// See PIG-1434
@Test
public void testScalarAliasesFilterClause() throws Exception {
String[] input = {"1\t5", "2\t10", "3\t20", "4\t12", "5\t8"};
// Test the use of scalars in expressions
String inputPath = BUILD_TEST_TMP + "table_testScalarAliasesFilterClause";
TestScalarAliases.createLocalInputFile(inputPath, input);
// Test in script mode
pigServer.registerQuery("A = LOAD '" + inputPath + "' as (a0, a1);");
pigServer.registerQuery("G = group A all;");
pigServer.registerQuery("C = foreach G generate AVG(A.$1) as average;");
pigServer.registerQuery("Y = filter A by a1 > C.average;");
Iterator<Tuple> iter = pigServer.openIterator("Y");
// Average is 11
Tuple t = iter.next();
assertTrue(t.toString().equals("(3,20)"));
t = iter.next();
assertTrue(t.toString().equals("(4,12)"));
assertFalse(iter.hasNext());
}
// See PIG-1434
@Test
public void testScalarWithNoSchemaDollarProj() throws Exception {
String[] scalarInput = {"1\t5"};
String[] input = {"1\t5", "2\t10", "3\t20"};
String inputPath = BUILD_TEST_TMP + "table_testScalarWithNoSchemaDollarProj";
TestScalarAliases.createLocalInputFile(inputPath, input);
String inputPathScalar = BUILD_TEST_TMP + "table_testScalarWithNoSchemaDollarProjScalar";
TestScalarAliases.createLocalInputFile(inputPathScalar, scalarInput);
// Load A as a scalar
pigServer.registerQuery("A = LOAD '" + inputPath + "';");
pigServer.registerQuery("scalar = LOAD '" + inputPathScalar + "';");
pigServer.registerQuery("B = foreach A generate 5 / scalar.$1;");
Iterator<Tuple> iter = pigServer.openIterator("B");
Tuple t = iter.next();
assertTrue(t.get(0).toString().equals("1"));
t = iter.next();
assertTrue(t.get(0).toString().equals("1"));
t = iter.next();
assertTrue(t.get(0).toString().equals("1"));
assertFalse(iter.hasNext());
}
public void testRecursiveFileListing() throws IOException {
String LOCAL_INPUT_FILE = "test.dat";
String INPUT_FILE = "foo/bar/test.dat";
PrintWriter w = new PrintWriter(new FileWriter(LOCAL_INPUT_FILE));
w.println("1");
w.println("2");
w.println("3");
w.println("5");
w.close();
Util.copyFromLocalToCluster(cluster, LOCAL_INPUT_FILE, INPUT_FILE);
pigServer.registerQuery("a = load 'foo' as (nums:chararray);");
pigServer.registerQuery("b = load 'foo' as (nums:chararray);");
pigServer.registerQuery("d = join a by nums, b by nums USING 'skewed';");
Iterator<Tuple> iter = pigServer.openIterator("d");
int count = 0;
while (iter.hasNext()) {
iter.next();
count++;
}
Assert.assertEquals(4, count);
new File(LOCAL_INPUT_FILE).delete();
Util.deleteFile(cluster, INPUT_FILE);
}
/**
* Tests that '\n', '\r' and '\r\n' are treated as record delims when using bzip just like they
* are when using uncompressed text
*/
@Test
public void testRecordDelims() throws Exception {
String[] inputData =
new String[] {
"1\t2\r3\t4", // '\r' case - this will be split into two tuples
"5\t6\r", // '\r\n' case
"7\t8", // '\n' case
"9\t10\r" // '\r\n' at the end of file
};
// bzip compressed input
File in = File.createTempFile("junit", ".bz2");
String compressedInputFileName = in.getAbsolutePath();
in.deleteOnExit();
String clusterCompressedFilePath = Util.removeColon(compressedInputFileName);
String unCompressedInputFileName = "testRecordDelims-uncomp.txt";
Util.createInputFile(cluster, unCompressedInputFileName, inputData);
try {
CBZip2OutputStream cos = new CBZip2OutputStream(new FileOutputStream(in));
for (int i = 0; i < inputData.length; i++) {
StringBuffer sb = new StringBuffer();
sb.append(inputData[i]).append("\n");
