switched automatic code formatting plugin to net.revelc.code.formatter:formatter-maven-plugin

This commit is contained in:
Claudio Atzori 2020-04-27 14:52:31 +02:00
parent d3fd05e3c5
commit a0bdbacdae
220 changed files with 20416 additions and 19639 deletions

View File

@ -1,3 +1,4 @@
package eu.dnetlib.maven.plugin.properties;
import java.io.File;
@ -20,18 +21,21 @@ public class GenerateOoziePropertiesMojo extends AbstractMojo {
public static final String PROPERTY_NAME_WF_SOURCE_DIR = "workflow.source.dir";
public static final String PROPERTY_NAME_SANDBOX_NAME = "sandboxName";
private final String[] limiters = {"dhp", "dnetlib", "eu"};
private final String[] limiters = {
"dhp", "dnetlib", "eu"
};
@Override
public void execute() throws MojoExecutionException, MojoFailureException {
if (System.getProperties().containsKey(PROPERTY_NAME_WF_SOURCE_DIR)
&& !System.getProperties().containsKey(PROPERTY_NAME_SANDBOX_NAME)) {
String generatedSandboxName =
generateSandboxName(System.getProperties().getProperty(PROPERTY_NAME_WF_SOURCE_DIR));
String generatedSandboxName = generateSandboxName(
System.getProperties().getProperty(PROPERTY_NAME_WF_SOURCE_DIR));
if (generatedSandboxName != null) {
System.getProperties().setProperty(PROPERTY_NAME_SANDBOX_NAME, generatedSandboxName);
} else {
System.out.println(
System.out
.println(
"unable to generate sandbox name from path: "
+ System.getProperties().getProperty(PROPERTY_NAME_WF_SOURCE_DIR));
}

