| package de.ids_mannheim.korap.query.serialize; |
| |
| import java.lang.reflect.Method; |
| import java.util.ArrayList; |
| import java.util.Arrays; |
| import java.util.HashMap; |
| import java.util.LinkedHashMap; |
| import java.util.LinkedList; |
| import java.util.List; |
| import java.util.Map; |
| import java.util.NoSuchElementException; |
| |
| import org.antlr.v4.runtime.ANTLRInputStream; |
| import org.antlr.v4.runtime.BailErrorStrategy; |
| import org.antlr.v4.runtime.CharStream; |
| import org.antlr.v4.runtime.CommonTokenStream; |
| import org.antlr.v4.runtime.Lexer; |
| import org.antlr.v4.runtime.ParserRuleContext; |
| import org.antlr.v4.runtime.tree.ParseTree; |
| import org.slf4j.LoggerFactory; |
| import org.slf4j.Logger; |
| |
| import de.ids_mannheim.korap.query.annis.AqlLexer; |
| import de.ids_mannheim.korap.query.annis.AqlParser; |
| import de.ids_mannheim.korap.util.QueryException; |
| |
| /** |
| * Map representation of ANNIS QL syntax tree as returned by ANTLR |
| * @author joachim |
| * |
| */ |
| public class AqlTree extends Antlr4AbstractSyntaxTree { |
| private static Logger log = LoggerFactory.getLogger(AqlTree.class); |
| /** |
| * Flag that indicates whether token fields or meta fields are currently being processed |
| */ |
| boolean inMeta = false; |
| /** |
| * Keeps track of operands that are to be integrated into yet uncreated objects. |
| */ |
| LinkedList<LinkedHashMap<String,Object>> operandStack = new LinkedList<LinkedHashMap<String,Object>>(); |
| /** |
| * Keeps track of explicitly (by #-var definition) or implicitly (number as reference) introduced entities (for later reference by #-operator) |
| */ |
| Map<String, LinkedHashMap<String,Object>> variableReferences = new LinkedHashMap<String, LinkedHashMap<String,Object>>(); |
| /** |
| * Counter for variable definitions. |
| */ |
| Integer variableCounter = 1; |
| /** |
| * Marks the currently active token in order to know where to add flags (might already have been taken away from token stack). |
| */ |
| LinkedHashMap<String,Object> curToken = new LinkedHashMap<String,Object>(); |
| /** |
| * Keeps track of operands lists that are to be serialised in an inverted |
| * order (e.g. the IN() operator) compared to their AST representation. |
| */ |
| private LinkedList<ArrayList<Object>> invertedOperandsLists = new LinkedList<ArrayList<Object>>(); |
| /** |
| * Keeps track of operation:class numbers. |
| */ |
| int classCounter = 0; |
| /** |
| * Keeps track of numers of relations processed (important when dealing with multiple predications). |
| */ |
| int relationCounter = 0; |
| /** |
| * Keeps track of references to nodes that are operands of groups (e.g. tree relations). Those nodes appear on the top level of the parse tree |
| * but are to be integrated into the AqlTree at a later point (namely as operands of the respective group). Therefore, store references to these |
| * nodes here and exclude the operands from being written into the query map individually. |
| */ |
| private LinkedList<String> operandOnlyNodeRefs = new LinkedList<String>(); |
| private List<ParseTree> globalLingTermNodes = new ArrayList<ParseTree>(); |
| private int totalRelationCount; |
| /** |
| * Keeps a record of reference-class-mapping, i.e. which 'class' has been assigned to which #n reference. This is important when introducing korap:reference |
| * spans to refer back to previously established classes for entities. |
| */ |
| private LinkedHashMap<String, Integer> refClassMapping = new LinkedHashMap<String, Integer>(); |
| private LinkedHashMap<String, Integer> nodeReferencesTotal = new LinkedHashMap<String, Integer>(); |
| private LinkedHashMap<String, Integer> nodeReferencesProcessed = new LinkedHashMap<String, Integer>(); |
| |
| /** |
| * |
| * @param tree The syntax tree as returned by ANTLR |
