Examples with SDocumentGraph org.corpus_tools.salt.common.SDocumentGraph used on opensource projects

Example 16 with SDocumentGraph

use of org.corpus_tools.salt.common.SDocumentGraph in project ANNIS by korpling.

the class VisJsComponent method computeDisplayedRelAnnotations.

/**
 * Returns the annotations to display according to the mappings configuration.
 *
 * This will check the "relation" parameter for determining the annotations to display.
 * It also iterates over all nodes of the graph
 * matching the type.
 *
 * @param input The input for the visualizer.
 * @param type Which type of relations to include
 * @return
 */
private static List<String> computeDisplayedRelAnnotations(VisualizerInput input, String relAnnosConfiguration, Class<? extends SRelation> type) {
    if (input == null) {
        return new LinkedList<>();
    }
    SDocumentGraph graph = input.getDocument().getDocumentGraph();
    Set<String> annotationPool = getRelationLevelSet(graph, null, type);
    List<String> confAnnotations = new LinkedList<>(annotationPool);
    if (relAnnosConfiguration != null && relAnnosConfiguration.trim().length() > 0) {
        String[] confSplit = relAnnosConfiguration.split(",");
        confAnnotations.clear();
        for (String entry : confSplit) {
            entry = entry.trim();
            // is regular expression?
            if (entry.startsWith("/") && entry.endsWith("/")) {
                // go over all remaining items in our pool of all annotations and
                // check if they match
                Pattern regex = Pattern.compile(StringUtils.strip(entry, "/"));
                LinkedList<String> matchingAnnotations = new LinkedList<>();
                for (String anno : annotationPool) {
                    if (regex.matcher(anno).matches()) {
                        matchingAnnotations.add(anno);
                    }
                }
                confAnnotations.addAll(matchingAnnotations);
                annotationPool.removeAll(matchingAnnotations);
            } else {
                confAnnotations.add(entry);
                annotationPool.remove(entry);
            }
        }
    }
    return confAnnotations;
}

Example 17 with SDocumentGraph

use of org.corpus_tools.salt.common.SDocumentGraph in project ANNIS by korpling.

the class SaltAnnotateExtractorTest method testLayerNames.

@Test
public void testLayerNames() throws SQLException {
    SaltProject project = instance.extractData(resultSetProviderSingleText.getResultSet());
    assertNotNull(project);
    SDocumentGraph g = project.getCorpusGraphs().get(0).getDocuments().get(0).getDocumentGraph();
    List<SLayer> layers = new ArrayList<>(g.getLayers());
    Collections.sort(layers, new NameComparator());
    assertEquals(6, layers.size());
    assertEquals("default_ns", layers.get(0).getName());
    assertEquals("dep", layers.get(1).getName());
    assertEquals("exmaralda", layers.get(2).getName());
    assertEquals("mmax", layers.get(3).getName());
    assertEquals("rst", layers.get(4).getName());
    assertEquals("tiger", layers.get(5).getName());
}

Example 18 with SDocumentGraph

use of org.corpus_tools.salt.common.SDocumentGraph in project ANNIS by korpling.

the class SaltAnnotateExtractorTest method testLayerRelations.

