Examples with NoSuchParseException edu.stanford.nlp.parser.common.NoSuchParseException used on opensource projects

Example 1 with NoSuchParseException

use of edu.stanford.nlp.parser.common.NoSuchParseException in project CoreNLP by stanfordnlp.

the class LexicalizedParserQuery method getBestParse.

Tree getBestParse(boolean stripSubcat) {
    if (parseSkipped) {
        return null;
    }
    if (bparser != null && parseSucceeded) {
        Tree binaryTree = bparser.getBestParse();
        Tree tree = debinarizer.transformTree(binaryTree);
        if (op.nodePrune) {
            NodePruner np = new NodePruner(pparser, debinarizer);
            tree = np.prune(tree);
        }
        if (stripSubcat) {
            tree = subcategoryStripper.transformTree(tree);
        }
        restoreOriginalWords(tree);
        return tree;
    } else if (pparser != null && pparser.hasParse() && fallbackToPCFG) {
        return getBestPCFGParse();
    } else if (dparser != null && dparser.hasParse()) {
        // return subcategoryStripper.transformTree(getBestDependencyParse(true));
        return getBestDependencyParse(true);
    } else {
        throw new NoSuchParseException();
    }
}

Example 2 with NoSuchParseException

use of edu.stanford.nlp.parser.common.NoSuchParseException in project CoreNLP by stanfordnlp.

the class LexicalizedParserQuery method getKBestParses.

/**
   * Return the k best parses of the sentence most recently parsed.
   *
   * NB: The dependency parser does not implement a k-best method
   * and the factored parser's method seems to be broken and therefore
   * this method always returns a list of size 1 if either of these
   * two parsers was used.
   *
   * @return A list of scored trees
   * @throws NoSuchParseException If no previously successfully parsed
   *                                sentence   */
@Override
public List<ScoredObject<Tree>> getKBestParses(int k) {
    if (parseSkipped) {
        return null;
    }
    if (bparser != null && parseSucceeded) {
        //The getKGoodParses seems to be broken, so just return the best parse
        Tree binaryTree = bparser.getBestParse();
        Tree tree = debinarizer.transformTree(binaryTree);
        if (op.nodePrune) {
            NodePruner np = new NodePruner(pparser, debinarizer);
            tree = np.prune(tree);
        }
        tree = subcategoryStripper.transformTree(tree);
        restoreOriginalWords(tree);
        double score = dparser.getBestScore();
        ScoredObject<Tree> so = new ScoredObject<>(tree, score);
        List<ScoredObject<Tree>> trees = new ArrayList<>(1);
        trees.add(so);
        return trees;
    } else if (pparser != null && pparser.hasParse() && fallbackToPCFG) {
        return this.getKBestPCFGParses(k);
    } else if (dparser != null && dparser.hasParse()) {
        // && fallbackToDG
        // The dependency parser doesn't support k-best parse extraction, so just
        // return the best parse
        Tree tree = this.getBestDependencyParse(true);
        double score = dparser.getBestScore();
        ScoredObject<Tree> so = new ScoredObject<>(tree, score);
        List<ScoredObject<Tree>> trees = new ArrayList<>(1);
        trees.add(so);
        return trees;
    } else {
        throw new NoSuchParseException();
    }
}

Example 3 with NoSuchParseException

use of edu.stanford.nlp.parser.common.NoSuchParseException in project CoreNLP by stanfordnlp.

the class ParserAnnotator method doOneSentence.

private List<Tree> doOneSentence(List<ParserConstraint> constraints, List<CoreLabel> words) {
    ParserQuery pq = parser.parserQuery();
    pq.setConstraints(constraints);
    pq.parse(words);
    List<Tree> trees = Generics.newLinkedList();
    try {
        // Use bestParse if kBest is set to 1.
        if (this.kBest == 1) {
            Tree t = pq.getBestParse();
            if (t == null) {
                log.warn("Parsing of sentence failed.  " + "Will ignore and continue: " + SentenceUtils.listToString(words));
            } else {
                double score = pq.getBestScore();
                t.setScore(score % -10000.0);
                trees.add(t);
            }
        } else {
            List<ScoredObject<Tree>> scoredObjects = pq.getKBestParses(this.kBest);
            if (scoredObjects == null || scoredObjects.size() < 1) {
                log.info("WARNING: Parsing of sentence failed.  " + "Will ignore and continue: " + SentenceUtils.listToString(words));
            } else {
                for (ScoredObject<Tree> so : scoredObjects) {
                    // -10000 denotes unknown words
                    Tree tree = so.object();
                    tree.setScore(so.score() % -10000.0);
                    trees.add(tree);
                }
            }
        }
    } catch (OutOfMemoryError e) {
        Runtime.getRuntime().gc();
        log.info("WARNING: Parsing of sentence ran out of memory (length=" + words.size() + ").  " + "Will ignore and continue.");
    } catch (NoSuchParseException e) {
        log.info("WARNING: Parsing of sentence failed, possibly because of out of memory.  " + "Will ignore and continue: " + SentenceUtils.listToString(words));
    }
    return trees;
}

