Examples with TaxonFuzzySearchIndex org.eol.globi.taxon.TaxonFuzzySearchIndex used on opensource projects

Example 1 with TaxonFuzzySearchIndex

use of org.eol.globi.taxon.TaxonFuzzySearchIndex in project eol-globi-data by jhpoelen.

the class LinkerTaxonIndex method link.

public void link() {
    Index<Node> taxons = graphDb.index().forNodes("taxons");
    Index<Node> ids = graphDb.index().forNodes(INDEX_TAXON_NAMES_AND_IDS, MapUtil.stringMap(IndexManager.PROVIDER, "lucene", "type", "fulltext"));
    TaxonFuzzySearchIndex fuzzySearchIndex = new TaxonFuzzySearchIndex(graphDb);
    IndexHits<Node> hits = taxons.query("*:*");
    for (Node hit : hits) {
        List<String> taxonIds = new ArrayList<>();
        List<String> taxonPathIdsAndNames = new ArrayList<>();
        TaxonNode taxonNode = new TaxonNode(hit);
        addTaxonId(taxonIds, taxonNode);
        addPathIdAndNames(taxonPathIdsAndNames, taxonNode);
        addToFuzzyIndex(graphDb, fuzzySearchIndex, hit, taxonNode);
        Iterable<Relationship> rels = hit.getRelationships(Direction.OUTGOING, NodeUtil.asNeo4j(RelTypes.SAME_AS));
        for (Relationship rel : rels) {
            TaxonNode sameAsTaxon = new TaxonNode(rel.getEndNode());
            addTaxonId(taxonIds, sameAsTaxon);
            addPathIdAndNames(taxonPathIdsAndNames, sameAsTaxon);
            addToFuzzyIndex(graphDb, fuzzySearchIndex, hit, sameAsTaxon);
        }
        Transaction tx = graphDb.beginTx();
        try {
            taxonPathIdsAndNames.addAll(taxonIds);
            String aggregateIds = StringUtils.join(taxonPathIdsAndNames, CharsetConstant.SEPARATOR);
            ids.add(hit, PropertyAndValueDictionary.PATH, aggregateIds);
            hit.setProperty(PropertyAndValueDictionary.EXTERNAL_IDS, aggregateIds);
            String aggregateTaxonIds = StringUtils.join(taxonIds, CharsetConstant.SEPARATOR);
            hit.setProperty(PropertyAndValueDictionary.NAME_IDS, aggregateTaxonIds);
            tx.success();
        } finally {
            tx.finish();
        }
    }
    hits.close();
}

Example 2 with TaxonFuzzySearchIndex

use of org.eol.globi.taxon.TaxonFuzzySearchIndex in project eol-globi-data by jhpoelen.

the class LinkerTaxonIndexTest method linking.

@Test
public void linking() throws NodeFactoryException {
    Taxon taxonFound = new TaxonImpl("Homo sapiens", "Bar:123");
    taxonFound.setPath("Animalia | Mammalia | Homo sapiens");
    Taxon taxon = taxonIndex.getOrCreateTaxon(taxonFound);
    TaxonImpl taxon1 = new TaxonImpl("Homo sapiens also", "FOO:444");
    taxon1.setPathIds("BARZ:111 | FOOZ:777");
    NodeUtil.connectTaxa(taxon1, (TaxonNode) taxon, getGraphDb(), RelTypes.SAME_AS);
    taxon = taxonIndex.getOrCreateTaxon(new TaxonImpl("Bla blaus", null));
    taxon.setExternalId("FOO 1234");
    resolveNames();
    new LinkerTaxonIndex(getGraphDb()).link();
    IndexHits<Node> hits = getGraphDb().index().forNodes(LinkerTaxonIndex.INDEX_TAXON_NAMES_AND_IDS).query("*:*");
    Node next = hits.next();
    assertThat(new TaxonNode(next).getName(), is("Homo sapiens"));
    assertThat(hits.hasNext(), is(true));
    hits.close();
    assertSingleHit(PropertyAndValueDictionary.PATH + ":BAR\\:123");
    assertSingleHit(PropertyAndValueDictionary.PATH + ":FOO\\:444");
    assertSingleHit(PropertyAndValueDictionary.PATH + ":FOO\\:444 " + PropertyAndValueDictionary.PATH + ":BAR\\:123");
    assertSingleHit(PropertyAndValueDictionary.PATH + ":BAR\\:*");
    assertSingleHit(PropertyAndValueDictionary.PATH + ":Homo");
    assertSingleHit(PropertyAndValueDictionary.PATH + ":\"Homo sapiens\"");
    Taxon node = taxonIndex.findTaxonByName("Homo sapiens");
    assertThat(((NodeBacked) node).getUnderlyingNode().getProperty(PropertyAndValueDictionary.EXTERNAL_IDS).toString(), is("Animalia | Mammalia | Homo sapiens | BARZ:111 | FOOZ:777 | Bar:123 | FOO:444"));
    assertThat(((NodeBacked) node).getUnderlyingNode().getProperty(PropertyAndValueDictionary.NAME_IDS).toString(), is("Bar:123 | FOO:444"));
    assertThat(new TaxonFuzzySearchIndex(getGraphDb()).query("name:sapienz~").size(), is(1));
    assertThat(new TaxonFuzzySearchIndex(getGraphDb()).query("name:sapienz").size(), is(0));
}