Path path,
    HFileContext fileContext) throws IOException {
  this.fileSize = fileSize;
  this.hfs = hfs;
  if (path != null) {
    this.pathName = path.toString();
  }
  this.fileContext = fileContext;
  this.hdrSize = headerSize(fileContext.isUseHBaseChecksum());

  this.streamWrapper = stream;
  // Older versions of HBase didn't support checksum.
  this.streamWrapper.prepareForBlockReader(!fileContext.isUseHBaseChecksum());
  defaultDecodingCtx = new HFileBlockDefaultDecodingContext(fileContext);
  encodedBlockDecodingCtx = defaultDecodingCtx;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
@Test
public void testNewBlocksHaveDefaultChecksum() throws IOException {
  Path path = new Path(TEST_UTIL.getDataTestDir(), "default_checksum");
  FSDataOutputStream os = fs.create(path);
  HFileContext meta = new HFileContextBuilder().build();
  HFileBlock.Writer hbw = new HFileBlock.Writer(null, meta);
  DataOutputStream dos = hbw.startWriting(BlockType.DATA);
  for (int i = 0; i < 1000; ++i)
    dos.writeInt(i);
  hbw.writeHeaderAndData(os);
  int totalSize = hbw.getOnDiskSizeWithHeader();
  os.close();

  // Use hbase checksums.
  assertEquals(true, hfs.useHBaseChecksum());

