private DLinqQueryNode VisitGroupBy(QueryNodeInfo source,
LambdaExpression keySelectExpr,
LambdaExpression elemSelectExpr,
LambdaExpression resultSelectExpr,
Expression comparerExpr,
Expression queryExpr)
{
DLinqQueryNode child = this.Visit(source);
ExpressionInfo einfo = new ExpressionInfo(keySelectExpr);
if (einfo.IsExpensive)
{
// Any method call that is not tagged as "expensive=false" will be deemed expensive.
// if the keySelector is expensive, we rewrite the query so that the key-function is invoked only once
// and the record key passed around via a Pair<TKey,TRecord>.
// keyFunc becomes pair=>pair.Key
// elementSelector must be rewritten so that references to (record) become (pair.Value)
Type[] vkTypes = keySelectExpr.Type.GetGenericArguments();
Type pairType = typeof(Pair<,>).MakeGenericType(vkTypes[1], vkTypes[0]);
ParameterExpression pairParam = Expression.Parameter(pairType, "e");
// Add Select(x => new Pair<K,S>(key(x), x))
ParameterExpression valueParam = keySelectExpr.Parameters[0];
Expression body = Expression.New(pairType.GetConstructors()[0], keySelectExpr.Body, valueParam);
Type delegateType = typeof(Func<,>).MakeGenericType(valueParam.Type, body.Type);
LambdaExpression selectExpr = Expression.Lambda(delegateType, body, valueParam);
child = new DLinqSelectNode(QueryNodeType.Select, selectExpr, null, queryExpr, child);
// Change keySelector to e => e.Key
PropertyInfo keyInfo = pairParam.Type.GetProperty("Key");
body = Expression.Property(pairParam, keyInfo);
delegateType = typeof(Func<,>).MakeGenericType(pairParam.Type, body.Type);
keySelectExpr = Expression.Lambda(delegateType, body, pairParam);
// Add or change elementSelector with e.Value
PropertyInfo valueInfo = pairParam.Type.GetProperty("Value");
body = Expression.Property(pairParam, valueInfo);
if (elemSelectExpr != null)
{
ParameterSubst subst = new ParameterSubst(elemSelectExpr.Parameters[0], body);
body = subst.Visit(elemSelectExpr.Body);
}
delegateType = typeof(Func<,>).MakeGenericType(pairParam.Type, body.Type);
elemSelectExpr = Expression.Lambda(delegateType, body, pairParam);
}
Type keyType = keySelectExpr.Type.GetGenericArguments()[1];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(keyType))
{
throw DryadLinqException.Create(DryadLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable, keyType),
queryExpr);
}
Type elemType;
if (elemSelectExpr == null)
{
elemType = keySelectExpr.Type.GetGenericArguments()[0];
}
else
{
elemType = elemSelectExpr.Type.GetGenericArguments()[1];
}
// The comparer object:
object comparer = null;
if (comparerExpr != null)
{
ExpressionSimplifier<object> evaluator = new ExpressionSimplifier<object>();
comparer = evaluator.Eval(comparerExpr);
}
LambdaExpression keySelectExpr1 = keySelectExpr;
LambdaExpression elemSelectExpr1 = elemSelectExpr;
LambdaExpression resultSelectExpr1 = resultSelectExpr;
LambdaExpression seedExpr1 = null;
LambdaExpression accumulateExpr1 = null;
List<DecompositionInfo> dInfoList = null;
if (resultSelectExpr != null)
{
dInfoList = Decomposition.GetDecompositionInfoList(resultSelectExpr, this.m_codeGen);
}
String groupByOpName = "GroupBy";
DLinqQueryNode groupByNode = child;
bool isPartitioned = child.OutputPartition.IsPartitionedBy(keySelectExpr, comparer);
if (dInfoList != null)
{
// ** Decomposable GroupBy-Reduce
// This block creates the first GroupByNode and does some preparation for subsequent nodes.
