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

Example 6 with SCorpusGraph

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

the class SaltAnnotateExtractorTest method testCorpusGraph.

@Test
public void testCorpusGraph() throws Exception {
    SaltProject project = instance.extractData(resultSetProviderSingleText.getResultSet());
    assertNotNull(project);
    assertEquals(1, project.getCorpusGraphs().size());
    SCorpusGraph corpusGraph = project.getCorpusGraphs().get(0);
    assertEquals(1, corpusGraph.getCorpora().size());
    assertEquals("pcc2", corpusGraph.getCorpora().get(0).getName());
    assertEquals(1, corpusGraph.getDocuments().size());
    assertEquals("4282", corpusGraph.getDocuments().get(0).getName());
}

Example 7 with SCorpusGraph

use of org.corpus_tools.salt.common.SCorpusGraph 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;
}

Example 8 with SCorpusGraph

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

the class AutoTokQuery method analyzingQuery.

@Override
public void analyzingQuery(SaltProject saltProject) {
    List<SToken> tokens = new ArrayList<>();
    for (SCorpusGraph g : saltProject.getCorpusGraphs()) {
        if (g != null) {
            for (SDocument doc : g.getDocuments()) {
                SDocumentGraph docGraph = doc.getDocumentGraph();
                List<SNode> sNodes = docGraph.getNodes();
                if (sNodes != null) {
                    for (SNode n : sNodes) {
                        if (n instanceof SToken) {
                            tokens.add((SToken) n);
                        }
                    }
                }
            }
        }
    }
    // select one random token from the result
    if (!tokens.isEmpty()) {
        int tries = 10;
        int r = new Random().nextInt(tokens.size() - 1);
        String text = CommonHelper.getSpannedText(tokens.get(r));
        while ("".equals(text) && tries > 0) {
            r = new Random().nextInt(tokens.size() - 1);
            text = CommonHelper.getSpannedText(tokens.get(r));
            tries--;
        }
        if ("".equals(text)) {
            finalAql = null;
        } else {
            finalAql = "\"" + text + "\"";
        }
    }
}

Example 9 with SCorpusGraph

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

the class ResultViewPanel method createPanels.

private List<SingleResultPanel> createPanels(SaltProject p, int localMatchIndex, long globalOffset) {
    List<SingleResultPanel> result = new LinkedList<>();
    int i = 0;
    for (SCorpusGraph corpusGraph : p.getCorpusGraphs()) {
        SDocument doc = corpusGraph.getDocuments().get(0);
        Match m = new Match();
        if (allMatches != null && localMatchIndex >= 0 && localMatchIndex < allMatches.size()) {
            m = allMatches.get(localMatchIndex);
        }
        SingleResultPanel panel = new SingleResultPanel(doc, m, i + globalOffset, new ResolverProviderImpl(cacheResolver), ps, sui, getVisibleTokenAnnos(), segmentationName, controller, instanceConfig, initialQuery);
        i++;
        panel.setWidth("100%");
        panel.setHeight("-1px");
        result.add(panel);
    }
    return result;
}

Example 10 with SCorpusGraph

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

the class CommonHelper method getTokenAnnotationLevelSet.

public static Set<String> getTokenAnnotationLevelSet(SaltProject p) {
    Set<String> result = new TreeSet<String>();
    for (SCorpusGraph corpusGraphs : p.getCorpusGraphs()) {
        for (SDocument doc : corpusGraphs.getDocuments()) {
            SDocumentGraph g = doc.getDocumentGraph();
            result.addAll(getTokenAnnotationLevelSet(g));
        }
    }
    return result;
}