// Basic plan: (reverse all partitions) then (reverse data in each partition)
// The main complication is to perform the first step.
// Approach:
// - tee the input.
// - have a dummy apply node that produces the singleton {0} at each partition
// - merge to get a seq {0,0,..} whose length = nPartition.
// - convert that seq to { (0,n), (1,n), ...}
// - hash-partition to send one item to each of the n workers.
// - use binary-apply to attach targetIndex to each source item
// Apply( seq1 = indexCountPair, seq2 = original data) => ({tgt, item0}, {tgt, item1}, .. )
// - hash-partition to move items to target partition.
// - use local LINQ reverse to do the local data reversal.
private DryadQueryNode VisitReverse(QueryNodeInfo source, Expression queryExpr)
{
DryadQueryNode child = this.Visit(source);
if (child.IsDynamic)
{
throw new DryadLinqException("Reverse is only supported for static partition count");
}
child.IsForked = true;
// Apply node for s => ValueZero(s)
Type paramType = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param = Expression.Parameter(paramType, "s");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("ValueZero");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
Expression body = Expression.Call(minfo, param);
Type funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode valueZeroNode = new DryadApplyNode(procFunc, queryExpr, child);
// Apply node for s => ReverseIndex(s)
paramType = typeof(IEnumerable<>).MakeGenericType(typeof(int));
param = Expression.Parameter(paramType, "s");
minfo = typeof(HpcLinqHelper).GetMethod("MakeIndexCountPairs");
body = Expression.Call(minfo, param);
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode mergeZeroNode = new DryadMergeNode(true, true, queryExpr, valueZeroNode);
DryadQueryNode indexCountNode = new DryadApplyNode(procFunc, queryExpr, mergeZeroNode);
// HashPartition to distribute the indexCounts -- one to each partition.
// each partition will receive (myPartitionID, pcount).
int pcount = child.OutputPartition.Count;
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
Expression keySelectBody = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, keySelectBody.Type);
LambdaExpression keySelectExpr = Expression.Lambda(funcType, keySelectBody, param);
DryadQueryNode hdistNode = new DryadHashPartitionNode(keySelectExpr,
null,
pcount,
queryExpr,
indexCountNode);
// Apply node for (x, y) => AddIndexForReverse(x, y)
ParameterExpression param1 = Expression.Parameter(body.Type, "x");
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param2 = Expression.Parameter(paramType2, "y");
minfo = typeof(HpcLinqHelper).GetMethod("AddIndexForReverse");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
body = Expression.Call(minfo, param1, param2);
funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression addIndexFunc = Expression.Lambda(funcType, body, param1, param2);
DryadQueryNode addIndexNode = new DryadApplyNode(addIndexFunc, queryExpr, hdistNode, child);
// HashPartition(x => x.index, x => x.value, pcount)
// Moves all data to correct target partition. (each worker will direct all its items to one target partition)
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
body = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
keySelectExpr = Expression.Lambda(funcType, body, param);
body = Expression.Property(param, "Value");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression resultSelectExpr = Expression.Lambda(funcType, body, param);
DryadQueryNode reversePartitionNode = new DryadHashPartitionNode(
keySelectExpr, resultSelectExpr, null,
pcount, false, queryExpr, addIndexNode);
// Reverse node
paramType = typeof(IEnumerable<>).MakeGenericType(reversePartitionNode.OutputTypes[0]);
param = Expression.Parameter(paramType, "x");
minfo = typeof(HpcLinqVertex).GetMethod("Reverse");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
body = Expression.Call(minfo, param);
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode resNode = new DryadMergeNode(true, true, queryExpr, reversePartitionNode);
resNode = new DryadApplyNode(procFunc, queryExpr, resNode);
return resNode;
}
private DryadQueryNode VisitSequenceEqual(QueryNodeInfo source1,
QueryNodeInfo source2,
Expression comparerExpr,
Expression queryExpr)
{
DryadQueryNode child1 = this.Visit(source1);
DryadQueryNode child2 = this.Visit(source2);
Type elemType = child1.OutputTypes[0];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(elemType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerExpressionMustBeSpecifiedOrElementTypeMustBeIEquatable,
String.Format(SR.ComparerExpressionMustBeSpecifiedOrElementTypeMustBeIEquatable, elemType),
queryExpr);
}
// Well, let us do it on a single machine for now
child1 = new DryadMergeNode(true, false, queryExpr, child1);
child2 = new DryadMergeNode(true, false, queryExpr, child2);
// Apply node for (x, y) => SequenceEqual(x, y, c)
Type paramType = typeof(IEnumerable<>).MakeGenericType(elemType);
ParameterExpression param1 = Expression.Parameter(paramType, "s1");
ParameterExpression param2 = Expression.Parameter(paramType, "s2");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("SequenceEqual");
minfo = minfo.MakeGenericMethod(elemType);
if (comparerExpr == null)
{
comparerExpr = Expression.Constant(null, typeof(IEqualityComparer<>).MakeGenericType(elemType));
}
Expression body = Expression.Call(minfo, param1, param2, comparerExpr);
Type funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(funcType, body, param1, param2);
return new DryadApplyNode(procFunc, queryExpr, child1, child2);
}
// Phase 1 of the query optimization
internal void GenerateQueryPlanPhase1()
{
if (this.m_queryPlan1 != null) return;
// Apply some simple rewrite rules
SimpleRewriter rewriter = new SimpleRewriter(this.m_exprNodeInfoMap.Values.ToList());
rewriter.Rewrite();
// Generate the query plan of phase1
var referencedNodes = this.m_referencedQueryMap.Values;
this.m_queryPlan1 = new DryadQueryNode[this.m_queryExprs.Length + referencedNodes.Count];
for (int i = 0; i < this.m_queryExprs.Length; i++)
{
this.m_queryPlan1[i] = this.Visit(this.m_queryNodeInfos[i].children[0].child);
}
int idx = this.m_queryExprs.Length;
foreach (QueryNodeInfo nodeInfo in referencedNodes)
{
// Add a Tee'd Merge
this.m_queryPlan1[idx] = this.Visit(nodeInfo.children[0].child);
DryadQueryNode mergeNode = new DryadMergeNode(true, false, nodeInfo.queryExpression,
this.m_queryPlan1[idx]);
this.m_queryPlan1[idx] = new DryadTeeNode(mergeNode.OutputTypes[0], true,
mergeNode.QueryExpression, mergeNode);
nodeInfo.queryNode = this.m_queryPlan1[idx];
idx++;
}
// Finally, add the output nodes.