Example 4 with NoSuchParseException

use of edu.stanford.nlp.parser.common.NoSuchParseException in project CoreNLP by stanfordnlp.

the class EvaluateTreebank method processResults.

public void processResults(ParserQuery pq, Tree goldTree, PrintWriter pwErr, PrintWriter pwOut, PrintWriter pwFileOut, PrintWriter pwStats, TreePrint treePrint) {
    if (pq.saidMemMessage()) {
        saidMemMessage = true;
    }
    Tree tree;
    List<? extends HasWord> sentence = pq.originalSentence();
    try {
        tree = pq.getBestParse();
    } catch (NoSuchParseException e) {
        tree = null;
    }
    List<ScoredObject<Tree>> kbestPCFGTrees = null;
    if (tree != null && kbestPCFG > 0) {
        kbestPCFGTrees = pq.getKBestPCFGParses(kbestPCFG);
    }
    //combo parse goes to pwOut (System.out)
    if (op.testOptions.verbose) {
        pwOut.println("ComboParser best");
        Tree ot = tree;
        if (ot != null && !op.tlpParams.treebankLanguagePack().isStartSymbol(ot.value())) {
            ot = ot.treeFactory().newTreeNode(op.tlpParams.treebankLanguagePack().startSymbol(), Collections.singletonList(ot));
        }
        treePrint.printTree(ot, pwOut);
    } else {
        treePrint.printTree(tree, pwOut);
    }
    // print various statistics
    if (tree != null) {
        if (op.testOptions.printAllBestParses) {
            List<ScoredObject<Tree>> parses = pq.getBestPCFGParses();
            int sz = parses.size();
            if (sz > 1) {
                pwOut.println("There were " + sz + " best PCFG parses with score " + parses.get(0).score() + '.');
                Tree transGoldTree = collinizer.transformTree(goldTree);
                int iii = 0;
                for (ScoredObject<Tree> sot : parses) {
                    iii++;
                    Tree tb = sot.object();
                    Tree tbd = debinarizer.transformTree(tb);
                    tbd = subcategoryStripper.transformTree(tbd);
                    pq.restoreOriginalWords(tbd);
                    pwOut.println("PCFG Parse #" + iii + " with score " + tbd.score());
                    tbd.pennPrint(pwOut);
                    Tree tbtr = collinizer.transformTree(tbd);
                    // pwOut.println("Tree size = " + tbtr.size() + "; depth = " + tbtr.depth());
                    kGoodLB.evaluate(tbtr, transGoldTree, pwErr);
                }
            }
        } else // Huang and Chiang (2006) Algorithm 3 output from the PCFG parser
        if (op.testOptions.printPCFGkBest > 0 && op.testOptions.outputkBestEquivocation == null) {
            List<ScoredObject<Tree>> trees = kbestPCFGTrees.subList(0, op.testOptions.printPCFGkBest);
            Tree transGoldTree = collinizer.transformTree(goldTree);
            int i = 0;
            for (ScoredObject<Tree> tp : trees) {
                i++;
                pwOut.println("PCFG Parse #" + i + " with score " + tp.score());
                Tree tbd = tp.object();
                tbd.pennPrint(pwOut);
                Tree tbtr = collinizer.transformTree(tbd);
                kGoodLB.evaluate(tbtr, transGoldTree, pwErr);
            }
        } else // Chart parser (factored) n-best list
        if (op.testOptions.printFactoredKGood > 0 && pq.hasFactoredParse()) {
            // DZ: debug n best trees
            List<ScoredObject<Tree>> trees = pq.getKGoodFactoredParses(op.testOptions.printFactoredKGood);
            Tree transGoldTree = collinizer.transformTree(goldTree);
            int ii = 0;
            for (ScoredObject<Tree> tp : trees) {
                ii++;
                pwOut.println("Factored Parse #" + ii + " with score " + tp.score());
                Tree tbd = tp.object();
                tbd.pennPrint(pwOut);
                Tree tbtr = collinizer.transformTree(tbd);
                kGoodLB.evaluate(tbtr, transGoldTree, pwOut);
            }
        } else //1-best output
        if (pwFileOut != null) {
            pwFileOut.println(tree.toString());
        }
        //Print the derivational entropy
        if (op.testOptions.outputkBestEquivocation != null && op.testOptions.printPCFGkBest > 0) {
            List<ScoredObject<Tree>> trees = kbestPCFGTrees.subList(0, op.testOptions.printPCFGkBest);
            double[] logScores = new double[trees.size()];
            int treeId = 0;
            for (ScoredObject<Tree> kBestTree : trees) logScores[treeId++] = kBestTree.score();
            //Re-normalize
            double entropy = 0.0;
            double denom = ArrayMath.logSum(logScores);
            for (double logScore : logScores) {
                double logPr = logScore - denom;
                entropy += Math.exp(logPr) * (logPr / Math.log(2));
            }
            //Convert to bits
            entropy *= -1;
            pwStats.printf("%f\t%d\t%d\n", entropy, trees.size(), sentence.size());
        }
    }
    // Perform various evaluations specified by the user
    if (tree != null) {
        //Strip subcategories and remove punctuation for evaluation
        tree = subcategoryStripper.transformTree(tree);
        Tree treeFact = collinizer.transformTree(tree);
        //Setup the gold tree
        if (op.testOptions.verbose) {
            pwOut.println("Correct parse");
            treePrint.printTree(goldTree, pwOut);
        }
        Tree transGoldTree = collinizer.transformTree(goldTree);
        if (transGoldTree != null)
            transGoldTree = subcategoryStripper.transformTree(transGoldTree);
        //Can't do evaluation in these two cases
        if (transGoldTree == null) {
            pwErr.println("Couldn't transform gold tree for evaluation, skipping eval. Gold tree was:");
            goldTree.pennPrint(pwErr);
            numSkippedEvals++;
