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

Example 1 with NonResolvingTaxonIndex

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

the class ExporterSiteMapForNamesTest method writeSiteMapWithNames.

@Test
public void writeSiteMapWithNames() throws StudyImporterException, IOException {
    taxonIndex = new NonResolvingTaxonIndex(getGraphDb());
    Study study = nodeFactory.getOrCreateStudy(new StudyImpl("title", "source", null, "citation 123"));
    TaxonImpl homoSapiens = new TaxonImpl("Homo sapiens", "homoSapiensId");
    homoSapiens.setPath("one two three");
    final Specimen human = nodeFactory.createSpecimen(study, homoSapiens);
    TaxonImpl dogTaxon = new TaxonImpl("Canis familiaris", null);
    final Specimen dog = nodeFactory.createSpecimen(study, dogTaxon);
    human.ate(dog);
    resolveNames();
    final File baseDirNames = createBaseDir("target/sitemap/names");
    final GraphExporter siteMapForNames = new ExporterSiteMapForNames();
    siteMapForNames.export(getGraphDb(), baseDirNames.getAbsolutePath());
    assertSiteMap(baseDirNames, "http://www.globalbioticinteractions.org/?interactionType=interactsWith&sourceTaxon=Homo%20sapiens", "https://depot.globalbioticinteractions.org/snapshot/target/data/sitemap/names/sitemap.xml.gz");
}

Example 2 with NonResolvingTaxonIndex

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

the class LinkerTaxonIndexTest method findByStringWithWhitespaces.

@Test
public void findByStringWithWhitespaces() throws NodeFactoryException {
    NonResolvingTaxonIndex taxonService = new NonResolvingTaxonIndex(getGraphDb());
    taxonService.getOrCreateTaxon(setTaxonProps(new TaxonImpl("Homo sapiens")));
    resolveNames();
    new LinkerTaxonIndex(getGraphDb()).link();
    assertThat(getGraphDb().index().existsForNodes("taxonNameSuggestions"), is(true));
    Index<Node> index = getGraphDb().index().forNodes("taxonNameSuggestions");
    Query query = new TermQuery(new Term("name", "name"));
    IndexHits<Node> hits = index.query(query);
    assertThat(hits.size(), is(1));
    hits = index.query("name", "s nme~");
    assertThat(hits.size(), is(1));
    hits = index.query("name", "geRman~");
    assertThat(hits.size(), is(1));
    hits = index.query("name:geRman~ AND name:som~");
    assertThat(hits.size(), is(1));
    hits = index.query("name:hmo~ AND name:SApiens~");
    assertThat(hits.size(), is(1));
    hits = index.query("name:hmo~ AND name:sapiens~");
    assertThat(hits.size(), is(1));
    // queries are case sensitive . . . should all be lower cased.
    hits = index.query("name:HMO~ AND name:saPIENS~");
    assertThat(hits.size(), is(0));
}

Example 3 with NonResolvingTaxonIndex

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

the class NameResolverTest method iNaturalistTaxon.

@Test
public void iNaturalistTaxon() throws NodeFactoryException {
    Specimen someOtherOrganism = nodeFactory.createSpecimen(nodeFactory.createStudy(new StudyImpl("bla", null, null, null)), new TaxonImpl("Blaus bla", "INAT_TAXON:58831"));
    Specimen someOtherOrganism2 = nodeFactory.createSpecimen(nodeFactory.createStudy(new StudyImpl("bla", null, null, null)), new TaxonImpl("Redus rha", "INAT_TAXON:126777"));
    someOtherOrganism.ate(someOtherOrganism2);
    final NameResolver nameResolver = new NameResolver(getGraphDb(), new NonResolvingTaxonIndex(getGraphDb()));
    nameResolver.setBatchSize(1L);
    nameResolver.resolve();
    Taxon resolvedTaxon = taxonIndex.findTaxonById("INAT_TAXON:58831");
    assertThat(resolvedTaxon, is(notNullValue()));
    assertThat(resolvedTaxon.getExternalId(), is("INAT_TAXON:58831"));
    assertThat(resolvedTaxon.getName(), is("Blaus bla"));
    Taxon resolvedTaxon2 = taxonIndex.findTaxonByName("Blaus bla");
    assertThat(resolvedTaxon2, is(notNullValue()));
    assertThat(resolvedTaxon2.getExternalId(), is("INAT_TAXON:58831"));
}

Example 4 with NonResolvingTaxonIndex

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

the class Normalizer method resolveAndLinkTaxa.

