Search in sources :

Example 21 with InteractionListener

use of org.eol.globi.process.InteractionListener in project eol-globi-data by jhpoelen.

the class DatasetImporterForDwCATest method importRecordsFromMCZ.

@Test
public void importRecordsFromMCZ() throws StudyImporterException, URISyntaxException {
    StringBuilder actualMessage = new StringBuilder();
    URL resource = getClass().getResource("/org/globalbioticinteractions/dataset/mcz/meta.xml");
    URI archiveRoot = new File(resource.toURI()).getParentFile().toURI();
    AtomicInteger recordCounter = new AtomicInteger(0);
    DatasetImporterForDwCA studyImporterForDwCA = new DatasetImporterForDwCA(null, null);
    studyImporterForDwCA.setLogger(new NullImportLogger() {

        @Override
        public void severe(LogContext ctx, String message) {
            actualMessage.append(message);
        }
    });
    studyImporterForDwCA.setDataset(new DatasetImpl("some/namespace", archiveRoot, inStream -> inStream));
    studyImporterForDwCA.setInteractionListener(new InteractionListener() {

        @Override
        public void on(Map<String, String> interaction) throws StudyImporterException {
            for (String expectedProperty : new String[] {}) {
                assertThat("no [" + expectedProperty + "] found in " + interaction, interaction.containsKey(expectedProperty), is(true));
                assertThat("no value of [" + expectedProperty + "] found in " + interaction, interaction.get(expectedProperty), is(notNullValue()));
            }
            assertThat(interaction.get(DatasetImporterForTSV.RESOURCE_TYPES), is("http://rs.tdwg.org/dwc/terms/ResourceRelationship | http://rs.tdwg.org/dwc/terms/Occurrence"));
            recordCounter.incrementAndGet();
        }
    });
    studyImporterForDwCA.importStudy();
    assertThat(recordCounter.get(), is(0));
    assertThat(actualMessage.toString(), startsWith("[failed to handle dwc record]"));
}
Also used : NullImportLogger(org.eol.globi.tool.NullImportLogger) TaxonUtil(org.eol.globi.service.TaxonUtil) Arrays(java.util.Arrays) INTERACTION_TYPE_NAME(org.eol.globi.data.DatasetImporterForTSV.INTERACTION_TYPE_NAME) URL(java.net.URL) URISyntaxException(java.net.URISyntaxException) INTERACTION_TYPE_ID(org.eol.globi.data.DatasetImporterForTSV.INTERACTION_TYPE_ID) InteractionListener(org.eol.globi.process.InteractionListener) StringUtils(org.apache.commons.lang3.StringUtils) NullImportLogger(org.eol.globi.tool.NullImportLogger) InteractType(org.eol.globi.domain.InteractType) Is(org.hamcrest.core.Is) REFERENCE_URL(org.eol.globi.data.DatasetImporterForTSV.REFERENCE_URL) CoreMatchers.notNullValue(org.hamcrest.CoreMatchers.notNullValue) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) Map(java.util.Map) Is.is(org.hamcrest.core.Is.is) SOURCE_LIFE_STAGE_NAME(org.eol.globi.data.DatasetImporterForTSV.SOURCE_LIFE_STAGE_NAME) JsonNode(com.fasterxml.jackson.databind.JsonNode) URI(java.net.URI) DatasetImporterForDwCA.parseAssociatedSequences(org.eol.globi.data.DatasetImporterForDwCA.parseAssociatedSequences) DatasetImporterForDwCA.importAssociatedTaxaExtension(org.eol.globi.data.DatasetImporterForDwCA.importAssociatedTaxaExtension) DatasetImporterForDwCA.mapReferenceInfo(org.eol.globi.data.DatasetImporterForDwCA.mapReferenceInfo) TARGET_OCCURRENCE_ID(org.eol.globi.data.DatasetImporterForTSV.TARGET_OCCURRENCE_ID) EXTENSION_RESOURCE_RELATIONSHIP(org.eol.globi.data.DatasetImporterForDwCA.EXTENSION_RESOURCE_RELATIONSHIP) SOURCE_TAXON_NAME(org.eol.globi.service.TaxonUtil.SOURCE_TAXON_NAME) DwcTerm(org.gbif.dwc.terms.DwcTerm) IsNot.not(org.hamcrest.core.IsNot.not) TARGET_FIELD_NUMBER(org.eol.globi.data.DatasetImporterForTSV.TARGET_FIELD_NUMBER) DatasetImporterForDwCA.importResourceRelationshipExtension(org.eol.globi.data.DatasetImporterForDwCA.importResourceRelationshipExtension) CoreMatchers.containsString(org.hamcrest.CoreMatchers.containsString) REFERENCE_ID(org.eol.globi.data.DatasetImporterForTSV.REFERENCE_ID) Record(org.gbif.dwc.record.Record) Set(java.util.Set) TestCase.assertNull(junit.framework.TestCase.assertNull) Matchers.startsWith(org.hamcrest.Matchers.startsWith) DatasetImporterForDwCA.parseDynamicPropertiesForInteractionsOnly(org.eol.globi.data.DatasetImporterForDwCA.parseDynamicPropertiesForInteractionsOnly) List(java.util.List) Matchers.containsInAnyOrder(org.hamcrest.Matchers.containsInAnyOrder) EXTENSION_ASSOCIATED_TAXA(org.eol.globi.data.DatasetImporterForDwCA.EXTENSION_ASSOCIATED_TAXA) Matchers.greaterThan(org.hamcrest.Matchers.greaterThan) TARGET_BODY_PART_NAME(org.eol.globi.data.DatasetImporterForTSV.TARGET_BODY_PART_NAME) DATASET_CITATION(org.eol.globi.data.DatasetImporterForTSV.DATASET_CITATION) CoreMatchers.anyOf(org.hamcrest.CoreMatchers.anyOf) LogContext(org.eol.globi.domain.LogContext) DatasetImpl(org.globalbioticinteractions.dataset.DatasetImpl) TARGET_CATALOG_NUMBER(org.eol.globi.data.DatasetImporterForTSV.TARGET_CATALOG_NUMBER) DwcTerm.relatedResourceID(org.gbif.dwc.terms.DwcTerm.relatedResourceID) AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) SOURCE_TAXON_FAMILY(org.eol.globi.service.TaxonUtil.SOURCE_TAXON_FAMILY) HashMap(java.util.HashMap) TreeSet(java.util.TreeSet) ArrayList(java.util.ArrayList) RESOURCE_TYPES(org.eol.globi.data.DatasetImporterForTSV.RESOURCE_TYPES) MatcherAssert.assertThat(org.hamcrest.MatcherAssert.assertThat) SOURCE_OCCURRENCE_ID(org.eol.globi.data.DatasetImporterForTSV.SOURCE_OCCURRENCE_ID) CoreMatchers.nullValue(org.hamcrest.CoreMatchers.nullValue) Term(org.gbif.dwc.terms.Term) MapUtils(org.apache.commons.collections4.MapUtils) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Assert.assertTrue(org.junit.Assert.assertTrue) Archive(org.gbif.dwc.Archive) DwCAUtil(org.globalbioticinteractions.dataset.DwCAUtil) Test(org.junit.Test) IOException(java.io.IOException) TARGET_TAXON_NAME(org.eol.globi.service.TaxonUtil.TARGET_TAXON_NAME) File(java.io.File) DatasetImporterForDwCA.parseAssociatedOccurrences(org.eol.globi.data.DatasetImporterForDwCA.parseAssociatedOccurrences) TARGET_SEX_NAME(org.eol.globi.data.DatasetImporterForTSV.TARGET_SEX_NAME) TreeMap(java.util.TreeMap) DcTerm(org.gbif.dwc.terms.DcTerm) REFERENCE_CITATION(org.eol.globi.data.DatasetImporterForTSV.REFERENCE_CITATION) LogContext(org.eol.globi.domain.LogContext) CoreMatchers.containsString(org.hamcrest.CoreMatchers.containsString) DatasetImpl(org.globalbioticinteractions.dataset.DatasetImpl) URI(java.net.URI) URL(java.net.URL) REFERENCE_URL(org.eol.globi.data.DatasetImporterForTSV.REFERENCE_URL) InteractionListener(org.eol.globi.process.InteractionListener) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) File(java.io.File) Test(org.junit.Test)

Example 22 with InteractionListener

use of org.eol.globi.process.InteractionListener in project eol-globi-data by jhpoelen.

the class DatasetImporterForDwCATest method importRecords.

