private static QueryParserNode ProcessTerm()
{
QueryParserNode nodeTerm;
QueryParserNode result = nodeTerm = ProcessPrimaryLevel();
GetNextToken();
if (strToken == "[")
{
SectionNode nodeOp = new SectionNode();
GetNextToken();
nodeOp.SectionName = strToken;
nodeOp.AddOperand(nodeTerm);
_errorMessage = "Error in the query syntax - expected ']' symbol in document section specifier";
Expect("]");
_errorMessage = null;
result = nodeOp;
}
else
{
BackToken();
}
return(result);
}
public static QueryPostfixForm ParseQuery(string query)
{
Initialize();
_query = query.Trim().ToLower();
QueryPostfixForm form;
try
{
QueryParserNode nodeRoot = ParseExpression();
QueryParserNode newRoot = null;
PropagateSectionOpInside(ref newRoot, nodeRoot);
if (newRoot == null)
{
newRoot = nodeRoot;
}
form = new QueryPostfixForm();
Tree2Postfix(newRoot, form);
}
catch (Exception)
{
form = null;
}
return(form);
}
private static void PropagateSectionOpInside(ref QueryParserNode parent, QueryParserNode subtreeRoot)
{
if (subtreeRoot.NodeType == QueryParserNode.Type.eoTerm)
{
return;
}
if (subtreeRoot.NodeType == QueryParserNode.Type.eoSection)
{
Debug.Assert(((OpNode)subtreeRoot).BranchesNumber == 1, "Illegal parsing of section operator");
QueryParserNode child = ((OpNode)subtreeRoot)[0];
ArrayList branches;
// if section op is applied to the tree - perform conversion -
// propagate section op into every operand of the tree, and
// tree root becomes new local root.
if (child.NodeType != QueryParserNode.Type.eoTerm &&
child.NodeType != QueryParserNode.Type.eoSection)
{
//-------------------------------------------------------------
if (parent == null)
{
parent = child;
}
else
{
((OpNode)parent).RemoveOperand(subtreeRoot);
((OpNode)parent).AddOperand(child);
}
//-------------------------------------------------------------
branches = ((OpNode)child).Branches();
for (int i = 0; i < branches.Count; i++)
{
SectionNode newSectionOp = new SectionNode();
newSectionOp.SectionName = ((SectionNode)subtreeRoot).SectionName;
newSectionOp.AddOperand((QueryParserNode)branches[i]);
branches[i] = newSectionOp;
}
((OpNode)child).SetOperands(branches);
}
//-----------------------------------------------------------------
if (child.NodeType != QueryParserNode.Type.eoTerm)
{
branches = ((OpNode)child).Branches();
for (int i = 0; i < branches.Count; i++)
{
PropagateSectionOpInside(ref child, (QueryParserNode)branches[i]);
}
}
}
}
private static QueryParserNode ParseExpression()
{
QueryParserNode nodeLeftOperand = ProcessTerm();
GetNextToken();
//---------------------------------------------------------------------
// if next token == "" => EOS, no more processing, return left operand
// as result,
// if next token == ")" => we reached the endo of parenthed expression,
// return left operand as result, else process two operands and Op,
// return Op as root.
//---------------------------------------------------------------------
if ((strToken == "") || (strToken == ")") || (strToken == "+\""))
{
BackToken();
return(nodeLeftOperand);
}
QueryParserNode.Type typeNode;
if (!isPhrasalMode)
{
switch (strToken)
{
case "and": typeNode = QueryParserNode.Type.eoAnd; break;
case "or": typeNode = QueryParserNode.Type.eoOr; break;
case "near": typeNode = QueryParserNode.Type.eoNear; break;
default: typeNode = QueryParserNode.Type.eoAnd; BackToken(); break;
}
}
else
{
typeNode = QueryParserNode.Type.eoPhraseNear;
BackToken();
}
OpNode nodeOp = new OpNode(typeNode);
QueryParserNode nodeRightOperand = ParseExpression();
nodeOp.AddOperand(nodeLeftOperand);
nodeOp.AddOperand(nodeRightOperand);
return(nodeOp);
}
private static void Tree2Postfix(QueryParserNode root, QueryPostfixForm form)
{
if (root == null)
{
return;
}
if (root.NodeType != QueryParserNode.Type.eoTerm)
{
foreach (QueryParserNode node in ((OpNode)root).Branches())
{
Tree2Postfix(node, form);
}
}
form.Add(root);
}
private static void BinaryOp(QueryParserNode node, Stack opStack)
{
#region Preconditions
if (opStack.Count < 2)
{
throw new ApplicationException("QueryProcessor -- Insufficient number of operands in the operating stack");
}
#endregion Preconditions
// First, check One or both arguments to be stopwords.