byte bytes[] = sb.toString().getBytes();
cos.write(bytes);
}
cos.close();
Util.copyFromLocalToCluster(cluster, compressedInputFileName, clusterCompressedFilePath);
// pig script to read uncompressed input
String script = "a = load '" + unCompressedInputFileName + "';";
PigServer pig = new PigServer(ExecType.MAPREDUCE, cluster.getProperties());
pig.registerQuery(script);
Iterator<Tuple> it1 = pig.openIterator("a");
// pig script to read compressed input
script = "a = load '" + Util.encodeEscape(clusterCompressedFilePath) + "';";
pig.registerQuery(script);
Iterator<Tuple> it2 = pig.openIterator("a");
while (it1.hasNext()) {
Tuple t1 = it1.next();
Tuple t2 = it2.next();
Assert.assertEquals(t1, t2);
}
Assert.assertFalse(it2.hasNext());
} finally {
in.delete();
Util.deleteFile(cluster, unCompressedInputFileName);
Util.deleteFile(cluster, clusterCompressedFilePath);
}
}
public void testShouldWorkWithWKT() throws Exception {
ArrayList<String[]> data = new ArrayList<String[]>();
data.add(new String[] {"1", "1", "0", "POINT (0.0 0.0)"});
data.add(new String[] {"1", "2", "1", "POINT (0.0 3.0)"});
data.add(new String[] {"1", "3", "2", "POINT (4.0 5.0)"});
data.add(new String[] {"1", "4", "3", "POINT (10.0 0.0)"});
data.add(new String[] {"2", "5", "0", "POINT (5.0 6.0)"});
data.add(new String[] {"2", "6", "1", "POINT (10.0 3.0)"});
data.add(new String[] {"2", "7", "2", "POINT (7.0 13.0)"});
data.add(new String[] {"3", "1", "0", "POINT (0.0 0.0)"});
data.add(new String[] {"3", "8", "1", "POINT (10.0 10.0)"});
data.add(new String[] {"3", "9", "2", "POINT (18.0 5.0)"});
data.add(new String[] {"3", "1", "3", "POINT (0.0 0.0)"});
String datafile = TestHelper.createTempFile(data, "\t");
datafile = datafile.replace("\\", "\\\\");
PigServer pig = new PigServer(LOCAL);
String query =
"A = LOAD 'file:"
+ datafile
+ "' as (geom_id: int, point_id: int, point_pos: int, point);\n"
+ "B = ORDER A BY point_pos;"
+ "C = GROUP B BY geom_id;"
+ "D = FOREACH C GENERATE group, FLATTEN("
+ MakeSegments.class.getName()
+ "(B.point_id, B.point));";
pig.registerQuery(query);
Iterator<?> it = pig.openIterator("D");
ArrayList<String[]> expectedResults = new ArrayList<String[]>();
expectedResults.add(new String[] {"1", "0", "1", "0.0", "0.0", "2", "0.0", "3.0"});
expectedResults.add(new String[] {"1", "1", "2", "0.0", "3.0", "3", "4.0", "5.0"});
expectedResults.add(new String[] {"1", "2", "3", "4.0", "5.0", "4", "10.0", "0.0"});
expectedResults.add(new String[] {"2", "0", "5", "5.0", "6.0", "6", "10.0", "3.0"});
expectedResults.add(new String[] {"2", "1", "6", "10.0", "3.0", "7", "7.0", "13.0"});
expectedResults.add(new String[] {"3", "0", "1", "0.0", "0.0", "8", "10.0", "10.0"});
expectedResults.add(new String[] {"3", "1", "8", "10.0", "10.0", "9", "18.0", "5.0"});
expectedResults.add(new String[] {"3", "2", "9", "18.0", "5.0", "1", "0.0", "0.0"});
Iterator<String[]> expectedResultIter = expectedResults.iterator();
int count = 0;
while (it.hasNext() && expectedResultIter.hasNext()) {
Tuple tuple = (Tuple) it.next();
String[] expectedResult = expectedResultIter.next();
if (tuple == null) break;
assertEquals(Integer.parseInt(expectedResult[0]), tuple.get(0));
assertEquals(Integer.parseInt(expectedResult[1]), tuple.get(1));
assertEquals(Long.parseLong(expectedResult[2]), tuple.get(2));
assertEquals(Double.parseDouble(expectedResult[3]), tuple.get(3));
assertEquals(Double.parseDouble(expectedResult[4]), tuple.get(4));