| * @param parser The ANTLR parser instance that generated the parse tree |
| */ |
| public AqlTree(String query) { |
| try { |
| process(query); |
| } catch (QueryException e) { |
| e.printStackTrace(); |
| } |
| System.out.println(">>> "+requestMap.get("query")+" <<<"); |
| } |
| |
| @Override |
| public void process(String query) throws QueryException { |
| ParseTree tree = parseAnnisQuery(query); |
| if (this.parser != null) { |
| super.parser = this.parser; |
| } else { |
| throw new NullPointerException("Parser has not been instantiated!"); |
| } |
| log.info("Processing Annis query."); |
| log.info("AST is: "+tree.toStringTree(parser)); |
| System.out.println("Processing Annis QL"); |
| if (verbose) System.out.println(tree.toStringTree(parser)); |
| processNode(tree); |
| log.info(requestMap.toString()); |
| } |
| |
| @SuppressWarnings("unchecked") |
| private void processNode(ParseTree node) { |
| // Top-down processing |
| if (visited.contains(node)) return; |
| else visited.add(node); |
| |
| String nodeCat = getNodeCat(node); |
| openNodeCats.push(nodeCat); |
| |
| stackedObjects = 0; |
| |
| if (verbose) { |
| System.err.println(" "+objectStack); |
| System.out.println(openNodeCats); |
| } |
| |
| /* |
| **************************************************************** |
| **************************************************************** |
| * Processing individual node categories * |
| **************************************************************** |
| **************************************************************** |
| */ |
| if (nodeCat.equals("exprTop")) { |
| List<ParseTree> andTopExprs = getChildrenWithCat(node, "andTopExpr"); |
| if (andTopExprs.size() > 1) { |
| LinkedHashMap<String, Object> topOr = makeGroup("or"); |
| requestMap.put("query", topOr); |
| objectStack.push(topOr); |
| } |
| } |
| |
| if (nodeCat.equals("andTopExpr")) { |
| // Before processing any child expr node, check if it has one or more "*ary_linguistic_term" nodes. |
| // Those nodes may use references to earlier established operand nodes. |
| // Those operand nodes are not to be included into the query map individually but |
| // naturally as operands of the relations/groups introduced by the |
| // *node. For that purpose, this section mines all used references |
| // and stores them in a list for later reference. |
| for (ParseTree exprNode : getChildrenWithCat(node,"expr")) { |
| // Pre-process any 'variableExpr' such that the variableReferences map can be filled |
| List<ParseTree> definitionNodes = new ArrayList<ParseTree>(); |
| definitionNodes.addAll(getChildrenWithCat(exprNode, "variableExpr")); |
| for (ParseTree definitionNode : definitionNodes) { |
| processNode(definitionNode); |
| } |
| // Then, mine all relations between nodes |
| List<ParseTree> lingTermNodes = new ArrayList<ParseTree>(); |
| lingTermNodes.addAll(getChildrenWithCat(exprNode, "n_ary_linguistic_term")); |
| globalLingTermNodes.addAll(lingTermNodes); |
| totalRelationCount = globalLingTermNodes.size(); |
| // Traverse refOrNode nodes under *ary_linguistic_term nodes and extract references |
| for (ParseTree lingTermNode : lingTermNodes) { |
| for (ParseTree refOrNode : getChildrenWithCat(lingTermNode, "refOrNode")) { |
| String refOrNodeString = refOrNode.getChild(0).toStringTree(parser); |
| if (refOrNodeString.startsWith("#")) { |
| String ref = refOrNode.getChild(0).toStringTree(parser).substring(1); |
| if (nodeReferencesTotal.containsKey(ref)) { |
| nodeReferencesTotal.put(ref, nodeReferencesTotal.get(ref)+1); |
| } else { |
| nodeReferencesTotal.put(ref, 1); |
| nodeReferencesProcessed.put(ref, 0); |
| } |
| } |
| } |
| } |
| } |
| } |
| |
| if (nodeCat.equals("unary_linguistic_term")) { |
| LinkedHashMap<String, Object> unaryOperator = parseUnaryOperator(node); |
