Examples with TsvDocument de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument used on opensource projects

Example 1 with TsvDocument

use of de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument in project webanno by webanno.

the class Tsv3XCasDocumentBuilder method of.

public static TsvDocument of(TsvSchema aSchema, JCas aJCas) {
    TsvFormatHeader format = new TsvFormatHeader("WebAnno TSV", "3.2");
    TsvDocument doc = new TsvDocument(format, aSchema, aJCas);
    // Fill document with all the sentences and tokens
    for (Sentence uimaSentence : select(aJCas, Sentence.class)) {
        TsvSentence sentence = doc.createSentence(uimaSentence);
        for (Token uimaToken : selectCovered(Token.class, uimaSentence)) {
            sentence.createToken(uimaToken);
        }
    }
    // Scan for chains
    for (Type headType : aSchema.getChainHeadTypes()) {
        for (FeatureStructure chainHead : CasUtil.selectFS(aJCas.getCas(), headType)) {
            List<AnnotationFS> elements = new ArrayList<>();
            AnnotationFS link = getFeature(chainHead, CHAIN_FIRST_FEAT, AnnotationFS.class);
            while (link != null) {
                elements.add(link);
                link = getFeature(link, CHAIN_NEXT_FEAT, AnnotationFS.class);
            }
            if (!elements.isEmpty()) {
                Type elementType = headType.getFeatureByBaseName(CHAIN_FIRST_FEAT).getRange();
                doc.createChain(headType, elementType, elements);
            }
        }
    }
    // Build indexes over the token start and end positions such that we can quickly locate
    // tokens based on their offsets.
    NavigableMap<Integer, TsvToken> tokenBeginIndex = new TreeMap<>();
    NavigableMap<Integer, TsvToken> tokenEndIndex = new TreeMap<>();
    List<TsvToken> tokens = new ArrayList<>();
    for (TsvSentence sentence : doc.getSentences()) {
        for (TsvToken token : sentence.getTokens()) {
            tokenBeginIndex.put(token.getBegin(), token);
            tokenEndIndex.put(token.getEnd(), token);
            tokens.add(token);
        }
    }
    // units.
    for (Type type : aSchema.getUimaTypes()) {
        LayerType layerType = aSchema.getLayerType(type);
        boolean addDisambiguationIdIfStacked = SPAN.equals(layerType);
        for (AnnotationFS annotation : CasUtil.select(aJCas.getCas(), type)) {
            doc.activateType(annotation.getType());
            // Get the relevant begin and end offsets for the current annotation
            int begin = annotation.getBegin();
            int end = annotation.getEnd();
            // to be sure.
            if (RELATION.equals(layerType)) {
                AnnotationFS targetFS = getFeature(annotation, FEAT_REL_TARGET, AnnotationFS.class);
                begin = targetFS.getBegin();
                end = targetFS.getEnd();
            }
            TsvToken beginToken = tokenBeginIndex.floorEntry(begin).getValue();
            TsvToken endToken = tokenEndIndex.ceilingEntry(end).getValue();
            // value obtained from the tokenBeginIndex.
            if (begin == end) {
                beginToken = endToken;
            }
            boolean singleToken = beginToken == endToken;
            boolean zeroWitdh = begin == end;
            boolean multiTokenCapable = SPAN.equals(layerType) || CHAIN.equals(layerType);
            // in either case.
            if (beginToken.getBegin() == begin && endToken.getEnd() == end) {
                doc.mapFS2Unit(annotation, beginToken);
                beginToken.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                if (multiTokenCapable) {
                    endToken.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                }
            } else if (zeroWitdh) {
                TsvSubToken t = beginToken.createSubToken(begin, min(beginToken.getEnd(), end));
                doc.mapFS2Unit(annotation, t);
                t.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
            } else {
                // the annotation.
                if (beginToken.getBegin() < begin) {
                    TsvSubToken t = beginToken.createSubToken(begin, min(beginToken.getEnd(), end));
                    doc.mapFS2Unit(annotation, t);
                    t.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                } else // If not the sub-token is ID-defining, then the begin token is ID-defining
                {
                    beginToken.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                    doc.mapFS2Unit(annotation, beginToken);
                }
                // checking if if singleToke is true.
                if (endToken.getEnd() > end) {
                    TsvSubToken t = endToken.createSubToken(max(endToken.getBegin(), begin), end);
                    t.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                    if (!singleToken) {
                        doc.mapFS2Unit(annotation, t);
                    }
                } else if (!singleToken && multiTokenCapable) {
                    endToken.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                }
            }
            // the end token
            if (multiTokenCapable && !singleToken) {
                ListIterator<TsvToken> i = tokens.listIterator(tokens.indexOf(beginToken));
                TsvToken t;
                while ((t = i.next()) != endToken) {
                    if (t != beginToken) {
                        t.addUimaAnnotation(annotation, addDisambiguationIdIfStacked);
                    }
                }
            }
            // Multi-token span annotations must get a disambiguation ID
            if (SPAN.equals(layerType) && !singleToken) {
                doc.addDisambiguationId(annotation);
            }
        }
    }
    // Scan all created units to see which columns actually contains values
    for (TsvSentence sentence : doc.getSentences()) {
        for (TsvToken token : sentence.getTokens()) {
            scanUnitForActiveColumns(token);
            scanUnitForAmbiguousSlotReferences(token);
            for (TsvSubToken subToken : token.getSubTokens()) {
                scanUnitForActiveColumns(subToken);
                scanUnitForAmbiguousSlotReferences(subToken);
            }
        }
    }
    // Activate the placeholder columns for any active types for which no other columns are
    // active.
    Set<Type> activeTypesNeedingPlaceholders = new HashSet<>(doc.getActiveTypes());
    for (TsvColumn col : doc.getActiveColumns()) {
        activeTypesNeedingPlaceholders.remove(col.uimaType);
    }
    for (TsvColumn col : doc.getSchema().getColumns()) {
        if (PLACEHOLDER.equals(col.featureType) && activeTypesNeedingPlaceholders.contains(col.uimaType)) {
            doc.activateColumn(col);
        }
    }
    return doc;
}