assertEquals(Long.parseLong(expectedResult[5]), tuple.get(5));
assertEquals(Double.parseDouble(expectedResult[6]), tuple.get(6));
assertEquals(Double.parseDouble(expectedResult[7]), tuple.get(7));
count++;
}
assertEquals(expectedResults.size(), count);
}
@Test
public void testErrorMessageUndefinedAliasInGroupByStatement() throws Exception {
String queryA = "A = load 'nosuchfile' using PigStorage() as (f1:chararray,f2:chararray);";
String queryB = "B = GROUP B by f1;";
PigServer ps = new PigServer(ExecType.LOCAL);
ps.registerQuery(queryA);
try {
ps.registerQuery(queryB);
} catch (FrontendException e) {
Assert.assertTrue(e.getMessage().contains("Undefined alias:"));
return;
}
Assert.fail();
}
@Test
public void readWriteNullValuesRead() throws IOException {
registerLoadQuery();
tempFilename = tempFilename + "-2";
pigServer.registerQuery(
String.format(
"STORE A INTO 'file:%s' USING %s('-c %s', '-c %s');",
tempFilename,
SequenceFileStorage.class.getName(),
IntWritableConverter.class.getName(),
NullWritableConverter.class.getName()));
registerLoadQuery(IntWritableConverter.class, NullWritableConverter.class, null);
validateIndex(pigServer.openIterator("A"), 2, 0, 0);
}
@Test
public void testBzipStoreInMultiQuery2() throws Exception {
String[] inputData = new String[] {"1\t2\r3\t4"};
String inputFileName = "input2.txt";
Util.createInputFile(cluster, inputFileName, inputData);
PigServer pig = new PigServer(ExecType.MAPREDUCE, cluster.getProperties());
PigContext pigContext = pig.getPigContext();
pigContext.getProperties().setProperty("output.compression.enabled", "true");
pigContext
.getProperties()
.setProperty("output.compression.codec", "org.apache.hadoop.io.compress.BZip2Codec");
pig.setBatchOn();
pig.registerQuery("a = load '" + inputFileName + "';");
pig.registerQuery("store a into 'output2.bz2';");
pig.registerQuery("store a into 'output2';");
pig.executeBatch();
FileSystem fs =
FileSystem.get(ConfigurationUtil.toConfiguration(pig.getPigContext().getProperties()));
FileStatus stat = fs.getFileStatus(new Path("output2/part-m-00000.bz2"));
assertTrue(stat.getLen() > 0);
stat = fs.getFileStatus(new Path("output2.bz2/part-m-00000.bz2"));
assertTrue(stat.getLen() > 0);
}
@Test(expected = IOException.class)
public void writeUnsupportedConversion() throws IOException {
registerLoadQuery();
// swap ordering of key and value
pigServer.registerQuery("A = FOREACH A GENERATE TOTUPLE(key), value;");
// the following should die because IntWritableConverter doesn't support conversion of Tuple to
// IntWritable
pigServer.registerQuery(
String.format(
"STORE A INTO 'file:%s-2' USING %s('-c %s', '-c %s');",
tempFilename,
SequenceFileStorage.class.getName(),
IntWritableConverter.class.getName(),
TextConverter.class.getName()));
}
@Test(expected = Exception.class)
public void readByteArraysWriteByteArraysWithoutTypeRead() throws IOException {
registerLoadQuery(
GenericWritableConverter.class, TextConverter.class, "key:bytearray, value:bytearray");
tempFilename = tempFilename + "-2";
pigServer.registerQuery(
String.format(
"STORE A INTO 'file:%s' USING %s('-c %s', '-c %s');",
tempFilename,
SequenceFileStorage.class.getName(),
GenericWritableConverter.class.getName(),
TextConverter.class.getName()));
registerLoadQuery();
validate(pigServer.openIterator("A"));
}
@Test
public void writeTextConversion() throws IOException {
registerLoadQuery();
tempFilename = tempFilename + "-2";
// rely on TextConverter for conversion of int to Text
pigServer.registerQuery(
