root/SPARQLBuilderWWW/src/java/org/biohackathon/SPARQLBuilder/OWL/OWLClassGraph.java @ 257

/*
 * To change this template, choose Tools | Templates
 * and open the template in the editor.
 */
package org.biohackathon.SPARQLBuilder.OWL;

/**
 *
 * @author atsuko
 */
import java.util.*;

public class OWLClassGraph extends LabeledMultiDigraph{
    int nsteps = 4;
    int limit = 100;
    int th = 1;
    double cth = 1.0; // 0.0(no path) - 1.0(all paths)
    
    List<String> nodeType;
    ArrayList<HashSet<Integer>> connectionTable;
    String sparqlEndpoint;
    Set<Integer> visited;
    List<Map<Integer, Integer>> edgeweight;
    List<Integer> nodeweight;
    Map<String, Boolean> checkedpaths;
    
    public class LinkAndPath{
        String originalClassURI; // originalClasssURI -classLink.propertyURI-> classLink.linkedClassURL
        ClassLink classLink;
        List<ClassLink> path;
        Set<String> classURIs; // apearing class URIs in the path
        
        
        public LinkAndPath(ClassLink classLink, List<ClassLink> path){
           this.classLink = classLink;
           this.path = path;
        }
        
        public LinkAndPath(ClassLink classLink, List<ClassLink> path, String originalClassURI){
           this.classLink = classLink;
           this.path = path;
           this.originalClassURI = originalClassURI;
        }

        public LinkAndPath(ClassLink classLink, List<ClassLink> path, String originalClassURI, Set<String> classURIs){
           this.classLink = classLink;
           this.path = path;
           this.originalClassURI = originalClassURI;
           this.classURIs = classURIs;
        }
    }

    public OWLClassGraph(){ // not used
        super();
        nodeType = new LinkedList<String>();
    }
        
    public OWLClassGraph(RDFSchemaAnalyzer rdfsa){ // for experiment
        super();
        nodeType = new LinkedList<String>();
        setClassGraph(rdfsa);
    }
    
    public OWLClassGraph(RDFSchemaAnalyzer rdfsa, String sparqlEndpoint, String startClass){ // used
        super();
        nodeType = new LinkedList<String>();
        setPartClassGraph(rdfsa, sparqlEndpoint, startClass);
    }

    public int getNumberOfEdge(String url){
        Integer node = labelednodes.get(url);
        if (node == null){
            return 0;
        }
        return adjlist.get(node).size();
    }
    
    public boolean visitedNode(String classURI){
        if ( visited.contains(labelednodes.get(classURI)) ){
            return true;
        }
        return false;
    }
    
    public Path[] getPaths(String startClass, String endClass){
        List<List<ClassLink>> paths = searchPaths(startClass, endClass);

        List<Path> sortedpaths = new LinkedList<Path>();
        ListIterator<List<ClassLink>> pit = paths.listIterator();
        int j = 0;
        while ( pit.hasNext() ){
            Path path = new Path();
            path.setStartClass(startClass);
            List<ClassLink> crrpath = pit.next();
            path.setClassLinks(crrpath);
            ListIterator<ClassLink> cit = crrpath.listIterator();
            int min = Integer.MAX_VALUE;
            while ( cit.hasNext() ){
                ClassLink cl = cit.next();
                if ( cl.getNumOfLinks() < min ){
                    min = cl.getNumOfLinks();
                }
            }
            // using length of path
            int rankwidth = (int) ( ( min * nsteps )/ crrpath.size() );
            path.setWidth(rankwidth);
            sortedpaths.add(path);
            j++;
        }
        Path[] patharray = new Path[paths.size()];
        Collections.sort(sortedpaths);
        Iterator<Path> pait = sortedpaths.listIterator();
        int i = 0;
        while ( pait.hasNext() ){
            patharray[paths.size()-i-1] = pait.next();
            i++;
        }
        return patharray;
    }
    
    private List<List<ClassLink>> searchPaths(String startClass, String endClass){
        //int asked = 0;
        checkedpaths = new HashMap<String, Boolean>();
        List<List<ClassLink>> paths = new ArrayList<>();
        Integer snode = labelednodes.get(startClass);
        Integer enode = labelednodes.get(endClass);
        List<List<Integer>> simplePaths = searchSimplePaths(snode, enode);
        