if (child.IsOrderedBy(keySelectExpr, comparer))
{
groupByOpName = "OrderedGroupBy";
}
int dcnt = dInfoList.Count;
ParameterExpression keyParam;
if (resultSelectExpr.Parameters.Count == 1)
{
keyParam = Expression.Parameter(keyType, DryadLinqCodeGen.MakeUniqueName("k"));
}
else
{
keyParam = resultSelectExpr.Parameters[0];
}
// Seed:
ParameterExpression param2 = Expression.Parameter(
elemType, DryadLinqCodeGen.MakeUniqueName("e"));
Expression zeroExpr = Expression.Constant(0, typeof(int));
Expression seedBody = zeroExpr;
if (dcnt != 0)
{
LambdaExpression seed = dInfoList[dcnt-1].Seed;
ParameterSubst subst = new ParameterSubst(seed.Parameters[0], param2);
seedBody = subst.Visit(seed.Body);
for (int i = dcnt - 2; i >= 0; i--)
{
seed = dInfoList[i].Seed;
subst = new ParameterSubst(seed.Parameters[0], param2);
Expression firstExpr = subst.Visit(seed.Body);
Type newPairType = typeof(Pair<,>).MakeGenericType(firstExpr.Type, seedBody.Type);
seedBody = Expression.New(newPairType.GetConstructors()[0], firstExpr, seedBody);
}
}
LambdaExpression seedExpr = Expression.Lambda(seedBody, param2);
// Accumulate:
ParameterExpression param1 = Expression.Parameter(
seedBody.Type, DryadLinqCodeGen.MakeUniqueName("a"));
Expression accumulateBody = zeroExpr;
if (dcnt != 0)
{
accumulateBody = Decomposition.AccumulateList(param1, param2, dInfoList, 0);
}
LambdaExpression accumulateExpr = Expression.Lambda(accumulateBody, param1, param2);
// Now prepare for the merge-aggregator and/or in the secondary group-by.
// keySelectExpr1: e => e.Key
Type reducerResType = typeof(Pair<,>).MakeGenericType(keyParam.Type, accumulateBody.Type);
ParameterExpression reducerResParam = Expression.Parameter(reducerResType, "e");
PropertyInfo keyInfo = reducerResParam.Type.GetProperty("Key");
Expression body = Expression.Property(reducerResParam, keyInfo);
Type delegateType = typeof(Func<,>).MakeGenericType(reducerResParam.Type, body.Type);
keySelectExpr1 = Expression.Lambda(delegateType, body, reducerResParam);
// elemSelectExpr1: e => e.Value
PropertyInfo valueInfo = reducerResParam.Type.GetProperty("Value");
body = Expression.Property(reducerResParam, valueInfo);
delegateType = typeof(Func<,>).MakeGenericType(reducerResParam.Type, body.Type);
elemSelectExpr1 = Expression.Lambda(delegateType, body, reducerResParam);
// SeedExpr1
param2 = Expression.Parameter(elemSelectExpr1.Body.Type,
DryadLinqCodeGen.MakeUniqueName("e"));
seedExpr1 = Expression.Lambda(param2, param2);
// AccumulateExpr1
Expression recursiveAccumulateBody = zeroExpr;
if (dcnt != 0)
{
recursiveAccumulateBody = Decomposition.RecursiveAccumulateList(param1, param2, dInfoList, 0);
}
accumulateExpr1 = Expression.Lambda(recursiveAccumulateBody, param1, param2);
// resultSelectExpr1
resultSelectExpr1 = null;
// The first groupByNode.
// If the input was already correctly partitioned, this will be the only groupByNode.
bool isPartial = StaticConfig.GroupByLocalAggregationIsPartial && !isPartitioned;
groupByNode = new DLinqGroupByNode(
groupByOpName, keySelectExpr, elemSelectExpr, null,
seedExpr, accumulateExpr, accumulateExpr1, comparerExpr,
isPartial, queryExpr, child);
}
else
{
// Can't do partial aggregation.