| String reference = node.getChild(0).toStringTree(parser).substring(1); |
| LinkedHashMap<String, Object> object = variableReferences.get(reference); |
| object.putAll(unaryOperator); |
| } |
| |
| if (nodeCat.equals("n_ary_linguistic_term")) { |
| processN_ary_linguistic_term(node); |
| } |
| |
| if (nodeCat.equals("variableExpr")) { |
| // simplex word or complex assignment (like qname = textSpec)? |
| String firstChildNodeCat = getNodeCat(node.getChild(0)); |
| LinkedHashMap<String, Object> object = null; |
| if (firstChildNodeCat.equals("node")) { |
| object = makeSpan(); |
| } else if (firstChildNodeCat.equals("tok")) { |
| object = makeToken(); |
| if (node.getChildCount() > 1) { // empty tokens do not wrap a term |
| LinkedHashMap<String, Object> term = makeTerm(); |
| term.put("layer", "orth"); |
| object.put("wrap", term); |
| } |
| } else if (firstChildNodeCat.equals("qName")) { // only (foundry/)?layer specified |
| // may be token or span, depending on indicated layer! (e.g. cnx/cat=NP or mate/pos=NN) |
| HashMap<String, Object> qNameParse = parseQNameNode(node.getChild(0)); |
| if (Arrays.asList(new String[]{"pos", "lemma", "morph", "tok"}).contains(qNameParse.get("layer"))) { |
| object = makeToken(); |
| LinkedHashMap<String, Object> term = makeTerm(); |
| object.put("wrap", term); |
| term.putAll(qNameParse); |
| } else { |
| object = makeSpan(); |
| object.putAll(qNameParse); |
| } |
| } else if (firstChildNodeCat.equals("textSpec")) { |
| object = makeToken(); |
| LinkedHashMap<String, Object> term = makeTerm(); |
| object.put("wrap", term); |
| term.put("layer", "orth"); |
| term.putAll(parseTextSpec(node.getChild(0))); |
| } |
| |
| if (node.getChildCount() == 3) { // (foundry/)?layer=key specification |
| if (object.get("@type").equals("korap:token")) { |
| HashMap<String, Object> term = (HashMap<String, Object>) object.get("wrap"); |
| term.putAll(parseTextSpec(node.getChild(2))); |
| term.put("match", parseMatchOperator(node.getChild(1))); |
| } else { |
| object.putAll(parseTextSpec(node.getChild(2))); |
| object.put("match", parseMatchOperator(node.getChild(1))); |
| } |
| } |
| |
| if (object != null) { |
| if (! operandOnlyNodeRefs.contains(variableCounter.toString())) { |
| putIntoSuperObject(object); |
| } |
| ParseTree parentsFirstChild = node.getParent().getChild(0); |
| if (getNodeCat(parentsFirstChild).endsWith("#")) { |
| variableReferences.put(getNodeCat(parentsFirstChild).replaceAll("#", ""), object); |
| } |
| variableReferences.put(variableCounter.toString(), object); |
| variableCounter++; |
| System.out.println(variableReferences); |
| } |
| } |
| |
| objectsToPop.push(stackedObjects); |
| |
| /* |
| **************************************************************** |
| **************************************************************** |
| * recursion until 'request' node (root of tree) is processed * |
| **************************************************************** |
| **************************************************************** |
| */ |
| for (int i=0; i<node.getChildCount(); i++) { |
| ParseTree child = node.getChild(i); |
| processNode(child); |
| } |
| |
| /* |
| ************************************************************** |
| * Stuff that happens after processing the children of a node * |
| ************************************************************** |
| */ |
| if (!objectsToPop.isEmpty()) { |
| for (int i=0; i<objectsToPop.pop(); i++) { |
| objectStack.pop(); |
| } |
| } |
| openNodeCats.pop(); |
| } |
| |
| |
| |
| /** |
| * Processes an operand node, creating a map for the operand containing all its information |
| * given in the node definition (referenced via '#'). If this node has been referred to and used earlier, |
| * a korap:reference is created in its place. |
| * The operand will be wrapped in a class group if necessary. |
| * @param operandTree |