String.format(
"STORE A INTO 'file:%s' USING %s('-c %s', '-c %s');",
tempFilename,
SequenceFileStorage.class.getName(),
TextConverter.class.getName(),
TextConverter.class.getName()));
registerLoadQuery(TextConverter.class, TextConverter.class, "key:chararray, value:chararray");
validate(pigServer.openIterator("A"));
}
private void testCount(
String inputFileName, Long expectedCount, int splitSize, String loadFuncSpec)
throws IOException {
String outputFile = "/tmp/bz-output";
// simple load-store script to verify that the bzip input is getting
// split
String scriptToTestSplitting =
"a = load '"
+ inputFileName
+ "' using "
+ loadFuncSpec
+ "; store a into '"
+ outputFile
+ "';";
String script =
"a = load '"
+ inputFileName
+ "';"
+ "b = group a all;"
+ "c = foreach b generate COUNT_STAR(a);";
Properties props = new Properties();
for (Entry<Object, Object> entry : cluster.getProperties().entrySet()) {
props.put(entry.getKey(), entry.getValue());
}
props.setProperty("mapred.max.split.size", Integer.toString(splitSize));
PigContext pigContext = new PigContext(ExecType.MAPREDUCE, props);
PigServer pig = new PigServer(pigContext);
FileSystem fs = FileSystem.get(ConfigurationUtil.toConfiguration(props));
fs.delete(new Path(outputFile), true);
Util.registerMultiLineQuery(pig, scriptToTestSplitting);
// verify that > 1 maps were launched due to splitting of the bzip input
FileStatus[] files = fs.listStatus(new Path(outputFile));
int numPartFiles = 0;
for (FileStatus fileStatus : files) {
if (fileStatus.getPath().getName().startsWith("part")) {
numPartFiles++;
}
}
assertEquals(true, numPartFiles > 0);
// verify record count to verify we read bzip data correctly
Util.registerMultiLineQuery(pig, script);
Iterator<Tuple> it = pig.openIterator("c");
Long result = (Long) it.next().get(0);
assertEquals(expectedCount, result);
}
// See PIG-1434
@Test
public void testScalarAliasesBatchNobatch() throws Exception {
String[] input = {"1\t5", "2\t10", "3\t20"};
String output = BUILD_TEST_TMP + "table_testScalarAliasesDir";
TestScalarAliases.deleteDirectory(new File(output));
// Test the use of scalars in expressions
String inputPath = BUILD_TEST_TMP + "table_testScalarAliasesBatch";
TestScalarAliases.createLocalInputFile(inputPath, input);
// Test in script mode
pigServer.setBatchOn();
pigServer.registerQuery("A = LOAD '" + inputPath + "' as (a0: long, a1: double);");
pigServer.registerQuery("B = group A all;");
pigServer.registerQuery("C = foreach B generate COUNT(A) as count, MAX(A.$1) as max;");
pigServer.registerQuery("Y = foreach A generate (a0 * C.count), (a1 / C.max);");
pigServer.registerQuery("Store Y into '" + output + "';");
pigServer.executeBatch();
// Check output
pigServer.registerQuery("Z = LOAD '" + output + "' as (a0: int, a1: double);");
Iterator<Tuple> iter;
Tuple t;
iter = pigServer.openIterator("Z");
t = iter.next();
assertTrue(t.toString().equals("(3,0.25)"));
t = iter.next();
assertTrue(t.toString().equals("(6,0.5)"));
t = iter.next();
assertTrue(t.toString().equals("(9,1.0)"));
assertFalse(iter.hasNext());
iter = pigServer.openIterator("Y");
t = iter.next();
assertTrue(t.toString().equals("(3,0.25)"));
t = iter.next();
assertTrue(t.toString().equals("(6,0.5)"));
t = iter.next();
assertTrue(t.toString().equals("(9,1.0)"));
assertFalse(iter.hasNext());
}
public void testSkewedJoinReducers() throws IOException {
pigServer.registerQuery("A = LOAD '" + INPUT_FILE1 + "' as (id, name, n);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE2 + "' as (id, name);");