Example 2 with TsvDocument

use of de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument in project webanno by webanno.

the class Tsv3XDeserializer method read.

public void read(LineNumberReader aIn, JCas aJCas) throws IOException {
    deferredActions.set(new ArrayList<>());
    TsvFormatHeader format = readFormat(aIn);
    TsvSchema schema = readSchema(aIn, aJCas);
    // Read the extra blank line after the schema declaration
    String emptyLine = aIn.readLine();
    assert isEmpty(emptyLine);
    TsvDocument doc = new TsvDocument(format, schema, aJCas);
    for (TsvColumn column : schema.getColumns()) {
        doc.activateColumn(column);
        doc.activateType(column.uimaType);
    }
    readContent(aIn, doc);
    // Complete the addition of the chains
    CAS cas = aJCas.getCas();
    for (TsvChain chain : doc.getChains()) {
        if (chain.getElements().isEmpty()) {
            continue;
        }
        Iterator<AnnotationFS> linkIterator = chain.getElements().iterator();
        AnnotationFS link = linkIterator.next();
        // Create the chain head
        FeatureStructure head = cas.createFS(chain.getHeadType());
        setFeature(head, CHAIN_FIRST_FEAT, link);
        cas.addFsToIndexes(head);
        // Connect the links to each other
        AnnotationFS prevLink = link;
        while (linkIterator.hasNext()) {
            link = linkIterator.next();
            setFeature(prevLink, CHAIN_NEXT_FEAT, link);
            prevLink = link;
        }
    }
    // Run deferred actions
    for (Runnable action : deferredActions.get()) {
        action.run();
    }
}

Example 3 with TsvDocument

use of de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument in project webanno by webanno.

the class Tsv3XSerializerTest method testSingleSubTokenWithValue.