// Use sort, mergesort, and ordered groupby, if TKey implements IComparable.
if ((comparer != null && TypeSystem.IsComparer(comparer, keyType)) ||
(comparer == null && TypeSystem.HasDefaultComparer(keyType)))
{
if (!child.IsOrderedBy(keySelectExpr, comparer))
{
groupByNode = new DLinqOrderByNode(keySelectExpr, comparerExpr, true, queryExpr, child);
}
groupByOpName = "OrderedGroupBy";
}
// Add a GroupByNode if it is partitioned or has elementSelector.
// If the input was already correctly partitioned, this will be the only groupByNode.
if (isPartitioned)
{
groupByNode = new DLinqGroupByNode(groupByOpName,
keySelectExpr,
elemSelectExpr,
resultSelectExpr,
null, // seed
null, // accumulate
null, // recursiveAccumulate
comparerExpr,
false, // isPartial
queryExpr,
groupByNode);
}
else if (elemSelectExpr != null)
{
// Local GroupBy without resultSelector:
groupByNode = new DLinqGroupByNode(groupByOpName,
keySelectExpr,
elemSelectExpr,
null, // resultSelect
null, // seed
null, // accumulate
null, // recursiveAccumulate
comparerExpr,
StaticConfig.GroupByLocalAggregationIsPartial, // isPartial
queryExpr,
groupByNode);
// keySelectExpr1: g => g.Key
ParameterExpression groupParam = Expression.Parameter(groupByNode.OutputTypes[0], "g");
PropertyInfo keyInfo = groupParam.Type.GetProperty("Key");
Expression body = Expression.Property(groupParam, keyInfo);
Type delegateType = typeof(Func<,>).MakeGenericType(groupParam.Type, body.Type);
keySelectExpr1 = Expression.Lambda(delegateType, body, groupParam);
// No elementSelector
elemSelectExpr1 = null;
// resultSelectExpr1
ParameterExpression keyParam;
Type groupType = typeof(IEnumerable<>).MakeGenericType(groupByNode.OutputTypes[0]);
groupParam = Expression.Parameter(groupType, DryadLinqCodeGen.MakeUniqueName("g"));
if (resultSelectExpr == null)
{
// resultSelectExpr1: (k, g) => MakeDryadLinqGroup(k, g)
keyParam = Expression.Parameter(keySelectExpr1.Body.Type, DryadLinqCodeGen.MakeUniqueName("k"));
MethodInfo groupingInfo = typeof(DryadLinqEnumerable).GetMethod("MakeDryadLinqGroup");
groupingInfo = groupingInfo.MakeGenericMethod(keyParam.Type, elemType);
body = Expression.Call(groupingInfo, keyParam, groupParam);
}
else
{
// resultSelectExpr1: (k, g) => resultSelectExpr(k, FlattenGroups(g))
keyParam = resultSelectExpr.Parameters[0];
MethodInfo flattenInfo = typeof(DryadLinqEnumerable).GetMethod("FlattenGroups");
flattenInfo = flattenInfo.MakeGenericMethod(keyParam.Type, elemType);
Expression groupExpr = Expression.Call(flattenInfo, groupParam);
ParameterSubst subst = new ParameterSubst(resultSelectExpr.Parameters[1], groupExpr);
body = subst.Visit(resultSelectExpr.Body);
}
delegateType = typeof(Func<,,>).MakeGenericType(keyParam.Type, groupParam.Type, body.Type);
resultSelectExpr1 = Expression.Lambda(delegateType, body, keyParam, groupParam);
}
}
// At this point, the first GroupByNode has been created.
