private void run(PhysicalPlan pp, String expectedFile) throws Exception {
String compiledPlan, goldenPlan = null;
int MAX_SIZE = 100000;
MRCompiler comp = new MRCompiler(pp, pc);
comp.compile();
MROperPlan mrp = comp.getMRPlan();
PlanPrinter ppp = new PlanPrinter(mrp);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
ppp.print(baos);
compiledPlan = baos.toString();
if (generate) {
FileOutputStream fos = new FileOutputStream(expectedFile);
fos.write(baos.toByteArray());
return;
}
FileInputStream fis = new FileInputStream(expectedFile);
byte[] b = new byte[MAX_SIZE];
int len = fis.read(b);
goldenPlan = new String(b, 0, len);
if (goldenPlan.charAt(len - 1) == '\n') goldenPlan = goldenPlan.substring(0, len - 1);
pp.explain(System.out);
System.out.println();
System.out.println("<<<" + compiledPlan + ">>>");
System.out.println("-------------");
System.out.println("Golden");
System.out.println("<<<" + goldenPlan + ">>>");
System.out.println("-------------");
assertEquals(goldenPlan, compiledPlan);
}
@Test
public void testMergeJoin() throws Exception {
String query =
"a = load '/tmp/input1';"
+ "b = load '/tmp/input2';"
+ "c = join a by $0, b by $0 using 'merge';"
+ "store c into '/tmp/output1';";
PhysicalPlan pp = Util.buildPp(pigServer, query);
MRCompiler comp = new MRCompiler(pp, pc);
comp.compile();
MROperPlan mrp = comp.getMRPlan();
assertTrue(mrp.size() == 2);
MapReduceOper mrOp0 = mrp.getRoots().get(0);
assertTrue(mrOp0.mapPlan.size() == 2);
PhysicalOperator load0 = mrOp0.mapPlan.getRoots().get(0);
MergeJoinIndexer func =
(MergeJoinIndexer)
PigContext.instantiateFuncFromSpec(((POLoad) load0).getLFile().getFuncSpec());
Field lrField = MergeJoinIndexer.class.getDeclaredField("lr");
lrField.setAccessible(true);
POLocalRearrange lr = (POLocalRearrange) lrField.get(func);
List<PhysicalPlan> innerPlans = lr.getPlans();
PhysicalOperator localrearrange0 = mrOp0.mapPlan.getSuccessors(load0).get(0);
assertTrue(localrearrange0 instanceof POLocalRearrange);
assertTrue(mrOp0.reducePlan.size() == 3);
PhysicalOperator pack0 = mrOp0.reducePlan.getRoots().get(0);
assertTrue(pack0 instanceof POPackage);
PhysicalOperator foreach0 = mrOp0.reducePlan.getSuccessors(pack0).get(0);
assertTrue(foreach0 instanceof POForEach);
PhysicalOperator store0 = mrOp0.reducePlan.getSuccessors(foreach0).get(0);
assertTrue(store0 instanceof POStore);
assertTrue(innerPlans.size() == 1);
PhysicalPlan innerPlan = innerPlans.get(0);
assertTrue(innerPlan.size() == 1);
PhysicalOperator project = innerPlan.getRoots().get(0);
assertTrue(project instanceof POProject);
assertTrue(((POProject) project).getColumn() == 0);
MapReduceOper mrOp1 = mrp.getSuccessors(mrOp0).get(0);
assertTrue(mrOp1.mapPlan.size() == 3);
PhysicalOperator load1 = mrOp1.mapPlan.getRoots().get(0);
assertTrue(load1 instanceof POLoad);
PhysicalOperator mergejoin1 = mrOp1.mapPlan.getSuccessors(load1).get(0);
assertTrue(mergejoin1 instanceof POMergeJoin);
PhysicalOperator store1 = mrOp1.mapPlan.getSuccessors(mergejoin1).get(0);
assertTrue(store1 instanceof POStore);
assertTrue(mrOp1.reducePlan.isEmpty());
}