// -------------------------------------------------------------------------
//
// findNodes() Locate all the nodes of the specified type, starting
// at the specified root.
//
// -------------------------------------------------------------------------
internal void FindNodes(IList dest, int kind)
{
if (fType == kind)
{
ILOG.J2CsMapping.Collections.Generics.Collections.Add(dest, this);
}
if (fLeftChild != null)
{
fLeftChild.FindNodes(dest, kind);
}
if (fRightChild != null)
{
fRightChild.FindNodes(dest, kind);
}
}
// -----------------------------------------------------------------------------
//
// calcChainedFollowPos. Modify the previously calculated followPos sets
// to implement rule chaining. NOT described by Aho
//
// -----------------------------------------------------------------------------
internal void CalcChainedFollowPos(RBBINode tree)
{
IList endMarkerNodes = new ArrayList();
IList leafNodes = new ArrayList();
// get a list of all endmarker nodes.
tree.FindNodes(endMarkerNodes, IBM.ICU.Text.RBBINode.endMark);
// get a list all leaf nodes
tree.FindNodes(leafNodes, IBM.ICU.Text.RBBINode.leafChar);
// Get all nodes that can be the start a match, which is FirstPosition()
// of the portion of the tree corresponding to user-written rules.
// See the tree description in bofFixup().
RBBINode userRuleRoot = tree;
if (fRB.fSetBuilder.SawBOF())
{
userRuleRoot = tree.fLeftChild.fRightChild;
}
IBM.ICU.Impl.Assert.Assrt(userRuleRoot != null);
ILOG.J2CsMapping.Collections.ISet matchStartNodes = userRuleRoot.fFirstPosSet;
// Iteratate over all leaf nodes,
//
IIterator endNodeIx = new ILOG.J2CsMapping.Collections.IteratorAdapter(leafNodes.GetEnumerator());
while (endNodeIx.HasNext())
{
RBBINode tNode = (RBBINode)endNodeIx.Next();
RBBINode endNode = null;
// Identify leaf nodes that correspond to overall rule match
// positions.
// These include an endMarkerNode in their followPos sets.
IIterator i = new ILOG.J2CsMapping.Collections.IteratorAdapter(endMarkerNodes.GetEnumerator());
while (i.HasNext())
{
RBBINode endMarkerNode = (RBBINode)i.Next();
if (ILOG.J2CsMapping.Collections.Collections.Contains(endMarkerNode, tNode.fFollowPos))
{
endNode = tNode;
break;
}
}
if (endNode == null)
{
// node wasn't an end node. Try again with the next.
continue;
}
// We've got a node that can end a match.
// Line Break Specific hack: If this node's val correspond to the
// $CM char class,
// don't chain from it.
// TODO: Add rule syntax for this behavior, get specifics out of
// here and
// into the rule file.
if (fRB.fLBCMNoChain)
{
int c = this.fRB.fSetBuilder.GetFirstChar(endNode.fVal);
if (c != -1)
{
// c == -1 occurs with sets containing only the {eof} marker
// string.
int cLBProp = IBM.ICU.Lang.UCharacter.GetIntPropertyValue(c,
IBM.ICU.Lang.UProperty_Constants.LINE_BREAK);
if (cLBProp == IBM.ICU.Lang.UCharacter.LineBreak.COMBINING_MARK)
{
continue;
}
}
}
// Now iterate over the nodes that can start a match, looking for
// ones
// with the same char class as our ending node.
RBBINode startNode;
IIterator startNodeIx = new ILOG.J2CsMapping.Collections.IteratorAdapter(matchStartNodes.GetEnumerator());
while (startNodeIx.HasNext())
{
startNode = (RBBINode)startNodeIx.Next();
if (startNode.fType != IBM.ICU.Text.RBBINode.leafChar)
{
continue;
}
if (endNode.fVal == startNode.fVal)
{
// The end val (character class) of one possible match is
// the
// same as the start of another.
// Add all nodes from the followPos of the start node to the
// followPos set of the end node, which will have the effect
// of
// letting matches transition from a match state at endNode
// to the second char of a match starting with startNode.
ILOG.J2CsMapping.Collections.Generics.Collections.AddAll(startNode.fFollowPos, endNode.fFollowPos);
}
}
}
}
