// 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-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());
}
@Override
public Block next() throws IOException, InterruptedException {
Tuple metaDataTuple = matchingMetaBlock.next();
if (metaDataTuple == null) return null; // Done
System.out.println("Collate Vector: metadata tuple = " + metaDataTuple.toString());
return generateVectorBlock(metaDataTuple);
}
// See PIG-1434
@Test
public void testFilteredScalarDollarProj() throws Exception {
String output = BUILD_TEST_TMP + "table_testFilteredScalarDollarProjDir";
TestScalarAliases.deleteDirectory(new File(output));
String[] input = {
"1\t5\t[state#maine,city#portland]\t{(a),(b)}\t(a,b)", "2\t10\t\t\t", "3\t20\t\t\t"
};
// Test the use of scalars in expressions
String inputPath = BUILD_TEST_TMP + "table_testFilteredScalarDollarProj";
TestScalarAliases.createLocalInputFile(inputPath, input);
// Test in script mode
pigServer.setBatchOn();
pigServer.registerQuery(
"A = LOAD '"
+ inputPath
+ "'"
+ " as (a0: long, a1: double, a2 : bytearray, "
+ "a3: bag{ t : tuple(tc : chararray)}, "
+ "a4: tuple(c1 : chararray, c2 : chararray) );");
pigServer.registerQuery("B = filter A by $1 < 8;");
pigServer.registerQuery(
"Y = foreach A generate (a0 * B.$0), (a1 / B.$1), B.$2, B.$2#'state', B.$3, B.a4;");
pigServer.registerQuery("Store Y into '" + output + "';");
pigServer.explain("Y", System.err);
pigServer.executeBatch();
// Check output
pigServer.registerQuery("Z = LOAD '" + output + "' as (a0: int, a1: double);");
pigServer.explain("Z", System.err);
Iterator<Tuple> iter = pigServer.openIterator("Z");
Tuple t = iter.next();
assertTrue(t.toString().equals("(1,1.0)"));
t = iter.next();
assertTrue(t.toString().equals("(2,2.0)"));
t = iter.next();
assertTrue(t.toString().equals("(3,4.0)"));
assertFalse(iter.hasNext());
// Check in non-batch mode
iter = pigServer.openIterator("Y");
t = iter.next();
assertEquals(t.toString(), "(1,1.0,[state#maine,city#portland],maine,{(a),(b)},(a,b))");
t = iter.next();
assertEquals(t.toString(), "(2,2.0,[state#maine,city#portland],maine,{(a),(b)},(a,b))");
t = iter.next();
assertEquals(t.toString(), "(3,4.0,[state#maine,city#portland],maine,{(a),(b)},(a,b))");
assertFalse(iter.hasNext());
}
@Test
public void test() throws IOException, URISyntaxException {
// Prepare Resource File
URL metricRes = EndpointGroupsTest.class.getResource("/avro/poem_sync_v2.avro");
File metricAvro = new File(metricRes.toURI());
UnwindServiceMetrics uw = new UnwindServiceMetrics("", "test");
uw.mpsMgr.loadAvro(metricAvro);
TupleFactory tf = TupleFactory.getInstance();
Tuple inTuple = tf.newTuple();
inTuple.append("SRMv2");
inTuple.append("se01.afroditi.hellasgrid.gr");
String jsonStr =
IOUtils.toString(this.getClass().getResourceAsStream("/ar/missing_endpoint.json"), "UTF-8");
Tuple expTuple = JsonToPig.jsonToTuple(jsonStr);
Tuple outTuple = uw.exec(inTuple);
assertTrue(expTuple.toString().equals(outTuple.toString()));
}
@Test
public void testExecNestedTuple() throws IOException {
Tuple input = tupleFactory.newTuple();
input.append("{\"stacks\":[[[4,3],[2,1]], [[1,2],[3,4]]]}");
Map<String, Object> myMap = jsonMap.exec(input);
Tuple stacks = (Tuple) myMap.get("stacks");
System.out.println(stacks);
Tuple reference = getTestTuple();
assertEquals(reference.toString(), stacks.toString());
assertEquals(reference.size(), stacks.size());
for (int i = 0; i < reference.size(); i++) {
Tuple r = (Tuple) reference.get(i);
Tuple s = (Tuple) stacks.get(i);
assertEquals(r.size(), s.size());
for (int j = 0; j < r.size(); j++) {
System.out.println("Checking if " + r.get(j) + " == " + s.get(j));
assertEquals(r.get(j), s.get(j));
}
}