Dictionary<DryadQueryNode, int> forkCounts = new Dictionary<DryadQueryNode, int>();
for (int i = 0; i < this.m_queryExprs.Length; i++)
{
DryadQueryNode queryNode = this.m_queryPlan1[i];
int cnt;
if (!forkCounts.TryGetValue(queryNode, out cnt))
{
cnt = queryNode.Parents.Count;
}
forkCounts[queryNode] = cnt + 1;
}
for (int i = 0; i < this.m_queryExprs.Length; i++)
{
HpcClientSideLog.Add("Query " + i + " Output: " + this.m_outputDatapaths[i]);
DryadQueryNode queryNode = this.m_queryPlan1[i];
if (TypeSystem.IsAnonymousType(queryNode.OutputTypes[0]))
{
throw new DryadLinqException(HpcLinqErrorCode.OutputTypeCannotBeAnonymous,
SR.OutputTypeCannotBeAnonymous);
}
// Add dummy Apply to make Dryad happy (it doesn't like to hook inputs straight to outputs)
if ((queryNode is DryadInputNode) || (forkCounts[queryNode] > 1))
{
// Add a dummy Apply
Type paramType = typeof(IEnumerable<>).MakeGenericType(queryNode.OutputTypes[0]);
ParameterExpression param = Expression.Parameter(paramType, "x");
Type type = typeof(Func<,>).MakeGenericType(paramType, paramType);
LambdaExpression applyExpr = Expression.Lambda(type, param, param);
DryadQueryNode applyNode = new DryadApplyNode(applyExpr, this.m_queryExprs[i], queryNode);
applyNode.OutputDataSetInfo = queryNode.OutputDataSetInfo;
queryNode = applyNode;
}
if (queryNode is DryadConcatNode)
{
// Again, we add dummy Apply in certain cases to make Dryad happy
((DryadConcatNode)queryNode).FixInputs();
}
// Add the output node
DscCompressionScheme outputScheme = this.m_context.Configuration.OutputDataCompressionScheme;
DryadOutputNode outputNode = new DryadOutputNode(this.m_context,
this.m_outputDatapaths[i],
this.m_isTempOutput[i],
outputScheme,
this.m_queryExprs[i],
queryNode);
this.m_queryPlan1[i] = outputNode;
if (this.m_outputUriMap.ContainsKey(this.m_outputDatapaths[i].ToLower()))
{
throw new DryadLinqException(HpcLinqErrorCode.MultipleOutputsWithSameDscUri,
String.Format(SR.MultipleOutputsWithSameDscUri, this.m_outputDatapaths[i]));
}
this.m_outputUriMap.Add(this.m_outputDatapaths[i].ToLower(), outputNode);
this.m_outputTypes[i] = this.m_queryPlan1[i].OutputTypes[0];
// Remove useless Tees to make Dryad happy
if ((queryNode is DryadTeeNode) && (forkCounts[queryNode] == 1))
{
DryadQueryNode teeChild = queryNode.Children[0];
teeChild.UpdateParent(queryNode, outputNode);
outputNode.UpdateChildren(queryNode, teeChild);
}
}
}
private DryadQueryNode VisitRangePartition(QueryNodeInfo source,
LambdaExpression keySelectExpr,
Expression keysExpr,
Expression comparerExpr,
Expression isDescendingExpr,
Expression partitionCountExpr,
Expression queryExpr)
{
DryadQueryNode child = this.Visit(source);
Type keyType = keySelectExpr.Type.GetGenericArguments()[1];
if (comparerExpr == null && !TypeSystem.HasDefaultComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIComparable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIComparable, keyType),
queryExpr);
}
DryadQueryNode resNode;
if (keysExpr == null)
{
// case: no keys are provided -- create range partitioner with auto-separator-selection
bool dynamicOptEnabled = (StaticConfig.DynamicOptLevel & StaticConfig.DynamicRangePartitionLevel) != 0;
bool useDynamic = (dynamicOptEnabled || child.IsDynamic);
// NOTE: for MayRTM, useDynamic should always be false
resNode = this.CreateRangePartition(useDynamic, keySelectExpr, null, comparerExpr, isDescendingExpr,
queryExpr, partitionCountExpr, child);
}
else
{
// case: keys are local enum (eg an array) -- create range partitioner with keys input via Object-store.
resNode = new DryadRangePartitionNode(keySelectExpr, null, keysExpr, comparerExpr,
isDescendingExpr, null, queryExpr, child);
resNode = new DryadMergeNode(false, true, queryExpr, resNode);
}
return resNode;
}
// Creates an "auto-sampling range-partition sub-query"
private DryadQueryNode CreateRangePartition(bool isDynamic,
LambdaExpression keySelectExpr,
LambdaExpression resultSelectExpr,
Expression comparerExpr,
Expression isDescendingExpr,
Expression queryExpr,
Expression partitionCountExpr,
DryadQueryNode child)
{
// Make child a Tee node
child.IsForked = true;
// The partition count
Expression countExpr = null;
if (isDescendingExpr == null)
{
isDescendingExpr = Expression.Constant(false, typeof(bool)); //default for isDescending is false.
}
// NOTE: for MayRTM, isDynamic should never be true
if (!isDynamic)
{
if (partitionCountExpr != null)
{
countExpr = partitionCountExpr;
}
else
{
// If partitionCount was not explicitly set, use the child's partition count.
countExpr = Expression.Constant(child.OutputPartition.Count);
}
}
Type recordType = child.OutputTypes[0];
Type keyType = keySelectExpr.Type.GetGenericArguments()[1];
// Create x => Phase1Sampling(x_1, keySelector, denv)
Type lambdaParamType1 = typeof(IEnumerable<>).MakeGenericType(recordType);
ParameterExpression lambdaParam1 = Expression.Parameter(lambdaParamType1, "x_1");
ParameterExpression denvParam = Expression.Parameter(typeof(HpcLinqVertexEnv), "denv");
MethodInfo minfo = typeof(HpcLinqSampler).GetMethod("Phase1Sampling");
Expression body = Expression.Call(minfo.MakeGenericMethod(recordType, keyType),
lambdaParam1, keySelectExpr, denvParam);
Type type = typeof(Func<,>).MakeGenericType(lambdaParam1.Type, body.Type);
LambdaExpression samplingExpr = Expression.Lambda(type, body, lambdaParam1);
// Create the Sampling node
DryadApplyNode samplingNode = new DryadApplyNode(samplingExpr, queryExpr, child);
// Create x => RangeSampler(x, keySelectExpr, comparer, isDescendingExpr)
Type lambdaParamType = typeof(IEnumerable<>).MakeGenericType(keyType);
ParameterExpression lambdaParam = Expression.Parameter(lambdaParamType, "x_2");
//For RTM, isDynamic should never be true.