| * @return A map object with the appropriate CQLF representation of the operand |
| */ |
| private LinkedHashMap<String, Object> retrieveOperand(ParseTree operandTree) { |
| LinkedHashMap<String, Object> operand = null; |
| if (!getNodeCat(operandTree.getChild(0)).equals("variableExpr")) { |
| String ref = operandTree.getChild(0).toStringTree(parser).substring(1); |
| operand = variableReferences.get(ref); |
| if (nodeReferencesTotal.get(ref) > 1) { |
| if (nodeReferencesProcessed.get(ref)==0) { |
| refClassMapping.put(ref, classCounter); |
| operand = wrapInClass(operand, classCounter++); |
| nodeReferencesProcessed.put(ref, nodeReferencesProcessed.get(ref)+1); |
| } else if (nodeReferencesProcessed.get(ref)>0 && nodeReferencesTotal.get(ref)>1) { |
| try { |
| operand = wrapInReference(operandStack.pop(), refClassMapping.get(ref)); |
| } catch (NoSuchElementException e) { |
| operand = makeReference(refClassMapping.get(ref)); |
| } |
| } |
| } |
| } |
| return operand; |
| } |
| |
| @SuppressWarnings("unchecked") |
| private void processN_ary_linguistic_term(ParseTree node) { |
| relationCounter++; |
| // get operator and determine type of group (sequence/treeRelation/relation/...) |
| // It's possible in Annis QL to concatenate operators, so there may be several operators under one n_ary_linguistic_term node. |
| // Counter 'i' will iteratively point to all operator nodes (odd-numbered) under this node. |
| for (int i=1; i<node.getChildCount(); i = i+2) { |
| ParseTree operandTree1 = node.getChild(i-1); |
| ParseTree operandTree2 = node.getChild(i+1); |
| String reltype = getNodeCat(node.getChild(i).getChild(0)); |
| |
| LinkedHashMap<String,Object> group = null; |
| ArrayList<Object> operands = null; |
| // Retrieve operands. |
| LinkedHashMap<String, Object> operand1 = retrieveOperand(operandTree1); |
| LinkedHashMap<String, Object> operand2 = retrieveOperand(operandTree2); |
| |
| // 'Proper' n_ary_linguistic_operators receive a considerably different serialisation than 'commonparent' and 'commonancestor'. |
| // For the latter cases, a dummy span is introduced and declared as a span class that has a dominance relation towards |
| // the two operands, one after the other, thus resulting in two nested relations! A Poliqarp+ equivalent for A $ B would be |
| // contains(focus(1:contains({1:<>},A)), B). |
| // This is modeled here... |
| if (reltype.equals("commonparent") || reltype.equals("commonancestor")) { |
| // make an (outer) group and an inner group containing the dummy node or previous relations |
| group = makeGroup("relation"); |
| LinkedHashMap<String,Object> innerGroup = makeGroup("relation"); |
| LinkedHashMap<String,Object> relation = makeRelation(); |
| LinkedHashMap<String,Object> term = makeTerm(); |
| term.put("layer", "c"); |
| relation.put("wrap", term); |
| // commonancestor is an indirect commonparent relation |
| if (reltype.equals("commonancestor")) relation.put("boundary", makeBoundary(1, null)); |
| group.put("relation", relation); |
| innerGroup.put("relation", relation); |
| // Get operands list before possible re-assignment of 'group' (see following 'if') |
| ArrayList<Object> outerOperands = (ArrayList<Object>) group.get("operands"); |
| ArrayList<Object> innerOperands = (ArrayList<Object>) innerGroup.get("operands"); |
| // for lowest level, add the underspecified node as first operand and wrap it in a class group |
| if (i == 1) { |
| innerOperands.add(wrapInClass(makeSpan(), classCounter)); |
| // add the first operand and wrap the whole group in a focusing reference |
| innerOperands.add(operand1); |
| innerGroup = wrapInReference(innerGroup, classCounter); |
| outerOperands.add(innerGroup); |
| } else { |
| outerOperands.add(operandStack.pop()); |
| } |
| // Lookahead: if next operator is not commonparent or commonancestor, wrap in class for accessibility |
