Examples of org.apache.pig.builtin.PigStorage

org.apache.pig.builtin.PigStorage
A load function that parses a line of input into fields using a delimiter to set the fields. The delimiter is given as a regular expression. See {@link java.lang.String#split(String)} and {@link java.util.regex.Pattern}for more information.


    @Test
    public void testStoreComplexData() throws Exception {
        inpDB = GenRandomData.genRandFullTupTextDataBag(new Random(), 10, 100);
        storeAndCopyLocally(inpDB);
        PigStorage ps = new PigStorage("\t");
        int size = 0;
        BufferedReader br = new BufferedReader(new FileReader(outputFileName));
        for(String line=br.readLine();line!=null;line=br.readLine()){
            String[] flds = line.split("\t",-1);
            Tuple t = new DefaultTuple();


            ResourceFieldSchema mapfs = GenRandomData.getRandMapFieldSchema();
            ResourceFieldSchema bagfs = GenRandomData.getSmallTupDataBagFieldSchema();
            ResourceFieldSchema tuplefs = GenRandomData.getSmallTupleFieldSchema();


            t.append(flds[0].compareTo("")!=0 ? ps.getLoadCaster().bytesToBag(flds[0].getBytes(), bagfs) : null);
            t.append(flds[1].compareTo("")!=0 ? new DataByteArray(flds[1].getBytes()) : null);
            t.append(flds[2].compareTo("")!=0 ? ps.getLoadCaster().bytesToCharArray(flds[2].getBytes()) : null);
            t.append(flds[3].compareTo("")!=0 ? ps.getLoadCaster().bytesToDouble(flds[3].getBytes()) : null);
            t.append(flds[4].compareTo("")!=0 ? ps.getLoadCaster().bytesToFloat(flds[4].getBytes()) : null);
            t.append(flds[5].compareTo("")!=0 ? ps.getLoadCaster().bytesToInteger(flds[5].getBytes()) : null);
            t.append(flds[6].compareTo("")!=0 ? ps.getLoadCaster().bytesToLong(flds[6].getBytes()) : null);
            t.append(flds[7].compareTo("")!=0 ? ps.getLoadCaster().bytesToMap(flds[7].getBytes(), mapfs) : null);
            t.append(flds[8].compareTo("")!=0 ? ps.getLoadCaster().bytesToTuple(flds[8].getBytes(), tuplefs) : null);
            t.append(flds[9].compareTo("")!=0 ? ps.getLoadCaster().bytesToBoolean(flds[9].getBytes()) : null);
            t.append(flds[10].compareTo("")!=0 ? ps.getLoadCaster().bytesToDateTime(flds[10].getBytes()) : null);
            assertEquals(true, TestHelper.bagContains(inpDB, t));
            ++size;
        }
        assertEquals(true, size==inpDB.size());
        br.close();

View Full Code Here

    public void testStoreComplexDataWithNull() throws Exception {
        Tuple inputTuple = GenRandomData.genRandSmallBagTextTupleWithNulls(new Random(), 10, 100);
        inpDB = DefaultBagFactory.getInstance().newDefaultBag();
        inpDB.add(inputTuple);
        storeAndCopyLocally(inpDB);
        PigStorage ps = new PigStorage("\t");
        BufferedReader br = new BufferedReader(new FileReader(outputFileName));
        for(String line=br.readLine();line!=null;line=br.readLine()){
            System.err.println("Complex data: ");
            System.err.println(line);
            String[] flds = line.split("\t",-1);
            Tuple t = new DefaultTuple();


            ResourceFieldSchema stringfs = new ResourceFieldSchema();
            stringfs.setType(DataType.CHARARRAY);
            ResourceFieldSchema intfs = new ResourceFieldSchema();
            intfs.setType(DataType.INTEGER);
            ResourceFieldSchema bytefs = new ResourceFieldSchema();
            bytefs.setType(DataType.BYTEARRAY);


            ResourceSchema tupleSchema = new ResourceSchema();
            tupleSchema.setFields(new ResourceFieldSchema[]{stringfs, intfs});
            ResourceFieldSchema tuplefs = new ResourceFieldSchema();
            tuplefs.setSchema(tupleSchema);
            tuplefs.setType(DataType.TUPLE);


            ResourceSchema bagSchema = new ResourceSchema();
            bagSchema.setFields(new ResourceFieldSchema[]{tuplefs});
            ResourceFieldSchema bagfs = new ResourceFieldSchema();
            bagfs.setSchema(bagSchema);
            bagfs.setType(DataType.BAG);