//string methodName = (isDynamic) ? "RangeSampler_Dynamic" : "RangeSampler_Static";
Debug.Assert(isDynamic == false, "Internal error: isDynamic is true.");
string methodName = "RangeSampler_Static";
minfo = typeof(HpcLinqSampler).GetMethod(methodName);
minfo = minfo.MakeGenericMethod(keyType);
Expression comparerArgExpr = comparerExpr;
if (comparerExpr == null)
{
if (!TypeSystem.HasDefaultComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIComparable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIComparable, keyType),
queryExpr);
}
comparerArgExpr = Expression.Constant(null, typeof(IComparer<>).MakeGenericType(keyType));
}
Expression lastArg;
if (isDynamic)
{
lastArg = denvParam;
}
else
{
lastArg = countExpr;
}
body = Expression.Call(minfo, lambdaParam, comparerArgExpr, isDescendingExpr, lastArg);
type = typeof(Func<,>).MakeGenericType(lambdaParam.Type, body.Type);
LambdaExpression samplerExpr = Expression.Lambda(type, body, lambdaParam);
// Create the sample node
DryadQueryNode sampleDataNode = new DryadMergeNode(false, true, queryExpr, samplingNode);
DryadQueryNode sampleNode = new DryadApplyNode(samplerExpr, queryExpr, sampleDataNode);
sampleNode.IsForked = true;
// Create the range distribute node
DryadQueryNode resNode = new DryadRangePartitionNode(keySelectExpr,
resultSelectExpr,
null,
comparerExpr,
isDescendingExpr,
countExpr,
queryExpr,
child,
sampleNode);
resNode = new DryadMergeNode(false, true, queryExpr, resNode);
// Set the dynamic manager for sampleNode
if (isDynamic)
{
sampleDataNode.DynamicManager = new DynamicRangeDistributor(resNode);
}
return resNode;
}
private DryadQueryNode VisitPartitionOp(string opName,
QueryNodeInfo source,
QueryNodeType nodeType,
Expression controlExpr,
MethodCallExpression queryExpr)
{
DryadQueryNode resNode;
if (nodeType == QueryNodeType.TakeWhile &&
controlExpr.Type.GetGenericArguments().Length != 2)
{
// The "indexed" version.
resNode = this.Visit(source);
// The following block used to be skipped for resNode.OutputPartition.Count == 1,
// which causes compilation error (bug 13593)
// @@TODO[p3] : implement a working optimization for nPartition==1 that calls
// directly to Linq TakeWhile.
// Note: the test is: if (resNode.IsDynamic || resNode.OutputPartition.Count > 1)
{
resNode.IsForked = true;
bool isLong = (queryExpr.Method.Name == "LongTakeWhile");
DryadQueryNode offsetNode = this.CreateOffset(isLong, queryExpr, resNode);
// Create (x, y) => GroupIndexedTakeWhile(x, y, controlExpr)
Type ptype1 = typeof(IEnumerable<>).MakeGenericType(resNode.OutputTypes[0]);
Type ptype2 = typeof(IEnumerable<>).MakeGenericType(typeof(IndexedValue<long>));
ParameterExpression param1 = Expression.Parameter(ptype1, HpcLinqCodeGen.MakeUniqueName("x"));
ParameterExpression param2 = Expression.Parameter(ptype2, HpcLinqCodeGen.MakeUniqueName("y"));
string methodName = "GroupIndexed" + queryExpr.Method.Name;
MethodInfo minfo = typeof(HpcLinqEnumerable).GetMethod(methodName);
minfo = minfo.MakeGenericMethod(resNode.OutputTypes[0]);
Expression body = Expression.Call(minfo, param1, param2, controlExpr);
Type type = typeof(Func<,,>).MakeGenericType(ptype1, ptype2, body.Type);
LambdaExpression procFunc = Expression.Lambda(type, body, param1, param2);
resNode = new DryadApplyNode(procFunc, queryExpr, resNode, offsetNode);
}
}
else if (!source.IsForked &&
(nodeType == QueryNodeType.Take || nodeType == QueryNodeType.TakeWhile) &&
(source.OperatorName == "OrderBy" || source.OperatorName == "OrderByDescending"))
{
resNode = this.Visit(source.children[0].child);
bool isDescending = (source.OperatorName == "OrderByDescending");
MethodCallExpression sourceQueryExpr = (MethodCallExpression)source.queryExpression;
LambdaExpression keySelectExpr = HpcLinqExpression.GetLambda(sourceQueryExpr.Arguments[1]);
Expression comparerExpr = null;
if (sourceQueryExpr.Arguments.Count == 3)
{
comparerExpr = sourceQueryExpr.Arguments[2];
}
resNode = this.PromoteConcat(
source.children[0].child,
resNode,
delegate(DryadQueryNode x) {
DryadQueryNode y = new DryadOrderByNode(keySelectExpr, comparerExpr,
isDescending, sourceQueryExpr, x);
return FirstStagePartitionOp(opName, nodeType, controlExpr, queryExpr, y);
});
if (resNode.IsDynamic || resNode.OutputPartition.Count > 1)
{
// Need a mergesort
resNode = new DryadMergeNode(keySelectExpr, comparerExpr, isDescending, false, sourceQueryExpr, resNode);
}
}
else
{
resNode = this.Visit(source);
if (nodeType == QueryNodeType.Take || nodeType == QueryNodeType.TakeWhile)
{
resNode = this.PromoteConcat(
source, resNode,
x => FirstStagePartitionOp(opName, nodeType, controlExpr, queryExpr, x));
}
}
resNode = new DryadPartitionOpNode(opName, nodeType, controlExpr, false, queryExpr, resNode);
return resNode;
}
private DryadQueryNode VisitHashPartition(QueryNodeInfo source,
LambdaExpression keySelectExpr,
Expression comparerExpr,
Expression countExpr,
Expression queryExpr)
{
DryadQueryNode child = this.Visit(source);
Type keyType = keySelectExpr.Type.GetGenericArguments()[1];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable, keyType),
queryExpr);
}
bool isDynamic = (StaticConfig.DynamicOptLevel & StaticConfig.DynamicHashPartitionLevel) != 0;
int nOutputPartitions;
if (countExpr != null)
{
ExpressionSimplifier<int> evaluator = new ExpressionSimplifier<int>();
nOutputPartitions = evaluator.Eval(countExpr);
isDynamic = false;
}
else
{
// Note: For MayRTM, isDynamic will never be true.
