root/SPARQLBuilderWWW/src/java/org/biohackathon/SPARQLBuilder/OWL/OWLClassGraph.java @ 229

/*
 * 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{
    //String startClass;
    //String endClass;
    int nsteps = 4;
    int limit = 100;
    int th = 1;
    List<String> nodeType;
    ArrayList<HashSet<Integer>> connectionTable;

    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(String startClass, String endClass){
        super();
        
        // start & end
        this.startClass = startClass;
        this.endClass = endClass;
        
        // parameters
        nsteps = 3;
        limit = 1000;
        th = 5;
    }
*/

/*    
    public OWLClassGraph(String startClass, String endClass, int th){
        super();
        
        // start & end
        this.startClass = startClass;
        this.endClass = endClass;
        // th of instances
        this.th = th;
        
        // parameters
        nsteps = 3;
        limit = 1000;
    }
*/    

    public OWLClassGraph(){
        super();
        nodeType = new LinkedList<String>();
        //setClassGraph(rdfsa);
        //connectionTable = createConnectionTable();
    }
    
    
    public OWLClassGraph(RDFSchemaAnalyzer rdfsa){
        super();
        nodeType = new LinkedList<String>();
        //setClassGraph(rdfsa);
        //connectionTable = createConnectionTable();
    }
    
    public int getNumberOfEdge(String url){
        Integer node = labelednodes.get(url);
        if (node == null){
            return 0;
        }
        return adjlist.get(node).size();
    }
    
    public Path[] getPaths(String startClass, String endClass){
        List<List<ClassLink>> paths = searchPaths(startClass, endClass);

        NavigableSet<Path> sortedpath = new TreeSet<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);
            sortedpath.add(path);
            j++;
        }
        Path[] patharray = new Path[paths.size()];
        Iterator<Path> pait = sortedpath.descendingIterator();
        int i = 0;
        while ( pait.hasNext() ){
            patharray[i] = pait.next();
            i++;
        }
        return patharray;
    }
    
    public HashSet<Integer> getConnectionList(Integer node){
        return connectionTable.get(node);
    }
    
    public Path[] getPaths_old(RDFSchemaAnalyzer rdfsa, boolean countLink, String startClass, String endClass){
        List<List<ClassLink>> paths = null;
        paths = searchPathsbyVisitingNodes(rdfsa, countLink, startClass, endClass);
        NavigableSet<Path> sortedpath = new TreeSet<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();
                }
            }
            path.setWidth(min);
            sortedpath.add(path);
            j++;
        }
        Path[] patharray = new Path[paths.size()];
        Iterator<Path> pait = sortedpath.descendingIterator();
        int i = 0;
        while ( pait.hasNext() ){
            patharray[i] = pait.next();
            i++;
        }
        return patharray;
    }

    private List<List<ClassLink>> searchPaths(String startClass, String endClass){

        List<List<ClassLink>> paths = new ArrayList<>();
        List<List<Integer>> simplePaths = new LinkedList<>();
        Integer snode = labelednodes.get(startClass);
        Integer enode = labelednodes.get(endClass);
        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;
        }
        
        ListIterator<List<Integer>> pit = simplePaths.listIterator();
        while( pit.hasNext()){
            List<Integer> spath = pit.next();
            List<List<ClassLink>> convertedPaths = convertSimplePathToPaths(spath);
            paths.addAll(convertedPaths);
        }
        return paths;
    }
    
    private List<List<ClassLink>> convertSimplePathToPaths(List<Integer> simplePath){
        List<List<ClassLink>> paths = new LinkedList<List<ClassLink>>();
        //List<List<LabeledEdge>> multiedges = new LinkedList<List<LabeledEdge>>();
        ListIterator<Integer> spit = simplePath.listIterator();
        Integer start = spit.next();
        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);
                    tmppaths.add(addedpath);
                }
            }
            paths = tmppaths;
            start = end;
        }        
        return paths;
    }
    
