src/main/java/org/xerial/util/xml/index/StrongDataGuide.java

   1 /*--------------------------------------------------------------------------
   2  *  Copyright 2004 Taro L. Saito
   3  *
   4  *  Licensed under the Apache License, Version 2.0 (the "License");
   5  *  you may not use this file except in compliance with the License.
   6  *  You may obtain a copy of the License at
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  *  Unless required by applicable law or agreed to in writing, software
  11  *  distributed under the License is distributed on an "AS IS" BASIS,
  12  *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  *  See the License for the specific language governing permissions and
  14  *  limitations under the License.
  15  *--------------------------------------------------------------------------*/
  16 //--------------------------------------
  17 // XerialJ Project
  18 //
  19 // StrongDataGuide.java
  20 // Since: 2005/06/02
  21 //
  22 // $URL$
  23 // $Author$
  24 //--------------------------------------
  25 package org.xerial.util.xml.index;
  26
  27 import static org.xmlpull.v1.XmlPullParser.END_DOCUMENT;
  28 import static org.xmlpull.v1.XmlPullParser.END_TAG;
  29 import static org.xmlpull.v1.XmlPullParser.START_TAG;
  30 import static org.xmlpull.v1.XmlPullParser.TEXT;
  31
  32 import java.io.BufferedReader;
  33 import java.io.FileNotFoundException;
  34 import java.io.FileReader;
  35 import java.io.IOException;
  36 import java.io.OutputStream;
  37 import java.io.PrintWriter;
  38 import java.io.Reader;
  39 import java.util.Collection;
  40 import java.util.Stack;
  41
  42 import org.xerial.core.XerialException;
  43 import org.xerial.util.StringUtil;
  44 import org.xerial.util.cui.OptionParser;
  45 import org.xerial.util.cui.OptionParserException;
  46 import org.xerial.util.graph.AdjacencyList;
  47 import org.xerial.util.graph.Edge;
  48 import org.xerial.util.graph.Graph;
  49 import org.xerial.util.log.Logger;
  50 import org.xerial.util.xml.SinglePath;
  51 import org.xerial.util.xml.XMLErrorCode;
  52 import org.xerial.util.xml.XMLException;
  53 import org.xerial.util.xml.pullparser.PullParserUtil;
  54 import org.xmlpull.v1.XmlPullParser;
  55 import org.xmlpull.v1.XmlPullParserException;
  56
  57 /**
  58  * StrongDataGuide, which aggregates XML nodes based on same paths
  59  *
  60  * @author leo
  61  *
  62  */
  63 public class StrongDataGuide
  64 {
  65     private Graph<SinglePath, String> _graph         = new AdjacencyList<SinglePath, String>();
  66     private int                       _currentPathID = 0;
  67     private Stack<Integer>            _cursorHistory = new Stack<Integer>();
  68     private SinglePath                _currentPath;
  69     private SinglePath                _rootPath;
  70     private int                       _rootPathID;
  71     private Logger                    _logger        = Logger.getLogger(StrongDataGuide.class);
  72
  73     /**
  74      *
  75      */
  76     public StrongDataGuide()
  77     {
  78         super();
  79         _rootPath = SinglePath.rootPath();
  80         _rootPathID = _graph.addNode(_rootPath);
  81     }
  82
  83     /**
  84      * Generates the StrongDataGuide from the XML file
  85      *
  86      * @param xmlFile
  87      * @throws FileNotFoundException
  88      * @throws XMLParserException
  89      * @throws XMLException
  90      * @throws IOException
  91      * @throws XerialException
  92      */
  93     public void generateFrom(String xmlFile) throws FileNotFoundException, XMLException, IOException, XerialException
  94     {
  95         Reader reader = new BufferedReader(new FileReader(xmlFile));
  96         generateFrom(reader);
  97     }
  98
  99     /**
 100      * Generates the {@link StrongDataGuide} from the {@link Reader}
 101      *
 102      * @param xmlReader
 103      * @throws FileNotFoundException
 104      * @throws XMLParserException
 105      */
 106     public void generateFrom(Reader xmlReader) throws XMLException, IOException, XerialException
 107     {
 108         // initialize
 109         _currentPathID = _rootPathID;
 110         _currentPath = _rootPath;
 111
 112         XmlPullParser parser = PullParserUtil.newParser(xmlReader);
 113         try
 114         {