nOutputPartitions = (isDynamic) ? 1 : child.OutputPartition.Count;
}
DryadQueryNode resNode = new DryadHashPartitionNode(
keySelectExpr, null, comparerExpr, nOutputPartitions,
isDynamic, queryExpr, child);
resNode = new DryadMergeNode(false, true, queryExpr, resNode);
return resNode;
}
private DryadQueryNode VisitZip(QueryNodeInfo first,
QueryNodeInfo second,
LambdaExpression resultSelector,
MethodCallExpression queryExpr)
{
DryadQueryNode child1 = this.Visit(first);
DryadQueryNode child2 = this.Visit(second);
if (child1.IsDynamic || child2.IsDynamic)
{
// Well, let us for now do it on a single machine
child1 = new DryadMergeNode(true, false, queryExpr, child1);
child2 = new DryadMergeNode(true, false, queryExpr, child2);
// Apply node for (x, y) => Zip(x, y, resultSelector)
Type paramType1 = typeof(IEnumerable<>).MakeGenericType(child1.OutputTypes[0]);
ParameterExpression param1 = Expression.Parameter(paramType1, "s1");
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(child2.OutputTypes[0]);
ParameterExpression param2 = Expression.Parameter(paramType2, "s2");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("Zip");
minfo = minfo.MakeGenericMethod(child1.OutputTypes[0]);
Expression body = Expression.Call(minfo, param1, param2, resultSelector);
Type funcType = typeof(Func<,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(funcType, body, param1, param2);
return new DryadApplyNode(procFunc, queryExpr, child1, child2);
}
else
{
int parCount1 = child1.OutputPartition.Count;
int parCount2 = child2.OutputPartition.Count;
// Count nodes
DryadQueryNode countNode1 = new DryadBasicAggregateNode(null, AggregateOpType.LongCount,
true, false, queryExpr, child1);
DryadQueryNode countNode2 = new DryadBasicAggregateNode(null, AggregateOpType.LongCount,
true, false, queryExpr, child2);
countNode1 = new DryadMergeNode(true, false, queryExpr, countNode1);
countNode2 = new DryadMergeNode(true, false, queryExpr, countNode2);
// Apply node for (x, y) => ZipCount(x, y)
Type paramType1 = typeof(IEnumerable<>).MakeGenericType(typeof(long));
ParameterExpression param1 = Expression.Parameter(paramType1, "x");
ParameterExpression param2 = Expression.Parameter(paramType1, "y");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("ZipCount");
Expression body = Expression.Call(minfo, param1, param2);
Type funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression zipCount = Expression.Lambda(funcType, body, param1, param2);
DryadQueryNode indexedCountNode = new DryadApplyNode(zipCount, queryExpr, countNode1, countNode2);
// HashPartition(x => x.index, parCount2)
ParameterExpression param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
Expression keySelectBody = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, keySelectBody.Type);
LambdaExpression keySelectExpr = Expression.Lambda(funcType, keySelectBody, param);
DryadQueryNode distCountNode = new DryadHashPartitionNode(keySelectExpr,
null,
parCount2,
queryExpr,
indexedCountNode);
// Apply node for (x, y) => AssignPartitionIndex(x, y)
param1 = Expression.Parameter(body.Type, "x");
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(child2.OutputTypes[0]);
param2 = Expression.Parameter(paramType2, "y");
minfo = typeof(HpcLinqHelper).GetMethod("AssignPartitionIndex");
minfo = minfo.MakeGenericMethod(child2.OutputTypes[0]);
body = Expression.Call(minfo, param1, param2);
funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression assignIndex = Expression.Lambda(funcType, body, param1, param2);
DryadQueryNode addIndexNode = new DryadApplyNode(assignIndex, queryExpr, distCountNode, child2);
// HashPartition(x => x.index, x => x.value, parCount1)
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
body = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
keySelectExpr = Expression.Lambda(funcType, body, param);
body = Expression.Property(param, "Value");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression resultSelectExpr = Expression.Lambda(funcType, body, param);
DryadQueryNode newChild2 = new DryadHashPartitionNode(keySelectExpr,
resultSelectExpr,
null,
parCount1,
false,
queryExpr,
addIndexNode);
newChild2 = new DryadMergeNode(true, true, queryExpr, newChild2);
// Finally the zip node
return new DryadZipNode(resultSelector, queryExpr, child1, newChild2);
}
}
private DryadQueryNode VisitDistinct(QueryNodeInfo source,
Expression comparerExpr,
Expression queryExpr)
{
DryadQueryNode child = this.Visit(source);
Type keyType = child.OutputTypes[0];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable, keyType),
queryExpr);
}
object comparer = null;
if (comparerExpr != null)
{
ExpressionSimplifier<object> evaluator = new ExpressionSimplifier<object>();
comparer = evaluator.Eval(comparerExpr);
}
LambdaExpression keySelectExpr = IdentityFunction.Instance(keyType);
if (!child.OutputPartition.IsPartitionedBy(keySelectExpr, comparer))
{
if (child.IsDynamic || child.OutputPartition.Count > 1)
{
child = new DryadDistinctNode(true, comparerExpr, queryExpr, child);
bool isDynamic = (StaticConfig.DynamicOptLevel & StaticConfig.DynamicHashPartitionLevel) != 0;
child = new DryadHashPartitionNode(keySelectExpr,
comparerExpr,
child.OutputPartition.Count,
isDynamic,
queryExpr,
child);
child = new DryadMergeNode(false, false, queryExpr, child);
}
}
DryadQueryNode resNode = new DryadDistinctNode(false, comparerExpr, queryExpr, child);
return resNode;
}
private DryadQueryNode VisitGroupBy(QueryNodeInfo source,
LambdaExpression keySelectExpr,
LambdaExpression elemSelectExpr,
LambdaExpression resultSelectExpr,
Expression comparerExpr,
Expression queryExpr)
{
DryadQueryNode 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 DryadSelectNode(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(HpcLinqErrorCode.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";
DryadQueryNode 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, HpcLinqCodeGen.MakeUniqueName("k"));
}
else
{
keyParam = resultSelectExpr.Parameters[0];
}
// Seed:
ParameterExpression param2 = Expression.Parameter(
elemType, HpcLinqCodeGen.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, HpcLinqCodeGen.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,
HpcLinqCodeGen.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 DryadGroupByNode(
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 DryadOrderByNode(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 DryadGroupByNode(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 DryadGroupByNode(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, HpcLinqCodeGen.MakeUniqueName("g"));
if (resultSelectExpr == null)
{
// resultSelectExpr1: (k, g) => MakeDryadLinqGroup(k, g)
keyParam = Expression.Parameter(keySelectExpr1.Body.Type, HpcLinqCodeGen.MakeUniqueName("k"));
MethodInfo groupingInfo = typeof(HpcLinqEnumerable).GetMethod("MakeHpcLinqGroup");
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(HpcLinqEnumerable).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.