try {
DataBag dbfrj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("C = join A by id, B by id using \"skewed\" parallel 1;");
Iterator<Tuple> iter = pigServer.openIterator("C");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
} catch (Exception e) {
fail("Should not throw exception, should continue execution");
}
}
@Test(expected = Exception.class)
public void readWithMissingWritableConverterArguments() throws IOException {
registerLoadQuery(
FixedArgsConstructorIntWritableConverter.class,
TextConverter.class,
"key: int, value: chararray");
validate(pigServer.openIterator("A"));
}
public void testSkewedJoinWithGroup() throws IOException {
pigServer.registerQuery("A = LOAD '" + INPUT_FILE1 + "' as (id, name, n);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE2 + "' as (id, name);");
pigServer.registerQuery("C = GROUP A by id;");
pigServer.registerQuery("D = GROUP B by id;");
DataBag dbfrj = BagFactory.getInstance().newDefaultBag(),
dbshj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("E = join C by group, D by group using \"skewed\" parallel 5;");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
{
pigServer.registerQuery("E = join C by group, D by group;");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbshj.add(iter.next());
}
}
Assert.assertTrue(dbfrj.size() > 0 && dbshj.size() > 0);
Assert.assertEquals(true, TestHelper.compareBags(dbfrj, dbshj));
}
// See PIG-1434
@Test
public void testScalarAliasesJoinClause() throws Exception {
String[] inputA = {"1\t5", "2\t10", "3\t20"};
String[] inputB = {"Total3\tthree", "Total2\ttwo", "Total1\tone"};
// Test the use of scalars in expressions
String inputPathA = BUILD_TEST_TMP + "table_testScalarAliasesJoinClauseA";
TestScalarAliases.createLocalInputFile(inputPathA, inputA);
String inputPathB = BUILD_TEST_TMP + "table_testScalarAliasesJoinClauseB";
TestScalarAliases.createLocalInputFile(inputPathB, inputB);
// Test in script mode
pigServer.registerQuery("A = LOAD '" + inputPathA + "' as (a0, a1);");
pigServer.registerQuery("G = group A all;");
pigServer.registerQuery("C = foreach G generate COUNT(A) as count;");
pigServer.registerQuery("B = LOAD '" + inputPathB + "' as (b0:chararray, b1:chararray);");
pigServer.registerQuery("Y = join A by CONCAT('Total', (chararray)C.count), B by $0;");
Iterator<Tuple> iter = pigServer.openIterator("Y");
String[] expected =
new String[] {"(1,5,Total3,three)", "(2,10,Total3,three)", "(3,20,Total3,three)"};
Util.checkQueryOutputsAfterSortRecursive(
iter,
expected,
org.apache.pig.newplan.logical.Util.translateSchema(pigServer.dumpSchema("Y")));
}
@Test
public void testSplitWithNotEvalCondition() throws Exception {
String defineQ =
"define minelogs org.apache.pig.test.RegexGroupCount('www\\\\.xyz\\\\.com/sports');";
String defineL = "a = load 'nosuchfile' " + " using PigStorage() as (source : chararray);";
String defineSplit =
"SPLIT a INTO a1 IF (minelogs(source) > 0 ), a2 IF (NOT (minelogs(source)>0));"; // (NOT
// (
// minelogs(source) ) > 0) ;";
PigServer ps = new PigServer(ExecType.LOCAL);
ps.registerQuery(defineQ);
ps.registerQuery(defineL);
try {
ps.registerQuery(defineSplit);
} catch (FrontendException e) {
Assert.fail(e.getMessage());
}
}
/** Tests the end-to-end writing and reading of a BZip file. */
@Test
public void testBzipInPig() throws Exception {
PigServer pig = new PigServer(ExecType.MAPREDUCE, cluster.getProperties());