| if (i < node.getChildCount()-2 && !getNodeCat(node.getChild(i+2).getChild(0)).startsWith("common")) { |
| operand2 = wrapInClass(operand2, ++classCounter); |
| } |
| outerOperands.add(operand2); |
| |
| // Wrap in another reference object in case other relations are following |
| if (i < node.getChildCount()-2) { |
| group = wrapInReference(group, classCounter); |
| } |
| // All other n-ary linguistic relations have special 'relation' attributes defined in CQLF and can be |
| // handled more easily... |
| } else { |
| LinkedHashMap<String, Object> operatorGroup = parseOperatorNode(node.getChild(i).getChild(0)); |
| String groupType; |
| try { |
| groupType = (String) operatorGroup.get("groupType"); |
| } catch (ClassCastException | NullPointerException n) { |
| groupType = "relation"; |
| } |
| if (groupType.equals("relation") || groupType.equals("treeRelation")) { |
| group = makeGroup(groupType); |
| LinkedHashMap<String, Object> relation = new LinkedHashMap<String, Object>(); |
| putAllButGroupType(relation, operatorGroup); |
| System.err.println(relation); |
| group.put("relation", relation); |
| } else if (groupType.equals("sequence")) { |
| group = makeGroup(groupType); |
| putAllButGroupType(group, operatorGroup); |
| } else if (groupType.equals("position")) { |
| group = new LinkedHashMap<String,Object>(); |
| putAllButGroupType(group, operatorGroup); |
| } |
| |
| // Get operands list before possible re-assignment of 'group' (see following 'if') |
| operands = (ArrayList<Object>) group.get("operands"); |
| // Wrap in reference object in case other relations are following |
| if (i < node.getChildCount()-2) { |
| group = wrapInReference(group, classCounter); |
| } |
| |
| // Inject operands. |
| // -> Case distinction: |
| if (node.getChildCount()==3) { |
| // Things are easy when there's just one operator (thus 3 children incl. operands)... |
| if (operand1 != null) operands.add(operand1); |
| if (operand2 != null) operands.add(operand2); |
| } else { |
| // ... but things get a little more complicated here. The AST is of this form: (operand1 operator 1 operand2 operator2 operand3 operator3 ...) |
| // but we'll have to serialize it in a nested, binary way: (((operand1 operator1 operand2) operator2 operand3) operator3 ...) |
| // the following code will do just that: |
| if (i == 1) { |
| // for the first operator, include both operands |
| if (operand1 != null) operands.add(operand1); |
| if (operand2 != null) operands.add(wrapInClass(operand2, classCounter++)); |
| // Don't put this into the super object directly but store on operandStack |
| // (because this group will have to be an operand of a subsequent operator) |
| operandStack.push(group); |
| // for all subsequent operators, only take the 2nd operand (first was already added by previous operator) |
| } else if (i < node.getChildCount()-2) { |
| // for all intermediate operators, include other previous groups and 2nd operand. Store this on the operandStack, too. |
| if (operand2 != null) operands.add(wrapInClass(operand2, classCounter++)); |
| operands.add(0, operandStack.pop()); |
| operandStack.push(group); |
| } else if (i == node.getChildCount()-2) { |
| // This is the last operator. Include 2nd operand only |
| if (operand2 != null) operands.add(operand2); |
| } |
| } |
| } |
| // Final step: decide what to do with the 'group' object, depending on whether all relations have been processed |
| if (i == node.getChildCount()-2 && relationCounter == totalRelationCount) { |
| putIntoSuperObject(group); |
| if (!operandStack.isEmpty()) { |
| operands.add(0, operandStack.pop()); |
| } |
| objectStack.push(group); |
| stackedObjects++; |
| } else { |
| operandStack.push(group); |
| } |
| } |
| } |
| |
| |
| |
| /** |
| * Parses a unary_linguistic_operator node. Possible operators are: root, arity, tokenarity. |