@Test
public void importRecords() throws StudyImporterException, URISyntaxException, IOException {
    URL resource = getClass().getResource("/org/globalbioticinteractions/dataset/dwca.zip");
    DatasetImporterForDwCA studyImporterForDwCA = new DatasetImporterForDwCA(null, null);
    DatasetImpl dataset = new DatasetImpl("some/namespace", resource.toURI(), inStream -> inStream);
    dataset.setConfig(new ObjectMapper().readTree("{ \"citation\": \"some citation\" }"));
    studyImporterForDwCA.setDataset(dataset);
    AtomicBoolean someRecords = new AtomicBoolean(false);
    Set<String> resourceTypes = new TreeSet<>();
    studyImporterForDwCA.setInteractionListener(new InteractionListener() {

        @Override
        public void on(Map<String, String> interaction) throws StudyImporterException {
            String associatedTaxa = interaction.get("http://rs.tdwg.org/dwc/terms/associatedTaxa");
            String dynamicProperties = interaction.get("http://rs.tdwg.org/dwc/terms/dynamicProperties");
            assertThat(StringUtils.isNotBlank(associatedTaxa) || StringUtils.isNotBlank(dynamicProperties), is(true));
            assertThat(interaction.get(SOURCE_TAXON_NAME), is(not(nullValue())));
            assertThat(interaction.get(TaxonUtil.TARGET_TAXON_NAME), is(not(nullValue())));
            assertThat(interaction.get(INTERACTION_TYPE_NAME), is(not(nullValue())));
            assertThat(interaction.get(DatasetImporterForTSV.DATASET_CITATION), containsString("some citation"));
            assertThat(interaction.get(DatasetImporterForTSV.DATASET_CITATION), containsString("Accessed at"));
            assertThat(interaction.get(DatasetImporterForTSV.DATASET_CITATION), containsString("dataset/dwca.zip"));
            assertThat(interaction.get(REFERENCE_ID), is(not(nullValue())));
            assertThat(interaction.get(DatasetImporterForTSV.REFERENCE_CITATION), is(not(nullValue())));
            assertThat(interaction.get(REFERENCE_URL), is(not(nullValue())));
            resourceTypes.addAll(Arrays.asList(splitByPipes(interaction.get(RESOURCE_TYPES))));
            someRecords.set(true);
        }
    });
    studyImporterForDwCA.importStudy();
    assertThat(someRecords.get(), is(true));
    assertThat(resourceTypes, containsInAnyOrder("http://rs.tdwg.org/dwc/terms/dynamicProperties", "http://rs.tdwg.org/dwc/terms/Occurrence", "http://rs.tdwg.org/dwc/terms/associatedTaxa"));
}
Also used : AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) InteractionListener(org.eol.globi.process.InteractionListener) TreeSet(java.util.TreeSet) DatasetImpl(org.globalbioticinteractions.dataset.DatasetImpl) CoreMatchers.containsString(org.hamcrest.CoreMatchers.containsString) URL(java.net.URL) REFERENCE_URL(org.eol.globi.data.DatasetImporterForTSV.REFERENCE_URL) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Test(org.junit.Test)

Example 23 with InteractionListener

use of org.eol.globi.process.InteractionListener in project eol-globi-data by jhpoelen.

the class DatasetImporterForZenodoMetadataIT method importStudy.