@Test
public void testLayerRelations() throws SQLException {
    SaltProject project = instance.extractData(resultSetProviderSingleText.getResultSet());
    assertNotNull(project);
    SDocumentGraph g = project.getCorpusGraphs().get(0).getDocuments().get(0).getDocumentGraph();
    // dep //
    List<SRelation<SNode, SNode>> e = new ArrayList<>(g.getLayerByName("dep").get(0).getRelations());
    Collections.sort(e, new EdgeComparator());
    assertEquals(9, e.size());
    assertEquals("tok_150", e.get(0).getSource().getName());
    assertEquals("tok_151", e.get(0).getTarget().getName());
    assertEquals("tok_152", e.get(1).getSource().getName());
    assertEquals("tok_153", e.get(1).getTarget().getName());
    assertEquals("tok_156", e.get(2).getSource().getName());
    assertEquals("tok_154", e.get(2).getTarget().getName());
    assertEquals("tok_156", e.get(3).getSource().getName());
    assertEquals("tok_155", e.get(3).getTarget().getName());
    assertEquals("tok_156", e.get(4).getSource().getName());
    assertEquals("tok_157", e.get(4).getTarget().getName());
    assertEquals("tok_157", e.get(5).getSource().getName());
    assertEquals("tok_158", e.get(5).getTarget().getName());
    assertEquals("tok_158", e.get(6).getSource().getName());
    assertEquals("tok_160", e.get(6).getTarget().getName());
    assertEquals("tok_160", e.get(7).getSource().getName());
    assertEquals("tok_159", e.get(7).getTarget().getName());
    assertEquals("tok_160", e.get(8).getSource().getName());
    assertEquals("tok_161", e.get(8).getTarget().getName());
    // exmaralda //
    e = new ArrayList<>(g.getLayerByName("exmaralda").get(0).getRelations());
    Collections.sort(e, new EdgeComparator());
    assertEquals(30, e.size());
    assertEquals("Focus_newInfSeg_10", e.get(0).getSource().getName());
    assertEquals("tok_154", e.get(0).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(1).getSource().getName());
    assertEquals("tok_155", e.get(1).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(2).getSource().getName());
    assertEquals("tok_156", e.get(2).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(3).getSource().getName());
    assertEquals("tok_157", e.get(3).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(4).getSource().getName());
    assertEquals("tok_158", e.get(4).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(5).getSource().getName());
    assertEquals("tok_159", e.get(5).getTarget().getName());
    assertEquals("Focus_newInfSeg_10", e.get(6).getSource().getName());
    assertEquals("tok_160", e.get(6).getTarget().getName());
    assertEquals("Focus_newInfSeg_9", e.get(7).getSource().getName());
    assertEquals("tok_150", e.get(7).getTarget().getName());
    assertEquals("Focus_newInfSeg_9", e.get(8).getSource().getName());
    assertEquals("tok_151", e.get(8).getTarget().getName());
    assertEquals("Focus_newInfSeg_9", e.get(9).getSource().getName());
    assertEquals("tok_152", e.get(9).getTarget().getName());
    assertEquals("Inf-StatSeg_29", e.get(10).getSource().getName());
    assertEquals("tok_150", e.get(10).getTarget().getName());
    assertEquals("Inf-StatSeg_29", e.get(11).getSource().getName());
    assertEquals("tok_151", e.get(11).getTarget().getName());
    assertEquals("Inf-StatSeg_30", e.get(12).getSource().getName());
    assertEquals("tok_155", e.get(12).getTarget().getName());
    assertEquals("NPSeg_29", e.get(13).getSource().getName());
    assertEquals("tok_150", e.get(13).getTarget().getName());
    assertEquals("NPSeg_29", e.get(14).getSource().getName());
    assertEquals("tok_151", e.get(14).getTarget().getName());
    assertEquals("NPSeg_30", e.get(15).getSource().getName());
    assertEquals("tok_155", e.get(15).getTarget().getName());
    assertEquals("PPSeg_7", e.get(16).getSource().getName());
    assertEquals("tok_150", e.get(16).getTarget().getName());
    assertEquals("PPSeg_7", e.get(17).getSource().getName());
    assertEquals("tok_151", e.get(17).getTarget().getName());
    assertEquals("SentSeg_10", e.get(18).getSource().getName());
    assertEquals("tok_154", e.get(18).getTarget().getName());
    assertEquals("SentSeg_10", e.get(19).getSource().getName());
    assertEquals("tok_155", e.get(19).getTarget().getName());
    assertEquals("SentSeg_10", e.get(20).getSource().getName());
    assertEquals("tok_156", e.get(20).getTarget().getName());
    assertEquals("SentSeg_10", e.get(21).getSource().getName());
    assertEquals("tok_157", e.get(21).getTarget().getName());
    assertEquals("SentSeg_10", e.get(22).getSource().getName());
    assertEquals("tok_158", e.get(22).getTarget().getName());
    assertEquals("SentSeg_10", e.get(23).getSource().getName());
    assertEquals("tok_159", e.get(23).getTarget().getName());
    assertEquals("SentSeg_10", e.get(24).getSource().getName());
    assertEquals("tok_160", e.get(24).getTarget().getName());
    assertEquals("SentSeg_10", e.get(25).getSource().getName());
    assertEquals("tok_161", e.get(25).getTarget().getName());
    assertEquals("SentSeg_9", e.get(26).getSource().getName());
    assertEquals("tok_150", e.get(26).getTarget().getName());
    assertEquals("SentSeg_9", e.get(27).getSource().getName());
    assertEquals("tok_151", e.get(27).getTarget().getName());
    assertEquals("SentSeg_9", e.get(28).getSource().getName());
    assertEquals("tok_152", e.get(28).getTarget().getName());
    assertEquals("SentSeg_9", e.get(29).getSource().getName());
    assertEquals("tok_153", e.get(29).getTarget().getName());
    // mmax, only control samples //
    e = new ArrayList<>(g.getLayerByName("mmax").get(0).getRelations());
    Collections.sort(e, new EdgeComparator());
    assertEquals(14, e.size());
    assertEquals("primmarkSeg_60", e.get(1).getSource().getName());
    assertEquals("tok_150", e.get(1).getTarget().getName());
    assertEquals("sentenceSeg_50010", e.get(7).getSource().getName());
    assertEquals("tok_158", e.get(7).getTarget().getName());
    // tiger, only control samples //
    e = new ArrayList<>(g.getLayerByName("tiger").get(0).getRelations());
    Collections.sort(e, new EdgeComparator());
    assertEquals(17, e.size());
    assertEquals("const_59", e.get(9).getSource().getName());
    assertEquals("tok_160", e.get(9).getTarget().getName());
    assertEquals("const_61", e.get(16).getSource().getName());
    assertEquals("tok_156", e.get(16).getTarget().getName());
    // urml, only control samples //
    e = new ArrayList<>(g.getLayerByName("rst").get(0).getRelations());
    Collections.sort(e, new EdgeComparator());
    assertEquals(20, e.size());
    assertEquals("u0", e.get(0).getSource().getName());
    assertEquals("u28", e.get(0).getTarget().getName());
    assertEquals("u11", e.get(5).getSource().getName());
    assertEquals("tok_153", e.get(5).getTarget().getName());
}

