2020-04-27 14:52:31 +02:00
|
|
|
|
2019-04-11 15:39:29 +02:00
|
|
|
package eu.dnetlib.dhp.transformation;
|
|
|
|
|
2021-02-08 12:19:38 +01:00
|
|
|
import static eu.dnetlib.dhp.common.Constants.MDSTORE_DATA_PATH;
|
2021-02-04 14:06:02 +01:00
|
|
|
import static org.junit.jupiter.api.Assertions.assertEquals;
|
|
|
|
|
|
|
|
import java.io.IOException;
|
|
|
|
import java.nio.file.Path;
|
|
|
|
import java.util.Map;
|
|
|
|
import java.util.stream.Collectors;
|
|
|
|
import java.util.stream.Stream;
|
|
|
|
|
2019-04-11 15:39:29 +02:00
|
|
|
import org.apache.commons.io.IOUtils;
|
2021-01-28 09:51:17 +01:00
|
|
|
import org.apache.commons.lang3.StringUtils;
|
2020-05-05 12:39:04 +02:00
|
|
|
import org.apache.spark.SparkConf;
|
2021-01-28 09:51:17 +01:00
|
|
|
import org.apache.spark.api.java.function.FilterFunction;
|
|
|
|
import org.apache.spark.sql.Dataset;
|
|
|
|
import org.apache.spark.sql.Encoder;
|
|
|
|
import org.apache.spark.sql.Encoders;
|
2020-05-05 12:39:04 +02:00
|
|
|
import org.apache.spark.sql.SparkSession;
|
2019-04-11 15:39:29 +02:00
|
|
|
import org.apache.spark.util.LongAccumulator;
|
2021-01-27 15:43:08 +01:00
|
|
|
import org.junit.jupiter.api.*;
|
2020-03-25 17:59:45 +01:00
|
|
|
import org.junit.jupiter.api.extension.ExtendWith;
|
|
|
|
import org.junit.jupiter.api.io.TempDir;
|
|
|
|
import org.mockito.junit.jupiter.MockitoExtension;
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-02-04 14:06:02 +01:00
|
|
|
import eu.dnetlib.dhp.aggregation.AbstractVocabularyTest;
|
|
|
|
import eu.dnetlib.dhp.aggregation.common.AggregationCounter;
|
|
|
|
import eu.dnetlib.dhp.model.mdstore.MetadataRecord;
|
2021-02-12 17:27:08 +01:00
|
|
|
import eu.dnetlib.dhp.transformation.xslt.DateCleaner;
|
2021-02-04 14:06:02 +01:00
|
|
|
import eu.dnetlib.dhp.transformation.xslt.XSLTTransformationFunction;
|
|
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
2020-04-28 11:23:29 +02:00
|
|
|
|
2020-03-25 17:59:45 +01:00
|
|
|
@ExtendWith(MockitoExtension.class)
|
2021-02-04 10:30:49 +01:00
|
|
|
public class TransformationJobTest extends AbstractVocabularyTest {
|
2019-11-04 17:41:01 +01:00
|
|
|
|
2021-02-04 10:30:49 +01:00
|
|
|
@BeforeEach
|
|
|
|
public void setUp() throws IOException, ISLookUpException {
|
|
|
|
setUpVocabulary();
|
|
|
|
}
|
|
|
|
|
2021-02-12 16:34:52 +01:00
|
|
|
@Test
|
|
|
|
@DisplayName("Test Date cleaner")
|
|
|
|
public void testDateCleaner() throws Exception {
|
|
|
|
DateCleaner dc = new DateCleaner();
|
2021-02-12 17:27:08 +01:00
|
|
|
assertEquals(dc.clean("20/09/1982"), "1982-09-20");
|
|
|
|
assertEquals(dc.clean("20-09-2002"), "2002-09-20");
|
|
|
|
assertEquals(dc.clean("2002-09-20"), "2002-09-20");
|
|
|
|
assertEquals(dc.clean("2002-9"), "2002-09-01");
|
|
|
|
assertEquals(dc.clean("2021"), "2021-01-01");
|
2021-02-12 16:34:52 +01:00
|
|
|
}
|
|
|
|
|
2020-04-27 14:52:31 +02:00
|
|
|
@Test
|
2021-01-27 15:43:08 +01:00
|
|
|
@DisplayName("Test Transform Single XML using XSLTTransformator")
|
2020-04-27 14:52:31 +02:00
|
|
|
public void testTransformSaxonHE() throws Exception {
|
|
|
|
|
2021-01-27 15:43:08 +01:00
|
|
|
// We Set the input Record getting the XML from the classpath
|
|
|
|
final MetadataRecord mr = new MetadataRecord();
|
2021-02-12 11:48:11 +01:00
|
|
|
mr.setBody(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dhp/transform/input_zenodo.xml")));
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-01-28 09:51:17 +01:00
|
|
|
// We Load the XSLT transformation Rule from the classpath
|
2021-02-12 11:48:11 +01:00
|
|
|
XSLTTransformationFunction tr = loadTransformationRule("/eu/dnetlib/dhp/transform/zenodo_tr.xslt");
|
|
|
|
|
|
|
|
MetadataRecord result = tr.call(mr);