| * Operators are embedded into a korap:term, in turn wrapped by an 'attr' property in a korap:span. |
| * @param node The unary_linguistic_operator node |
| * @return A map containing the attr key, to be inserted into korap:span |
| */ |
| private LinkedHashMap<String, Object> parseUnaryOperator(ParseTree node) { |
| LinkedHashMap<String, Object> attr = new LinkedHashMap<String, Object>(); |
| LinkedHashMap<String, Object> term = makeTerm(); |
| String op = node.getChild(1).toStringTree(parser).substring(1); |
| if (op.equals("arity") || op.equals("tokenarity")) { |
| LinkedHashMap<String, Object> boundary = boundaryFromRangeSpec(node.getChild(3), false); |
| term.put(op, boundary); |
| } else { |
| term.put(op, true); |
| } |
| attr.put("attr", term); |
| return attr; |
| } |
| |
| private LinkedHashMap<String, Object> parseOperatorNode(ParseTree operatorNode) { |
| LinkedHashMap<String, Object> relation = null; |
| String operator = getNodeCat(operatorNode); |
| // DOMINANCE |
| if (operator.equals("dominance")) { |
| relation = makeRelation(); |
| relation.put("groupType", "relation"); |
| ParseTree leftChildSpec = getFirstChildWithCat(operatorNode, "@l"); |
| ParseTree rightChildSpec = getFirstChildWithCat(operatorNode, "@r"); |
| ParseTree qName = getFirstChildWithCat(operatorNode, "qName"); |
| ParseTree edgeSpec = getFirstChildWithCat(operatorNode, "edgeSpec"); |
| ParseTree star = getFirstChildWithCat(operatorNode, "*"); |
| ParseTree rangeSpec = getFirstChildWithCat(operatorNode, "rangeSpec"); |
| LinkedHashMap<String,Object> term = makeTerm(); |
| term.put("layer", "c"); |
| if (leftChildSpec != null) relation.put("index", 0); |
| if (rightChildSpec != null) relation.put("index", -1); |
| if (qName != null) term = parseQNameNode(qName); |
| if (edgeSpec != null) term.putAll(parseEdgeSpec(edgeSpec)); |
| if (star != null) relation.put("boundary", makeBoundary(0, null)); |
| if (rangeSpec != null) relation.put("boundary", boundaryFromRangeSpec(rangeSpec)); |
| relation.put("wrap", term); |
| } |
| else if (operator.equals("pointing")) { |
| // String reltype = operatorNode.getChild(1).toStringTree(parser); |
| relation = makeRelation(); |
| relation.put("groupType", "relation"); |
| ParseTree qName = getFirstChildWithCat(operatorNode, "qName"); |
| ParseTree edgeSpec = getFirstChildWithCat(operatorNode, "edgeSpec"); |
| ParseTree star = getFirstChildWithCat(operatorNode, "*"); |
| ParseTree rangeSpec = getFirstChildWithCat(operatorNode, "rangeSpec"); |
| // if (qName != null) relation.putAll(parseQNameNode(qName)); |
| LinkedHashMap<String,Object> term = makeTerm(); |
| if (qName != null) term.putAll(parseQNameNode(qName)); |
| if (edgeSpec != null) term.putAll(parseEdgeSpec(edgeSpec)); |
| if (star != null) relation.put("boundary", makeBoundary(0, null)); |
| if (rangeSpec != null) relation.put("boundary", boundaryFromRangeSpec(rangeSpec)); |
| relation.put("wrap", term); |
| } |
| else if (operator.equals("precedence")) { |
| relation = new LinkedHashMap<String, Object>(); |
| relation.put("groupType", "sequence"); |
| ParseTree rangeSpec = getFirstChildWithCat(operatorNode, "rangeSpec"); |
| ParseTree star = getFirstChildWithCat(operatorNode, "*"); |
| ArrayList<Object> distances = new ArrayList<Object>(); |
| if (star != null) { |
| distances.add(makeDistance("w", 0, null)); |
| relation.put("distances", distances); |
| } |
| if (rangeSpec != null) { |
| distances.add(parseDistance(rangeSpec)); |
| relation.put("distances", distances); |
| } |
| relation.put("inOrder", true); |
| } |
| else if (operator.equals("spanrelation")) { |
| // relation = makeGroup("position"); |
| // relation.put("groupType", "position"); |
| String reltype = operatorNode.getChild(0).toStringTree(parser); |
| String[] frames = new String[]{}; |
| switch (reltype) { |
| case "_=_": |
| frames = new String[]{"frames:matches"}; |
| break; |
| case "_l_": |
| frames = new String[]{"frames:startswith"}; |
| break; |
| case "_r_": |
| frames = new String[]{"frames:endswith"}; |
| break; |
| case "_i_": |
| frames = new String[]{"frames:contains"};break; |
| case "_o_": |
| frames = new String[]{"frames:overlapsLeft", "frames:overlapsRight"}; |
| break; |
| case "_ol_": |
| frames = new String[]{"frames:overlapsLeft"}; |
| break; |
| case "_or_": |
| frames = new String[]{"frames:overlapsRight"}; |
| break; |
| } |
| // relation.put("frames", frames); |
| // relation.put("sharedClasses", sharedClasses); |
| relation = makePosition(frames, new String[]{}); |
| relation.put("groupType", "position"); |
| } |
| else if (operator.equals("identity")) { |
| //TODO |
| } |
| else if (operator.equals("equalvalue")) { |
| //TODO |
| } |
| else if (operator.equals("notequalvalue")) { |
| //TODO |
| } |
| return relation; |
| } |
| |
| @SuppressWarnings("unchecked") |
| private LinkedHashMap<String,Object> parseEdgeSpec(ParseTree edgeSpec) { |
| List<ParseTree> annos = getChildrenWithCat(edgeSpec, "edgeAnno"); |
| if (annos.size() == 1) return parseEdgeAnno(annos.get(0)); |
| else { |
| LinkedHashMap<String,Object> termGroup = makeTermGroup("and"); |
| ArrayList<Object> operands = (ArrayList<Object>) termGroup.get("operands"); |
| for (ParseTree anno : annos) { |
| operands.add(parseEdgeAnno(anno)); |
| } |
| return termGroup; |
| } |
| } |
| |
| private LinkedHashMap<String, Object> parseEdgeAnno( |
| ParseTree edgeAnnoSpec) { |
| LinkedHashMap<String, Object> edgeAnno = new LinkedHashMap<String, Object>(); |
| edgeAnno.put("@type", "korap:term"); |
| ParseTree qNameNode = edgeAnnoSpec.getChild(0); |
| ParseTree matchOperatorNode = edgeAnnoSpec.getChild(1); |
| ParseTree textSpecNode = edgeAnnoSpec.getChild(2); |
| ParseTree layerNode = getFirstChildWithCat(qNameNode, "layer"); |
| ParseTree foundryNode = getFirstChildWithCat(qNameNode, "foundry"); |
| if (foundryNode!=null) edgeAnno.put("foundry", foundryNode.getChild(0).toStringTree(parser)); |
| if (layerNode!=null) edgeAnno.put("layer", layerNode.getChild(0).toStringTree(parser)); |
| edgeAnno.putAll(parseTextSpec(textSpecNode)); |
| edgeAnno.put("match", parseMatchOperator(matchOperatorNode)); |
| return edgeAnno; |
| } |
| |
| private LinkedHashMap<String, Object> boundaryFromRangeSpec(ParseTree rangeSpec) { |
| return boundaryFromRangeSpec(rangeSpec, true); |
| } |
| |
| private LinkedHashMap<String, Object> boundaryFromRangeSpec(ParseTree rangeSpec, boolean expandToMax) { |
| Integer min = Integer.parseInt(rangeSpec.getChild(0).toStringTree(parser)); |
| Integer max = min; |
| if (expandToMax) max = null; |
| if (rangeSpec.getChildCount()==3) |
| max = Integer.parseInt(rangeSpec.getChild(2).toStringTree(parser)); |
| return makeBoundary(min, max); |
| } |
| |
| private LinkedHashMap<String, Object> parseDistance(ParseTree rangeSpec) { |
| Integer min = Integer.parseInt(rangeSpec.getChild(0).toStringTree(parser)); |
| Integer max = null; |
| if (rangeSpec.getChildCount()==3) |
| max = Integer.parseInt(rangeSpec.getChild(2).toStringTree(parser)); |
| return makeDistance("w", min, max); |
| } |
| |
| private LinkedHashMap<String, Object> parseTextSpec(ParseTree node) { |
| LinkedHashMap<String, Object> term = new LinkedHashMap<String, Object>(); |
| if (hasChild(node, "regex")) { |
| term.put("type", "type:regex"); |
| term.put("key", node.getChild(0).getChild(0).toStringTree(parser).replaceAll("/", "")); |
| } else { |
| term.put("key", node.getChild(1).toStringTree(parser)); |
| } |
| term.put("match", "match:eq"); |
| return term; |
| } |
| |
| /** |
| * Parses the match operator (= or !=) |
| * @param node |
| * @return |
| */ |
| private String parseMatchOperator(ParseTree node) { |
