public void printData(HashMap<Object, List<Tuple>> data) throws ExecException {
for (Object o : data.keySet()) {
System.out.println(o);
for (Tuple t : data.get(o)) {
System.out.println("\t" + t.toDelimitedString(", "));
}
}
}
@Override
public DataBag exec(Tuple input) throws IOException {
retrieveContextValues();
ArrayList<String> joinKeyNames = new ArrayList<String>();
for (int i = 1; i < input.size(); i += 2) {
joinKeyNames.add((String) input.get(i));
}
JoinCollector collector = new JoinCollector();
// the first bag is the outer bag
String leftBagName = bagNames.get(0);
DataBag leftBag = getBag(input, leftBagName);
String leftBagJoinKeyName =
getPrefixedAliasName(bagNameToJoinKeyPrefix.get(leftBagName), joinKeyNames.get(0));
collector.setJoinData(collector.groupTuples(leftBag, leftBagJoinKeyName));
// now, for each additional bag, group up the tuples by the join key, then join them in
if (bagNames.size() > 1) {
for (int i = 1; i < bagNames.size(); i++) {
String bagName = bagNames.get(i);
DataBag bag = getBag(input, bagName);
String joinKeyName =
getPrefixedAliasName(bagNameToJoinKeyPrefix.get(bagName), joinKeyNames.get(i));
int tupleSize = bagNameToSize.get(bagName);
if (bag == null)
throw new IOException(
"Error in instance: "
+ getInstanceName()
+ " with properties: "
+ getInstanceProperties()
+ " and tuple: "
+ input.toDelimitedString(", ")
+ " -- Expected bag, got null");
HashMap<Object, List<Tuple>> groupedData = collector.groupTuples(bag, joinKeyName);
// outer join, so go back in and add nulls;
groupedData = collector.insertNullTuples(groupedData, tupleSize);
for (Map.Entry<Object, List<Tuple>> entry : groupedData.entrySet()) {
collector.joinTuples(entry.getKey(), entry.getValue());
}
}
}
// assemble output bag
DataBag outputBag = BagFactory.getInstance().newDefaultBag();
for (List<Tuple> tuples : collector.getJoinData().values()) {
for (Tuple tuple : tuples) {
outputBag.add(tuple);
}
}
return outputBag;
}
@Test
public void testProtoToPig() throws IOException {
AddressBook abProto = Fixtures.buildAddressBookProto();
Tuple abProtoTuple = tf_.newTuple(new DataByteArray(abProto.toByteArray()));
ProtobufBytesToTuple abProtoToPig =
new ProtobufBytesToTuple(AddressBook.class.getCanonicalName());
Tuple abTuple = abProtoToPig.exec(abProtoTuple);
assertEquals(
"{(Elephant Bird,123,[email protected],{(415-999-9999,HOME),(415-666-6666,MOBILE),(415-333-3333,WORK)}),(Elephant Bird,123,[email protected],{(415-999-9999,HOME),(415-666-6666,MOBILE),(415-333-3333,WORK)})},",
abTuple.toDelimitedString(","));
}