Example 19 with SDocumentGraph

use of org.corpus_tools.salt.common.SDocumentGraph in project ANNIS by korpling.

the class SaltAnnotateExtractorTest method testRelationType.

@Test
public void testRelationType() throws SQLException {
    SaltProject project = instance.extractData(resultSetProviderSingleText.getResultSet());
    assertNotNull(project);
    SDocumentGraph g = project.getCorpusGraphs().get(0).getDocuments().get(0).getDocumentGraph();
    for (SRelation<? extends SNode, ? extends SNode> r : g.getRelations()) {
        if (!(r instanceof STextualRelation)) {
            assertEquals(1, r.getLayers().size());
            String layerName = r.getLayers().iterator().next().getName();
            if ("exmaralda".equals(layerName) || "urml".equals(layerName) || "mmax".equals(layerName)) {
                assertTrue("instance of SSpanningRelation", r instanceof SSpanningRelation);
            } else if ("dep".equals(layerName)) {
                assertTrue("instance of SPointingRelation", r instanceof SPointingRelation);
            } else if ("tiger".equals(layerName)) {
                assertTrue("instance of SDominanceRelation", r instanceof SDominanceRelation);
            }
        }
    }
}

Example 20 with SDocumentGraph

use of org.corpus_tools.salt.common.SDocumentGraph in project ANNIS by korpling.

the class SaltAnnotateExtractor method extractData.

