public void testForeachPlan2() throws Exception {
LogicalPlanTester lpt = new LogicalPlanTester();
lpt.buildPlan("a = load '/test/d.txt' as (id, d:bag{t:(id:int, s)});");
lpt.buildPlan("b = foreach a generate id, FLATTEN(d);");
LogicalPlan plan = lpt.buildPlan("store b into '/test/empty';");
// check basics
org.apache.pig.experimental.logical.relational.LogicalPlan newPlan = migratePlan(plan);
org.apache.pig.experimental.logical.relational.LogicalPlan expected =
new org.apache.pig.experimental.logical.relational.LogicalPlan();
LogicalSchema aschema = new LogicalSchema();
aschema.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
LogicalSchema aschema2 = new LogicalSchema();
LogicalSchema aschema3 = new LogicalSchema();
aschema3.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.INTEGER));
aschema3.addField(new LogicalSchema.LogicalFieldSchema("s", null, DataType.BYTEARRAY));
aschema2.addField(new LogicalSchema.LogicalFieldSchema("t", aschema3, DataType.TUPLE));
aschema.addField(new LogicalSchema.LogicalFieldSchema("d", aschema2, DataType.BAG));
LOLoad load =
new LOLoad(
new FileSpec("file:///test/d.txt", new FuncSpec("org.apache.pig.builtin.PigStorage")),
aschema,
expected);
expected.add(load);
LOForEach foreach2 = new LOForEach(expected);
org.apache.pig.experimental.logical.relational.LogicalPlan innerPlan =
new org.apache.pig.experimental.logical.relational.LogicalPlan();
LOInnerLoad l1 = new LOInnerLoad(innerPlan, foreach2, 0);
innerPlan.add(l1);
LOInnerLoad l2 = new LOInnerLoad(innerPlan, foreach2, 1);
List<LogicalExpressionPlan> eps = new ArrayList<LogicalExpressionPlan>();
LogicalExpressionPlan p1 = new LogicalExpressionPlan();
new ProjectExpression(p1, DataType.BYTEARRAY, 0, 0);
LogicalExpressionPlan p2 = new LogicalExpressionPlan();
new ProjectExpression(p2, DataType.BAG, 1, 0);
eps.add(p1);
eps.add(p2);
LOGenerate gen = new LOGenerate(innerPlan, eps, new boolean[] {false, true});
innerPlan.add(gen);
innerPlan.connect(l1, gen);
innerPlan.connect(l2, gen);
foreach2.setInnerPlan(innerPlan);
expected.add(foreach2);
LOStore s =
new LOStore(
expected,
new FileSpec("file:///test/empty", new FuncSpec("org.apache.pig.builtin.PigStorage")));
expected.add(s);
expected.connect(load, foreach2);
expected.connect(foreach2, s);
try {
UidStamper stamper = new UidStamper(expected);
stamper.visit();
} catch (Exception e) {
throw new VisitorException(e);
}
assertTrue(expected.isEqual(newPlan));
LogicalSchema schema = foreach2.getSchema();
aschema = new LogicalSchema();
aschema.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
aschema.addField(new LogicalSchema.LogicalFieldSchema("d::id", null, DataType.INTEGER));
aschema.addField(new LogicalSchema.LogicalFieldSchema("s", null, DataType.BYTEARRAY));
assertTrue(schema.isEqual(aschema));
assertTrue(schema.getField("id") == schema.getField(0));
assertTrue(schema.getField("d::id") == schema.getField(1));
}
public void testForeachSchema() throws Exception {
// test flatten
LogicalPlanTester lpt = new LogicalPlanTester();
lpt.buildPlan("a = load '/test/d.txt' as (id, d:tuple(v, s));");
LogicalPlan plan = lpt.buildPlan("b = foreach a generate id, FLATTEN(d);");
org.apache.pig.experimental.logical.relational.LogicalPlan newPlan = migratePlan(plan);
LogicalRelationalOperator op = (LogicalRelationalOperator) newPlan.getSinks().get(0);
LogicalSchema s2 = new LogicalSchema();
s2.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
s2.addField(new LogicalSchema.LogicalFieldSchema("v", null, DataType.BYTEARRAY));
s2.addField(new LogicalSchema.LogicalFieldSchema("s", null, DataType.BYTEARRAY));
assertTrue(s2.isEqual(op.getSchema()));
// test no flatten
lpt = new LogicalPlanTester();
lpt.buildPlan("a = load '/test/d.txt' as (id, d:bag{t:(v, s)});");
plan = lpt.buildPlan("b = foreach a generate id, d;");
newPlan = migratePlan(plan);
op = (LogicalRelationalOperator) newPlan.getSinks().get(0);
LogicalSchema aschema = new LogicalSchema();
aschema.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
LogicalSchema aschema2 = new LogicalSchema();
LogicalSchema aschema3 = new LogicalSchema();
aschema3.addField(new LogicalSchema.LogicalFieldSchema("v", null, DataType.BYTEARRAY));
aschema3.addField(new LogicalSchema.LogicalFieldSchema("s", null, DataType.BYTEARRAY));
aschema2.addField(new LogicalSchema.LogicalFieldSchema("t", aschema3, DataType.TUPLE));
aschema.addField(new LogicalSchema.LogicalFieldSchema("d", aschema2, DataType.BAG));
assertTrue(aschema.isEqual(op.getSchema()));
// check with defined data type
lpt = new LogicalPlanTester();
lpt.buildPlan("a = load '/test/d.txt' as (id, d:bag{t:(v:int, s)});");
lpt.buildPlan("b = foreach a generate id, FLATTEN(d);");
plan = lpt.buildPlan("store b into '/test/empty';");
newPlan = migratePlan(plan);
op = (LogicalRelationalOperator) newPlan.getSinks().get(0);
op = (LogicalRelationalOperator) newPlan.getPredecessors(op).get(0);
LogicalSchema schema = op.getSchema();
aschema = new LogicalSchema();
aschema.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
aschema.addField(new LogicalSchema.LogicalFieldSchema("v", null, DataType.INTEGER));
aschema.addField(new LogicalSchema.LogicalFieldSchema("s", null, DataType.BYTEARRAY));
assertTrue(schema.isEqual(aschema));
// test with add
lpt = new LogicalPlanTester();
lpt.buildPlan("a = load '/test/d.txt' as (id, v:int, s:int);");
lpt.buildPlan("b = foreach a generate id, v+s;");
plan = lpt.buildPlan("store b into '/test/empty';");
newPlan = migratePlan(plan);
op = (LogicalRelationalOperator) newPlan.getSinks().get(0);
op = (LogicalRelationalOperator) newPlan.getPredecessors(op).get(0);
schema = op.getSchema();
aschema = new LogicalSchema();
aschema.addField(new LogicalSchema.LogicalFieldSchema("id", null, DataType.BYTEARRAY));
aschema.addField(new LogicalSchema.LogicalFieldSchema(null, null, DataType.INTEGER));
assertTrue(schema.isEqual(aschema));
}