File in = File.createTempFile("junit", ".bz2");
in.deleteOnExit();
File out = File.createTempFile("junit", ".bz2");
out.delete();
String clusterOutput = Util.removeColon(out.getAbsolutePath());
CBZip2OutputStream cos = new CBZip2OutputStream(new FileOutputStream(in));
for (int i = 1; i < 100; i++) {
StringBuffer sb = new StringBuffer();
sb.append(i).append("\n").append(-i).append("\n");
byte bytes[] = sb.toString().getBytes();
cos.write(bytes);
}
cos.close();
pig.registerQuery(
"AA = load '"
+ Util.generateURI(Util.encodeEscape(in.getAbsolutePath()), pig.getPigContext())
+ "';");
pig.registerQuery("A = foreach (group (filter AA by $0 > 0) all) generate flatten($1);");
pig.registerQuery("store A into '" + Util.encodeEscape(clusterOutput) + "';");
FileSystem fs =
FileSystem.get(ConfigurationUtil.toConfiguration(pig.getPigContext().getProperties()));
FSDataInputStream is = fs.open(new Path(clusterOutput + "/part-r-00000.bz2"));
CBZip2InputStream cis = new CBZip2InputStream(is, -1, out.length());
// Just a sanity check, to make sure it was a bzip file; we
// will do the value verification later
assertEquals(100, cis.read(new byte[100]));
cis.close();
pig.registerQuery("B = load '" + Util.encodeEscape(clusterOutput) + "';");
Iterator<Tuple> i = pig.openIterator("B");
HashMap<Integer, Integer> map = new HashMap<Integer, Integer>();
while (i.hasNext()) {
Integer val = DataType.toInteger(i.next().get(0));
map.put(val, val);
}
assertEquals(new Integer(99), new Integer(map.keySet().size()));
for (int j = 1; j < 100; j++) {
assertEquals(new Integer(j), map.get(j));
}
in.delete();
Util.deleteFile(cluster, clusterOutput);
}
@Test
public void readWriteUnexpectedNullValuesRead() throws IOException {
registerLoadQuery();
tempFilename = tempFilename + "-2";
// swap last value with null; this pair should not be stored
pigServer.registerQuery(
String.format("A = FOREACH A GENERATE key, (key == 2 ? null : value) AS value;"));
pigServer.registerQuery(
String.format(
"STORE A INTO 'file:%s' USING %s('-c %s', '-c %s');",
tempFilename,
SequenceFileStorage.class.getName(),
IntWritableConverter.class.getName(),
TextConverter.class.getName()));
registerLoadQuery();
// validation against expected pairs will succeed, with expected number of pairs one less than
// usual (the last pair wasn't stored due to null value)
validate(pigServer.openIterator("A"), DATA.length - 1);
}
@Test
public void readWithoutSchemaTestSchema() throws IOException {
registerLoadQuery(IntWritableConverter.class, TextConverter.class, null);
Schema schema = pigServer.dumpSchema("A");
Assert.assertNotNull(schema);
Assert.assertEquals("key", schema.getField(0).alias);
Assert.assertEquals(DataType.INTEGER, schema.getField(0).type);
Assert.assertEquals("value", schema.getField(1).alias);
Assert.assertEquals(DataType.CHARARRAY, schema.getField(1).type);
}
public void testSchemaSerialization() throws IOException {
MiniCluster cluster = MiniCluster.buildCluster();
PigServer pigServer = new PigServer(ExecType.MAPREDUCE, cluster.getProperties());
String inputFileName = "testSchemaSerialization-input.txt";
String[] inputData = new String[] {"foo\t1", "hello\t2"};
Util.createInputFile(cluster, inputFileName, inputData);
String script =
"a = load '"
+ inputFileName
+ "' as (f1:chararray, f2:int);"
+ " b = group a all; c = foreach b generate org.apache.pig.test.InputSchemaUDF(a);";
Util.registerMultiLineQuery(pigServer, script);
Iterator<Tuple> it = pigServer.openIterator("c");