DryadMergeNode mergeNode = null;
DryadQueryNode groupByNode1 = groupByNode;
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;
DryadQueryNode hdistNode = new DryadHashPartitionNode(keySelectExpr1, comparerExpr, parCount,
isDynamic, queryExpr, groupByNode);
// Create the Merge Node
if (groupByOpName == "OrderedGroupBy")
{
// Mergesort with the same keySelector of the hash partition
mergeNode = new DryadMergeNode(keySelectExpr1, comparerExpr, true, false, queryExpr, hdistNode);
}
else
{
// Random merge
mergeNode = new DryadMergeNode(false, false, queryExpr, hdistNode);
}
groupByNode1 = new DryadGroupByNode(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,
HpcLinqCodeGen.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 DryadSelectNode(QueryNodeType.Select, selectExpr, null, queryExpr, groupByNode1);
}
return groupByNode1;
}
private DryadQueryNode VisitConcat(QueryNodeInfo source, MethodCallExpression queryExpr)
{
DryadQueryNode[] childs = new DryadQueryNode[source.children.Count];
for (int i = 0; i < source.children.Count; ++i)
{
childs[i] = this.Visit(source.children[i].child);
}
DryadQueryNode resNode = new DryadConcatNode(queryExpr, childs);
int parCount = resNode.OutputPartition.Count;
if (!resNode.IsDynamic && parCount > StaticConfig.MaxPartitionCount)
{
// Too many partitions, need to repartition
int newParCount = parCount / 2;
DryadQueryNode countNode = new DryadBasicAggregateNode(null, AggregateOpType.LongCount,
true, false, queryExpr, resNode);
DryadQueryNode mergeCountNode = new DryadMergeNode(true, false, queryExpr, countNode);
// Apply node for s => IndexedCount(s)
Type paramType = typeof(IEnumerable<>).MakeGenericType(typeof(long));
ParameterExpression param = Expression.Parameter(paramType, "s");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("IndexedCount");
minfo = minfo.MakeGenericMethod(typeof(long));
Expression body = Expression.Call(minfo, param);
Type funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression indexedCountFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode indexedCountNode = new DryadApplyNode(indexedCountFunc, queryExpr, mergeCountNode);
// HashPartition(x => x.index, parCount)
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
Expression keySelectBody = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, keySelectBody.Type);
LambdaExpression keySelectExpr = Expression.Lambda(funcType, keySelectBody, param);
DryadQueryNode distCountNode = new DryadHashPartitionNode(keySelectExpr,
null,
parCount,
queryExpr,
indexedCountNode);
// Apply node for (x, y) => AddPartitionIndex(x, y, newParCount)
ParameterExpression param1 = Expression.Parameter(body.Type, "x");
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(resNode.OutputTypes[0]);
ParameterExpression param2 = Expression.Parameter(paramType2, "y");
minfo = typeof(HpcLinqHelper).GetMethod("AddPartitionIndex");
minfo = minfo.MakeGenericMethod(resNode.OutputTypes[0]);
body = Expression.Call(minfo, param1, param2, Expression.Constant(newParCount));
funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
LambdaExpression addIndexFunc = Expression.Lambda(funcType, body, param1, param2);
DryadQueryNode addIndexNode = new DryadApplyNode(addIndexFunc, queryExpr, distCountNode, resNode);
// HashPartition(x => x.index, x => x.value, newParCount)
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
body = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
keySelectExpr = Expression.Lambda(funcType, body, param);
body = Expression.Property(param, "Value");
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression resultSelectExpr = Expression.Lambda(funcType, body, param);
resNode = new DryadHashPartitionNode(keySelectExpr,
resultSelectExpr,
null,
newParCount,
false,
queryExpr,
addIndexNode);
resNode = new DryadMergeNode(true, true, queryExpr, resNode);
}
return resNode;
}
private DryadQueryNode VisitApplyWithPartitionIndex(
QueryNodeInfo source,
LambdaExpression procLambda,
Expression queryExpr)
{
// var indices = source.Apply(s => ValueZero(s)).Apply(s => AssignIndex(s));
// indices.Apply(source, (x, y) => ProcessWithIndex(x, y, procFunc));
DryadQueryNode child = this.Visit(source);
if (child.IsDynamic)
{
throw new DryadLinqException("ApplyWithPartitionIndex is only supported for static partition count");
}
child.IsForked = true;
// Apply node for s => ValueZero(s)
Type paramType = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param = Expression.Parameter(paramType, "s");
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("ValueZero");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
Expression body = Expression.Call(minfo, param);
Type funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode valueZeroNode = new DryadApplyNode(procFunc, queryExpr, child);
valueZeroNode = new DryadMergeNode(true, true, queryExpr, valueZeroNode);
// Apply node for s => AssignIndex(s)
paramType = typeof(IEnumerable<>).MakeGenericType(typeof(int));
param = Expression.Parameter(paramType, "s");
minfo = typeof(HpcLinqHelper).GetMethod("AssignIndex");
body = Expression.Call(minfo, param);
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode assignIndexNode = new DryadApplyNode(procFunc, queryExpr, valueZeroNode);
// HashPartition to distribute the indices -- one to each partition.
int pcount = child.OutputPartition.Count;
param = Expression.Parameter(body.Type, "x");
funcType = typeof(Func<,>).MakeGenericType(param.Type, param.Type);
LambdaExpression keySelectExpr = Expression.Lambda(funcType, param, param);
DryadQueryNode hdistNode = new DryadHashPartitionNode(keySelectExpr,
null,
pcount,
queryExpr,
assignIndexNode);
// Apply node for (x, y) => ProcessWithIndex(x, y, procLambda));
Type paramType1 = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param1 = Expression.Parameter(paramType1, HpcLinqCodeGen.MakeUniqueName("x"));
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(typeof(int));
ParameterExpression param2 = Expression.Parameter(paramType2, HpcLinqCodeGen.MakeUniqueName("y"));
minfo = typeof(HpcLinqHelper).GetMethod("ProcessWithIndex");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0], procLambda.Body.Type);
body = Expression.Call(minfo, param1, param2, procLambda);
funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param1, param2);
return new DryadApplyNode(procFunc, queryExpr, child, hdistNode);
}
private DryadQueryNode VisitApply(QueryNodeInfo source1,
QueryNodeInfo source2,
LambdaExpression procLambda,
bool perPartition,
bool isFirstOnly,
Expression queryExpr)
{
DryadQueryNode child1 = this.Visit(source1);
DryadQueryNode applyNode;
if (source2 == null)
{
// Unary-apply case:
if (perPartition)
{
//homomorphic
applyNode = this.PromoteConcat(source1, child1, x => new DryadApplyNode(procLambda, queryExpr, x));
}
else
{
//non-homomorphic
if (child1.IsDynamic || child1.OutputPartition.Count > 1)
{
child1 = new DryadMergeNode(true, false, queryExpr, child1);
}
applyNode = new DryadApplyNode(procLambda, queryExpr, child1);
}
}
else
{
// Binary-apply case:
DryadQueryNode child2 = this.Visit(source2);
if (perPartition && isFirstOnly)
{
// The function is left homomorphic:
if (!child2.IsForked && (child1.IsDynamic || child1.OutputPartition.Count > 1))
{
// The normal cases..
if (IsMergeNodeNeeded(child2))
{
if (child1.IsDynamic)
{
child2 = new DryadMergeNode(true, false, queryExpr, child2);
child2.IsForked = true;
}
else
{
// Rather than do full merge and broadcast, which has lots of data-movement
// 1. Tee output2 with output cross-product
// 2. Do a merge-stage which will have input1.nPartition nodes each performing a merge.
