* Licensed to the Apache Software Foundation (ASF) under one or more

* contributor license agreemnets.  See the NOTICE file distributed with

* this work for additional information regarding copyright ownership.

* The ASF licenses this file to You under the Apache License, Version 2.0

* (the "License"); you may not use this file except in compliance with

* the License. You may obtain a copy of the License at

*

*     http://www.apache.org/licenses/LICENSE-2.0

*

* Unless required by applicable law or agreed to in writing, software

* distributed under the License is distributed on an "AS IS" BASIS,

* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

* See the License for the specific language governing permissions and

* limitations under the License.

*/

* Class for a shift reduce style parser based on Adwait Ratnaparkhi's 1998 thesis.

*/

  private MaxentModel buildModel;

  private MaxentModel checkModel;

  private BuildContextGenerator buildContextGenerator;

  private CheckContextGenerator checkContextGenerator;

  private double[] bprobs;

  private double[] cprobs;

  private static final String TOP_START = START + TOP_NODE;

  private int topStartIndex;

  private Map<String, String> startTypeMap;

  private Map<String, String> contTypeMap;

  private int completeIndex;

  private int incompleteIndex;

  public Parser(ParserModel model, int beamSize, double advancePercentage) {

    this(model.getBuildModel(), model.getCheckModel(),

        new POSTaggerME(model.getParserTaggerModel(), 10, 0),

        new ChunkerME(model.getParserChunkerModel(),

            ChunkerME.DEFAULT_BEAM_SIZE,

            new ParserChunkerSequenceValidator(model.getParserChunkerModel()),

            new ChunkContextGenerator(ChunkerME.DEFAULT_BEAM_SIZE)),

            model.getHeadRules(), beamSize, advancePercentage);

  }

 

  public Parser(ParserModel model) {

    this(model, defaultBeamSize, defaultAdvancePercentage);

  }

  /**

   * Creates a new parser using the specified models and head rules.

   * @param buildModel The model to assign constituent labels.

   * @param checkModel The model to determine a constituent is complete.

   * @param tagger The model to assign pos-tags.

   * @param chunker The model to assign flat constituent labels.

   * @param headRules The head rules for head word perculation.

   */

  @Deprecated

  public Parser(MaxentModel buildModel, MaxentModel checkModel, POSTagger tagger, Chunker chunker, HeadRules headRules) {

    this(buildModel,checkModel,tagger,chunker,headRules,defaultBeamSize,defaultAdvancePercentage);

  }

  /**

   * Creates a new parser using the specified models and head rules using the specified beam size and advance percentage.

   * @param buildModel The model to assign constituent labels.

   * @param checkModel The model to determine a constituent is complete.

   * @param tagger The model to assign pos-tags.

   * @param chunker The model to assign flat constituent labels.

   * @param headRules The head rules for head word perculation.

   * @param beamSize The number of different parses kept during parsing.

   * @param advancePercentage The minimal amount of probability mass which advanced outcomes must represent.

   * Only outcomes which contribute to the top "advancePercentage" will be explored.

   */

  @Deprecated

  public Parser(MaxentModel buildModel, MaxentModel checkModel, POSTagger tagger, Chunker chunker, HeadRules headRules, int beamSize, double advancePercentage) {

    super(tagger, chunker, headRules, beamSize, advancePercentage);

    this.buildModel = buildModel;

    this.checkModel = checkModel;

    bprobs = new double[buildModel.getNumOutcomes()];

    cprobs = new double[checkModel.getNumOutcomes()];

    this.buildContextGenerator = new BuildContextGenerator();

    this.checkContextGenerator = new CheckContextGenerator();

    startTypeMap = new HashMap<String, String>();

    contTypeMap = new HashMap<String, String>();

    for (int boi = 0, bon = buildModel.getNumOutcomes(); boi < bon; boi++) {

      String outcome = buildModel.getOutcome(boi);

      if (outcome.startsWith(START)) {

        //System.err.println("startMap "+outcome+"->"+outcome.substring(START.length()));

        startTypeMap.put(outcome, outcome.substring(START.length()));

      }

      else if (outcome.startsWith(CONT)) {

        //System.err.println("contMap "+outcome+"->"+outcome.substring(CONT.length()));

        contTypeMap.put(outcome, outcome.substring(CONT.length()));

      }

    }

    topStartIndex = buildModel.getIndex(TOP_START);

    completeIndex = checkModel.getIndex(COMPLETE);

    incompleteIndex = checkModel.getIndex(INCOMPLETE);

  }

  @Override

  protected void advanceTop(Parse p) {

    buildModel.eval(buildContextGenerator.getContext(p.getChildren(), 0), bprobs);

    p.addProb(Math.log(bprobs[topStartIndex]));

    checkModel.eval(checkContextGenerator.getContext(p.getChildren(), TOP_NODE, 0, 0), cprobs);

    p.addProb(Math.log(cprobs[completeIndex]));

    p.setType(TOP_NODE);