while (it.hasNext()) {
Tuple t = it.next();
Assert.assertEquals("{a: {(f1: chararray,f2: int)}}", t.get(0));
}
cluster.shutDown();
}
public void testSkewedJoinKeyPartition() throws IOException {
try {
Util.deleteFile(cluster, "skewedjoin");
} catch (Exception e) {
// it is ok if directory not exist
}
pigServer.registerQuery("A = LOAD '" + INPUT_FILE1 + "' as (id, name, n);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE2 + "' as (id, name);");
pigServer.registerQuery("E = join A by id, B by id using \"skewed\" parallel 7;");
pigServer.store("E", "skewedjoin");
int[][] lineCount = new int[3][7];
new File("skewedjoin").mkdir();
// check how many times a key appear in each part- file
for (int i = 0; i < 7; i++) {
Util.copyFromClusterToLocal(
cluster, "skewedjoin/part-r-0000" + i, "skewedjoin/part-r-0000" + i);
BufferedReader reader = new BufferedReader(new FileReader("skewedjoin/part-r-0000" + i));
String line = null;
while ((line = reader.readLine()) != null) {
String[] cols = line.split("\t");
int key = Integer.parseInt(cols[0]) / 100 - 1;
lineCount[key][i]++;
}
}
int fc = 0;
for (int i = 0; i < 3; i++) {
for (int j = 0; j < 7; j++) {
if (lineCount[i][j] > 0) {
fc++;
}
}
}
// atleast one key should be a skewed key
// check atleast one key should appear in more than 1 part- file
assertTrue(fc > 3);
}
// See PIG-1434
@Test
public void testScalarAliasesGrammarNegative() throws Exception {
String[] input = {"1\t5", "2\t10", "3\t20"};
String inputPath = BUILD_TEST_TMP + "table_testScalarAliasesGrammar";
TestScalarAliases.createLocalInputFile(inputPath, input);
try {
pigServer.registerQuery("A = LOAD '" + inputPath + "' as (a0: long, a1: double);");
pigServer.registerQuery("B = group A all;");
pigServer.registerQuery("C = foreach B generate COUNT(A);");
// Only projections of C are supported
pigServer.registerQuery("Y = foreach A generate C;");
pigServer.openIterator("Y");
// Control should not reach here
fail("Scalar projections are only supported");
} catch (IOException pe) {
assertTrue(pe.getMessage().contains("Invalid scalar projection: C"));
}
}
public void testSkewedJoin3Way() throws IOException {
pigServer.registerQuery("A = LOAD '" + INPUT_FILE1 + "' as (id, name, n);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE2 + "' as (id, name);");
pigServer.registerQuery("C = LOAD '" + INPUT_FILE3 + "' as (id, name);");
try {
DataBag dbfrj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("D = join A by id, B by id, C by id using \"skewed\" parallel 5;");
Iterator<Tuple> iter = pigServer.openIterator("D");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
} catch (Exception e) {
return;
}
fail("Should throw exception, do not support 3 way join");
}
public void testSkewedJoinNullKeys() throws IOException {
pigServer.registerQuery("A = LOAD '" + INPUT_FILE5 + "' as (id,name);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE5 + "' as (id,name);");
try {
DataBag dbfrj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("C = join A by id, B by id using \"skewed\";");
Iterator<Tuple> iter = pigServer.openIterator("C");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
} catch (Exception e) {
System.out.println(e.getMessage());
e.printStackTrace();
fail("Should support null keys in skewed join");
}
return;
}
public void testSkewedJoinManyReducers() throws IOException {
pigServer.getPigContext().getProperties().setProperty("pig.skewedjoin.reduce.maxtuple", "2");