 115             int state;
 116             while ((state = parser.next()) != END_DOCUMENT)
 117             {
 118                 switch (state)
 119                 {
 120                 case START_TAG:
 121                     String name = parser.getName();
 122                     _logger.trace("start tag: " + name);
 123                     int pathID = getPathID(name);
 124                     _logger.trace("path ID  : " + pathID);
 125                     moveCursor(pathID);
 126                     // process attributes
 127                     for (int i = 0; i < parser.getAttributeCount(); i++)
 128                     {
 129                         int attributeID = getPathID(String.format("%s@%s", name, parser.getAttributeName(i)));
 130                         moveCursor(attributeID);
 131                         traceBack();
 132                     }
 133                     break;
 134                 case END_TAG:
 135                     traceBack();
 136                     break;
 137                 case TEXT:
 138                     break;
 139                 }
 140             }
 141
 142         }
 143         catch (XmlPullParserException e)
 144         {
 145             throw new XMLException(XMLErrorCode.PARSE_ERROR, e);
 146         }
 147     }
 148
 149     private void moveCursor(int pathID)
 150     {
 151         Collection<Integer> destNodeID = _graph.getDestNodeIDSetOf(_currentPathID);
 152         if (!destNodeID.contains(pathID))
 153         {
 154             _graph.addEdge(new Edge(_currentPathID, pathID), "edge");
 155         }
 156         _cursorHistory.push(_currentPathID);
 157         _currentPathID = pathID;
 158         _currentPath = _graph.getNodeLabel(pathID);
 159     }
 160
 161     private void traceBack()
 162     {
 163         assert !_cursorHistory.empty();
 164         _currentPathID = _cursorHistory.pop();
 165         _currentPath = _graph.getNodeLabel(_currentPathID);
 166     }
 167
 168     private int getPathID(String tagName)
 169     {
 170         SinglePath path = new SinglePath(_currentPath, tagName);
 171         int pathID = _graph.getNodeID(path);
 172         if (pathID == -1)
 173         {
 174             pathID = _graph.addNode(path);
 175         }
 176         return pathID;
 177     }
 178
 179     public void outputGraphviz(OutputStream out)
 180     {
 181         PrintWriter gout = new PrintWriter(out);
 182         gout.println("digraph G {");
 183         // output node labels
 184         for (int pathID : _graph.getNodeIDSet())
 185         {
 186             SinglePath path = _graph.getNodeLabel(pathID);
 187             gout.println(pathID + " [label=" + StringUtil.quote(path.getLeaf(), "\"") + "];");
 188         }
 189         for (int pathID : _graph.getNodeIDSet())
 190         {
 191             for (int destNodeID : _graph.getDestNodeIDSetOf(pathID))
 192             {
 193                 gout.println(pathID + " -> " + destNodeID + ";");
 194             }
 195         }
 196
 197         gout.println("}");
 198         gout.flush();
 199     }
 200
 201     public static enum Opt {
 202         help
 203     }
 204
 205     public static void main(String[] args) throws OptionParserException
 206     {
 207         OptionParser<Opt> opt = new OptionParser<Opt>();
 208         opt.addOption(Opt.help, "h", "help", "display help messages");
 209
 210         opt.parse(args);
 211         if (opt.isSet(Opt.help) || opt.getArgumentLength() < 1)
 212         {
 213             printHelpMessage(opt);
 214             return;
 215         }
 216
 217         StrongDataGuide sdg = new StrongDataGuide();
 218         try
 219         {
 220             sdg.generateFrom(opt.getArgument(0));
 221             sdg.outputGraphviz(System.out);
 222         }
 223         catch (XerialException e)
 224         {
 225             System.err.println(e.getMessage());
 226         }
 227         catch (IOException e)
 228         {
 229             System.err.println(e.getMessage());
 230         }
 231     }
 232
 233     private static void printHelpMessage(OptionParser<Opt> opt)
 234     {
 235         System.out.println("usage: > java -jar StrongDataGuide.jar [option] xml_file");
 236         System.out.println(opt.helpMessage());
 237
 238     }
 239
 240 }