    /*
    private List<List<ClassLink>> searchPaths(RDFSchemaAnalyzer rdfsa, boolean countLinks){
        List<List<ClassLink>> paths = new ArrayList<>();
        List<LinkAndPath> lp = new LinkedList<>();
        lp.add(new LinkAndPath(new ClassLink("",startClass,null,Direction.both,0,0,0,0,0,false,false), new LinkedList<ClassLink>(), ""));
        try{
          for ( int i = 0; i < nsteps; i++ ){
              ListIterator<LinkAndPath> lit = lp.listIterator();
              List<LinkAndPath> nextlp = new LinkedList<>();
              while ( lit.hasNext() ){
                  LinkAndPath crrlp = lit.next();
                  ClassLink[] classLinks = rdfsa.getNextClass(null, crrlp.classLink.getLinkedClassURI(), limit, countLinks);
                  for ( int j = 0 ; j < classLinks.length; j++ ){
                      List<ClassLink> crrpath = new LinkedList<>(crrlp.path);
                      crrpath.add(classLinks[j]);
                      if ( classLinks[j].getLinkedClassURI() == null ){ continue; }
                      if ( classLinks[j].getLinkedClassURI().equals(endClass) ){
                          paths.add(new LinkedList<>(crrpath));
                          continue;
                      }
                      if ( countLinks == true && classLinks[j].getNumOfLinks() <= th){
                          continue;
                      }
                      if ( i >= 2 ){
                          if ( crrlp.classLink.getPropertyURI().equals(classLinks[j].getPropertyURI()) &&
                           crrlp.classLink.getDirection() != classLinks[j].getDirection() &&
                           crrlp.originalClassURI.equals( classLinks[j].getLinkedClassURI()) ){
                              continue;
                          }
                      }
                      nextlp.add(new LinkAndPath(classLinks[j], crrpath, crrlp.classLink.getLinkedClassURI()));
                  }
              }
              lp = nextlp;
          }
        }catch(Exception e){ 
            System.err.println(e);
        }
        return paths;  
    }
    */
    
    private List<List<ClassLink>> searchPathsbyVisitingNodes(RDFSchemaAnalyzer rdfsa, boolean countLinks,
            String startClass, String endClass){
        List<List<ClassLink>> paths = new ArrayList<>();
        List<LinkAndPath> lp = new LinkedList<>();
        Set<String> urls = new HashSet<String>();
        urls.add(startClass);
        lp.add(new LinkAndPath(new ClassLink("",startClass,null,Direction.both,0,0,0,0,0,false,false), 
                                 new LinkedList<ClassLink>(), startClass, urls));
        try{
          for ( int i = 0; i < nsteps; i++ ){
              ListIterator<LinkAndPath> lit = lp.listIterator();
              List<LinkAndPath> nextlp = new LinkedList<>();
              while ( lit.hasNext() ){
                  LinkAndPath crrlp = lit.next();
                  ClassLink[] classLinks = rdfsa.getNextClass(null, crrlp.classLink.getLinkedClassURI(), limit, countLinks);
                  // limit
                  for ( int j = 0 ; j < classLinks.length; j++ ){
                      if (crrlp.classURIs.contains(classLinks[j].getLinkedClassURI()) ){
                          continue; // visited
                      }
                      List<ClassLink> crrpath = new LinkedList<>(crrlp.path);
                      crrpath.add(classLinks[j]);
                      Set<String> crrurls = new HashSet<String>(crrlp.classURIs);
                      crrurls.add(classLinks[j].getLinkedClassURI());
                      if ( classLinks[j].getLinkedClassURI() == null ){ continue; }
                      if ( classLinks[j].getNumOfLinks() <= th){
                          continue;
                      }
                      if ( classLinks[j].getLinkedClassURI().equals(endClass) ){
                          paths.add(new LinkedList<>(crrpath));
                          continue;
                      }
                      //crrlp.classURIs.add()
                      nextlp.add(new LinkAndPath(classLinks[j], crrpath, crrlp.classLink.getLinkedClassURI(),crrurls));
                  }
              }
              lp = nextlp;
          }
        }catch(Exception e){ 
            System.err.println(e);
        }
        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]);
                       /*
                       ClassLink rev = new ClassLink( classLinks[j].getPropertyURI(),
                               //classLinks[j].getLinkedClassURI(),
                               classes[i].getClassURI(),
                               classLinks[j].getLinkedLiteralDatatypeURI(),
                               classLinks[j].getDirection(), classLinks[j].getNumOfLinks(), 
                               classLinks[j].getNumOfOriginInstances(), classLinks[j].getNumOfLinkedInstances(),
                               classLinks[j].getNumOfOriginClassInstances(), classLinks[j].getNumOfLinkedClassInstances(),
                               classLinks[j].isDomainClassLimitedQ(), classLinks[j].isRangeClassLimitedQ() );
                       rev.setDirection(Direction.reverse);
                       addEdge(n, i, rev);
                               */
                   }
               }
           }catch(Exception e){
               System.err.println(e);
           }
       }       
   }

