Examples with FileStatus org.apache.flink.core.fs.FileStatus used on opensource projects

Example 31 with FileStatus

use of org.apache.flink.core.fs.FileStatus in project flink by apache.

the class BinaryInputFormat method getStatistics.

@Override
public SequentialStatistics getStatistics(BaseStatistics cachedStats) {
    final FileBaseStatistics cachedFileStats = (cachedStats != null && cachedStats instanceof FileBaseStatistics) ? (FileBaseStatistics) cachedStats : null;
    try {
        final Path filePath = this.filePath;
        // get the filesystem
        final FileSystem fs = FileSystem.get(filePath.toUri());
        final ArrayList<FileStatus> allFiles = new ArrayList<FileStatus>(1);
        // let the file input format deal with the up-to-date check and the basic size
        final FileBaseStatistics stats = getFileStats(cachedFileStats, filePath, fs, allFiles);
        if (stats == null) {
            return null;
        }
        // check whether the file stats are still sequential stats (in that case they are still valid)
        if (stats instanceof SequentialStatistics) {
            return (SequentialStatistics) stats;
        }
        return createStatistics(allFiles, stats);
    } catch (IOException ioex) {
        if (LOG.isWarnEnabled()) {
            LOG.warn(String.format("Could not determine complete statistics for file '%s' due to an I/O error", this.filePath), ioex);
        }
    } catch (Throwable t) {
        if (LOG.isErrorEnabled()) {
            LOG.error(String.format("Unexpected problem while getting the file statistics for file '%s'", this.filePath), t);
        }
    }
    // no stats available
    return null;
}

Example 32 with FileStatus

use of org.apache.flink.core.fs.FileStatus in project flink by apache.

the class HadoopInputFormatBase method getStatistics.

@Override
public BaseStatistics getStatistics(BaseStatistics cachedStats) throws IOException {
    // only gather base statistics for FileInputFormats
    if (!(mapreduceInputFormat instanceof FileInputFormat)) {
        return null;
    }
    JobContext jobContext = new JobContextImpl(configuration, null);
    final FileBaseStatistics cachedFileStats = (cachedStats instanceof FileBaseStatistics) ? (FileBaseStatistics) cachedStats : null;
    try {
        final org.apache.hadoop.fs.Path[] paths = FileInputFormat.getInputPaths(jobContext);
        return getFileStats(cachedFileStats, paths, new ArrayList<FileStatus>(1));
    } catch (IOException ioex) {
        if (LOG.isWarnEnabled()) {
            LOG.warn("Could not determine statistics due to an io error: " + ioex.getMessage());
        }
    } catch (Throwable t) {
        if (LOG.isErrorEnabled()) {
            LOG.error("Unexpected problem while getting the file statistics: " + t.getMessage(), t);
        }
    }
    // no statistics available
    return null;
}

Example 33 with FileStatus

use of org.apache.flink.core.fs.FileStatus in project flink by apache.

the class HadoopInputFormatBase method getFileStats.

// --------------------------------------------------------------------------------------------
// Helper methods
// --------------------------------------------------------------------------------------------
private FileBaseStatistics getFileStats(FileBaseStatistics cachedStats, org.apache.hadoop.fs.Path[] hadoopFilePaths, ArrayList<FileStatus> files) throws IOException {
    long latestModTime = 0L;
    // get the file info and check whether the cached statistics are still valid.
    for (org.apache.hadoop.fs.Path hadoopPath : hadoopFilePaths) {
        final Path filePath = new Path(hadoopPath.toUri());
        final FileSystem fs = FileSystem.get(filePath.toUri());
        final FileStatus file = fs.getFileStatus(filePath);
        latestModTime = Math.max(latestModTime, file.getModificationTime());
        // enumerate all files and check their modification time stamp.
        if (file.isDir()) {
            FileStatus[] fss = fs.listStatus(filePath);
            files.ensureCapacity(files.size() + fss.length);
            for (FileStatus s : fss) {
                if (!s.isDir()) {
                    files.add(s);
                    latestModTime = Math.max(s.getModificationTime(), latestModTime);
                }
            }
        } else {
            files.add(file);
        }
    }
    // check whether the cached statistics are still valid, if we have any
    if (cachedStats != null && latestModTime <= cachedStats.getLastModificationTime()) {
        return cachedStats;
    }
    // calculate the whole length
    long len = 0;
    for (FileStatus s : files) {
        len += s.getLen();
    }
    // sanity check
    if (len <= 0) {
        len = BaseStatistics.SIZE_UNKNOWN;
    }
    return new FileBaseStatistics(latestModTime, len, BaseStatistics.AVG_RECORD_BYTES_UNKNOWN);
}

Example 34 with FileStatus

use of org.apache.flink.core.fs.FileStatus in project flink by apache.

the class FileInputFormat method createInputSplits.