        ListIterator<List<Integer>> pit = simplePaths.listIterator();
        //System.out.println("SPATH:");
        //System.out.println(simplePaths.size());
        while( pit.hasNext()){
            List<Integer> spath = pit.next();
            List<List<ClassLink>> convertedPaths = convertSimplePathToPaths(spath);
            paths.addAll(convertedPaths);
        }
        //System.out.println("PATH:");
        //System.out.println(paths.size());
        return paths;
    }

    private List<List<Integer>> searchSimplePaths(Integer snode, Integer enode){
        List<List<Integer>> simplePaths = new LinkedList<>();
        List<List<Integer>> lp = new LinkedList<>();
        List<Integer> ini = new LinkedList<Integer>(); // initial path
        ini.add(snode);
        lp.add(ini);
        for (int i = 0; i < nsteps; i++ ){
            ListIterator<List<Integer>> lit = lp.listIterator();
            List<List<Integer>> nextlp = new LinkedList<>();
            while ( lit.hasNext() ){ 
                List<Integer> crrpath = lit.next();
                Integer crrnode = crrpath.get(crrpath.size()-1);
                Set<Integer> nexts = gadjlist.get(crrnode).keySet();
                Iterator<Integer> nit = nexts.iterator();
                while( nit.hasNext() ){
                    Integer nextnode = nit.next();
                    if ( crrpath.contains(nextnode) ){ continue; }
                    List<Integer> nextpath = new LinkedList<Integer>(crrpath); // copy
                    nextpath.add(nextnode);
                    if ( nextnode.equals(enode) ){
                        simplePaths.add(nextpath);
                        continue;
                    }
                    nextlp.add(nextpath);
                }
            }
            lp = nextlp;
        }        
        return simplePaths;
    }
    
    
    private List<List<ClassLink>> convertSimplePathToPaths(List<Integer> simplePath){
        List<List<ClassLink>> paths = new LinkedList<List<ClassLink>>();
        ListIterator<Integer> spit = simplePath.listIterator();
        Integer start = spit.next();
        String startClass = this.labels.get(start);
        Integer end = spit.next();
        List<LabeledEdge> edges = gadjlist.get(start).get(end);
        ListIterator<LabeledEdge> eit = edges.listIterator();
        while ( eit.hasNext() ){
            List<ClassLink> cl = new LinkedList<ClassLink>();
            cl.add((ClassLink)eit.next().getLabel());
            paths.add(cl);
        }
        start = end;
        while( spit.hasNext() ){
            end = spit.next();
            // start-end
            edges = gadjlist.get(start).get(end);
            List<List<ClassLink>> tmppaths = new LinkedList<List<ClassLink>>();            
            // current path
            ListIterator<List<ClassLink>> pit = paths.listIterator();
            while ( pit.hasNext() ){
                List<ClassLink> basepath = pit.next();
                eit = edges.listIterator();
                while ( eit.hasNext() ){
                    ClassLink cl = (ClassLink) eit.next().label;
                    List<ClassLink> addedpath = new LinkedList<ClassLink>(basepath);
                    addedpath.add(cl);
                    // check
                    //if (checkPath(startClass, addedpath)){
                        tmppaths.add(addedpath);
                    //}
                }
            }
            paths = tmppaths;
            start = end;
        }        
        return paths;
    }
    