  FSDataInputStreamWrapper is = new FSDataInputStreamWrapper(fs, path);
  meta = new HFileContextBuilder().withHBaseCheckSum(true).build();
  HFileBlock.FSReader hbr = new HFileBlock.FSReaderImpl(
      is, totalSize, (HFileSystem) fs, path, meta);
  HFileBlock b = hbr.readBlockData(0, -1, false, false);
  assertEquals(b.getChecksumType(), ChecksumType.getDefaultChecksumType().getCode());
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Read all blocks from {@code path} to populate {@code blockCache}.
 */
private static void cacheBlocks(Configuration conf, CacheConfig cacheConfig, FileSystem fs,
    Path path, HFileContext cxt) throws IOException {
  FSDataInputStreamWrapper fsdis = new FSDataInputStreamWrapper(fs, path);
  long fileSize = fs.getFileStatus(path).getLen();
  FixedFileTrailer trailer =
    FixedFileTrailer.readFromStream(fsdis.getStream(false), fileSize);
  HFile.Reader reader = new HFileReaderImpl(path, trailer, fsdis, fileSize, cacheConfig,
    fsdis.getHfs(), conf);
  reader.loadFileInfo();
  long offset = trailer.getFirstDataBlockOffset(),
    max = trailer.getLastDataBlockOffset();
  List<HFileBlock> blocks = new ArrayList<>(4);
  HFileBlock block;
  while (offset <= max) {
    block = reader.readBlock(offset, -1, /* cacheBlock */ true, /* pread */ false,
    /* isCompaction */ false, /* updateCacheMetrics */ true, null, null);
    offset += block.getOnDiskSizeWithHeader();
    blocks.add(block);
  }
  LOG.info("read " + Iterables.toString(blocks));
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Method returns the reader given the specified arguments.
 * TODO This is a bad abstraction.  See HBASE-6635.
 *
 * @param path hfile's path
 * @param fsdis stream of path's file
 * @param size max size of the trailer.
 * @param cacheConf Cache configuation values, cannot be null.
 * @param hfs
 * @return an appropriate instance of HFileReader
 * @throws IOException If file is invalid, will throw CorruptHFileException flavored IOException
 */
private static Reader pickReaderVersion(Path path, FSDataInputStreamWrapper fsdis,
    long size, CacheConfig cacheConf, HFileSystem hfs) throws IOException {
  FixedFileTrailer trailer = null;
  try {
    boolean isHBaseChecksum = fsdis.shouldUseHBaseChecksum();
    assert !isHBaseChecksum; // Initially we must read with FS checksum.
    trailer = FixedFileTrailer.readFromStream(fsdis.getStream(isHBaseChecksum), size);
    switch (trailer.getMajorVersion()) {
    case 2:
      return new HFileReaderV2(
        path, trailer, fsdis, size, cacheConf, hfs);
    default:
      throw new CorruptHFileException("Invalid HFile version " + trailer.getMajorVersion());
    }
  } catch (Throwable t) {
    try {
      fsdis.close();
    } catch (Throwable t2) {
      LOG.warn("Error closing fsdis FSDataInputStreamWrapper", t2);
    }
    throw new CorruptHFileException("Problem reading HFile Trailer from file " + path, t);
  }
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * @param fs fileystem to read from
 * @param p path to the file
 * @param in {@link FSDataInputStreamWrapper}
 * @param size Full size of the file
 * @param cacheConf
 * @param r original reference file. This will be not null only when reading a split file.
 * @return a Reader instance to use instead of the base reader if overriding
 * default behavior, null otherwise
 * @throws IOException
 */
public StoreFile.Reader preStoreFileReaderOpen(final FileSystem fs, final Path p,
    final FSDataInputStreamWrapper in, final long size, final CacheConfig cacheConf,
    final Reference r) throws IOException {
  return execOperationWithResult(null,
      coprocessors.isEmpty() ? null : new RegionOperationWithResult<StoreFile.Reader>() {
    @Override
    public void call(RegionObserver oserver, ObserverContext<RegionCoprocessorEnvironment> ctx)
        throws IOException {
      setResult(oserver.preStoreFileReaderOpen(ctx, fs, p, in, size, cacheConf, r, getResult()));
    }
  });
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * @param fs fileystem to read from
 * @param p path to the file
 * @param in {@link FSDataInputStreamWrapper}
 * @param size Full size of the file
 * @param cacheConf
 * @param r original reference file. This will be not null only when reading a split file.
 * @param reader the base reader instance
 * @return The reader to use
 * @throws IOException
 */
public StoreFile.Reader postStoreFileReaderOpen(final FileSystem fs, final Path p,
    final FSDataInputStreamWrapper in, final long size, final CacheConfig cacheConf,
    final Reference r, final StoreFile.Reader reader) throws IOException {
  return execOperationWithResult(reader,
      coprocessors.isEmpty() ? null : new RegionOperationWithResult<StoreFile.Reader>() {
    @Override
    public void call(RegionObserver oserver, ObserverContext<RegionCoprocessorEnvironment> ctx)
        throws IOException {
      setResult(oserver.postStoreFileReaderOpen(ctx, fs, p, in, size, cacheConf, r, getResult()));
    }
  });
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
public FSReaderImpl(FSDataInputStreamWrapper stream, long fileSize, HFileSystem hfs, Path path,
    HFileContext fileContext) throws IOException {
  super(fileSize, hfs, path, fileContext);
  this.streamWrapper = stream;
  // Older versions of HBase didn't support checksum.
  this.streamWrapper.prepareForBlockReader(!fileContext.isUseHBaseChecksum());
  defaultDecodingCtx = new HFileBlockDefaultDecodingContext(fileContext);
  encodedBlockDecodingCtx = defaultDecodingCtx;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
protected HFileContext createHFileContext(FSDataInputStreamWrapper fsdis, long fileSize,
    HFileSystem hfs, Path path, FixedFileTrailer trailer) throws IOException {
  return new HFileContextBuilder()
    .withIncludesMvcc(this.includesMemstoreTS)
    .withCompression(this.compressAlgo)
    .withHBaseCheckSum(trailer.getMinorVersion() >= MINOR_VERSION_WITH_CHECKSUM)
    .build();
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
@Override
public Reader preStoreFileReaderOpen(ObserverContext<RegionCoprocessorEnvironment> ctx,
    FileSystem fs, Path p, FSDataInputStreamWrapper in, long size, CacheConfig cacheConf,
    Reference r, Reader reader) throws IOException {
  ctPreStoreFileReaderOpen.incrementAndGet();
  return null;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
@Override
public Reader postStoreFileReaderOpen(ObserverContext<RegionCoprocessorEnvironment> ctx,
    FileSystem fs, Path p, FSDataInputStreamWrapper in, long size, CacheConfig cacheConf,
    Reference r, Reader reader) throws IOException {
  ctPostStoreFileReaderOpen.incrementAndGet();
  return reader;
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * Test all checksum types by writing and reading back blocks.
 */
@Test
public void testAllChecksumTypes() throws IOException {
  List<ChecksumType> cktypes = new ArrayList<>(Arrays.asList(ChecksumType.values()));
  for (Iterator<ChecksumType> itr = cktypes.iterator(); itr.hasNext(); ) {
    ChecksumType cktype = itr.next();
    Path path = new Path(TEST_UTIL.getDataTestDir(), "checksum" + cktype.getName());
    FSDataOutputStream os = fs.create(path);
    HFileContext meta = new HFileContextBuilder()
        .withChecksumType(cktype).build();
    HFileBlock.Writer hbw = new HFileBlock.Writer(null, meta);
    DataOutputStream dos = hbw.startWriting(BlockType.DATA);
    for (int i = 0; i < 1000; ++i)
      dos.writeInt(i);
    hbw.writeHeaderAndData(os);
    int totalSize = hbw.getOnDiskSizeWithHeader();
    os.close();

    // Use hbase checksums.
    assertEquals(true, hfs.useHBaseChecksum());

    FSDataInputStreamWrapper is = new FSDataInputStreamWrapper(fs, path);
    meta = new HFileContextBuilder().withHBaseCheckSum(true).build();
    HFileBlock.FSReader hbr = new HFileBlock.FSReaderImpl(
        is, totalSize, (HFileSystem) fs, path, meta);
    HFileBlock b = hbr.readBlockData(0, -1, -1, false);
    ByteBuffer data = b.getBufferWithoutHeader();
    for (int i = 0; i < 1000; i++) {
      assertEquals(i, data.getInt());
    }
    boolean exception_thrown = false;
    try {
      data.getInt();
    } catch (BufferUnderflowException e) {
      exception_thrown = true;
    }
    assertTrue(exception_thrown);
    assertEquals(0, HFile.getChecksumFailuresCount());
  }
}
 

import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; //导入依赖的package包/类
/**
 * This factory method is used only by unit tests
 */
static Reader createReaderFromStream(Path path,
    FSDataInputStream fsdis, long size, CacheConfig cacheConf, Configuration conf)
    throws IOException {
  FSDataInputStreamWrapper wrapper = new FSDataInputStreamWrapper(fsdis);
  return pickReaderVersion(path, wrapper, size, cacheConf, null, conf);
}