@Test
public void testSingleSubTokenWithValue() throws Exception {
    // Create test document
    JCas cas = makeJCasOneSentence("This is a test .");
    addNamedEntity(cas, 1, 3, "PER");
    // Set up TSV schema
    TsvSchema schema = new TsvSchema();
    Type namedEntityType = cas.getCasType(NamedEntity.type);
    schema.addColumn(new TsvColumn(namedEntityType, LayerType.SPAN, "value", FeatureType.PRIMITIVE));
    // Convert test document content to TSV model
    TsvDocument doc = Tsv3XCasDocumentBuilder.of(schema, cas);
    String expectedSentence = "#Text=This is a test .\n" + "1-1\t0-4\tThis\t_\t\n" + "1-1.1\t1-3\thi\tPER\t\n" + "1-2\t5-7\tis\t_\t\n" + "1-3\t8-9\ta\t_\t\n" + "1-4\t10-14\ttest\t_\t\n" + "1-5\t15-16\t.\t_\t\n";
    assertEquals(expectedSentence, doc.getSentences().get(0).toString());
}

Example 4 with TsvDocument

use of de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument in project webanno by webanno.

the class Tsv3XSerializerTest method testSingleZeroWidthTokenWithoutValue.

@Test
public void testSingleZeroWidthTokenWithoutValue() throws Exception {
    // Create test document
    JCas cas = makeJCasOneSentence("This is a test .");
    addNamedEntity(cas, 0, 0, null);
    // Set up TSV schema
    TsvSchema schema = new TsvSchema();
    Type namedEntityType = cas.getCasType(NamedEntity.type);
    schema.addColumn(new TsvColumn(namedEntityType, LayerType.SPAN, "value", FeatureType.PRIMITIVE));
    // Convert test document content to TSV model
    TsvDocument doc = Tsv3XCasDocumentBuilder.of(schema, cas);
    String expectedSentence = "#Text=This is a test .\n" + "1-1\t0-4\tThis\t_\t\n" + "1-1.1\t0-0\t\t*\t\n" + "1-2\t5-7\tis\t_\t\n" + "1-3\t8-9\ta\t_\t\n" + "1-4\t10-14\ttest\t_\t\n" + "1-5\t15-16\t.\t_\t\n";
    assertEquals(expectedSentence, doc.getSentences().get(0).toString());
}

Example 5 with TsvDocument

use of de.tudarmstadt.ukp.clarin.webanno.tsv.internal.tsv3x.model.TsvDocument in project webanno by webanno.

the class Tsv3XSerializerTest method testStackedSingleTokenWithValue.

@Test
public void testStackedSingleTokenWithValue() throws Exception {
    // Create test document
    JCas cas = makeJCasOneSentence("This is a test .");
    NamedEntity ne1 = addNamedEntity(cas, 0, 4, "PER");
    NamedEntity ne2 = addNamedEntity(cas, 0, 4, "ORG");
    // Set up TSV schema
    TsvSchema schema = new TsvSchema();
    Type namedEntityType = cas.getCasType(NamedEntity.type);
    schema.addColumn(new TsvColumn(namedEntityType, LayerType.SPAN, "value", FeatureType.PRIMITIVE));
    // Convert test document content to TSV model
    TsvDocument doc = Tsv3XCasDocumentBuilder.of(schema, cas);
    doc.getSentences().get(0).getTokens().get(0).addUimaAnnotation(ne1, true);
    doc.getSentences().get(0).getTokens().get(0).addUimaAnnotation(ne2, true);
    assertEquals("1-1\t0-4\tThis\tPER[1]|ORG[2]\t", doc.getSentences().get(0).getTokens().get(0).toString());
    String expectedSentence = "#Text=This is a test .\n" + "1-1\t0-4\tThis\tPER[1]|ORG[2]\t\n" + "1-2\t5-7\tis\t_\t\n" + "1-3\t8-9\ta\t_\t\n" + "1-4\t10-14\ttest\t_\t\n" + "1-5\t15-16\t.\t_\t\n";
    assertEquals(expectedSentence, doc.getSentences().get(0).toString());
}