@Override public FileInputSplit[] createInputSplits(int minNumSplits) throws IOException { final List<FileStatus> files = this.getFiles(); final List<FileInputSplit> inputSplits = new ArrayList<FileInputSplit>(minNumSplits); for (FileStatus file : files) { final FileSystem fs = file.getPath().getFileSystem(); final long blockSize = this.blockSize == NATIVE_BLOCK_SIZE ? fs.getDefaultBlockSize() : this.blockSize; for (long pos = 0, length = file.getLen(); pos < length; pos += blockSize) { long remainingLength = Math.min(pos + blockSize, length) - pos; // get the block locations and make sure they are in order with respect to their offset final BlockLocation[] blocks = fs.getFileBlockLocations(file, pos, remainingLength); Arrays.sort(blocks); inputSplits.add(new FileInputSplit(inputSplits.size(), file.getPath(), pos, remainingLength, blocks[0].getHosts())); } } if (inputSplits.size() < minNumSplits) { LOG.warn(String.format( "With the given block size %d, the files %s cannot be split into %d blocks. Filling up with empty splits...", blockSize, Arrays.toString(getFilePaths()), minNumSplits)); FileStatus last = files.get(files.size() - 1); final BlockLocation[] blocks = last.getPath().getFileSystem().getFileBlockLocations(last, 0, last.getLen()); for (int index = files.size(); index < minNumSplits; index++) { inputSplits.add(new FileInputSplit(index, last.getPath(), last.getLen(), 0, blocks[0].getHosts())); } } return inputSplits.toArray(new FileInputSplit[inputSplits.size()]); }
Set<String> hosts = new HashSet<String>(); for(BlockLocation block : blocks) { hosts.addAll(Arrays.asList(block.getHosts())); blocks[blockIndex].getHosts()); inputSplits.add(fis); blockIndex = getBlockIndexForPosition(blocks, position, halfSplit, blockIndex); final FileInputSplit fis = new FileInputSplit(splitNum++, file.getPath(), position, bytesUnassigned, blocks[blockIndex].getHosts()); inputSplits.add(fis); String[] hosts; if (blocks.length > 0) { hosts = blocks[0].getHosts(); } else { hosts = new String[0];
private void createSplitForUnSplittableFile( LocatedFileStatus file, BlockingQueue<FileInputSplit> inputSplits, AtomicInteger splitNumber) throws IOException { final FileSystem fs = file.getPath().getFileSystem(); final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, file.getLen()); final Set<String> hosts = new HashSet<>(); for (BlockLocation block : blocks) { hosts.addAll(Arrays.asList(block.getHosts())); } long len = file.getLen(); if (testForUnsplittable(file)) { len = READ_WHOLE_SPLIT_FLAG; } FileInputSplit fis = new FileInputSplit(splitNumber.incrementAndGet(), file.getPath(), 0, len, hosts.toArray(new String[0])); inputSplits.add(fis); }
@Override public FileInputSplit[] createInputSplits(int minNumSplits) throws IOException { final List<FileStatus> files = this.getFiles(); final List<FileInputSplit> inputSplits = new ArrayList<FileInputSplit>(minNumSplits); for (FileStatus file : files) { final FileSystem fs = file.getPath().getFileSystem(); final long blockSize = this.blockSize == NATIVE_BLOCK_SIZE ? fs.getDefaultBlockSize() : this.blockSize; for (long pos = 0, length = file.getLen(); pos < length; pos += blockSize) { long remainingLength = Math.min(pos + blockSize, length) - pos; // get the block locations and make sure they are in order with respect to their offset final BlockLocation[] blocks = fs.getFileBlockLocations(file, pos, remainingLength); Arrays.sort(blocks); inputSplits.add(new FileInputSplit(inputSplits.size(), file.getPath(), pos, remainingLength, blocks[0].getHosts())); } } if (inputSplits.size() < minNumSplits) { LOG.warn(String.format( "With the given block size %d, the files %s cannot be split into %d blocks. Filling up with empty splits...", blockSize, Arrays.toString(getFilePaths()), minNumSplits)); FileStatus last = files.get(files.size() - 1); final BlockLocation[] blocks = last.getPath().getFileSystem().getFileBlockLocations(last, 0, last.getLen()); for (int index = files.size(); index < minNumSplits; index++) { inputSplits.add(new FileInputSplit(index, last.getPath(), last.getLen(), 0, blocks[0].getHosts())); } } return inputSplits.toArray(new FileInputSplit[inputSplits.size()]); }
@Override public FileInputSplit[] createInputSplits(int minNumSplits) throws IOException { final List<FileStatus> files = this.getFiles(); final List<FileInputSplit> inputSplits = new ArrayList<FileInputSplit>(minNumSplits); for (FileStatus file : files) { final FileSystem fs = file.getPath().getFileSystem(); final long blockSize = this.blockSize == NATIVE_BLOCK_SIZE ? fs.getDefaultBlockSize() : this.blockSize; for (long pos = 0, length = file.getLen(); pos < length; pos += blockSize) { long remainingLength = Math.min(pos + blockSize, length) - pos; // get the block locations and make sure they are in order with respect to their offset final BlockLocation[] blocks = fs.getFileBlockLocations(file, pos, remainingLength); Arrays.sort(blocks); inputSplits.add(new FileInputSplit(inputSplits.size(), file.getPath(), pos, remainingLength, blocks[0].getHosts())); } } if (inputSplits.size() < minNumSplits) { LOG.warn(String.format( "With the given block size %d, the files %s cannot be split into %d blocks. Filling up with empty splits...", blockSize, Arrays.toString(getFilePaths()), minNumSplits)); FileStatus last = files.get(files.size() - 1); final BlockLocation[] blocks = last.getPath().getFileSystem().getFileBlockLocations(last, 0, last.getLen()); for (int index = files.size(); index < minNumSplits; index++) { inputSplits.add(new FileInputSplit(index, last.getPath(), last.getLen(), 0, blocks[0].getHosts())); } } return inputSplits.toArray(new FileInputSplit[inputSplits.size()]); }
blocks[blockIndex].getHosts()); inputSplits.add(fis); blockIndex = getBlockIndexForPosition(blocks, position, halfSplit, blockIndex); final FileInputSplit fis = new FileInputSplit(splitNumber.incrementAndGet(), file.getPath(), position, bytesUnassigned, blocks[blockIndex].getHosts()); inputSplits.add(fis); final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, 0); final String[] hosts = (blocks.length > 0) ? blocks[0].getHosts() : new String[0];
Set<String> hosts = new HashSet<String>(); for(BlockLocation block : blocks) { hosts.addAll(Arrays.asList(block.getHosts())); blocks[blockIndex].getHosts()); inputSplits.add(fis); blockIndex = getBlockIndexForPosition(blocks, position, halfSplit, blockIndex); final FileInputSplit fis = new FileInputSplit(splitNum++, file.getPath(), position, bytesUnassigned, blocks[blockIndex].getHosts()); inputSplits.add(fis); String[] hosts; if (blocks.length > 0) { hosts = blocks[0].getHosts(); } else { hosts = new String[0];