public ColumnStatistics GetLeafToLeafStatistics(MatchEdge nodeEdge, MatchEdge componentEdge)
{
var edgeTuple = new Tuple <string, string>(nodeEdge.EdgeAlias, componentEdge.EdgeAlias);
if (LeafToLeafSelectivity.ContainsKey(edgeTuple))
{
return(LeafToLeafSelectivity[edgeTuple]);
}
var mergedStatistics = ColumnStatistics.UpdateHistogram(Context.GetEdgeStatistics(nodeEdge), Context.GetEdgeStatistics(componentEdge));
LeafToLeafSelectivity[edgeTuple] = mergedStatistics;
return(mergedStatistics);
}
/// <summary>
/// Transit from current component to the new component in the next state given the Node Unit
/// </summary>
/// <param name="candidateTree"></param>
/// <param name="densityDict"></param>
/// <param name="subGraph"></param>
/// <param name="statisticsCalculator"></param>
/// <returns></returns>
public MatchComponent GetNextState(
OneHeightTree candidateTree,
Dictionary <string, double> densityDict,
IMatchJoinStatisticsCalculator statisticsCalculator)
{
var newComponent = new MatchComponent(this);
var root = candidateTree.TreeRoot;
WBooleanExpression joinCondition = null;
string nodeName = "";
// Update Nodes
if (newComponent.MaterializedNodeSplitCount.ContainsKey(root))
{
newComponent.MaterializedNodeSplitCount[root]++;
nodeName = newComponent.GetNodeRefName(root);
joinCondition = new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = root.RefAlias
},
new Identifier {
Value = "GlobalNodeId"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = nodeName
},
new Identifier {
Value = "GlobalNodeId"
}
),
},
ComparisonType = BooleanComparisonType.Equals
};
}
else
{
nodeName = root.RefAlias;
newComponent.Nodes.Add(root);
newComponent.MaterializedNodeSplitCount[root] = 0;
newComponent.StatisticsDict[root] = new ColumnStatistics {
Selectivity = 1.0 / root.TableRowCount
};
}
// Constructs table reference
WTableReference nodeTable = new WNamedTableReference
{
Alias = new Identifier {
Value = nodeName
},
TableObjectName = root.TableObjectName
};
WTableReference compTable = newComponent.TableRef;
// Updates join conditions
double selectivity = 1.0;
double degrees = 1.0;
var DensityCount = new Dictionary <string, int>(StringComparer.CurrentCultureIgnoreCase);
List <MatchEdge> inEdges;
if (newComponent.UnmaterializedNodeMapping.TryGetValue(root, out inEdges))
{
var firstEdge = inEdges.First();
bool materialized = newComponent.EdgeMaterilizedDict[firstEdge];
newComponent.UnmaterializedNodeMapping.Remove(root);
selectivity *= 1.0 / root.TableRowCount;
// Component materialized edge to root
if (materialized)
{
joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition, new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = firstEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = nodeName
},
new Identifier {
Value = "GlobalNodeId"
}
)
},
ComparisonType = BooleanComparisonType.Equals
});
//var statistics = ColumnStatistics.UpdateHistogram(newComponent.StatisticsDict[root],
// new ColumnStatistics {Selectivity = 1.0/root.TableRowCount});
//selectivity *= statistics.Selectivity;
//newComponent.StatisticsDict[root] = statistics;
if (DensityCount.ContainsKey(root.TableObjectName.ToString()))
{
DensityCount[root.TableObjectName.ToString()]++;
}
else
{
DensityCount[root.TableObjectName.ToString()] = 1;
}
}
// Component unmaterialized edge to root
else
{
ColumnStatistics statistics = null;
foreach (var edge in inEdges)
{
// Update component table
compTable = SpanTableRef(compTable, edge, newComponent.GetNodeRefName(edge.SourceNode));
newComponent.EdgeMaterilizedDict[edge] = true;
joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = edge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = nodeName
},
new Identifier {
Value = "GlobalNodeId"
}
)
},
ComparisonType = BooleanComparisonType.Equals
});
statistics = ColumnStatistics.UpdateHistogram(statistics,
newComponent.Context.GetEdgeStatistics(edge));
selectivity *= statistics.Selectivity;
}
newComponent.StatisticsDict[root] = statistics;
if (DensityCount.ContainsKey(root.TableObjectName.ToString()))
{
DensityCount[root.TableObjectName.ToString()] += inEdges.Count;
}
else
{
DensityCount[root.TableObjectName.ToString()] = inEdges.Count;
}
}
}
var jointEdges = candidateTree.MaterializedEdges;
int sinkToSinkCount = 0;
foreach (var jointEdge in jointEdges)
{
// Update node table