object o1 = opStack.Pop(), o2 = opStack.Peek();
opStack.Push(o1);
// If both arguments are stopwords - push them both, leave the stopword
// sign as the result for subsequent calculations.
// If only one argument is stopword - leave its counterpart as the result.
if (o1 is StopwordTerm)
{
opStack.Pop();
}
else
if (o2 is StopwordTerm)
{
opStack.Pop(); // o1
opStack.Pop(); // o2
opStack.Push(o1); // o1 again instead of the couple.
}
// both arguments are normal terms which might be met in the index.
else
{
Entry[] result;
Entry[] rightIndices = ExtractOperandFromStack(opStack);
Entry[] leftIndices = ExtractOperandFromStack(opStack);
if (node.NodeType == QueryParserNode.Type.eoOr)
{
result = Join(leftIndices, rightIndices);
}
else
{
result = Intercross(leftIndices, rightIndices, ((OpNode)node).RequiredProximity);
}
opStack.Push(result);
}
}
private static void IteratePostfixExpression(IList <QueryParserNode> postfixForm, IntHashTable tokens, Stack <List <long> > opStack)
{
for (int i = 0; i < postfixForm.Count; i++)
{
QueryParserNode node = postfixForm[i];
switch (node.NodeType)
{
case QueryParserNode.Type.eoTerm:
PushTermOnStack(((TermNode)node).Term, tokens, opStack); break;
case QueryParserNode.Type.eoSection:
UnarySectionOp(((SectionNode)node).SectionName, opStack); break;
default:
BinaryOp(node, opStack); break;
}
}
}
private static void IteratePostfixExpression(IList <QueryParserNode> postfixForm, TermIndexAccessor termIndex, Stack opStack)
{
for (int i = 0; i < postfixForm.Count; i++)
{
QueryParserNode node = postfixForm[i];
if (node.NodeType == QueryParserNode.Type.eoTerm)
{
PushTermOnStack(((TermNode)node).Term, termIndex, opStack);
}
else
if (node.NodeType == QueryParserNode.Type.eoSection)
{
UnarySectionOp(((SectionNode)node).SectionName, opStack);
}
else
{
BinaryOp(node, opStack);
}
}
}
private static QueryParserNode ProcessPrimaryLevel()
{
QueryParserNode nodePrimaryToken = null;
GetNextToken();
if (strToken != "")
{
if (strToken == "(")
{
nodePrimaryToken = ParseExpression();
_errorMessage = "Error in the query syntax - expected ')' symbol in expression";
Expect(")");
_errorMessage = null;
}
else
if (strToken == "\"+")
{
isPhrasalMode = true;
nodePrimaryToken = ParseExpression();
Expect("+\"");
isPhrasalMode = false;
}
else
{
string normToken = OMEnv.LexemeConstructor.GetNormalizedToken(strToken);
if (isDelimitableToken(normToken))
{
nodePrimaryToken = SplitTokenToTree(normToken);
}
else
{
nodePrimaryToken = new TermNode(normToken);
}
}
}
return(nodePrimaryToken);
}
private static void BinaryOp(QueryParserNode node, Stack <List <long> > opStack)
{
#region Preconditions
if (opStack.Count < 2)
{
throw new ApplicationException("QueryProcessor -- Insufficient number of operands in the operating stack");
}
#endregion Preconditions
List <long> result;
List <long> right = opStack.Pop(), left = opStack.Pop();
if (node.NodeType == QueryParserNode.Type.eoOr)
{
result = Join(left, right);
}
else
{
result = Intercross(left, right, ((OpNode)node).RequiredProximity);
}
opStack.Push(result);
}
public void RemoveOperand(QueryParserNode node)
{
aOperands.Remove(node);
}
public void AddOperand(QueryParserNode node)
{
aOperands.Add(node);
}