            ResourceSchema mapSchema = new ResourceSchema();
            mapSchema.setFields(new ResourceFieldSchema[]{bytefs});
            ResourceFieldSchema mapfs = new ResourceFieldSchema();
            mapfs.setSchema(mapSchema);
            mapfs.setType(DataType.MAP);


            t.append(flds[0].compareTo("")!=0 ? ps.getLoadCaster().bytesToBag(flds[0].getBytes(), bagfs) : null);
            t.append(flds[1].compareTo("")!=0 ? new DataByteArray(flds[1].getBytes()) : null);
            t.append(flds[2].compareTo("")!=0 ? ps.getLoadCaster().bytesToCharArray(flds[2].getBytes()) : null);
            t.append(flds[3].compareTo("")!=0 ? ps.getLoadCaster().bytesToDouble(flds[3].getBytes()) : null);
            t.append(flds[4].compareTo("")!=0 ? ps.getLoadCaster().bytesToFloat(flds[4].getBytes()) : null);
            t.append(flds[5].compareTo("")!=0 ? ps.getLoadCaster().bytesToInteger(flds[5].getBytes()) : null);
            t.append(flds[6].compareTo("")!=0 ? ps.getLoadCaster().bytesToLong(flds[6].getBytes()) : null);
            t.append(flds[7].compareTo("")!=0 ? ps.getLoadCaster().bytesToMap(flds[7].getBytes(), mapfs) : null);
            t.append(flds[8].compareTo("")!=0 ? ps.getLoadCaster().bytesToTuple(flds[8].getBytes(), tuplefs) : null);
            t.append(flds[9].compareTo("")!=0 ? ps.getLoadCaster().bytesToBoolean(flds[9].getBytes()) : null);
            t.append(flds[10].compareTo("")!=0 ? ps.getLoadCaster().bytesToDateTime(flds[10].getBytes()) : null);
            t.append(flds[11].compareTo("")!=0 ? ps.getLoadCaster().bytesToCharArray(flds[10].getBytes()) : null);
            assertTrue(TestHelper.tupleEquals(inputTuple, t));
        }
        br.close();
    }

View Full Code Here

    @Test
    public void testLFPig() throws Exception {
        Util.createInputFile(cluster, "input.txt", new String[]
            {"this:is:delimited:by:a:colon\n"});
        int arity1 = 6;
        LoadFunc lf = new PigStorage(":");
        LoadFunc p1 = new ReadToEndLoader(lf, ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f1 = p1.getNext();
        assertTrue(f1.size() == arity1);
        Util.deleteFile(cluster, "input.txt");


        int LOOP_COUNT = 100;
        String[] input = new String[LOOP_COUNT * LOOP_COUNT];
        int n = 0;
        for (int i = 0; i < LOOP_COUNT; i++) {
            for (int j = 0; j < LOOP_COUNT; j++) {
                input[n++] = (i + "\t" + i + "\t" + j % 2);
            }
        }
        Util.createInputFile(cluster, "input.txt", input);


        LoadFunc p15 = new ReadToEndLoader(new PigStorage(), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);


        int count = 0;
        while (true) {
            Tuple f15 = p15.getNext();
            if (f15 == null)
                break;
            count++;
            assertEquals(3, f15.size());
        }
        assertEquals(LOOP_COUNT * LOOP_COUNT, count);
        Util.deleteFile(cluster, "input.txt");


        String input2 = ":this:has:a:leading:colon\n";
        int arity2 = 6;
        Util.createInputFile(cluster, "input.txt", new String[] {input2});
        LoadFunc p2 = new ReadToEndLoader(new PigStorage(":"), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f2 = p2.getNext();
        assertTrue(f2.size() == arity2);
        Util.deleteFile(cluster, "input.txt");


        String input3 = "this:has:a:trailing:colon:\n";
        int arity3 = 6;
        Util.createInputFile(cluster, "input.txt", new String[] {input3});
        LoadFunc p3 = new ReadToEndLoader(new PigStorage(":"), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f3 = p3.getNext();
        assertTrue(f3.size() == arity3);
        Util.deleteFile(cluster, "input.txt");
    }

View Full Code Here

        String query = "a = load 'testSFPig-input.txt';" +
                "store a into '" + outputLocation + "';";
        mrPigServer.setBatchOn();
        Util.registerMultiLineQuery(mrPigServer, query);
        mrPigServer.executeBatch();
        LoadFunc lfunc = new ReadToEndLoader(new PigStorage(), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), outputLocation, 0);
        Tuple f2 = lfunc.getNext();
        Util.deleteFile(cluster, "testSFPig-input.txt");


        Util.deleteFile(cluster, outputLocation);