// This acheives a distribution of the entire input2 to the Apply nodes with least data-movement.
child2 = new DryadTeeNode(child2.OutputTypes[0], true, queryExpr, child2);
child2.ConOpType = ConnectionOpType.CrossProduct;
child2 = new DryadMergeNode(child1.OutputPartition.Count, queryExpr, child2);
}
}
else
{
// the right-data is alread a single partition, so just tee it.
// this will provide a copy to each of the apply nodes.
child2 = new DryadTeeNode(child2.OutputTypes[0], true, queryExpr, child2);
}
}
else
{
// Less common cases..
// a full merge of the right-data may be necessary.
if (child2.IsDynamic || child2.OutputPartition.Count > 1)
{
child2 = new DryadMergeNode(true, false, queryExpr, child2);
if (child1.IsDynamic || child1.OutputPartition.Count > 1)
{
child2.IsForked = true;
}
}
}
applyNode = new DryadApplyNode(procLambda, queryExpr, child1, child2);
}
else if (perPartition && !isFirstOnly && !child1.IsDynamic && !child2.IsDynamic)
{
// Full homomorphic
// No merging occurs.
// NOTE: We generally expect that both the left and right datasets have matching partitionCount.
// however, we don't test for it yet as users might know what they are doing, and it makes
// LocalDebug inconsistent as LocalDebug doesn't throw in that situation.
applyNode = new DryadApplyNode(procLambda, queryExpr, child1, child2);
}
else
{
// Non-homomorphic
// Full merges of both data sets is necessary.
if (child1.IsDynamic || child1.OutputPartition.Count > 1)
{
child1 = new DryadMergeNode(true, false, queryExpr, child1);
}
if (child2.IsDynamic || child2.OutputPartition.Count > 1)
{
child2 = new DryadMergeNode(true, false, queryExpr, child2);
}
applyNode = new DryadApplyNode(procLambda, queryExpr, child1, child2);
}
}
return applyNode;
}
private DryadQueryNode VisitSetOperation(QueryNodeInfo source1,
QueryNodeInfo source2,
QueryNodeType nodeType,
Expression comparerExpr,
Expression queryExpr)
{
DryadQueryNode child1 = this.Visit(source1);
DryadQueryNode child2 = this.Visit(source2);
DryadQueryNode resNode = null;
Type keyType = child1.OutputTypes[0];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable, keyType),
queryExpr);
}
// The comparer object:
object comparer = null;
if (comparerExpr != null)
{
ExpressionSimplifier<object> evaluator = new ExpressionSimplifier<object>();
comparer = evaluator.Eval(comparerExpr);
}
LambdaExpression keySelectExpr = IdentityFunction.Instance(keyType);
if (child1.IsDynamic || child2.IsDynamic)
{
// Well, let us do the simplest thing for now
child1 = new DryadHashPartitionNode(keySelectExpr,
null,
StaticConfig.DefaultPartitionCount,
queryExpr,
child1);
if (IsMergeNodeNeeded(child1))
{
child1 = new DryadMergeNode(false, false, queryExpr, child1);
}
child2 = new DryadHashPartitionNode(keySelectExpr,
null,
StaticConfig.DefaultPartitionCount,
queryExpr,
child2);
if (IsMergeNodeNeeded(child2))
{
child2 = new DryadMergeNode(false, false, queryExpr, child2);
}
resNode = new DryadSetOperationNode(nodeType, nodeType.ToString(), comparerExpr,
queryExpr, child1, child2);
return resNode;
}
bool isDescending1 = child1.OutputDataSetInfo.orderByInfo.IsDescending;
bool isDescending2 = child2.OutputDataSetInfo.orderByInfo.IsDescending;
// Partition child1 and child2 if needed
if (child1.OutputPartition.ParType == PartitionType.Range &&
child1.OutputPartition.HasKeys &&
child1.OutputPartition.IsPartitionedBy(keySelectExpr, comparer))
{
if (child2.OutputPartition.ParType != PartitionType.Range ||
!child2.OutputPartition.IsPartitionedBy(keySelectExpr, comparer) ||
child1.OutputPartition.IsSamePartition(child2.OutputPartition))
{
// Range distribute child2 using child1's partition
child2 = child1.OutputPartition.CreatePartitionNode(keySelectExpr, child2);
if (IsMergeNodeNeeded(child2))
{
child2 = new DryadMergeNode(false, false, queryExpr, child2);
}
}
}
else if (child2.OutputPartition.ParType == PartitionType.Range &&
child2.OutputPartition.HasKeys &&
child2.OutputPartition.IsPartitionedBy(keySelectExpr, comparer))
{
// Range distribute child1 using child2's partition
child1 = child2.OutputPartition.CreatePartitionNode(keySelectExpr, child1);
if (IsMergeNodeNeeded(child1))
{
child1 = new DryadMergeNode(false, false, queryExpr, child1);
}
}
else if (child1.OutputPartition.ParType == PartitionType.Hash &&
child1.OutputPartition.IsPartitionedBy(keySelectExpr, comparer))
{
if (child2.OutputPartition.ParType != PartitionType.Hash ||
!child2.OutputPartition.IsPartitionedBy(keySelectExpr, comparer) ||
!child1.OutputPartition.IsSamePartition(child2.OutputPartition))
{
// Hash distribute child2:
child2 = new DryadHashPartitionNode(keySelectExpr,
comparerExpr,
child1.OutputPartition.Count,
queryExpr,
child2);
if (IsMergeNodeNeeded(child2))
{
child2 = new DryadMergeNode(false, false, queryExpr, child2);
}
}
}
else if (child2.OutputPartition.ParType == PartitionType.Hash &&
child2.OutputPartition.IsPartitionedBy(keySelectExpr, comparer))
{
child1 = new DryadHashPartitionNode(keySelectExpr,
comparerExpr,
child2.OutputPartition.Count,
queryExpr,
child1);
if (IsMergeNodeNeeded(child1))
{
child1 = new DryadMergeNode(false, false, queryExpr, child1);
}
}
else
{
// No luck. Hash distribute both child1 and child2, then perform hash operation
int parCnt = Math.Max(child1.OutputPartition.Count, child2.OutputPartition.Count);
if (parCnt > 1)
{
child1 = new DryadHashPartitionNode(keySelectExpr, comparerExpr, parCnt, queryExpr, child1);
if (IsMergeNodeNeeded(child1))
{
child1 = new DryadMergeNode(false, false, queryExpr, child1);
}
child2 = new DryadHashPartitionNode(keySelectExpr, comparerExpr, parCnt, queryExpr, child2);
if (IsMergeNodeNeeded(child2))
{
child2 = new DryadMergeNode(false, false, queryExpr, child2);
}
}
}
// Perform either hash or ordered operation
string opName = "";
if (child1.IsOrderedBy(keySelectExpr, comparer))
{
if (!child1.IsOrderedBy(keySelectExpr, comparer) ||
isDescending1 != isDescending2)