pigServer.registerQuery("A = LOAD '" + INPUT_FILE6 + "' as (id,name);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE7 + "' as (id,name);");
DataBag dbfrj = BagFactory.getInstance().newDefaultBag(),
dbrj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("E = join A by id, B by id using \"skewed\" parallel 300;");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
{
pigServer.registerQuery("E = join A by id, B by id;");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbrj.add(iter.next());
}
}
Assert.assertEquals(dbfrj.size(), dbrj.size());
Assert.assertEquals(true, TestHelper.compareBags(dbfrj, dbrj));
}
// See PIG-1636
@Test
public void testScalarAliasesLimit() throws Exception {
String[] input = {"a\t1", "b\t2", "c\t3", "a\t4", "c\t5"};
// Test the use of scalars in expressions
String inputPath = BUILD_TEST_TMP + "table_testScalarAliasesLimit";
TestScalarAliases.createLocalInputFile(inputPath, input);
// Test in script mode
pigServer.registerQuery("A = LOAD '" + inputPath + "' as (a0:chararray, a1: int);");
pigServer.registerQuery("G = group A all;");
pigServer.registerQuery("C = foreach G generate SUM(A.$1) as total;");
pigServer.registerQuery("C1 = limit C 1;");
pigServer.registerQuery("Y = foreach A generate a0, a1 * (double)C1.total;");
Iterator<Tuple> iter = pigServer.openIterator("Y");
// Average is 11
Tuple t = iter.next();
assertTrue(t.toString().equals("(a,15.0)"));
t = iter.next();
assertTrue(t.toString().equals("(b,30.0)"));
t = iter.next();
assertTrue(t.toString().equals("(c,45.0)"));
t = iter.next();
assertTrue(t.toString().equals("(a,60.0)"));
t = iter.next();
assertTrue(t.toString().equals("(c,75.0)"));
assertFalse(iter.hasNext());
}
public void testSkewedJoinWithNoProperties() throws IOException {
pigServer = new PigServer(ExecType.MAPREDUCE, cluster.getProperties());
pigServer.registerQuery("A = LOAD '" + INPUT_FILE1 + "' as (id, name, n);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE2 + "' as (id, name);");
try {
DataBag dbfrj = BagFactory.getInstance().newDefaultBag();
DataBag dbshj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery(
"C = join A by (id, name), B by (id, name) using \"skewed\" parallel 5;");
Iterator<Tuple> iter = pigServer.openIterator("C");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
{
pigServer.registerQuery("E = join A by(id, name), B by (id, name);");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbshj.add(iter.next());
}
}
Assert.assertTrue(dbfrj.size() > 0 && dbshj.size() > 0);
Assert.assertEquals(true, TestHelper.compareBags(dbfrj, dbshj));
} catch (Exception e) {
fail(e.getMessage());
}
}
// pig 1048
public void testSkewedJoinOneValue() throws IOException {
pigServer.registerQuery("A = LOAD '" + INPUT_FILE3 + "' as (id,name);");
pigServer.registerQuery("B = LOAD '" + INPUT_FILE3 + "' as (id,name);");
// Filter key with a single value
pigServer.registerQuery("C = FILTER A by id == 400;");
pigServer.registerQuery("D = FILTER B by id == 400;");
DataBag dbfrj = BagFactory.getInstance().newDefaultBag(),
dbrj = BagFactory.getInstance().newDefaultBag();
{
pigServer.registerQuery("E = join C by id, D by id using \"skewed\";");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbfrj.add(iter.next());
}
}
{
pigServer.registerQuery("E = join C by id, D by id;");
Iterator<Tuple> iter = pigServer.openIterator("E");
while (iter.hasNext()) {
dbrj.add(iter.next());
}
}
Assert.assertEquals(dbfrj.size(), dbrj.size());
Assert.assertEquals(true, TestHelper.compareBags(dbfrj, dbrj));
}