View Full Code Here

            new Job(CONF), createMockPOLoadMapReduceOper(2L * 1000 * 1000 * 1001)));
    }


    private static MapReduceOper createMockPOLoadMapReduceOper(long size) throws Exception {
        MapReduceOper mro = new MapReduceOper(new OperatorKey());
        mro.mapPlan.add(createPOLoadWithSize(size, new PigStorage()));
        return mro;
    }

View Full Code Here


    @Test
    public void testPigStorageSchemaSearch() throws Exception {
        String globtestdir = "build/test/tmpglobbingdata/";
        ResourceSchema testSchema = new ResourceSchema(Utils.parseSchema("a0:chararray"));
        PigStorage pigStorage = new PigStorage();
        pigContext.connect();
        try{
            Util.deleteDirectory(new File(datadir));
            pig.mkdirs(globtestdir+"a");
            pig.mkdirs(globtestdir+"a/a0");
            putInputFile(globtestdir+"a/a0/input");
            pig.mkdirs(globtestdir+"a/b0");
            putInputFile(globtestdir+"a/b0/input");
            pig.mkdirs(globtestdir+"b");
        } catch (IOException e) {};


        // if schema file is not found, schema is null
        ResourceSchema schema = pigStorage.getSchema(globtestdir, new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(schema==null);


        // if .pig_schema is in the input directory
        putSchemaFile(globtestdir+"a/a0/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"a/a0", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(ResourceSchema.equals(schema, testSchema));
        new File(globtestdir+"a/a0/.pig_schema").delete();


        // .pig_schema in one of globStatus returned directory
        putSchemaFile(globtestdir+"a/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"*", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(ResourceSchema.equals(schema, testSchema));
        new File(globtestdir+"a/.pig_schema").delete();


        putSchemaFile(globtestdir+"b/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"*", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(ResourceSchema.equals(schema, testSchema));
        new File(globtestdir+"b/.pig_schema").delete();


        // if .pig_schema is deep in the globbing, it will not get used
        putSchemaFile(globtestdir+"a/a0/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"*", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(schema==null);
        putSchemaFile(globtestdir+"a/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"*", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(ResourceSchema.equals(schema, testSchema));
        new File(globtestdir+"a/a0/.pig_schema").delete();
        new File(globtestdir+"a/.pig_schema").delete();


        pigStorage = new PigStorage("\t", "-schema");
        putSchemaFile(globtestdir+"a/.pig_schema", testSchema);
        schema = pigStorage.getSchema(globtestdir+"{a,b}", new Job(ConfigurationUtil.toConfiguration(pigContext.getProperties())));
        Assert.assertTrue(ResourceSchema.equals(schema, testSchema));
    }

View Full Code Here

    @Test
    public void testLFPig() throws Exception {
        Util.createInputFile(cluster, "input.txt", new String[]
            {"this:is:delimited:by:a:colon\n"});
        int arity1 = 6;
        LoadFunc lf = new PigStorage(":");
        LoadFunc p1 = new ReadToEndLoader(lf, ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f1 = p1.getNext();
        assertTrue(f1.size() == arity1);
        Util.deleteFile(cluster, "input.txt");


        int LOOP_COUNT = 100;
        String[] input = new String[LOOP_COUNT * LOOP_COUNT];
        int n = 0;
        for (int i = 0; i < LOOP_COUNT; i++) {
            for (int j = 0; j < LOOP_COUNT; j++) {
                input[n++] = (i + "\t" + i + "\t" + j % 2);
            }
        }
        Util.createInputFile(cluster, "input.txt", input);


        LoadFunc p15 = new ReadToEndLoader(new PigStorage(), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);


        int count = 0;
        while (true) {
            Tuple f15 = p15.getNext();
            if (f15 == null)
                break;
            count++;
            assertEquals(3, f15.size());
        }
        assertEquals(LOOP_COUNT * LOOP_COUNT, count);
        Util.deleteFile(cluster, "input.txt");


        String input2 = ":this:has:a:leading:colon\n";
        int arity2 = 6;
        Util.createInputFile(cluster, "input.txt", new String[] {input2});
        LoadFunc p2 = new ReadToEndLoader(new PigStorage(":"), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f2 = p2.getNext();
        assertTrue(f2.size() == arity2);
        Util.deleteFile(cluster, "input.txt");


        String input3 = "this:has:a:trailing:colon:\n";
        int arity3 = 6;
        Util.createInputFile(cluster, "input.txt", new String[] {input2});
        LoadFunc p3 = new ReadToEndLoader(new PigStorage(":"), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), "input.txt", 0);
        Tuple f3 = p3.getNext();
        assertTrue(f3.size() == arity3);
        Util.deleteFile(cluster, "input.txt");
    }