DLinqQueryNode groupByNode1 = groupByNode;
DLinqMergeNode mergeNode = null;
if (!isPartitioned)
{
// Create HashPartitionNode, MergeNode, and second GroupByNode
// Note, if we are doing decomposable-GroupByReduce, there is still some work to go after this
// - attach the combiner to the first merge-node
// - attach the combiner to the merge-node
// - attach finalizer to second GroupBy
int parCount = (groupByNode.IsDynamic) ? StaticConfig.DefaultPartitionCount : groupByNode.OutputPartition.Count;
bool isDynamic = (StaticConfig.DynamicOptLevel & StaticConfig.DynamicHashPartitionLevel) != 0;
DLinqQueryNode hdistNode = new DLinqHashPartitionNode(keySelectExpr1, comparerExpr, parCount,
isDynamic, queryExpr, groupByNode);
// Create the Merge Node
if (groupByOpName == "OrderedGroupBy")
{
// Mergesort with the same keySelector of the hash partition
mergeNode = new DLinqMergeNode(keySelectExpr1, comparerExpr, true, queryExpr, hdistNode);
}
else
{
// Random merge
mergeNode = new DLinqMergeNode(false, queryExpr, hdistNode);
}
groupByNode1 = new DLinqGroupByNode(groupByOpName, keySelectExpr1, elemSelectExpr1,
resultSelectExpr1, seedExpr1, accumulateExpr1,
accumulateExpr1, comparerExpr, false, queryExpr,
mergeNode);
}
// Final tidy-up for decomposable GroupBy-Reduce pattern.
// - attach combiner to first GroupByNode
// - attache combiner to MergeNode as an aggregator
// - build a SelectNode to project out results and call finalizer on them.
if (dInfoList != null)
{
// Add dynamic aggregator to the merge-node, if applicable
if (StaticConfig.GroupByDynamicReduce && !isPartitioned)
{
mergeNode.AddAggregateNode(groupByNode1);
}
// Add the final Select node
Type keyResultPairType = typeof(Pair<,>).MakeGenericType(keyType, seedExpr1.Body.Type);
ParameterExpression keyResultPairParam = Expression.Parameter(keyResultPairType,
DryadLinqCodeGen.MakeUniqueName("e"));
PropertyInfo valuePropInfo_1 = keyResultPairType.GetProperty("Value");
Expression combinedValueExpr = Expression.Property(keyResultPairParam, valuePropInfo_1);
// First, build the combinerList
int dcnt = dInfoList.Count;
Expression[] combinerList = new Expression[dcnt];
for (int i = 0; i < dcnt; i++)
{
if (i + 1 == dcnt)
{
combinerList[i] = combinedValueExpr;
}
else
{
PropertyInfo keyPropInfo = combinedValueExpr.Type.GetProperty("Key");
combinerList[i] = Expression.Property(combinedValueExpr, keyPropInfo);
PropertyInfo valuePropInfo = combinedValueExpr.Type.GetProperty("Value");
combinedValueExpr = Expression.Property(combinedValueExpr, valuePropInfo);
}
LambdaExpression finalizerExpr = dInfoList[i].FinalReducer;
if (finalizerExpr != null)
{
ParameterSubst subst = new ParameterSubst(finalizerExpr.Parameters[0], combinerList[i]);
combinerList[i] = subst.Visit(finalizerExpr.Body);
}
}
// Build the funcList
Expression[] funcList = new Expression[dcnt];
for (int i = 0; i < dcnt; i++)
{
funcList[i] = dInfoList[i].Func;
}
// Apply the substitutions
CombinerSubst combinerSubst = new CombinerSubst(resultSelectExpr, keyResultPairParam, funcList, combinerList);
Expression finalizerSelectBody = combinerSubst.Visit();
// Finally, the Select node
Type delegateType = typeof(Func<,>).MakeGenericType(keyResultPairType, finalizerSelectBody.Type);
LambdaExpression selectExpr = Expression.Lambda(delegateType, finalizerSelectBody, keyResultPairParam);
groupByNode1 = new DLinqSelectNode(QueryNodeType.Select, selectExpr, null, queryExpr, groupByNode1);
}
return groupByNode1;
}
internal virtual string Visit(DLinqSelectNode node,
CodeMemberMethod vertexMethod,
string[] readerNames,
string[] writerNames)
{
return node.AddVertexCode(vertexMethod, readerNames, writerNames);
}
private DLinqQueryNode VisitSelect(QueryNodeInfo source,
QueryNodeType nodeType,
LambdaExpression selector,
LambdaExpression resultSelector,
MethodCallExpression queryExpr)
{
DLinqQueryNode selectNode;
if (selector.Type.GetGenericArguments().Length == 2)
{
// If this select's child is a groupby node, push this select into its child, if
// 1. The groupby node is not tee'd, and
// 2. The groupby node has no result selector, and
// 3. The selector is decomposable
if (!source.IsForked &&
IsGroupByWithoutResultSelector(source.QueryExpression) &&
Decomposition.GetDecompositionInfoList(selector, m_codeGen) != null)
{
MethodCallExpression expr = (MethodCallExpression)source.QueryExpression;
LambdaExpression keySelectExpr = DryadLinqExpression.GetLambda(expr.Arguments[1]);
// Figure out elemSelectExpr and comparerExpr
LambdaExpression elemSelectExpr = null;
Expression comparerExpr = null;
if (expr.Arguments.Count == 3)
{
elemSelectExpr = DryadLinqExpression.GetLambda(expr.Arguments[2]);
if (elemSelectExpr == null)
{
comparerExpr = expr.Arguments[2];
}
}
else if (expr.Arguments.Count == 4)
{
elemSelectExpr = DryadLinqExpression.GetLambda(expr.Arguments[2]);
comparerExpr = expr.Arguments[3];
}
// Construct new query expression by building result selector expression
// and pushing it to groupby node.
