Examples with PropertyEnricherException org.eol.globi.service.PropertyEnricherException used on opensource projects

Example 1 with PropertyEnricherException

use of org.eol.globi.service.PropertyEnricherException in project eol-globi-data by jhpoelen.

the class LinkerTermMatcher method handleBatch.

private void handleBatch(final GraphDatabaseService graphDb, TermMatcher termMatcher, final Map<Long, TaxonNode> nodeMap, int counter) {
    StopWatch stopWatch = new StopWatch();
    stopWatch.start();
    String msgPrefix = "batch #" + counter / BATCH_SIZE;
    LOG.info(msgPrefix + " preparing...");
    List<String> nodeIdAndNames = new ArrayList<String>();
    for (Map.Entry<Long, TaxonNode> entry : nodeMap.entrySet()) {
        String name = entry.getKey() + "|" + entry.getValue().getName();
        nodeIdAndNames.add(name);
    }
    try {
        if (nodeIdAndNames.size() > 0) {
            termMatcher.findTermsForNames(nodeIdAndNames, new TermMatchListener() {

                @Override
                public void foundTaxonForName(Long nodeId, String name, Taxon taxon, NameType relType) {
                    TaxonNode taxonNode = nodeMap.get(nodeId);
                    if (taxonNode != null && NameType.NONE != relType && !TaxonUtil.likelyHomonym(taxon, taxonNode)) {
                        NodeUtil.connectTaxa(taxon, taxonNode, graphDb, RelTypes.forType(relType));
                    }
                }
            });
        }
    } catch (PropertyEnricherException ex) {
        LOG.error(msgPrefix + " problem matching terms", ex);
    }
    stopWatch.stop();
    LOG.info(msgPrefix + " completed in [" + stopWatch.getTime() + "] ms (" + (1.0 * stopWatch.getTime() / BATCH_SIZE) + " ms/name )");
    nodeMap.clear();
}

Example 2 with PropertyEnricherException

use of org.eol.globi.service.PropertyEnricherException in project eol-globi-data by jhpoelen.

the class TaxonCacheService method initTaxonIdMap.

private void initTaxonIdMap() throws PropertyEnricherException {
    try {
        LOG.info("taxon lookup service instantiating...");
        File luceneDir = new File(getCacheDir().getAbsolutePath(), "lucene");
        boolean preexisting = luceneDir.exists();
        createCacheDir(luceneDir, isTemporary());
        TaxonLookupServiceImpl taxonLookupService = new TaxonLookupServiceImpl(new SimpleFSDirectory(luceneDir));
        taxonLookupService.setMaxHits(getMaxTaxonLinks());
        taxonLookupService.start();
        if (!isTemporary() && preexisting) {
            LOG.info("pre-existing taxon lookup index found, no need to re-index...");
        } else {
            LOG.info("no pre-existing taxon lookup index found, re-indexing...");
            int count = 0;
            LOG.info("taxon map loading [" + taxonMapResource + "] ...");
            StopWatch watch = new StopWatch();
            watch.start();
            BufferedReader reader = createBufferedReader(taxonMapResource);
            final LabeledCSVParser labeledCSVParser = CSVTSVUtil.createLabeledTSVParser(reader);
            while (labeledCSVParser.getLine() != null) {
                Taxon provided = TaxonMapParser.parseProvidedTaxon(labeledCSVParser);
                Taxon resolved = TaxonMapParser.parseResolvedTaxon(labeledCSVParser);
                addIfNeeded(taxonLookupService, provided.getExternalId(), resolved.getExternalId());
                addIfNeeded(taxonLookupService, provided.getName(), resolved.getExternalId());
                addIfNeeded(taxonLookupService, resolved.getName(), resolved.getExternalId());
                count++;
            }
            watch.stop();
            logCacheLoadStats(watch.getTime(), count);
            LOG.info("taxon map loading [" + taxonMapResource + "] done.");
        }
        taxonLookupService.finish();
        this.taxonLookupService = taxonLookupService;
        LOG.info("taxon lookup service instantiating done.");
    } catch (IOException e) {
        throw new PropertyEnricherException("problem initiating taxon cache index", e);
    }
}

Example 3 with PropertyEnricherException

use of org.eol.globi.service.PropertyEnricherException in project eol-globi-data by jhpoelen.