View File

@ -9,6 +9,7 @@
* express or implied. See the License for the specific language governing permissions and
* limitations under the License.
*/
package eu.dnetlib.maven.plugin.properties;
import edu.umd.cs.findbugs.annotations.SuppressFBWarnings;
@ -70,33 +71,31 @@ public class WritePredefinedProjectProperties extends AbstractMojo {
protected File outputFile;
/**
* If true, the plugin will silently ignore any non-existent properties files, and the build will
* continue
* If true, the plugin will silently ignore any non-existent properties files, and the build will continue
*
* @parameter property="properties.quiet" default-value="true"
*/
private boolean quiet;
/**
* Comma separated list of characters to escape when writing property values. cr=carriage return,
* lf=linefeed, tab=tab. Any other values are taken literally.
* Comma separated list of characters to escape when writing property values. cr=carriage return, lf=linefeed,
* tab=tab. Any other values are taken literally.
*
* @parameter default-value="cr,lf,tab" property="properties.escapeChars"
*/
private String escapeChars;
/**
* If true, the plugin will include system properties when writing the properties file. System
* properties override both environment variables and project properties.
* If true, the plugin will include system properties when writing the properties file. System properties override
* both environment variables and project properties.
*
* @parameter default-value="false" property="properties.includeSystemProperties"
*/
private boolean includeSystemProperties;
/**
* If true, the plugin will include environment variables when writing the properties file.
* Environment variables are prefixed with "env". Environment variables override project
* properties.
* If true, the plugin will include environment variables when writing the properties file. Environment variables
* are prefixed with "env". Environment variables override project properties.
*
* @parameter default-value="false" property="properties.includeEnvironmentVariables"
*/
@ -110,8 +109,8 @@ public class WritePredefinedProjectProperties extends AbstractMojo {
private String exclude;
/**
* Comma separated set of properties to write to the properties file. If provided, only the
* properties matching those supplied here will be written to the properties file.
* Comma separated set of properties to write to the properties file. If provided, only the properties matching
* those supplied here will be written to the properties file.
*
* @parameter property="properties.include"
*/
@ -122,7 +121,9 @@ public class WritePredefinedProjectProperties extends AbstractMojo {
* @see org.apache.maven.plugin.AbstractMojo#execute()
*/
@Override
@SuppressFBWarnings({"NP_UNWRITTEN_FIELD", "UWF_UNWRITTEN_FIELD"})
@SuppressFBWarnings({
"NP_UNWRITTEN_FIELD", "UWF_UNWRITTEN_FIELD"
})
public void execute() throws MojoExecutionException, MojoFailureException {
Properties properties = new Properties();
// Add project properties
@ -437,8 +438,8 @@ public class WritePredefinedProjectProperties extends AbstractMojo {
*/
public void setIncludePropertyKeysFromFiles(String[] includePropertyKeysFromFiles) {
if (includePropertyKeysFromFiles != null) {
this.includePropertyKeysFromFiles =
Arrays.copyOf(includePropertyKeysFromFiles, includePropertyKeysFromFiles.length);
this.includePropertyKeysFromFiles = Arrays
.copyOf(includePropertyKeysFromFiles, includePropertyKeysFromFiles.length);
}
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.maven.plugin.properties;
import static eu.dnetlib.maven.plugin.properties.GenerateOoziePropertiesMojo.PROPERTY_NAME_SANDBOX_NAME;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.maven.plugin.properties;
import static eu.dnetlib.maven.plugin.properties.WritePredefinedProjectProperties.PROPERTY_PREFIX_ENV;
@ -20,7 +21,8 @@ import org.mockito.junit.jupiter.MockitoExtension;
@ExtendWith(MockitoExtension.class)
public class WritePredefinedProjectPropertiesTest {
@Mock private MavenProject mavenProject;
@Mock
private MavenProject mavenProject;
private WritePredefinedProjectProperties mojo;
@ -145,7 +147,9 @@ public class WritePredefinedProjectPropertiesTest {
includedProperties.setProperty(includedKey, "irrelevantValue");
includedProperties.store(new FileWriter(includedPropertiesFile), null);
mojo.setIncludePropertyKeysFromFiles(new String[] {includedPropertiesFile.getAbsolutePath()});
mojo.setIncludePropertyKeysFromFiles(new String[] {
includedPropertiesFile.getAbsolutePath()
});
// execute
mojo.execute();
@ -171,8 +175,11 @@ public class WritePredefinedProjectPropertiesTest {
projectProperties.setProperty(includedKey, includedValue);
doReturn(projectProperties).when(mavenProject).getProperties();
mojo.setIncludePropertyKeysFromFiles(
new String[] {"/eu/dnetlib/maven/plugin/properties/included.properties"});
mojo
.setIncludePropertyKeysFromFiles(
new String[] {
"/eu/dnetlib/maven/plugin/properties/included.properties"
});
// execute
mojo.execute();
@ -197,7 +204,9 @@ public class WritePredefinedProjectPropertiesTest {
projectProperties.setProperty(includedKey, includedValue);
doReturn(projectProperties).when(mavenProject).getProperties();
mojo.setIncludePropertyKeysFromFiles(new String[] {""});
mojo.setIncludePropertyKeysFromFiles(new String[] {
""
});
// execute
Assertions.assertThrows(MojoExecutionException.class, () -> mojo.execute());
@ -221,7 +230,9 @@ public class WritePredefinedProjectPropertiesTest {
includedProperties.setProperty(includedKey, "irrelevantValue");
includedProperties.storeToXML(new FileOutputStream(includedPropertiesFile), null);
mojo.setIncludePropertyKeysFromFiles(new String[] {includedPropertiesFile.getAbsolutePath()});
mojo.setIncludePropertyKeysFromFiles(new String[] {
includedPropertiesFile.getAbsolutePath()
});
// execute
mojo.execute();
@ -252,7 +263,9 @@ public class WritePredefinedProjectPropertiesTest {
includedProperties.setProperty(includedKey, "irrelevantValue");
includedProperties.store(new FileOutputStream(includedPropertiesFile), null);
mojo.setIncludePropertyKeysFromFiles(new String[] {includedPropertiesFile.getAbsolutePath()});
mojo.setIncludePropertyKeysFromFiles(new String[] {
includedPropertiesFile.getAbsolutePath()
});
// execute
Assertions.assertThrows(MojoExecutionException.class, () -> mojo.execute());
@ -262,7 +275,9 @@ public class WritePredefinedProjectPropertiesTest {
public void testExecuteWithQuietModeOn(@TempDir File testFolder) throws Exception {
// given
mojo.setQuiet(true);
mojo.setIncludePropertyKeysFromFiles(new String[] {"invalid location"});
mojo.setIncludePropertyKeysFromFiles(new String[] {
"invalid location"
});
// execute
mojo.execute();
@ -276,7 +291,9 @@ public class WritePredefinedProjectPropertiesTest {
@Test
public void testExecuteIncludingPropertyKeysFromInvalidFile() {
// given
mojo.setIncludePropertyKeysFromFiles(new String[] {"invalid location"});
mojo.setIncludePropertyKeysFromFiles(new String[] {
"invalid location"
});
// execute
Assertions.assertThrows(MojoExecutionException.class, () -> mojo.execute());