    public void setPartClassGraph(RDFSchemaAnalyzer rdfsa, String startClass){
        // 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");
        }
        int nedge = 0;
        Integer snode = labelednodes.get(startClass);
        Set<Integer> nodes = new HashSet<Integer>();
        nodes.add(snode);
        Set<Integer> visited = new HashSet<Integer>();
        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();
                visited.add(crr);
                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 ( !visited.contains(nn) ){
                            nextnodes.add(nn);
                        }
                        if ( nn != null ){
                            addEdge(crr, nn, classLinks[j]);
                        /*
                        }else{
                            nn = labelednodes.get(classLinks[j].getLinkedLiteralDatatypeURI());
                            if ( nn == null ){
                                addNode(classLinks[j].getLinkedLiteralDatatypeURI());
                                n = nodeType.size();
                                nodeType.add("literal");
                            }
                            addEdge(i, n, classLinks[j]);
                            /*
                            ClassLink rev = new ClassLink( classLinks[j].getPropertyURI(),
                               //classLinks[j].getLinkedClassURI(),
                               classes[i].getClassURI(),
                               classLinks[j].getLinkedLiteralDatatypeURI(),
                               classLinks[j].getDirection(), classLinks[j].getNumOfLinks(), 
                               classLinks[j].getNumOfOriginInstances(), classLinks[j].getNumOfLinkedInstances(),
                               classLinks[j].getNumOfOriginClassInstances(), classLinks[j].getNumOfLinkedClassInstances(),
                               classLinks[j].isDomainClassLimitedQ(), classLinks[j].isRangeClassLimitedQ() );
                            rev.setDirection(Direction.reverse);
                            addEdge(n, i, rev);
                        */     
                        }
                    }
                }catch(Exception e){
                    e.printStackTrace();
                }
            }
            nodes = nextnodes;
        }
    }
   
   private ArrayList<HashSet<Integer>> createConnectionTable(){
       ArrayList<HashSet<Integer>> ct = new ArrayList<HashSet<Integer>>();
       for (int i = 0; i < labels.size(); i++ ){ // each node
           HashSet<Integer> cl = createConnectionList(i);
           ct.add(cl);
       }
       return ct;
   }
   
   private HashSet<Integer> createConnectionList(Integer node){
       HashSet<Integer> cl = new HashSet<Integer>();
       HashSet<Integer> crrnodes = new HashSet<Integer>();
       crrnodes.add(node);
       cl.add(node);
       for ( int i = 0 ; i < nsteps; i++ ){
           Set<Integer> nextnodes = new HashSet<Integer>();
           Iterator<Integer> cit = crrnodes.iterator();
           while ( cit.hasNext() ){
               Integer crr = cit.next();
               Set<Integer> nexts = gadjlist.get(crr).keySet();
               Iterator<Integer> nit = nexts.iterator();
               while ( nit.hasNext() ){
                   Integer next = nit.next();
                   if ( !cl.contains(next) ){
                       nextnodes.add(next);
                       cl.add(next);
                   }
               }
           }
           crrnodes.clear();
           crrnodes.addAll(nextnodes);
       }
       return cl;
   }
}