Examples with StringComparator org.apache.flink.api.common.typeutils.base.StringComparator used on opensource projects

Example 1 with StringComparator

use of org.apache.flink.api.common.typeutils.base.StringComparator in project flink by apache.

the class MassiveStringSorting method testStringSorting.

public void testStringSorting() {
    File input = null;
    File sorted = null;
    try {
        // the source file
        input = generateFileWithStrings(300000, "http://some-uri.com/that/is/a/common/prefix/to/all");
        // the sorted file
        sorted = File.createTempFile("sorted_strings", "txt");
        String[] command = { "/bin/bash", "-c", "export LC_ALL=\"C\" && cat \"" + input.getAbsolutePath() + "\" | sort > \"" + sorted.getAbsolutePath() + "\"" };
        Process p = null;
        try {
            p = Runtime.getRuntime().exec(command);
            int retCode = p.waitFor();
            if (retCode != 0) {
                throw new Exception("Command failed with return code " + retCode);
            }
            p = null;
        } finally {
            if (p != null) {
                p.destroy();
            }
        }
        // sort the data
        UnilateralSortMerger<String> sorter = null;
        BufferedReader reader = null;
        BufferedReader verifyReader = null;
        try {
            MemoryManager mm = new MemoryManager(1024 * 1024, 1);
            IOManager ioMan = new IOManagerAsync();
            TypeSerializer<String> serializer = StringSerializer.INSTANCE;
            TypeComparator<String> comparator = new StringComparator(true);
            reader = new BufferedReader(new FileReader(input));
            MutableObjectIterator<String> inputIterator = new StringReaderMutableObjectIterator(reader);
            sorter = new UnilateralSortMerger<String>(mm, ioMan, inputIterator, new DummyInvokable(), new RuntimeSerializerFactory<String>(serializer, String.class), comparator, 1.0, 4, 0.8f, true, /* use large record handler */
            false);
            MutableObjectIterator<String> sortedData = sorter.getIterator();
            reader.close();
            // verify
            verifyReader = new BufferedReader(new FileReader(sorted));
            String next;
            while ((next = verifyReader.readLine()) != null) {
                String nextFromStratoSort = sortedData.next("");
                Assert.assertNotNull(nextFromStratoSort);
                Assert.assertEquals(next, nextFromStratoSort);
            }
        } finally {
            if (reader != null) {
                reader.close();
            }
            if (verifyReader != null) {
                verifyReader.close();
            }
            if (sorter != null) {
                sorter.close();
            }
        }
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        Assert.fail(e.getMessage());
    } finally {
        if (input != null) {
            input.delete();
        }
        if (sorted != null) {
            sorted.delete();
        }
    }
}