            return;
        } else if (treeFact == null) {
            pwErr.println("Couldn't transform hypothesis tree for evaluation, skipping eval. Tree was:");
            tree.pennPrint(pwErr);
            numSkippedEvals++;
            return;
        } else if (treeFact.yield().size() != transGoldTree.yield().size()) {
            List<Label> fYield = treeFact.yield();
            List<Label> gYield = transGoldTree.yield();
            pwErr.println("WARNING: Evaluation could not be performed due to gold/parsed yield mismatch.");
            pwErr.printf("  sizes: gold: %d (transf) %d (orig); parsed: %d (transf) %d (orig).%n", gYield.size(), goldTree.yield().size(), fYield.size(), tree.yield().size());
            pwErr.println("  gold: " + SentenceUtils.listToString(gYield, true));
            pwErr.println("  pars: " + SentenceUtils.listToString(fYield, true));
            numSkippedEvals++;
            return;
        }
        if (topKEvals.size() > 0) {
            List<Tree> transGuesses = new ArrayList<>();
            int kbest = Math.min(op.testOptions.evalPCFGkBest, kbestPCFGTrees.size());
            for (ScoredObject<Tree> guess : kbestPCFGTrees.subList(0, kbest)) {
                transGuesses.add(collinizer.transformTree(guess.object()));
            }
            for (BestOfTopKEval eval : topKEvals) {
                eval.evaluate(transGuesses, transGoldTree, pwErr);
            }
        }
        //PCFG eval
        Tree treePCFG = pq.getBestPCFGParse();
        if (treePCFG != null) {
            Tree treePCFGeval = collinizer.transformTree(treePCFG);
            if (pcfgLB != null) {
                pcfgLB.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgChildSpecific != null) {
                pcfgChildSpecific.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgLA != null) {
                pcfgLA.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgCB != null) {
                pcfgCB.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgDA != null) {
                // Re-index the leaves after Collinization, stripping traces, etc.
                treePCFGeval.indexLeaves(true);
                transGoldTree.indexLeaves(true);
                pcfgDA.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgTA != null) {
                pcfgTA.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgLL != null && pq.getPCFGParser() != null) {
                pcfgLL.recordScore(pq.getPCFGParser(), pwErr);
            }
            if (pcfgRUO != null) {
                pcfgRUO.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgCUO != null) {
                pcfgCUO.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
            if (pcfgCatE != null) {
                pcfgCatE.evaluate(treePCFGeval, transGoldTree, pwErr);
            }
        }
        //Dependency eval
        // todo: is treeDep really useful here, or should we really use depDAEval tree (debinarized) throughout? We use it for parse, and it sure seems like we could use it for tag eval, but maybe not factDA?
        Tree treeDep = pq.getBestDependencyParse(false);
        if (treeDep != null) {
            Tree goldTreeB = binarizerOnly.transformTree(goldTree);
            Tree goldTreeEval = goldTree.deepCopy();
            goldTreeEval.indexLeaves(true);
            goldTreeEval.percolateHeads(op.langpack().headFinder());
            Tree depDAEval = pq.getBestDependencyParse(true);
            depDAEval.indexLeaves(true);
            depDAEval.percolateHeadIndices();
            if (depDA != null) {
                depDA.evaluate(depDAEval, goldTreeEval, pwErr);
            }
            if (depTA != null) {
                Tree undoneTree = debinarizer.transformTree(treeDep);
                undoneTree = subcategoryStripper.transformTree(undoneTree);
                pq.restoreOriginalWords(undoneTree);
                // pwErr.println("subcategoryStripped tree: " + undoneTree.toStructureDebugString());
                depTA.evaluate(undoneTree, goldTree, pwErr);
            }
            if (depLL != null && pq.getDependencyParser() != null) {
                depLL.recordScore(pq.getDependencyParser(), pwErr);
            }
            Tree factTreeB;
            if (pq.hasFactoredParse()) {
                factTreeB = pq.getBestFactoredParse();
            } else {
                factTreeB = treeDep;
            }
            if (factDA != null) {
                factDA.evaluate(factTreeB, goldTreeB, pwErr);
            }
        }
        //Factored parser (1best) eval
        if (factLB != null) {
            factLB.evaluate(treeFact, transGoldTree, pwErr);
        }
        if (factChildSpecific != null) {
            factChildSpecific.evaluate(treeFact, transGoldTree, pwErr);
        }
        if (factLA != null) {
            factLA.evaluate(treeFact, transGoldTree, pwErr);
        }
        if (factTA != null) {
            factTA.evaluate(tree, boundaryRemover.transformTree(goldTree), pwErr);
        }
        if (factLL != null && pq.getFactoredParser() != null) {
            factLL.recordScore(pq.getFactoredParser(), pwErr);
        }
        if (factCB != null) {
            factCB.evaluate(treeFact, transGoldTree, pwErr);
        }
        for (Eval eval : evals) {
            eval.evaluate(treeFact, transGoldTree, pwErr);
        }
        if (parserQueryEvals != null) {
            for (ParserQueryEval eval : parserQueryEvals) {
                eval.evaluate(pq, transGoldTree, pwErr);
            }
        }
        if (op.testOptions.evalb) {
            // empty out scores just in case
            nanScores(tree);
            EvalbFormatWriter.writeEVALBline(treeFact, transGoldTree);
        }
    }
    pwErr.println();
}