@Test
public void importStudy() throws IOException, StudyImporterException {
    List<Map<String, String>> links = new ArrayList<>();
    final DatasetImporterForZenodoMetadata importer = new DatasetImporterForZenodoMetadata(null, null);
    importer.setInteractionListener(new InteractionListener() {

        @Override
        public void on(Map<String, String> interaction) throws StudyImporterException {
            links.add(interaction);
        }
    });
    final DatasetImpl dataset = new DatasetImpl("some/namespace", URI.create("some:archive"), in -> in) {

        @Override
        public String getOrDefault(String key, String defaultValue) {
            return "https://sandbox.zenodo.org/api/records/?custom=%5Bobo%3ARO_0002453%5D%3A%5B%3A%5D";
        }
    };
    importer.setDataset(dataset);
    importer.importStudy();
    assertThat(links.size() > 0, Is.is(true));
}
Also used : InteractionListener(org.eol.globi.process.InteractionListener) ArrayList(java.util.ArrayList) DatasetImpl(org.globalbioticinteractions.dataset.DatasetImpl) Map(java.util.Map) Test(org.junit.Test)

Example 24 with InteractionListener

use of org.eol.globi.process.InteractionListener in project eol-globi-data by jhpoelen.

the class InteractionListenerIndexingTest method indexOnSourceOccurrenceIdOnly.

@Test
public void indexOnSourceOccurrenceIdOnly() throws StudyImporterException {
    TreeMap<Pair<String, String>, Map<String, String>> interactionsWithUnresolvedOccurrenceIds = new TreeMap<>();
    InteractionListener listener = new InteractionListenerIndexing(interactionsWithUnresolvedOccurrenceIds);
    listener.on(new TreeMap<String, String>() {

        {
            put(DatasetImporterForTSV.SOURCE_OCCURRENCE_ID, "target123");
            put(TaxonUtil.SOURCE_TAXON_NAME, "sourceName123");
        }
    });
    assertThat(interactionsWithUnresolvedOccurrenceIds.size(), Is.is(0));
}
Also used : InteractionListener(org.eol.globi.process.InteractionListener) TreeMap(java.util.TreeMap) TreeMap(java.util.TreeMap) Map(java.util.Map) Pair(org.apache.commons.lang3.tuple.Pair) Test(org.junit.Test)

Example 25 with InteractionListener

use of org.eol.globi.process.InteractionListener in project eol-globi-data by jhpoelen.

the class DatasetImporterForMetaTable method importStudy.

@Override
public void importStudy() throws StudyImporterException {
    try {
        for (JsonNode tableConfig : collectTables(dataset)) {
            Dataset datasetProxy = new DatasetProxy(dataset);
            datasetProxy.setConfig(tableConfig);
            InteractionListener interactionListener = getInteractionListener();
            final InteractionListener listener = new TableInteractionListenerProxy(datasetProxy, interactionListener);
            importTable(listener, new TableParserFactoryImpl(), tableConfig, datasetProxy, getLogger());
        }
    } catch (IOException | NodeFactoryException e) {
        String msg = "problem importing from [" + getBaseUrl() + "]";
        LogUtil.logError(getLogger(), msg, e);
        throw new StudyImporterException(msg, e);
    }
}
Also used : InteractionListener(org.eol.globi.process.InteractionListener) Dataset(org.globalbioticinteractions.dataset.Dataset) JsonNode(com.fasterxml.jackson.databind.JsonNode) IOException(java.io.IOException) DatasetProxy(org.globalbioticinteractions.dataset.DatasetProxy)

Aggregations

InteractionListener (org.eol.globi.process.InteractionListener)58 Test (org.junit.Test)51 Map (java.util.Map)46 ArrayList (java.util.ArrayList)36 TreeMap (java.util.TreeMap)26 URI (java.net.URI)24 IOException (java.io.IOException)22 JsonNode (com.fasterxml.jackson.databind.JsonNode)20 DatasetImpl (org.globalbioticinteractions.dataset.DatasetImpl)20 ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)19 AtomicInteger (java.util.concurrent.atomic.AtomicInteger)16 URL (java.net.URL)15 CoreMatchers.containsString (org.hamcrest.CoreMatchers.containsString)14 InputStream (java.io.InputStream)13 List (java.util.List)13 TaxonUtil (org.eol.globi.service.TaxonUtil)13 Archive (org.gbif.dwc.Archive)13 MatcherAssert.assertThat (org.hamcrest.MatcherAssert.assertThat)13 HashMap (java.util.HashMap)11 AtomicBoolean (java.util.concurrent.atomic.AtomicBoolean)10