View File

@ -1,3 +1,4 @@
package eu.dnetlib.collector.worker.model;
import java.util.HashMap;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.data.mdstore.manager.common.model;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.data.mdstore.manager.common.model;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.data.mdstore.manager.common.model;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.data.mdstore.manager.common.model;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.application;
import com.fasterxml.jackson.databind.ObjectMapper;
@ -21,8 +22,7 @@ public class ArgumentApplicationParser implements Serializable {
public ArgumentApplicationParser(final String json_configuration) throws Exception {
final ObjectMapper mapper = new ObjectMapper();
final OptionsParameter[] configuration =
mapper.readValue(json_configuration, OptionsParameter[].class);
final OptionsParameter[] configuration = mapper.readValue(json_configuration, OptionsParameter[].class);
createOptionMap(configuration);
}
@ -32,7 +32,8 @@ public class ArgumentApplicationParser implements Serializable {
private void createOptionMap(final OptionsParameter[] configuration) {
Arrays.stream(configuration)
Arrays
.stream(configuration)
.map(
conf -> {
final Option o = new Option(conf.getParamName(), true, conf.getParamDescription());
@ -74,10 +75,11 @@ public class ArgumentApplicationParser implements Serializable {
public void parseArgument(final String[] args) throws Exception {
CommandLineParser parser = new BasicParser();
CommandLine cmd = parser.parse(options, args);
Arrays.stream(cmd.getOptions())
Arrays
.stream(cmd.getOptions())
.forEach(
it ->
objectMap.put(
it -> objectMap
.put(
it.getLongOpt(),
compressedValues.contains(it.getLongOpt())
? decompressValue(it.getValue())

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.application;
public class OptionsParameter {
@ -8,7 +9,8 @@ public class OptionsParameter {
private boolean paramRequired;
private boolean compressed;
public OptionsParameter() {}
public OptionsParameter() {
}
public String getParamName() {
return paramName;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import java.io.Serializable;
@ -6,16 +7,18 @@ import java.util.function.Supplier;
/** Provides serializable and throwing extensions to standard functional interfaces. */
public class FunctionalInterfaceSupport {
private FunctionalInterfaceSupport() {}
private FunctionalInterfaceSupport() {
}
/**
* Serializable supplier of any kind of objects. To be used withing spark processing pipelines
* when supplying functions externally.
* Serializable supplier of any kind of objects. To be used withing spark processing pipelines when supplying
* functions externally.
*
* @param <T>
*/
@FunctionalInterface
public interface SerializableSupplier<T> extends Supplier<T>, Serializable {}
public interface SerializableSupplier<T> extends Supplier<T>, Serializable {
}
/**
* Extension of consumer accepting functions throwing an exception.

