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

Example 26 with FileSystem

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

the class HDFSTest method testHDFS.

@Test
public void testHDFS() {
    Path file = new Path(hdfsURI + hdPath);
    org.apache.hadoop.fs.Path result = new org.apache.hadoop.fs.Path(hdfsURI + "/result");
    try {
        FileSystem fs = file.getFileSystem();
        assertTrue("Must be HadoopFileSystem", fs instanceof HadoopFileSystem);
        DopOneTestEnvironment.setAsContext();
        try {
            WordCount.main(new String[] { "--input", file.toString(), "--output", result.toString() });
        } catch (Throwable t) {
            t.printStackTrace();
            Assert.fail("Test failed with " + t.getMessage());
        } finally {
            DopOneTestEnvironment.unsetAsContext();
        }
        assertTrue("No result file present", hdfs.exists(result));
        // validate output:
        org.apache.hadoop.fs.FSDataInputStream inStream = hdfs.open(result);
        StringWriter writer = new StringWriter();
        IOUtils.copy(inStream, writer);
        String resultString = writer.toString();
        Assert.assertEquals("hdfs 10\n" + "hello 10\n", resultString);
        inStream.close();
    } catch (IOException e) {
        e.printStackTrace();
        Assert.fail("Error in test: " + e.getMessage());
    }
}

Example 27 with FileSystem

use of org.apache.flink.core.fs.FileSystem 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 28 with FileSystem

use of org.apache.flink.core.fs.FileSystem 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 Path path = this.filePath;
    final List<FileInputSplit> inputSplits = new ArrayList<FileInputSplit>(minNumSplits);
    // get all the files that are involved in the splits
    List<FileStatus> files = new ArrayList<FileStatus>();
    long totalLength = 0;
    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 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 = (minNumSplits < 1) ? Long.MAX_VALUE : (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 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 29 with FileSystem

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

the class Plan method registerCachedFile.

/**
	 *  register cache files in program level
	 * @param entry contains all relevant information
	 * @param name user defined name of that file
	 * @throws java.io.IOException
	 */
public void registerCachedFile(String name, DistributedCacheEntry entry) throws IOException {
    if (!this.cacheFile.containsKey(name)) {
        try {
            URI u = new URI(entry.filePath);
            if (!u.getPath().startsWith("/")) {
                u = new File(entry.filePath).toURI();
            }
            FileSystem fs = FileSystem.get(u);
            if (fs.exists(new Path(u.getPath()))) {
                this.cacheFile.put(name, new DistributedCacheEntry(u.toString(), entry.isExecutable));
            } else {
                throw new IOException("File " + u.toString() + " doesn't exist.");
            }
        } catch (URISyntaxException ex) {
            throw new IOException("Invalid path: " + entry.filePath, ex);
        }
    } else {
        throw new IOException("cache file " + name + "already exists!");
    }
}

Example 30 with FileSystem

use of org.apache.flink.core.fs.FileSystem 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<FileStatus>();
    final FileSystem fs = this.filePath.getFileSystem();
    final FileStatus pathFile = fs.getFileStatus(this.filePath);
    if (pathFile.isDir()) {
        // input is directory. list all contained files
        final FileStatus[] partials = fs.listStatus(this.filePath);
        for (FileStatus partial : partials) {
            if (!partial.isDir()) {
                files.add(partial);
            }
        }
    } else {
        files.add(pathFile);
    }
    return files;
}