{
// Sort inner if unsorted
child2 = new DryadOrderByNode(keySelectExpr, comparerExpr, isDescending1, queryExpr, child2);
}
opName = "Ordered";
}
else if (child2.IsOrderedBy(keySelectExpr, comparer))
{
if (!child1.IsOrderedBy(keySelectExpr, comparer) ||
isDescending1 != isDescending2)
{
// Sort outer if unsorted
child1 = new DryadOrderByNode(keySelectExpr, comparerExpr, isDescending2, queryExpr, child1);
}
opName = "Ordered";
}
resNode = new DryadSetOperationNode(nodeType, opName + nodeType, comparerExpr, queryExpr, child1, child2);
return resNode;
}
private DryadQueryNode VisitJoin(QueryNodeInfo outerSource,
QueryNodeInfo innerSource,
QueryNodeType nodeType,
LambdaExpression outerKeySelector,
LambdaExpression innerKeySelector,
LambdaExpression resultSelector,
Expression comparerExpr,
Expression queryExpr)
{
DryadQueryNode outerChild = this.Visit(outerSource);
DryadQueryNode innerChild = this.Visit(innerSource);
DryadQueryNode joinNode = null;
Type keyType = outerKeySelector.Type.GetGenericArguments()[1];
if (comparerExpr == null && !TypeSystem.HasDefaultEqualityComparer(keyType))
{
throw DryadLinqException.Create(HpcLinqErrorCode.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable,
string.Format(SR.ComparerMustBeSpecifiedOrKeyTypeMustBeIEquatable, keyType),
queryExpr);
}
// The comparer object:
object comparer = null;
if (comparerExpr != null)
{
ExpressionSimplifier<object> evaluator = new ExpressionSimplifier<object>();
comparer = evaluator.Eval(comparerExpr);
}
if (outerChild.IsDynamic || innerChild.IsDynamic)
{
// Well, let us do the simplest thing for now
outerChild = new DryadHashPartitionNode(outerKeySelector,
comparerExpr,
StaticConfig.DefaultPartitionCount,
queryExpr,
outerChild);
if (IsMergeNodeNeeded(outerChild))
{
outerChild = new DryadMergeNode(false, false, queryExpr, outerChild);
}
innerChild = new DryadHashPartitionNode(innerKeySelector,
comparerExpr,
StaticConfig.DefaultPartitionCount,
queryExpr,
innerChild);
if (IsMergeNodeNeeded(innerChild))
{
innerChild = new DryadMergeNode(false, false, queryExpr, innerChild);
}
joinNode = new DryadJoinNode(nodeType,
"Hash" + nodeType,
outerKeySelector,
innerKeySelector,
resultSelector,
comparerExpr,
queryExpr,
outerChild,
innerChild);
return joinNode;
}
bool isOuterDescending = outerChild.OutputDataSetInfo.orderByInfo.IsDescending;
bool isInnerDescending = innerChild.OutputDataSetInfo.orderByInfo.IsDescending;
// Partition outer and inner if needed
if (outerChild.OutputPartition.ParType == PartitionType.Range &&
outerChild.OutputPartition.HasKeys &&
outerChild.OutputPartition.IsPartitionedBy(outerKeySelector, comparer))
{
if (innerChild.OutputPartition.ParType != PartitionType.Range ||
!innerChild.OutputPartition.IsPartitionedBy(innerKeySelector, comparer) ||
!outerChild.OutputPartition.IsSamePartition(innerChild.OutputPartition))
{
// Range distribute inner using outer's partition.
innerChild = outerChild.OutputPartition.CreatePartitionNode(innerKeySelector, innerChild);
if (IsMergeNodeNeeded(innerChild))
{
innerChild = new DryadMergeNode(false, false, queryExpr, innerChild);
}
}
}
else if (innerChild.OutputPartition.ParType == PartitionType.Range &&
innerChild.OutputPartition.HasKeys &&
innerChild.OutputPartition.IsPartitionedBy(innerKeySelector, comparer))
{
// Range distribute outer using inner's partition.
outerChild = innerChild.OutputPartition.CreatePartitionNode(outerKeySelector, outerChild);
if (IsMergeNodeNeeded(outerChild))
{
outerChild = new DryadMergeNode(false, false, queryExpr, outerChild);
}
}
else if (outerChild.OutputPartition.ParType == PartitionType.Hash &&
outerChild.OutputPartition.IsPartitionedBy(outerKeySelector, comparer))
{
if (innerChild.OutputPartition.ParType != PartitionType.Hash ||
!innerChild.OutputPartition.IsPartitionedBy(innerKeySelector, comparer) ||
!outerChild.OutputPartition.IsSamePartition(innerChild.OutputPartition))
{
innerChild = new DryadHashPartitionNode(innerKeySelector,
comparerExpr,
outerChild.OutputPartition.Count,
queryExpr,
innerChild);
if (IsMergeNodeNeeded(innerChild))
{
innerChild = new DryadMergeNode(false, false, queryExpr, innerChild);
}
}
}
else if (innerChild.OutputPartition.ParType == PartitionType.Hash &&
innerChild.OutputPartition.IsPartitionedBy(innerKeySelector, comparer))
{
outerChild = new DryadHashPartitionNode(outerKeySelector,
comparerExpr,
innerChild.OutputPartition.Count,
queryExpr,
outerChild);
if (IsMergeNodeNeeded(outerChild))
{
outerChild = new DryadMergeNode(false, false, queryExpr, outerChild);
}
}
else
{
// No luck. Hash partition both outer and inner
int parCnt = Math.Max(outerChild.OutputPartition.Count, innerChild.OutputPartition.Count);
if (parCnt > 1)
{
outerChild = new DryadHashPartitionNode(outerKeySelector,
comparerExpr,
parCnt,
queryExpr,
outerChild);
if (IsMergeNodeNeeded(outerChild))
{
outerChild = new DryadMergeNode(false, false, queryExpr, outerChild);
}
innerChild = new DryadHashPartitionNode(innerKeySelector,
comparerExpr,
parCnt,
queryExpr,
innerChild);
if (IsMergeNodeNeeded(innerChild))
{
innerChild = new DryadMergeNode(false, false, queryExpr, innerChild);
}
}
}
// Perform either merge or hash join
string opName = "Hash";
if (outerChild.IsOrderedBy(outerKeySelector, comparer))
{
if (!innerChild.IsOrderedBy(innerKeySelector, comparer) ||
isOuterDescending != isInnerDescending)
{
// Sort inner if unsorted
innerChild = new DryadOrderByNode(innerKeySelector, comparerExpr,
isOuterDescending, queryExpr, innerChild);
}
opName = "Merge";
}
else if (innerChild.IsOrderedBy(innerKeySelector, comparer))
{
if (!outerChild.IsOrderedBy(outerKeySelector, comparer) ||
isOuterDescending != isInnerDescending)
{
// Sort outer if unsorted
outerChild = new DryadOrderByNode(outerKeySelector, comparerExpr,
isInnerDescending, queryExpr, outerChild);
}
opName = "Merge";
}
joinNode = new DryadJoinNode(nodeType,
opName + nodeType,
outerKeySelector,
