Examples with FileSystem edu.iu.dsc.tws.api.data.FileSystem used on opensource projects

Example 16 with FileSystem

use of edu.iu.dsc.tws.api.data.FileSystem in project twister2 by DSC-SPIDAL.

the class FileSystemUtils method getFileSystem.

/**
 * For hadoop file system
 */
public static FileSystem getFileSystem(URI uri, Config config) throws IOException {
    FileSystem fs = null;
    URI asked = uri;
    URI curUri = uri;
    if (curUri == null) {
        throw new IOException("The URI " + curUri.toString() + " is not a vaild URI");
    }
    // TODO: check if the sycn is actually needed or can be scoped down
    synchronized (SYNCHRONIZATION_OBJECT) {
        if (curUri.getScheme() == null) {
            try {
                if (defaultScheme == null) {
                    defaultScheme = new URI(DataConstants.DEFAULT_FILESYSTEM_SCHEME);
                }
                curUri = new URI(defaultScheme.getScheme(), null, defaultScheme.getHost(), defaultScheme.getPort(), curUri.getPath(), null, null);
            } catch (URISyntaxException e) {
                try {
                    if (defaultScheme.getScheme().equals("file")) {
                        curUri = new URI("file", null, new Path(new File(curUri.getPath()).getAbsolutePath()).toUri().getPath(), null);
                    }
                } catch (URISyntaxException ex) {
                    // we tried to repair it, but could not. report the scheme error
                    throw new IOException("The URI '" + curUri.toString() + "' is not valid.");
                }
            }
        }
        if (curUri.getScheme() == null) {
            throw new IOException("The URI '" + curUri + "' is invalid.\n" + "The fs.default-scheme = " + defaultScheme + ", the requested URI = " + asked + ", and the final URI = " + curUri + ".");
        }
        if (curUri.getScheme().equals("file") && curUri.getAuthority() != null && !curUri.getAuthority().isEmpty()) {
            String supposedUri = "file:///" + curUri.getAuthority() + curUri.getPath();
            throw new IOException("Found local file path with authority '" + curUri.getAuthority() + "' in path '" + curUri.toString() + "'. Hint: Did you forget a slash? (correct path would be '" + supposedUri + "')");
        }
        // TODO : need to add cache that can save FileSystem Objects and return from cache if available
        if (!isSupportedScheme(curUri.getScheme())) {
        // TODO: handle when the system is not supported
        } else {
            String fsClass = SUPPORTEDFS.get(curUri.getScheme());
            if (DataContext.TWISTER2_HDFS_FILESYSTEM.equals(curUri.getScheme())) {
                try {
                    fs = instantiateFileSystem(fsClass, config);
                } catch (NoSuchMethodException e) {
                    throw new RuntimeException("No such method to invoke", e);
                } catch (InvocationTargetException e) {
                    throw new RuntimeException("Invocation exception occured", e);
                }
                fs.initialize(curUri);
            } else {
                fs = instantiateFileSystem(fsClass);
                fs.initialize(curUri);
            }
        }
    }
    return fs;
}

Example 17 with FileSystem

use of edu.iu.dsc.tws.api.data.FileSystem in project twister2 by DSC-SPIDAL.

the class FileSystemUtils method instantiateFileSystem.

private static FileSystem instantiateFileSystem(String className, Config config) throws IOException, NoSuchMethodException, InvocationTargetException {
    Class<?> fileSystemClass;
    Object newInstance;
    try {
        Configuration conf = new Configuration(true);
        conf.addResource(new org.apache.hadoop.fs.Path(HdfsDataContext.getHdfsConfigDirectory(config)));
        conf.addResource(new org.apache.hadoop.fs.Path(HdfsDataContext.getHdfsDataDirectory(config)));
        conf.set("fs.defaultFS", getHdfsURL(config));
        fileSystemClass = ClassLoader.getSystemClassLoader().loadClass(className);
        Constructor<?> classConstructor = fileSystemClass.getConstructor(Configuration.class, org.apache.hadoop.fs.FileSystem.class);
        newInstance = classConstructor.newInstance(new Object[] { conf, org.apache.hadoop.fs.FileSystem.get(conf) });
    } catch (InstantiationException e) {
        throw new IOException("instantiation exception occured:" + e.getMessage());
    } catch (ClassNotFoundException e) {
        throw new IOException("Could not load file system class '" + className + '\'', e);
    } catch (IllegalAccessException e) {
        throw new IOException("Illegal access exception: " + e.getMessage(), e);
    }
    return (FileSystem) newInstance;
}

Example 18 with FileSystem

use of edu.iu.dsc.tws.api.data.FileSystem in project twister2 by DSC-SPIDAL.

the class FileSystemUtils method getFileSystem.