|
|
|
|
|
2021-01-28 09:51:17 +01:00
|
|
|
// Print the record
|
2021-02-12 11:48:11 +01:00
|
|
|
System.out.println(result.getBody());
|
2021-01-28 09:51:17 +01:00
|
|
|
// TODO Create significant Assert
|
2020-04-27 14:52:31 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
@Test
|
2021-02-24 15:07:59 +01:00
|
|
|
@DisplayName("Test TransformSparkJobNode.main")
|
2020-04-27 14:52:31 +02:00
|
|
|
public void transformTest(@TempDir Path testDir) throws Exception {
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
conf.setAppName(TransformationJobTest.class.getSimpleName());
|
|
|
|
conf.setMaster("local");
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-02-25 12:28:56 +01:00
|
|
|
try (SparkSession spark = SparkSession.builder().config(conf).getOrCreate()) {
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-02-25 12:28:56 +01:00
|
|
|
final String mdstore_input = this
|
|
|
|
.getClass()
|
|
|
|
.getResource("/eu/dnetlib/dhp/transform/mdstorenative")
|
|
|
|
.getFile();
|
2021-02-24 15:07:59 +01:00
|
|
|
final String mdstore_output = testDir.toString() + "/version";
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
mockupTrasformationRule("simpleTRule", "/eu/dnetlib/dhp/transform/ext_simple.xsl");
|
2021-01-27 15:43:08 +01:00
|
|
|
|
2021-02-25 12:28:56 +01:00
|
|
|
final Map<String, String> parameters = Stream.of(new String[][] {
|
|
|
|
{
|
|
|
|
"dateOfTransformation", "1234"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
"transformationPlugin", "XSLT_TRANSFORM"
|
|
|
|
},
|
|
|
|
{
|
|
|
|
"transformationRuleId", "simpleTRule"
|
|
|
|
},
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
}).collect(Collectors.toMap(data -> data[0], data -> data[1]));
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
TransformSparkJobNode.transformRecords(parameters, isLookUpService, spark, mdstore_input, mdstore_output);
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
// TODO introduce useful assertions
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
final Encoder<MetadataRecord> encoder = Encoders.bean(MetadataRecord.class);
|
|
|
|
final Dataset<MetadataRecord> mOutput = spark
|
2021-02-25 12:28:56 +01:00
|
|
|
.read()
|
|
|
|
.format("parquet")
|
|
|
|
.load(mdstore_output + MDSTORE_DATA_PATH)
|
|
|
|
.as(encoder);
|
2020-05-05 12:39:04 +02:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
final Long total = mOutput.count();
|
2020-04-27 14:52:31 +02:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
final long recordTs = mOutput
|
2021-02-25 12:28:56 +01:00
|
|
|
.filter((FilterFunction<MetadataRecord>) p -> p.getDateOfTransformation() == 1234)
|
|
|
|
.count();
|
2020-04-27 14:52:31 +02:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
final long recordNotEmpty = mOutput
|
2021-02-25 12:28:56 +01:00
|
|
|
.filter((FilterFunction<MetadataRecord>) p -> !StringUtils.isBlank(p.getBody()))
|
|
|
|
.count();
|
2020-04-27 14:52:31 +02:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
assertEquals(total, recordTs);
|
2020-04-27 14:52:31 +02:00
|
|
|
|
2021-02-24 15:07:59 +01:00
|
|
|
assertEquals(total, recordNotEmpty);
|
|
|
|
}
|
2020-04-27 14:52:31 +02:00
|
|
|
}
|
|
|
|
|
2021-01-27 15:43:08 +01:00
|
|
|
private XSLTTransformationFunction loadTransformationRule(final String path) throws Exception {
|
|
|
|
final String trValue = IOUtils.toString(this.getClass().getResourceAsStream(path));
|
|
|
|
final LongAccumulator la = new LongAccumulator();
|
2021-01-28 09:51:17 +01:00
|
|
|
return new XSLTTransformationFunction(new AggregationCounter(la, la, la), trValue, 0, vocabularies);
|
2021-01-27 15:43:08 +01:00
|
|
|
}
|
2020-04-27 14:52:31 +02:00
|
|
|
|
2019-04-11 15:39:29 +02:00
|
|
|
}
|