nodeTable = SpanTableRef(nodeTable, jointEdge, nodeName);
degrees *= jointEdge.AverageDegree;
newComponent.EdgeMaterilizedDict[jointEdge] = true;
var sinkNode = jointEdge.SinkNode;
// Leaf to component materialized node
if (newComponent.MaterializedNodeSplitCount.ContainsKey(sinkNode))
{
joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = jointEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = sinkNode.RefAlias
},
new Identifier {
Value = "GlobalNodeId"
}
)
},
ComparisonType = BooleanComparisonType.Equals
});
var statistics = ColumnStatistics.UpdateHistogram(newComponent.StatisticsDict[sinkNode],
newComponent.Context.GetEdgeStatistics(jointEdge));
selectivity *= statistics.Selectivity;
newComponent.StatisticsDict[sinkNode] = statistics;
if (DensityCount.ContainsKey(sinkNode.TableObjectName.ToString()))
{
DensityCount[sinkNode.TableObjectName.ToString()]++;
}
else
{
DensityCount[sinkNode.TableObjectName.ToString()] = 1;
}
}
// Leaf to component unmaterialized node
else
{
inEdges = newComponent.UnmaterializedNodeMapping[sinkNode];
var firstEdge = inEdges.First();
bool materlizedEdge = newComponent.EdgeMaterilizedDict[firstEdge];
// Leaf to materialized leaf
if (materlizedEdge)
{
joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = jointEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = firstEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
)
},
ComparisonType = BooleanComparisonType.Equals
});
sinkToSinkCount++;
var statistics = ColumnStatistics.UpdateHistogram(newComponent.StatisticsDict[sinkNode],
newComponent.Context.GetEdgeStatistics(jointEdge));
selectivity *= statistics.Selectivity;
newComponent.StatisticsDict[sinkNode] = statistics;
}
// Leaf to unmaterialized leaf
else
{
ColumnStatistics compSinkNodeStatistics = null;
foreach (var inEdge in inEdges)
{
compTable = SpanTableRef(compTable, inEdge, newComponent.GetNodeRefName(inEdge.SourceNode));
newComponent.EdgeMaterilizedDict[inEdge] = true;
joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
new WBooleanComparisonExpression
{
FirstExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = jointEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
),
},
SecondExpr = new WColumnReferenceExpression
{
ColumnType = ColumnType.Regular,
MultiPartIdentifier = new WMultiPartIdentifier(
new Identifier {
Value = inEdge.EdgeAlias
},
new Identifier {
Value = "Sink"
}
)
},
ComparisonType = BooleanComparisonType.Equals
});
sinkToSinkCount++;
var leafToLeafStatistics = statisticsCalculator.GetLeafToLeafStatistics(jointEdge, inEdge);
selectivity *= leafToLeafStatistics.Selectivity;
compSinkNodeStatistics =
ColumnStatistics.UpdateHistogram(compSinkNodeStatistics,
newComponent.Context.GetEdgeStatistics(inEdge));
}
newComponent.StatisticsDict[sinkNode] = compSinkNodeStatistics;
}
}
}
var unmatEdges = candidateTree.UnmaterializedEdges;
foreach (var unmatEdge in unmatEdges)
{
newComponent.EdgeMaterilizedDict[unmatEdge] = false;
newComponent.Nodes.Add(unmatEdge.SinkNode);
var sinkNodeInEdges = newComponent.UnmaterializedNodeMapping.GetOrCreate(unmatEdge.SinkNode);
sinkNodeInEdges.Add(unmatEdge);
degrees *= unmatEdge.AverageDegree;
}
// Calculate Estimated Join Selectivity & Estimated Node Size
double estimatedSelectity = 1.0;
int count = 0;
bool sinkJoin = false;
foreach (var item in densityDict.Where(e => DensityCount.ContainsKey(e.Key)))
{
var density = item.Value;
var curJoinCount = DensityCount[item.Key];
var curJoinSelectitivy = Math.Pow(density, 2 - Math.Pow(2, 1 - curJoinCount));
if (!sinkJoin && ColumnStatistics.DefaultDensity < density)
{
var curSinkJoinSelectivity = Math.Pow(ColumnStatistics.DefaultDensity,
2 - Math.Pow(2, 1 - sinkToSinkCount));
estimatedSelectity *= Math.Pow(curSinkJoinSelectivity, Math.Pow(2, -count));
count += sinkToSinkCount;
sinkJoin = true;
}
estimatedSelectity *= Math.Pow(curJoinSelectitivy, Math.Pow(2, -count));
count += curJoinCount;
}
var estimatedNodeUnitSize = root.EstimatedRows *
Math.Pow(1000, candidateTree.MaterializedEdges.Count + candidateTree.UnmaterializedEdges.Count);
// Update Table Reference
newComponent.TableRef = GetPlanAndUpdateCost(candidateTree, newComponent, nodeTable, compTable, joinCondition,
degrees, selectivity, estimatedNodeUnitSize, estimatedSelectity);
return(newComponent);
}