| return node.toStringTree(parser).equals("=") ? "match:eq" : "match:ne"; |
| } |
| |
| private LinkedHashMap<String, Object> parseQNameNode(ParseTree node) { |
| LinkedHashMap<String, Object> fields = new LinkedHashMap<String, Object>(); |
| ParseTree layerNode = getFirstChildWithCat(node, "layer"); |
| ParseTree foundryNode = getFirstChildWithCat(node, "foundry"); |
| if (foundryNode != null) fields.put("foundry", foundryNode.getChild(0).toStringTree(parser)); |
| fields.put("layer", layerNode.getChild(0).toStringTree(parser)); |
| return fields; |
| } |
| |
| private void putIntoSuperObject(LinkedHashMap<String, Object> object) { |
| putIntoSuperObject(object, 0); |
| } |
| |
| @SuppressWarnings({ "unchecked" }) |
| private void putIntoSuperObject(LinkedHashMap<String, Object> object, int objStackPosition) { |
| if (objectStack.size()>objStackPosition) { |
| ArrayList<Object> topObjectOperands = (ArrayList<Object>) objectStack.get(objStackPosition).get("operands"); |
| if (!invertedOperandsLists.contains(topObjectOperands)) { |
| topObjectOperands.add(object); |
| } else { |
| topObjectOperands.add(0, object); |
| } |
| } else { |
| requestMap.put("query", object); |
| } |
| } |
| |
| private void putAllButGroupType(Map<String, Object> container, Map<String, Object> input) { |
| for (String key : input.keySet()) { |
| if (!key.equals("groupType")) { |
| container.put(key, input.get(key)); |
| } |
| } |
| } |
| |
| private ParserRuleContext parseAnnisQuery (String p) throws QueryException { |
| Lexer poliqarpLexer = new AqlLexer((CharStream)null); |
| ParserRuleContext tree = null; |
| // Like p. 111 |
| try { |
| |
| // Tokenize input data |
| ANTLRInputStream input = new ANTLRInputStream(p); |
| poliqarpLexer.setInputStream(input); |
| CommonTokenStream tokens = new CommonTokenStream(poliqarpLexer); |
| parser = new AqlParser(tokens); |
| |
| // Don't throw out erroneous stuff |
| parser.setErrorHandler(new BailErrorStrategy()); |
| parser.removeErrorListeners(); |
| |
| // Get starting rule from parser |
| Method startRule = AqlParser.class.getMethod("start"); |
| tree = (ParserRuleContext) startRule.invoke(parser, (Object[])null); |
| } |
| |
| // Some things went wrong ... |
| catch (Exception e) { |
| log.error(e.getMessage()); |
| System.err.println( e.getMessage() ); |
| } |
| |
| if (tree == null) { |
| log.error("Could not parse query. Make sure it is correct ANNIS QL syntax."); |
| throw new QueryException("Could not parse query. Make sure it is correct ANNIS QL syntax."); |
| } |
| |
| // Return the generated tree |
| return tree; |
| } |
| |
| public static void main(String[] args) { |
| /* |
| * For testing |
| */ |
| String[] queries = new String[] { |
| // "cat=\"NP\" & cat=\"VP\" & #1 $ #2 ", |
| // "Haus", |
| // "lemma=\"Haus\"", |
| // "Katze=\"Hund\"", |
| // "cnx/c=\"NP\"", |
| // "cat=\"NP\"", |
| // "node & node & #1 .+ #2", |
| // " #1 > #2 & cnx/cat=\"VP\" & cnx/cat=\"NP\"", |
| // "\"Mann\" & node & #2 >[cat=\"NP\"] #1", |
| // "node & node & #2 ->coref[val=\"true\"] #1", |
| // "cat=\"NP\" & cat=\"VP\" & cat=\"PP\" & #1 $ #2 > #3", |
| // "tok=\"Mann\" & tok=\"geht\" & #1 .* #2", |
| // "\"Sonne\"", |
| // "\"so\" & ( \"nicht\" | \"doch\" ) & #1 .1,6 #2", |
| // |
| // "NP#cat=\"NP\" & PP1#cat=\"PP\" . PP2#cat=\"PP\" & #NP > #PP1 & #NP > #PP2 ", |
| // "cat=\"NP\" > cat=\"VP\" & #1 _l_ #2", |
| // "cat=\"NP\" > cat=\"VP\" & #1 . tok=\"foo\"", |
| "cat=\"NP\" & cat=\"VP\" & #1 > #2 & #1 _l_ #2", |
| "tok" |
| }; |
| // AqlTree.verbose=true; |
| for (String q : queries) { |
| try { |
| System.out.println(q); |
| AqlTree at = new AqlTree(q); |
| System.out.println(at.parseAnnisQuery(q).toStringTree(at.parser)); |
| System.out.println(); |
| |
| } catch (NullPointerException | QueryException npe) { |
| npe.printStackTrace(); |
| } |
| } |
| } |
| |
| } |