    private void setClassGraph(RDFSchemaAnalyzer rdfsa){
        // setNodes
        SClass[] classes = null;
        try{
            classes = rdfsa.getOWLClasses(null, null, null, true);
        }catch(Exception e){
            System.err.println(e); return;
        }
        for (int i = 0 ; i < classes.length; i++){
            addNode(classes[i].getClassURI());
            nodeType.add("class");
        }
        // setEdges
        for (int i = 0 ; i < classes.length; i++ ){
            try{
                ClassLink[] classLinks = rdfsa.getNextClass(null, classes[i].getClassURI(), limit, true);
                for (int j = 0 ; j < classLinks.length; j++){
                    Integer n = labelednodes.get(classLinks[j].getLinkedClassURI());
                    if ( n != null ){
                        addEdge(i, n, classLinks[j]);
                    }else{
                        n = labelednodes.get(classLinks[j].getLinkedLiteralDatatypeURI());
                        if ( n == null ){
                           addNode(classLinks[j].getLinkedLiteralDatatypeURI());
                           n = nodeType.size();
                           nodeType.add("literal");
                        }
                        addEdge(i, n, classLinks[j]);
                    }
                }
            }catch(Exception e){
                System.err.println(e);
            }
        }       
    }

    public void setPartClassGraph(RDFSchemaAnalyzer rdfsa, String sparqlEndpoint, String startClass){
        // set endpoint
        this.sparqlEndpoint = sparqlEndpoint;
        visited = new HashSet<Integer>();
        edgeweight = new LinkedList<Map<Integer,Integer>>();
        nodeweight = new LinkedList<Integer>();
        // setNodes for all classes
        SClass[] classes = null;
        try{
           classes = rdfsa.getOWLClasses(null, null, null, true);
        }catch(Exception e){
           System.err.println(e); return;
        }
        for (int i = 0 ; i < classes.length; i++){
           addNode(classes[i].getClassURI());
           nodeType.add("class");
           edgeweight.add(new HashMap<Integer,Integer>());
           nodeweight.add(classes[i].getNumOfInstances());
        }
        // setEdges
        Integer snode = labelednodes.get(startClass);
        Set<Integer> nodes = new HashSet<Integer>();
        nodes.add(snode);
        visited.add(snode);
        for (int i = 0 ; i < nsteps; i++ ){
            Iterator<Integer> nit = nodes.iterator();
            Set<Integer> nextnodes = new HashSet<Integer>();
            while ( nit.hasNext() ){
                Integer crr = nit.next();
                try{
                    ClassLink[] classLinks = rdfsa.getNextClass(null, labels.get(crr), limit, true);
                    for (int j = 0 ; j < classLinks.length; j++){
                        Integer nn = labelednodes.get(classLinks[j].getLinkedClassURI());
                        if ( nn == null ){
                            continue;
                        }
                        if ( !visited.contains(nn) ){
                            nextnodes.add(nn);
                        }
                        addEdge(crr, nn, classLinks[j]);
                        updateWeight(crr, nn, classLinks[j]);
                    }
                }catch(Exception e){
                    e.printStackTrace();
                }
            }
            nodes = nextnodes;
            visited.addAll(nodes);
        }
        // cut visited         
        Iterator<Integer> nit = visited.iterator();
        while(nit.hasNext()){
            Integer node = nit.next();
            if ( ! node.equals(snode) ){
                List<List<Integer>> paths = searchSimplePaths(snode, node);
                if ( paths.isEmpty()){
                    nit.remove();
                }
            }
        }
    }       
   
    private void updateWeight(Integer node1, Integer node2, ClassLink edge){
        Map<Integer, Integer> weight = edgeweight.get(node1);
        Integer crr = weight.get(node2);
        if (crr == null ){
            crr = edge.getNumOfLinkedClassInstances();
            weight.put(node2, crr);           
        }
        if ( crr < edge.getNumOfLinkedClassInstances() ){
            crr = edge.getNumOfLinkedClassInstances();
            weight.put(node2, crr);
        }
        weight = edgeweight.get(node2);
        crr = weight.get(node1);
        if (crr == null ){
            crr = edge.getNumOfOriginClassInstances();
            weight.put(node1, crr);
        }
        if ( crr < edge.getNumOfOriginClassInstances() ){
            crr = edge.getNumOfOriginInstances();
            weight.put(node1, crr);
        }
    }
    
}