the class TaxonEnricherImpl method enrichTaxon.

private Map<String, String> enrichTaxon(Map<Class, Integer> errorCounts, PropertyEnricher service, Integer errorCount, Map<String, String> properties) throws PropertyEnricherException {
    try {
        Map<String, String> enrichedProperties = service.enrich(properties);
        resetErrorCount(errorCounts, service);
        return enrichedProperties;
    } catch (PropertyEnricherException ex) {
        LOG.warn("failed to find a match for [" + properties + "] in [" + service.getClass().getSimpleName() + "]", ex);
        incrementErrorCount(errorCounts, service, errorCount);
        throw new PropertyEnricherException("re-throwing", ex);
    }
}

Example 4 with PropertyEnricherException

use of org.eol.globi.service.PropertyEnricherException in project eol-globi-data by jhpoelen.

the class ResolvingTaxonIndexTest method synonymsAddedToIndexOnce.

@Test
public final void synonymsAddedToIndexOnce() throws NodeFactoryException {
    ResolvingTaxonIndex taxonService = createTaxonService(getGraphDb());
    taxonService.setEnricher(new PropertyEnricher() {

        private boolean firstTime = true;

        @Override
        public Map<String, String> enrich(Map<String, String> properties) throws PropertyEnricherException {
            Taxon taxon = TaxonUtil.mapToTaxon(properties);
            if ("not pref".equals(taxon.getName())) {
                if (!firstTime) {
                    fail("should already have indexed [" + taxon.getName() + "]...");
                }
                taxon.setName("preferred");
                taxon.setExternalId("bla:123");
                taxon.setPath("one | two | three");
                taxon.setPathIds("1 | 2 | 3");
                firstTime = false;
            }
            return TaxonUtil.taxonToMap(taxon);
        }

        @Override
        public void shutdown() {
        }
    });
    this.taxonService = taxonService;
    Taxon taxon2 = new TaxonImpl("not pref", null);
    taxon2.setPath(null);
    TaxonNode first = this.taxonService.getOrCreateTaxon(taxon2);
    assertThat(first.getName(), is("preferred"));
    assertThat(first.getPath(), is("one | two | three"));
    assertThat(first.getPathIds(), is("1 | 2 | 3"));
    Taxon taxon1 = new TaxonImpl("not pref", null);
    taxon1.setPath(null);
    TaxonNode second = this.taxonService.getOrCreateTaxon(taxon1);
    assertThat(second.getNodeID(), is(first.getNodeID()));
    TaxonNode third = this.taxonService.getOrCreateTaxon(new TaxonImpl("not pref"));
    assertThat(third.getNodeID(), is(first.getNodeID()));
    TaxonNode foundTaxon = this.taxonService.findTaxonByName("not pref");
    assertThat(foundTaxon.getNodeID(), is(first.getNodeID()));
    foundTaxon = this.taxonService.findTaxonByName("preferred");
    assertThat(foundTaxon.getNodeID(), is(first.getNodeID()));
}

Example 5 with PropertyEnricherException

use of org.eol.globi.service.PropertyEnricherException in project eol-globi-data by jhpoelen.

the class EOLService method getEOLPageId.

private Long getEOLPageId(String name, String externalId) throws PropertyEnricherException {
    Long eolPageId = null;
    TaxonomyProvider taxonomyProvider = ExternalIdUtil.taxonomyProviderFor(externalId);
    if (taxonomyProvider != null) {
        String idNoPrefix = ExternalIdUtil.stripPrefix(taxonomyProvider, externalId);
        if (taxonomyProvider == TaxonomyProvider.EOL) {
            try {
                eolPageId = Long.parseLong(idNoPrefix);
            } catch (NumberFormatException ex) {
                throw new PropertyEnricherException("failed to parse eol id [" + idNoPrefix + "]");
            }
        } else if (EOL_TAXON_PROVIDER_MAP.containsKey(taxonomyProvider)) {
            eolPageId = getPageIdFromProvider(EOL_TAXON_PROVIDER_MAP.get(taxonomyProvider), idNoPrefix);
        }
    } else if (StringUtils.isNotBlank(name) && !PropertyAndValueDictionary.NO_NAME.equals(name)) {
        eolPageId = getPageId(name, true);
    }
    return eolPageId;
}