/**
 * Computes the input splits for the file. By default, one file block is one split. If more
 * splits are requested than blocks are available, then a split may be a fraction of a block and
 * splits may cross block boundaries.
 *
 * @param minNumSplits The minimum desired number of file splits.
 * @return The computed file splits.
 * @see org.apache.flink.api.common.io.InputFormat#createInputSplits(int)
 */
@Override
public FileInputSplit[] createInputSplits(int minNumSplits) throws IOException {
    if (minNumSplits < 1) {
        throw new IllegalArgumentException("Number of input splits has to be at least 1.");
    }
    // take the desired number of splits into account
    minNumSplits = Math.max(minNumSplits, this.numSplits);
    final List<FileInputSplit> inputSplits = new ArrayList<FileInputSplit>(minNumSplits);
    // get all the files that are involved in the splits
    List<FileStatus> files = new ArrayList<>();
    long totalLength = 0;
    for (Path path : getFilePaths()) {
        final FileSystem fs = path.getFileSystem();
        final FileStatus pathFile = fs.getFileStatus(path);
        if (pathFile.isDir()) {
            totalLength += addFilesInDir(path, files, true);
        } else {
            testForUnsplittable(pathFile);
            files.add(pathFile);
            totalLength += pathFile.getLen();
        }
    }
    // returns if unsplittable
    if (unsplittable) {
        int splitNum = 0;
        for (final FileStatus file : files) {
            final FileSystem fs = file.getPath().getFileSystem();
            final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, file.getLen());
            Set<String> hosts = new HashSet<String>();
            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(splitNum++, file.getPath(), 0, len, hosts.toArray(new String[hosts.size()]));
            inputSplits.add(fis);
        }
        return inputSplits.toArray(new FileInputSplit[inputSplits.size()]);
    }
    final long maxSplitSize = totalLength / minNumSplits + (totalLength % minNumSplits == 0 ? 0 : 1);
    // now that we have the files, generate the splits
    int splitNum = 0;
    for (final FileStatus file : files) {
        final FileSystem fs = file.getPath().getFileSystem();
        final long len = file.getLen();
        final long blockSize = file.getBlockSize();
        final long minSplitSize;
        if (this.minSplitSize <= blockSize) {
            minSplitSize = this.minSplitSize;
        } else {
            if (LOG.isWarnEnabled()) {
                LOG.warn("Minimal split size of " + this.minSplitSize + " is larger than the block size of " + blockSize + ". Decreasing minimal split size to block size.");
            }
            minSplitSize = blockSize;
        }
        final long splitSize = Math.max(minSplitSize, Math.min(maxSplitSize, blockSize));
        final long halfSplit = splitSize >>> 1;
        final long maxBytesForLastSplit = (long) (splitSize * MAX_SPLIT_SIZE_DISCREPANCY);
        if (len > 0) {
            // get the block locations and make sure they are in order with respect to their
            // offset
            final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, len);
            Arrays.sort(blocks);
            long bytesUnassigned = len;
            long position = 0;
            int blockIndex = 0;
            while (bytesUnassigned > maxBytesForLastSplit) {
                // get the block containing the majority of the data
                blockIndex = getBlockIndexForPosition(blocks, position, halfSplit, blockIndex);
                // create a new split
                FileInputSplit fis = new FileInputSplit(splitNum++, file.getPath(), position, splitSize, blocks[blockIndex].getHosts());
                inputSplits.add(fis);
                // adjust the positions
                position += splitSize;
                bytesUnassigned -= splitSize;
            }
            // assign the last split
            if (bytesUnassigned > 0) {
                blockIndex = getBlockIndexForPosition(blocks, position, halfSplit, blockIndex);
                final FileInputSplit fis = new FileInputSplit(splitNum++, file.getPath(), position, bytesUnassigned, blocks[blockIndex].getHosts());
                inputSplits.add(fis);
            }
        } else {
            // special case with a file of zero bytes size
            final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, 0);
            String[] hosts;
            if (blocks.length > 0) {
                hosts = blocks[0].getHosts();
            } else {
                hosts = new String[0];
            }
            final FileInputSplit fis = new FileInputSplit(splitNum++, file.getPath(), 0, 0, hosts);
            inputSplits.add(fis);
        }
    }
    return inputSplits.toArray(new FileInputSplit[inputSplits.size()]);
}

Example 35 with FileStatus

use of org.apache.flink.core.fs.FileStatus in project flink by apache.

the class BinaryInputFormat method getFiles.

protected List<FileStatus> getFiles() throws IOException {
    // get all the files that are involved in the splits
    List<FileStatus> files = new ArrayList<>();
    for (Path filePath : getFilePaths()) {
        final FileSystem fs = filePath.getFileSystem();
        final FileStatus pathFile = fs.getFileStatus(filePath);
        if (pathFile.isDir()) {
            // input is directory. list all contained files
            final FileStatus[] partials = fs.listStatus(filePath);
            for (FileStatus partial : partials) {
                if (!partial.isDir()) {
                    files.add(partial);
                }
            }
        } else {
            files.add(pathFile);
        }
    }
    return files;
}