  }

  @Override

  protected Parse[] advanceParses(final Parse p, double probMass) {

    double q = 1 - probMass;

    /** The closest previous node which has been labeled as a start node. */

    Parse lastStartNode = null;

    /** The index of the closest previous node which has been labeled as a start node. */

    int lastStartIndex = -1;

    /** The type of the closest previous node which has been labeled as a start node. */

    String lastStartType = null;

    /** The index of the node which will be labeled in this iteration of advancing the parse. */

    int advanceNodeIndex;

    /** The node which will be labeled in this iteration of advancing the parse. */

    Parse advanceNode=null;

    Parse[] originalChildren = p.getChildren();

    Parse[] children = collapsePunctuation(originalChildren,punctSet);

    int numNodes = children.length;

    if (numNodes == 0) {

      return null;

    }

    //determines which node needs to be labeled and prior labels.

    for (advanceNodeIndex = 0; advanceNodeIndex < numNodes; advanceNodeIndex++) {

      advanceNode = children[advanceNodeIndex];

      if (advanceNode.getLabel() == null) {

        break;

      }

      else if (startTypeMap.containsKey(advanceNode.getLabel())) {

        lastStartType = startTypeMap.get(advanceNode.getLabel());

        lastStartNode = advanceNode;

        lastStartIndex = advanceNodeIndex;

        //System.err.println("lastStart "+i+" "+lastStart.label+" "+lastStart.prob);

      }

    }

    int originalAdvanceIndex = mapParseIndex(advanceNodeIndex,children,originalChildren);

    List<Parse> newParsesList = new ArrayList<Parse>(buildModel.getNumOutcomes());

    //call build

    buildModel.eval(buildContextGenerator.getContext(children, advanceNodeIndex), bprobs);

    double bprobSum = 0;

    while (bprobSum < probMass) {

      // The largest unadvanced labeling.

      int max = 0;

      for (int pi = 1; pi < bprobs.length; pi++) { //for each build outcome

        if (bprobs[pi] > bprobs[max]) {

          max = pi;

        }

      }

      if (bprobs[max] == 0) {

        break;

      }

      double bprob = bprobs[max];

      bprobs[max] = 0; //zero out so new max can be found

      bprobSum += bprob;

      String tag = buildModel.getOutcome(max);

      //System.out.println("trying "+tag+" "+bprobSum+" lst="+lst);

      if (max == topStartIndex) { // can't have top until complete

        continue;

      }

      //System.err.println(i+" "+tag+" "+bprob);

      if (startTypeMap.containsKey(tag)) { //update last start

        lastStartIndex = advanceNodeIndex;

        lastStartNode = advanceNode;

        lastStartType = startTypeMap.get(tag);

      }

      else if (contTypeMap.containsKey(tag)) {

        if (lastStartNode == null || !lastStartType.equals(contTypeMap.get(tag))) {

          continue; //Cont must match previous start or continue

        }

      }

      Parse newParse1 = (Parse) p.clone(); //clone parse

      if (createDerivationString) newParse1.getDerivation().append(max).append("-");

      newParse1.setChild(originalAdvanceIndex,tag); //replace constituent being labeled to create new derivation

      newParse1.addProb(Math.log(bprob));

      //check

      //String[] context = checkContextGenerator.getContext(newParse1.getChildren(), lastStartType, lastStartIndex, advanceNodeIndex);

      checkModel.eval(checkContextGenerator.getContext(collapsePunctuation(newParse1.getChildren(),punctSet), lastStartType, lastStartIndex, advanceNodeIndex), cprobs);

      //System.out.println("check "+lastStartType+" "+cprobs[completeIndex]+" "+cprobs[incompleteIndex]+" "+tag+" "+java.util.Arrays.asList(context));

      Parse newParse2 = newParse1;

      if (cprobs[completeIndex] > q) { //make sure a reduce is likely

        newParse2 = (Parse) newParse1.clone();

        if (createDerivationString) newParse2.getDerivation().append(1).append(".");

        newParse2.addProb(Math.log(cprobs[completeIndex]));

        Parse[] cons = new Parse[advanceNodeIndex - lastStartIndex + 1];

        boolean flat = true;

        //first

        cons[0] = lastStartNode;

        flat &= cons[0].isPosTag();

        //last

        cons[advanceNodeIndex - lastStartIndex] = advanceNode;

        flat &= cons[advanceNodeIndex - lastStartIndex].isPosTag();

        //middle

        for (int ci = 1; ci < advanceNodeIndex - lastStartIndex; ci++) {

          cons[ci] = children[ci + lastStartIndex];

          flat &= cons[ci].isPosTag();

        }

        if (!flat) { //flat chunks are done by chunker

          if (lastStartIndex == 0 && advanceNodeIndex == numNodes-1) { //check for top node to include end and begining punctuation

            //System.err.println("ParserME.advanceParses: reducing entire span: "+new Span(lastStartNode.getSpan().getStart(), advanceNode.getSpan().getEnd())+" "+lastStartType+" "+java.util.Arrays.asList(children));

            newParse2.insert(new Parse(p.getText(), p.getSpan(), lastStartType, cprobs[1], headRules.getHead(cons, lastStartType)));