View Full Code Here

        String query = "a = load 'testSFPig-input.txt';" +
            "store a into '" + outputLocation + "';";
        mrPigServer.setBatchOn();
        Util.registerMultiLineQuery(mrPigServer, query);
        mrPigServer.executeBatch();
        LoadFunc lfunc = new ReadToEndLoader(new PigStorage(), ConfigurationUtil.
            toConfiguration(cluster.getProperties()), outputLocation, 0);
        Tuple f2 = lfunc.getNext();
        Util.deleteFile(cluster, "testSFPig-input.txt");


        Util.deleteFile(cluster, outputLocation);

View Full Code Here


    @Test
    public void testGetInputSizeFromFs() throws Exception {
        long size = 2L * 1024 * 1024 * 1024;
        Assert.assertEquals(size, InputSizeReducerEstimator.getTotalInputFileSize(
                CONF, Lists.newArrayList(createPOLoadWithSize(size, new PigStorage())),
                new org.apache.hadoop.mapreduce.Job(CONF)));


        Assert.assertEquals(size, InputSizeReducerEstimator.getTotalInputFileSize(
                CONF,
                Lists.newArrayList(createPOLoadWithSize(size, new PigStorageWithStatistics())),
                new org.apache.hadoop.mapreduce.Job(CONF)));


        Assert.assertEquals(size * 2, InputSizeReducerEstimator.getTotalInputFileSize(
                CONF,
                Lists.newArrayList(
                        createPOLoadWithSize(size, new PigStorage()),
                        createPOLoadWithSize(size, new PigStorageWithStatistics())),
                new org.apache.hadoop.mapreduce.Job(CONF)));
    }

View Full Code Here


    @Test
    public void testStoreComplexData() throws Exception {
        inpDB = GenRandomData.genRandFullTupTextDataBag(new Random(), 10, 100);
        storeAndCopyLocally(inpDB);
        PigStorage ps = new PigStorage("\t");
        int size = 0;
        BufferedReader br = new BufferedReader(new FileReader(outputFileName));
        for(String line=br.readLine();line!=null;line=br.readLine()){
            String[] flds = line.split("\t",-1);
            Tuple t = new DefaultTuple();


            ResourceFieldSchema bagfs = GenRandomData.getSmallTupDataBagFieldSchema();
            ResourceFieldSchema tuplefs = GenRandomData.getSmallTupleFieldSchema();


            t.append(flds[0].compareTo("")!=0 ? ps.getLoadCaster().bytesToBag(flds[0].getBytes(), bagfs) : null);
            t.append(flds[1].compareTo("")!=0 ? new DataByteArray(flds[1].getBytes()) : null);
            t.append(flds[2].compareTo("")!=0 ? ps.getLoadCaster().bytesToCharArray(flds[2].getBytes()) : null);
            t.append(flds[3].compareTo("")!=0 ? ps.getLoadCaster().bytesToDouble(flds[3].getBytes()) : null);
            t.append(flds[4].compareTo("")!=0 ? ps.getLoadCaster().bytesToFloat(flds[4].getBytes()) : null);
            t.append(flds[5].compareTo("")!=0 ? ps.getLoadCaster().bytesToInteger(flds[5].getBytes()) : null);
            t.append(flds[6].compareTo("")!=0 ? ps.getLoadCaster().bytesToLong(flds[6].getBytes()) : null);
            t.append(flds[7].compareTo("")!=0 ? ps.getLoadCaster().bytesToMap(flds[7].getBytes()) : null);
            t.append(flds[8].compareTo("")!=0 ? ps.getLoadCaster().bytesToTuple(flds[8].getBytes(), tuplefs) : null);
            t.append(flds[9].compareTo("")!=0 ? ps.getLoadCaster().bytesToBoolean(flds[9].getBytes()) : null);
            t.append(flds[10].compareTo("")!=0 ? ps.getLoadCaster().bytesToDateTime(flds[10].getBytes()) : null);
            assertEquals(true, TestHelper.bagContains(inpDB, t));
            ++size;
        }
        assertEquals(true, size==inpDB.size());
    }

View Full Code Here

0 1 2 3 4 5 6 7 8

TOP

Related Classes of org.apache.pig.builtin.PigStorage

org.apache.commons.cli.CommandLine

org.apache.commons.cli.CommandLineParser

org.apache.commons.cli.GnuParser

org.apache.commons.cli.HelpFormatter

org.apache.commons.cli.Option

org.apache.commons.cli.Options

org.apache.hadoop.conf.Configuration

org.apache.hadoop.fs.FileSystem

org.apache.hadoop.fs.Path

org.apache.hadoop.io.Text

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.