}
// See PIG-1434
@Test
public void testScalarWithTwoBranches() throws Exception {
String[] inputA = {"1\t5", "2\t10", "3\t20"};
String[] inputX = {"pig", "hadoop", "rocks"};
String output = BUILD_TEST_TMP + "testScalarWithTwoBranchesDir";
TestScalarAliases.deleteDirectory(new File(output));
// Test the use of scalars in expressions
String inputPathA = BUILD_TEST_TMP + "testScalarWithTwoBranchesA";
TestScalarAliases.createLocalInputFile(inputPathA, inputA);
String inputPathX = BUILD_TEST_TMP + "testScalarWithTwoBranchesX";
TestScalarAliases.createLocalInputFile(inputPathX, inputX);
// Test in script mode
pigServer.setBatchOn();
pigServer.registerQuery("A = LOAD '" + inputPathA + "' 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("X = LOAD '" + inputPathX + "' as (names: chararray);");
pigServer.registerQuery("Y = foreach X generate names, C.max;");
pigServer.registerQuery("Store Y into '" + output + "';");
pigServer.executeBatch();
// Check output
pigServer.registerQuery("Z = LOAD '" + output + "' as (a0: chararray, a1: double);");
Iterator<Tuple> iter = pigServer.openIterator("Z");
Tuple t = iter.next();
assertTrue(t.toString().equals("(pig,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(hadoop,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(rocks,20.0)"));
assertFalse(iter.hasNext());
// Check in non-batch mode
iter = pigServer.openIterator("Y");
t = iter.next();
assertTrue(t.toString().equals("(pig,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(hadoop,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(rocks,20.0)"));
assertFalse(iter.hasNext());
pigServer.getPigContext().getProperties().remove("tez.am.inline.task.execution.max-tasks");
}
// 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());
}
// See PIG-1434
@Test
public void testUseScalarMultipleTimes() throws Exception {
String[] input = {"1\t5", "2\t10", "3\t20"};
String outputY = BUILD_TEST_TMP + "table_testUseScalarMultipleTimesOutY";
TestScalarAliases.deleteDirectory(new File(outputY));
String outputZ = BUILD_TEST_TMP + "table_testUseScalarMultipleTimesOutZ";
TestScalarAliases.deleteDirectory(new File(outputZ));
// Test the use of scalars in expressions
String inputPath = BUILD_TEST_TMP + "table_testUseScalarMultipleTimes";
TestScalarAliases.createLocalInputFile(inputPath, input);
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 '" + outputY + "';");
pigServer.registerQuery("Z = foreach A generate (a1 + C.count), (a0 * C.max);");
pigServer.registerQuery("Store Z into '" + outputZ + "';");
// Test Multiquery store
pigServer.executeBatch();
// Check output
pigServer.registerQuery("M = LOAD '" + outputY + "' as (a0: int, a1: double);");
Iterator<Tuple> iter;
Tuple t;
iter = pigServer.openIterator("M");
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());
// Check output
pigServer.registerQuery("N = LOAD '" + outputZ + "' as (a0: double, a1: double);");
iter = pigServer.openIterator("N");
t = iter.next();
assertTrue(t.toString().equals("(8.0,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(13.0,40.0)"));
t = iter.next();
assertTrue(t.toString().equals("(23.0,60.0)"));
assertFalse(iter.hasNext());
// Non batch mode
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());
// Check in non-batch mode
iter = pigServer.openIterator("Z");
t = iter.next();
assertTrue(t.toString().equals("(8.0,20.0)"));
t = iter.next();
assertTrue(t.toString().equals("(13.0,40.0)"));
t = iter.next();
assertTrue(t.toString().equals("(23.0,60.0)"));
assertFalse(iter.hasNext());
}
@Override
public String toString() {
return t.toString();
}
@Override
public byte[] toBytes(Tuple t) throws IOException {
return t.toString().getBytes();
}