private static void pruneReduceSinkOperator(
boolean[] retainFlags, ReduceSinkOperator reduce, ColumnPrunerProcCtx cppCtx)
throws SemanticException {
ReduceSinkDesc reduceConf = reduce.getConf();
Map<String, ExprNodeDesc> oldMap = reduce.getColumnExprMap();
LOG.info("RS " + reduce.getIdentifier() + " oldColExprMap: " + oldMap);
RowResolver oldRR = cppCtx.getOpToParseCtxMap().get(reduce).getRowResolver();
ArrayList<ColumnInfo> old_signature = oldRR.getRowSchema().getSignature();
ArrayList<ColumnInfo> signature = new ArrayList<ColumnInfo>(old_signature);
List<String> valueColNames = reduceConf.getOutputValueColumnNames();
ArrayList<String> newValueColNames = new ArrayList<String>();
List<ExprNodeDesc> keyExprs = reduceConf.getKeyCols();
List<ExprNodeDesc> valueExprs = reduceConf.getValueCols();
ArrayList<ExprNodeDesc> newValueExprs = new ArrayList<ExprNodeDesc>();
for (int i = 0; i < retainFlags.length; i++) {
String outputCol = valueColNames.get(i);
ExprNodeDesc outputColExpr = valueExprs.get(i);
if (!retainFlags[i]) {
String[] nm = oldRR.reverseLookup(outputCol);
if (nm == null) {
outputCol = Utilities.ReduceField.VALUE.toString() + "." + outputCol;
nm = oldRR.reverseLookup(outputCol);
}
// In case there are multiple columns referenced to the same column name, we won't
// do row resolve once more because the ColumnInfo in row resolver is already removed
if (nm == null) {
continue;
}
// Only remove information of a column if it is not a key,
// i.e. this column is not appearing in keyExprs of the RS
if (ExprNodeDescUtils.indexOf(outputColExpr, keyExprs) == -1) {
ColumnInfo colInfo = oldRR.getFieldMap(nm[0]).remove(nm[1]);
oldRR.getInvRslvMap().remove(colInfo.getInternalName());
oldMap.remove(outputCol);
signature.remove(colInfo);
}
} else {
newValueColNames.add(outputCol);
newValueExprs.add(outputColExpr);
}
}
oldRR.getRowSchema().setSignature(signature);
reduce.getSchema().setSignature(signature);
reduceConf.setOutputValueColumnNames(newValueColNames);
reduceConf.setValueCols(newValueExprs);
TableDesc newValueTable =
PlanUtils.getReduceValueTableDesc(
PlanUtils.getFieldSchemasFromColumnList(
reduceConf.getValueCols(), newValueColNames, 0, ""));
reduceConf.setValueSerializeInfo(newValueTable);
LOG.info("RS " + reduce.getIdentifier() + " newColExprMap: " + oldMap);
}
@Override
public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx ctx, Object... nodeOutputs)
throws SemanticException {
ReduceSinkOperator op = (ReduceSinkOperator) nd;
ColumnPrunerProcCtx cppCtx = (ColumnPrunerProcCtx) ctx;
RowResolver resolver = cppCtx.getOpToParseCtxMap().get(op).getRowResolver();
ReduceSinkDesc conf = op.getConf();
List<String> colLists = new ArrayList<String>();
ArrayList<ExprNodeDesc> keys = conf.getKeyCols();
LOG.debug("Reduce Sink Operator " + op.getIdentifier() + " key:" + keys);
for (ExprNodeDesc key : keys) {
colLists = Utilities.mergeUniqElems(colLists, key.getCols());
}
assert op.getNumChild() == 1;
Operator<? extends OperatorDesc> child = op.getChildOperators().get(0);
List<String> childCols;
if (child instanceof CommonJoinOperator) {
childCols = cppCtx.getJoinPrunedColLists().get(child).get((byte) conf.getTag());
} else {
childCols = cppCtx.getPrunedColList(child);
}
List<ExprNodeDesc> valCols = conf.getValueCols();
List<String> valColNames = conf.getOutputValueColumnNames();
if (childCols != null) {
boolean[] flags = new boolean[valCols.size()];
for (String childCol : childCols) {
int index = valColNames.indexOf(Utilities.removeValueTag(childCol));
if (index < 0) {
continue;
}
flags[index] = true;
colLists = Utilities.mergeUniqElems(colLists, valCols.get(index).getCols());
}
Collections.sort(colLists);
pruneReduceSinkOperator(flags, op, cppCtx);
cppCtx.getPrunedColLists().put(op, colLists);
return null;
}
// Reduce Sink contains the columns needed - no need to aggregate from
// children
for (ExprNodeDesc val : valCols) {
colLists = Utilities.mergeUniqElems(colLists, val.getCols());
}
cppCtx.getPrunedColLists().put(op, colLists);
return null;
}