selectNode = VisitGroupBy(source.Children[0].Child, keySelectExpr,
elemSelectExpr, selector, comparerExpr, queryExpr);
if (nodeType == QueryNodeType.SelectMany)
{
Type selectorRetType = selector.Type.GetGenericArguments()[1];
LambdaExpression id = IdentityFunction.Instance(selectorRetType);
selectNode = new DLinqSelectNode(nodeType, id, resultSelector, queryExpr, selectNode);
}
}
else
{
DLinqQueryNode child = this.Visit(source);
selectNode = this.PromoteConcat(
source, child,
x => new DLinqSelectNode(nodeType, selector, resultSelector, queryExpr, x));
}
}
else
{
// The "indexed" version
DLinqQueryNode child = this.Visit(source);
if (!child.IsDynamic && child.OutputPartition.Count == 1)
{
selectNode = this.PromoteConcat(
source, child,
x => new DLinqSelectNode(nodeType, selector, resultSelector, queryExpr, x));
}
else
{
child.IsForked = true;
// Create (x, y) => Select(x, y, selector)
Type ptype1 = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
Type ptype2 = typeof(IEnumerable<>).MakeGenericType(typeof(IndexedValue<long>));
ParameterExpression param1 = Expression.Parameter(ptype1, DryadLinqCodeGen.MakeUniqueName("x"));
ParameterExpression param2 = Expression.Parameter(ptype2, DryadLinqCodeGen.MakeUniqueName("y"));
string methodName = queryExpr.Method.Name;
Type[] selectorTypeArgs = selector.Type.GetGenericArguments();
Type typeArg2 = selectorTypeArgs[selectorTypeArgs.Length - 1];
if (nodeType == QueryNodeType.SelectMany)
{
if (resultSelector != null)
{
methodName += "Result";
}
typeArg2 = typeArg2.GetGenericArguments()[0];
}
string targetMethodName = methodName + "WithStartIndex";
MethodInfo minfo = typeof(DryadLinqEnumerable).GetMethod(targetMethodName);
Expression body;
if (resultSelector == null)
{
minfo = minfo.MakeGenericMethod(child.OutputTypes[0], typeArg2);
body = Expression.Call(minfo, param1, param2, selector);
}
else
{
minfo = minfo.MakeGenericMethod(child.OutputTypes[0], typeArg2, resultSelector.Body.Type);
body = Expression.Call(minfo, param1, param2, selector, resultSelector);
}
Type type = typeof(Func<,,>).MakeGenericType(ptype1, ptype2, body.Type);
LambdaExpression procFunc = Expression.Lambda(type, body, param1, param2);
bool isLong = methodName.StartsWith("Long", StringComparison.Ordinal);
DLinqQueryNode offsetNode = this.CreateOffset(isLong, queryExpr, child);
selectNode = new DLinqApplyNode(procFunc, queryExpr, child, offsetNode);
}
}
return selectNode;
}