@Override
public SaltProject extractData(ResultSet resultSet) throws SQLException, DataAccessException {
    SaltProject project = SaltFactory.createSaltProject();
    try {
        SCorpusGraph corpusGraph = null;
        SDocumentGraph graph = null;
        // fn: parent information (pre and component) id to node
        FastInverseMap<Long, SNode> nodeByRankID = new FastInverseMap<>();
        TreeSet<Long> allTextIDs = new TreeSet<>();
        TreeMap<Long, String> tokenTexts = new TreeMap<>();
        TreeMap<Long, SToken> tokenByIndex = new TreeMap<>();
        TreeMap<String, TreeMap<Long, String>> nodeBySegmentationPath = new TreeMap<>();
        Map<String, ComponentEntry> componentForSpan = new HashMap<>();
        // clear mapping functions for this graph
        // assumes that the result set is sorted by key, pre
        nodeByRankID.clear();
        SDocument document = null;
        AtomicInteger numberOfRelations = new AtomicInteger();
        int match_index = 0;
        SolutionKey<?> key = createSolutionKey();
        int counter = 0;
        while (resultSet.next()) {
            if (counter % 1000 == 0) {
                log.debug("handling resultset row {}", counter);
            }
            counter++;
            // List<String> annotationGraphKey =
            key.retrieveKey(resultSet);
            if (key.isNewKey()) {
                // create the text for the last graph
                if (graph != null && document != null) {
                    createMissingSpanningRelations(graph, nodeByRankID, tokenByIndex, componentForSpan, numberOfRelations);
                    createPrimaryTexts(graph, allTextIDs, tokenTexts, tokenByIndex);
                    addOrderingRelations(graph, nodeBySegmentationPath);
                }
                // new match, reset everything
                nodeByRankID.clear();
                tokenTexts.clear();
                tokenByIndex.clear();
                componentForSpan.clear();
                Integer matchstart = resultSet.getInt("matchstart");
                corpusGraph = SaltFactory.createSCorpusGraph();
                corpusGraph.setName("match_" + (match_index + matchstart));
                project.addCorpusGraph(corpusGraph);
                graph = SaltFactory.createSDocumentGraph();
                document = SaltFactory.createSDocument();
                document.setDocumentGraphLocation(org.eclipse.emf.common.util.URI.createFileURI(Files.createTempDir().getAbsolutePath()));
                List<String> path = corpusPathExtractor.extractCorpusPath(resultSet, "path");
                SCorpus toplevelCorpus = SaltFactory.createSCorpus();
                toplevelCorpus.setName(path.get(0));
                corpusGraph.addNode(toplevelCorpus);
                Validate.isTrue(path.size() >= 2, "Corpus path must be have at least two members (toplevel and document)");
                SCorpus corpus = toplevelCorpus;
                for (int i = 1; i < path.size() - 1; i++) {
                    SCorpus subcorpus = SaltFactory.createSCorpus();
                    subcorpus.setName(path.get(i));
                    corpusGraph.addSubCorpus(corpus, subcorpus);
                    corpus = subcorpus;
                }
                document.setName(path.get(path.size() - 1));
                document.setId("" + match_index);
                corpusGraph.addDocument(corpus, document);
                document.setDocumentGraph(graph);
                match_index++;
            }
            // end if new key
            // get node data
            SNode node = createOrFindNewNode(resultSet, graph, allTextIDs, tokenTexts, tokenByIndex, nodeBySegmentationPath, key, nodeByRankID);
            long rankID = longValue(resultSet, RANK_TABLE, "id");
            long componentID = longValue(resultSet, COMPONENT_TABLE, "id");
            if (!resultSet.wasNull()) {
                nodeByRankID.put(rankID, node);
                createRelation(resultSet, graph, nodeByRankID, node, numberOfRelations);
                if (node instanceof SSpan) {
                    componentForSpan.put(node.getId(), new ComponentEntry(componentID, 'c', stringValue(resultSet, COMPONENT_TABLE, "namespace"), stringValue(resultSet, COMPONENT_TABLE, "name")));
                }
            }
        }
        // the last match needs a primary text, too
        if (graph != null) {
            createMissingSpanningRelations(graph, nodeByRankID, tokenByIndex, componentForSpan, numberOfRelations);
            createPrimaryTexts(graph, allTextIDs, tokenTexts, tokenByIndex);
            addOrderingRelations(graph, nodeBySegmentationPath);
        }
    } catch (Exception ex) {
        log.error("could not map result set to SaltProject", ex);
    }
    return project;
}