/**
 * Returns a unsafe filesystem for the given uri
 */
public static FileSystem getFileSystem(URI uri) throws IOException {
    FileSystem fs = null;
    URI asked = uri;
    URI curUri = uri;
    if (curUri == null) {
        throw new IOException("The URI " + curUri.toString() + " is not a vaild URI");
    }
    // TODO: check if the sycn is actually needed or can be scoped down
    synchronized (SYNCHRONIZATION_OBJECT) {
        if (curUri.getScheme() == null) {
            try {
                if (defaultScheme == null) {
                    defaultScheme = new URI(DataConstants.DEFAULT_FILESYSTEM_SCHEME);
                }
                curUri = new URI(defaultScheme.getScheme(), null, defaultScheme.getHost(), defaultScheme.getPort(), curUri.getPath(), null, null);
            } catch (URISyntaxException e) {
                try {
                    if (defaultScheme.getScheme().equals("file")) {
                        curUri = new URI("file", null, new Path(new File(curUri.getPath()).getAbsolutePath()).toUri().getPath(), null);
                    }
                } catch (URISyntaxException ex) {
                    // we tried to repair it, but could not. report the scheme error
                    throw new IOException("The URI '" + curUri.toString() + "' is not valid.");
                }
            }
        }
        if (curUri.getScheme() == null) {
            throw new IOException("The URI '" + curUri + "' is invalid.\n" + "The fs.default-scheme = " + defaultScheme + ", the requested URI = " + asked + ", and the final URI = " + curUri + ".");
        }
        if (curUri.getScheme().equals("file") && curUri.getAuthority() != null && !curUri.getAuthority().isEmpty()) {
            String supposedUri = "file:///" + curUri.getAuthority() + curUri.getPath();
            throw new IOException("Found local file path with authority '" + curUri.getAuthority() + "' in path '" + curUri.toString() + "'. Hint: Did you forget a slash? (correct path would be '" + supposedUri + "')");
        }
        // TODO : need to add cache that can save FileSystem Objects and return from cache if available
        if (!isSupportedScheme(curUri.getScheme())) {
        // TODO: handle when the system is not supported
        } else {
            String fsClass = SUPPORTEDFS.get(curUri.getScheme());
            fs = instantiateFileSystem(fsClass);
            fs.initialize(curUri);
        }
    }
    return fs;
}

Example 19 with FileSystem

use of edu.iu.dsc.tws.api.data.FileSystem in project twister2 by DSC-SPIDAL.

the class CSVInputPartitioner method sumFilesInDir.

long sumFilesInDir(Path path, List<FileStatus> files, boolean logExcludedFiles) throws IOException {
    final FileSystem fs = FileSystemUtils.get(path);
    long length = 0;
    for (FileStatus file : fs.listFiles(path)) {
        if (file.isDir()) {
            if (acceptFile(file) && enumerateNestedFiles) {
                length += sumFilesInDir(file.getPath(), files, logExcludedFiles);
            } else {
                if (logExcludedFiles) {
                    LOG.log(Level.INFO, "Directory " + file.getPath().toString() + " did not pass the " + "file-filter and is excluded.");
                }
            }
        } else {
            if (acceptFile(file)) {
                files.add(file);
                length += file.getLen();
            } else {
                if (logExcludedFiles) {
                    LOG.log(Level.INFO, "Directory " + file.getPath().toString() + " did not pass the file-filter and is excluded.");
                }
            }
        }
    }
    return length;
}

Example 20 with FileSystem

use of edu.iu.dsc.tws.api.data.FileSystem in project twister2 by DSC-SPIDAL.

the class CompleteCSVInputPartitioner method createInputSplits.

/**
 * It creates the split for the complete file.
 *
 * @param minNumSplits Number of minimal input splits, as a hint.
 */
@Override
public FileInputSplit<OT>[] createInputSplits(int minNumSplits) throws IOException {
    if (minNumSplits < 1) {
        throw new IllegalArgumentException("Number of input splits has to be at least 1.");
    }
    int curminNumSplits = Math.max(minNumSplits, this.numSplits);
    final Path path = this.filePath;
    final List<FileInputSplit> inputSplits = new ArrayList<>(curminNumSplits);
    List<FileStatus> files = new ArrayList<>();
    long totalLength = 0;
    final FileSystem fs = FileSystemUtils.get(path, config);
    final FileStatus pathFile = fs.getFileStatus(path);
    if (pathFile.isDir()) {
        totalLength += sumFilesInDir(path, files, true);
    } else {
        files.add(pathFile);
        totalLength += pathFile.getLen();
    }
    final long maxSplitSize = totalLength;
    // Generate the splits
    int splitNum = 0;
    for (final FileStatus file : files) {
        final long len = file.getLen();
        final long blockSize = file.getBlockSize();
        final long localminSplitSize;
        if (this.minSplitSize <= blockSize) {
            localminSplitSize = this.minSplitSize;
        } else {
            LOG.log(Level.WARNING, "Minimal split size of " + this.minSplitSize + " is larger than the block size of " + blockSize + ". Decreasing minimal split size to block size.");
            localminSplitSize = blockSize;
        }
        final long splitSize = Math.max(localminSplitSize, Math.min(maxSplitSize, blockSize));
        if (len > 0) {
            final BlockLocation[] blocks = fs.getFileBlockLocations(file, 0, len);
            Arrays.sort(blocks);
            long position = 0;
            int blockIndex = 0;
            for (int i = 0; i < curminNumSplits; i++) {
                blockIndex = getBlockIndexForPosition(blocks, position, splitSize, blockIndex);
                final FileInputSplit fis = new CSVInputSplit(splitNum++, file.getPath(), position, splitSize, 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];
            }
            for (int i = 0; i < curminNumSplits; i++) {
                final FileInputSplit fis = new CSVInputSplit(splitNum++, file.getPath(), 0, 0, hosts);
                inputSplits.add(fis);
            }
        }
    }
    return inputSplits.toArray(new FileInputSplit[inputSplits.size()]);
}