private void resolveAndLinkTaxa(CommandLine cmdLine, GraphDatabaseService graphService) {
    if (cmdLine == null || !cmdLine.hasOption(OPTION_SKIP_RESOLVE_CITATIONS)) {
        LOG.info("resolving citations to DOIs ...");
        new LinkerDOI(graphService, new DOIResolverCache()).link();
        new LinkerDOI(graphService).link();
    } else {
        LOG.info("skipping citation resolving ...");
    }
    if (cmdLine == null || !cmdLine.hasOption(OPTION_SKIP_TAXON_CACHE)) {
        LOG.info("resolving names with taxon cache ...");
        final TaxonCacheService taxonCacheService = new TaxonCacheService("/taxa/taxonCache.tsv.gz", "/taxa/taxonMap.tsv.gz");
        try {
            ResolvingTaxonIndex index = new ResolvingTaxonIndex(taxonCacheService, graphService);
            index.setIndexResolvedTaxaOnly(true);
            TaxonFilter taxonCacheFilter = new TaxonFilter() {

                private KnownBadNameFilter knownBadNameFilter = new KnownBadNameFilter();

                @Override
                public boolean shouldInclude(Taxon taxon) {
                    return taxon != null && knownBadNameFilter.shouldInclude(taxon);
                }
            };
            new NameResolver(graphService, index, taxonCacheFilter).resolve();
            LOG.info("adding same and similar terms for resolved taxa...");
            List<Linker> linkers = new ArrayList<>();
            linkers.add(new LinkerTermMatcher(graphService, taxonCacheService));
            appendOpenTreeTaxonLinker(graphService, linkers);
            linkers.forEach(LinkUtil::doTimedLink);
            LOG.info("adding same and similar terms for resolved taxa done.");
        } finally {
            taxonCacheService.shutdown();
        }
        LOG.info("resolving names with taxon cache done.");
    } else {
        LOG.info("skipping taxon cache ...");
    }
    if (cmdLine == null || !cmdLine.hasOption(OPTION_SKIP_RESOLVE)) {
        new NameResolver(graphService, new NonResolvingTaxonIndex(graphService)).resolve();
        new TaxonInteractionIndexer(graphService).index();
    } else {
        LOG.info("skipping taxa resolving ...");
    }
    if (cmdLine == null || !cmdLine.hasOption(OPTION_SKIP_LINK)) {
        List<Linker> linkers = new ArrayList<>();
        linkers.add(new LinkerTaxonIndex(graphService));
        linkers.forEach(LinkUtil::doTimedLink);
    } else {
        LOG.info("skipping linking ...");
    }
    if (cmdLine == null || !cmdLine.hasOption(OPTION_SKIP_LINK_THUMBNAILS)) {
        LinkUtil.doTimedLink(new ImageLinker(graphService, null));
    } else {
        LOG.info("skipping linking of taxa to thumbnails ...");
    }
}

Example 5 with NonResolvingTaxonIndex

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

the class TaxonInteractionIndexerTest method indexNoNameNoMatch.

@Test
public void indexNoNameNoMatch() throws NodeFactoryException {
    Specimen human = nodeFactory.createSpecimen(nodeFactory.createStudy(new StudyImpl("bla", null, null, null)), new TaxonImpl("Homo sapiens", PropertyAndValueDictionary.NO_MATCH));
    Specimen animal = nodeFactory.createSpecimen(nodeFactory.createStudy(new StudyImpl("bla", null, null, null)), new TaxonImpl("Canis lupus", PropertyAndValueDictionary.NO_MATCH));
    human.ate(animal);
    for (int i = 0; i < 10; i++) {
        Specimen fish = nodeFactory.createSpecimen(nodeFactory.createStudy(new StudyImpl("bla", null, null, null)), new TaxonImpl("Arius felis", null));
        human.ate(fish);
    }
    assertNull(taxonIndex.findTaxonById(PropertyAndValueDictionary.NO_MATCH));
    assertNull(taxonIndex.findTaxonByName("Homo sapiens"));
    new NameResolver(getGraphDb(), new NonResolvingTaxonIndex(getGraphDb())).resolve();
    assertNotNull(taxonIndex.findTaxonByName("Homo sapiens"));
    assertNull(taxonIndex.findTaxonById(PropertyAndValueDictionary.NO_MATCH));
    assertNull(taxonIndex.findTaxonByName(PropertyAndValueDictionary.NO_NAME));
}