Example 5 with NoSuchParseException

use of edu.stanford.nlp.parser.common.NoSuchParseException in project CoreNLP by stanfordnlp.

the class DVParserCostAndGradient method forwardPropagateTree.

private void forwardPropagateTree(Tree tree, List<String> words, IdentityHashMap<Tree, SimpleMatrix> nodeVectors, IdentityHashMap<Tree, Double> scores) {
    if (tree.isLeaf()) {
        return;
    }
    if (tree.isPreTerminal()) {
        Tree wordNode = tree.children()[0];
        String word = wordNode.label().value();
        SimpleMatrix wordVector = dvModel.getWordVector(word);
        wordVector = NeuralUtils.elementwiseApplyTanh(wordVector);
        nodeVectors.put(tree, wordVector);
        return;
    }
    for (Tree child : tree.children()) {
        forwardPropagateTree(child, words, nodeVectors, scores);
    }
    // at this point, nodeVectors contains the vectors for all of
    // the children of tree
    SimpleMatrix childVec;
    if (tree.children().length == 2) {
        childVec = NeuralUtils.concatenateWithBias(nodeVectors.get(tree.children()[0]), nodeVectors.get(tree.children()[1]));
    } else {
        childVec = NeuralUtils.concatenateWithBias(nodeVectors.get(tree.children()[0]));
    }
    if (op.trainOptions.useContextWords) {
        childVec = concatenateContextWords(childVec, tree.getSpan(), words);
    }
    SimpleMatrix W = dvModel.getWForNode(tree);
    if (W == null) {
        String error = "Could not find W for tree " + tree;
        if (op.testOptions.verbose) {
            log.info(error);
        }
        throw new NoSuchParseException(error);
    }
    SimpleMatrix currentVector = W.mult(childVec);
    currentVector = NeuralUtils.elementwiseApplyTanh(currentVector);
    nodeVectors.put(tree, currentVector);
    SimpleMatrix scoreW = dvModel.getScoreWForNode(tree);
    if (scoreW == null) {
        String error = "Could not find scoreW for tree " + tree;
        if (op.testOptions.verbose) {
            log.info(error);
        }
        throw new NoSuchParseException(error);
    }
    double score = scoreW.dot(currentVector);
    //score = NeuralUtils.sigmoid(score);
    scores.put(tree, score);
//log.info(Double.toString(score)+" ");
}