          }

          else {

            newParse2.insert(new Parse(p.getText(), new Span(lastStartNode.getSpan().getStart(), advanceNode.getSpan().getEnd()), lastStartType, cprobs[1], headRules.getHead(cons, lastStartType)));

          }

          newParsesList.add(newParse2);

        }

      }

      if (cprobs[incompleteIndex] > q) { //make sure a shift is likely

        if (createDerivationString) newParse1.getDerivation().append(0).append(".");

        if (advanceNodeIndex != numNodes - 1) { //can't shift last element

          newParse1.addProb(Math.log(cprobs[incompleteIndex]));

          newParsesList.add(newParse1);

        }

      }

    }

    Parse[] newParses = new Parse[newParsesList.size()];

    newParsesList.toArray(newParses);

    return newParses;

  }

  /**

   * @deprecated Please do not use anymore, use the ObjectStream train methods instead! This method

   * will be removed soon.

   */

  @Deprecated

  public static AbstractModel train(opennlp.model.EventStream es, int iterations, int cut) throws java.io.IOException {

    return opennlp.maxent.GIS.trainModel(iterations, new TwoPassDataIndexer(es, cut));

  }

  public static void mergeReportIntoManifest(Map<String, String> manifest,

      Map<String, String> report, String namespace) {

   

    for (Map.Entry<String, String> entry : report.entrySet()) {

      manifest.put(namespace + "." + entry.getKey(), entry.getValue());

    }

  }

 

  public static ParserModel train(String languageCode, ObjectStream<Parse> parseSamples, HeadRules rules, TrainingParameters mlParams)

          throws IOException {

   

    System.err.println("Building dictionary");

   

    Dictionary mdict = buildDictionary(parseSamples, rules, mlParams);

   

    parseSamples.reset();

   

    Map<String, String> manifestInfoEntries = new HashMap<String, String>();

   

    // build

    System.err.println("Training builder");

    opennlp.model.EventStream bes = new ParserEventStream(parseSamples, rules, ParserEventTypeEnum.BUILD, mdict);

    Map<String, String> buildReportMap = new HashMap<String, String>();

    AbstractModel buildModel = TrainUtil.train(bes, mlParams.getSettings("build"), buildReportMap);

    mergeReportIntoManifest(manifestInfoEntries, buildReportMap, "build");

   

    parseSamples.reset();

   

    // tag

    POSModel posModel = POSTaggerME.train(languageCode, new PosSampleStream(parseSamples),

        mlParams.getParameters("tagger"), null, null);

   

    parseSamples.reset();

   

    // chunk

    ChunkerModel chunkModel = ChunkerME.train(languageCode,

        new ChunkSampleStream(parseSamples),

        new ChunkContextGenerator(), mlParams.getParameters("chunker"));

   

    parseSamples.reset();

   

    // check

    System.err.println("Training checker");

    opennlp.model.EventStream kes = new ParserEventStream(parseSamples, rules, ParserEventTypeEnum.CHECK);

    Map<String, String> checkReportMap = new HashMap<String, String>();

    AbstractModel checkModel = TrainUtil.train(kes, mlParams.getSettings("check"), checkReportMap);

    mergeReportIntoManifest(manifestInfoEntries, checkReportMap, "check");

    // TODO: Remove cast for HeadRules

    return new ParserModel(languageCode, buildModel, checkModel,

        posModel, chunkModel, (opennlp.tools.parser.lang.en.HeadRules) rules,

        ParserType.CHUNKING, manifestInfoEntries);

  }

  /**

  * @deprecated use {@link #train(String, ObjectStream, HeadRules, TrainingParameters)}

  * instead and pass in a TrainingParameters object.

  */

  @Deprecated

  public static ParserModel train(String languageCode, ObjectStream<Parse> parseSamples, HeadRules rules, int iterations, int cut)

      throws IOException {

   

    TrainingParameters params = new TrainingParameters();

    params.put("dict", TrainingParameters.CUTOFF_PARAM, Integer.toString(cut));

    params.put("tagger", TrainingParameters.CUTOFF_PARAM, Integer.toString(cut));

    params.put("tagger", TrainingParameters.ITERATIONS_PARAM, Integer.toString(iterations));

    params.put("chunker", TrainingParameters.CUTOFF_PARAM, Integer.toString(cut));

    params.put("chunker", TrainingParameters.ITERATIONS_PARAM, Integer.toString(iterations));

    params.put("check", TrainingParameters.CUTOFF_PARAM, Integer.toString(cut));

    params.put("check", TrainingParameters.ITERATIONS_PARAM, Integer.toString(iterations));

    params.put("build", TrainingParameters.CUTOFF_PARAM, Integer.toString(cut));

    params.put("build", TrainingParameters.ITERATIONS_PARAM, Integer.toString(iterations));

    return train(languageCode, parseSamples, rules, params);

  }

}