innerKeySelector,
resultSelector,
comparerExpr,
queryExpr,
outerChild,
innerChild);
return joinNode;
}
private DryadQueryNode VisitSlidingWindow(QueryNodeInfo source,
LambdaExpression procLambda,
Expression windowSizeExpr,
Expression queryExpr)
{
// var windows = source.Apply(s => HpcLinqHelper.Last(s, windowSize));
// var slided = windows.Apply(s => HpcLinqHelper.Slide(s)).HashPartition(x => x.Index);
// slided.Apply(source, (x, y) => HpcLinqHelper.ProcessWindows(x, y, procFunc, windowSize));
DryadQueryNode child = this.Visit(source);
if (child.IsDynamic)
{
throw new DryadLinqException("SlidingWindow is only supported for static partition count");
}
ExpressionSimplifier<int> evaluator = new ExpressionSimplifier<int>();
Expression windowSize = Expression.Constant(evaluator.Eval(windowSizeExpr), typeof(int));
child.IsForked = true;
// Apply node for s => Last(s, windowSize)
Type paramType = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param = Expression.Parameter(paramType, HpcLinqCodeGen.MakeUniqueName("s"));
MethodInfo minfo = typeof(HpcLinqHelper).GetMethod("Last");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
Expression body = Expression.Call(minfo, param, windowSize);
Type funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode lastNode = new DryadApplyNode(procFunc, queryExpr, child);
lastNode = new DryadMergeNode(true, true, queryExpr, lastNode);
// Apply node for s => Slide(s)
param = Expression.Parameter(body.Type, HpcLinqCodeGen.MakeUniqueName("s"));
minfo = typeof(HpcLinqHelper).GetMethod("Slide");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0]);
body = Expression.Call(minfo, param);
funcType = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param);
DryadQueryNode slideNode = new DryadApplyNode(procFunc, queryExpr, lastNode);
// Hash partition to distribute from partition i to i+1
int pcount = child.OutputPartition.Count;
param = Expression.Parameter(body.Type.GetGenericArguments()[0], "x");
Expression keySelectBody = Expression.Property(param, "Index");
funcType = typeof(Func<,>).MakeGenericType(param.Type, keySelectBody.Type);
LambdaExpression keySelectExpr = Expression.Lambda(funcType, keySelectBody, param);
DryadQueryNode hdistNode = new DryadHashPartitionNode(keySelectExpr,
null,
pcount,
queryExpr,
slideNode);
// Apply node for (x, y) => ProcessWindows(x, y, proclambda, windowSize)
Type paramType1 = typeof(IEnumerable<>).MakeGenericType(body.Type);
ParameterExpression param1 = Expression.Parameter(paramType1, HpcLinqCodeGen.MakeUniqueName("x"));
Type paramType2 = typeof(IEnumerable<>).MakeGenericType(child.OutputTypes[0]);
ParameterExpression param2 = Expression.Parameter(paramType2, HpcLinqCodeGen.MakeUniqueName("y"));
minfo = typeof(HpcLinqHelper).GetMethod("ProcessWindows");
minfo = minfo.MakeGenericMethod(child.OutputTypes[0], procLambda.Body.Type);
body = Expression.Call(minfo, param1, param2, procLambda, windowSize);
funcType = typeof(Func<,,>).MakeGenericType(param1.Type, param2.Type, body.Type);
procFunc = Expression.Lambda(funcType, body, param1, param2);
return new DryadApplyNode(procFunc, queryExpr, hdistNode, child);
}
private DryadQueryNode VisitMultiApply(QueryNodeInfo source,
LambdaExpression procLambda,
bool perPartition,
bool isFirstOnly,
MethodCallExpression queryExpr)
{
DryadQueryNode[] childs = new DryadQueryNode[source.children.Count];
for (int i = 0; i < source.children.Count; ++i)
{
childs[i] = this.Visit(source.children[i].child);
}
bool isDynamic = childs.Any(x => x.IsDynamic);
if (perPartition && !isDynamic)
{
// Homomorphic case.
if (isFirstOnly)
{
for (int i = 1; i < childs.Length; ++i)
{
childs[i] = new DryadTeeNode(childs[i].OutputTypes[0], true, queryExpr, childs[i]);
childs[i].ConOpType = ConnectionOpType.CrossProduct;
childs[i] = new DryadMergeNode(childs[0].OutputPartition.Count, queryExpr, childs[i]);
}
}
else
{
int count = childs[0].OutputPartition.Count;
for (int i = 1; i < childs.Length; ++i)
{
if (childs[i].OutputPartition.Count != count)
{
throw DryadLinqException.Create(HpcLinqErrorCode.HomomorphicApplyNeedsSamePartitionCount,
SR.HomomorphicApplyNeedsSamePartitionCount,
queryExpr);
}
}
}
}
else
{
// Non-homomorphic case.
for (int i = 0; i < childs.Length; ++i)
{
if (childs[i].IsDynamic || childs[i].OutputPartition.Count > 1)
{
childs[i] = new DryadMergeNode(true, false, queryExpr, childs[i]);
}
}
}
DryadQueryNode applyNode = new DryadApplyNode(procLambda, true, queryExpr, childs);
return applyNode;
}
private DryadQueryNode CreateOffset(bool isLong, Expression queryExpr, DryadQueryNode child)
{
// Count node
DryadQueryNode countNode = new DryadBasicAggregateNode(null, AggregateOpType.LongCount,
true, false, queryExpr, child);
// Apply node for x => Offsets(x)
Type paramType = typeof(IEnumerable<>).MakeGenericType(typeof(long));
ParameterExpression param = Expression.Parameter(paramType, "x");
MethodInfo minfo = typeof(HpcLinqEnumerable).GetMethod("Offsets");
Expression body = Expression.Call(minfo, param, Expression.Constant(isLong, typeof(bool)));
Type type = typeof(Func<,>).MakeGenericType(param.Type, body.Type);
LambdaExpression procFunc = Expression.Lambda(type, body, param);
DryadQueryNode mergeCountNode = new DryadMergeNode(true, true, queryExpr, countNode);
DryadQueryNode offsetsNode = new DryadApplyNode(procFunc, queryExpr, mergeCountNode);
// HashPartition
LambdaExpression keySelectExpr = IdentityFunction.Instance(typeof(IndexedValue<long>));
int pcount = child.OutputPartition.Count;
DryadQueryNode hdistNode = new DryadHashPartitionNode(keySelectExpr, null, null, pcount,
false, queryExpr, offsetsNode);
DryadQueryNode resNode = new DryadMergeNode(false, true, queryExpr, hdistNode);
return resNode;
}
internal virtual string Visit(DryadMergeNode node,
CodeMemberMethod vertexMethod,
string[] readerNames,
string[] writerNames)
{
return node.AddVertexCode(vertexMethod, readerNames, writerNames);
}