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import static eu.dnetlib.dhp.common.ThrowingSupport.rethrowAsRuntimeException;
@ -16,7 +17,8 @@ import org.slf4j.LoggerFactory;
public class HdfsSupport {
private static final Logger logger = LoggerFactory.getLogger(HdfsSupport.class);
private HdfsSupport() {}
private HdfsSupport() {
}
/**
* Checks a path (file or dir) exists on HDFS.
@ -62,8 +64,8 @@ public class HdfsSupport {
public static List<String> listFiles(String path, Configuration configuration) {
logger.info("Listing files in path: {}", path);
return rethrowAsRuntimeException(
() ->
Arrays.stream(FileSystem.get(configuration).listStatus(new Path(path)))
() -> Arrays
.stream(FileSystem.get(configuration).listStatus(new Path(path)))
.filter(FileStatus::isDirectory)
.map(x -> x.getPath().toString())
.collect(Collectors.toList()));

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.ThrowingConsumer;
@ -9,12 +10,12 @@ import org.apache.spark.sql.SparkSession;
/** SparkSession utility methods. */
public class SparkSessionSupport {
private SparkSessionSupport() {}
private SparkSessionSupport() {
}
/**
* Runs a given function using SparkSession created using default builder and supplied SparkConf.
* Stops SparkSession when SparkSession is managed. Allows to reuse SparkSession created
* externally.
* Runs a given function using SparkSession created using default builder and supplied SparkConf. Stops SparkSession
* when SparkSession is managed. Allows to reuse SparkSession created externally.
*
* @param conf SparkConf instance
* @param isSparkSessionManaged When true will stop SparkSession
@ -27,9 +28,8 @@ public class SparkSessionSupport {
}
/**
* Runs a given function using SparkSession created with hive support and using default builder
* and supplied SparkConf. Stops SparkSession when SparkSession is managed. Allows to reuse
* SparkSession created externally.
* Runs a given function using SparkSession created with hive support and using default builder and supplied
* SparkConf. Stops SparkSession when SparkSession is managed. Allows to reuse SparkSession created externally.
*
* @param conf SparkConf instance
* @param isSparkSessionManaged When true will stop SparkSession
@ -45,9 +45,8 @@ public class SparkSessionSupport {
}
/**
* Runs a given function using SparkSession created using supplied builder and supplied SparkConf.
* Stops SparkSession when SparkSession is managed. Allows to reuse SparkSession created
* externally.
* Runs a given function using SparkSession created using supplied builder and supplied SparkConf. Stops
* SparkSession when SparkSession is managed. Allows to reuse SparkSession created externally.
*
* @param sparkSessionBuilder Builder of SparkSession
* @param conf SparkConf instance

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.ThrowingRunnable;
@ -6,7 +7,8 @@ import eu.dnetlib.dhp.common.FunctionalInterfaceSupport.ThrowingSupplier;
/** Exception handling utility methods. */
public class ThrowingSupport {
private ThrowingSupport() {}
private ThrowingSupport() {
}
/**
* Executes given runnable and rethrows any exceptions as RuntimeException.

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.model.mdstore;
import eu.dnetlib.dhp.utils.DHPUtils;
@ -16,8 +17,7 @@ public class MetadataRecord implements Serializable {
private String encoding;
/**
* The information about the provenance of the record see @{@link Provenance} for the model of
* this information
* The information about the provenance of the record see @{@link Provenance} for the model of this information
*/
private Provenance provenance;

View File

@ -1,11 +1,13 @@
package eu.dnetlib.dhp.model.mdstore;
import java.io.Serializable;
/**
* @author Sandro La Bruzzo
* <p>Provenace class models the provenance of the record in the metadataStore It contains the
* identifier and the name of the datasource that gives the record
* <p>
* Provenace class models the provenance of the record in the metadataStore It contains the identifier and the
* name of the datasource that gives the record
*/
public class Provenance implements Serializable {
@ -15,7 +17,8 @@ public class Provenance implements Serializable {
private String nsPrefix;
public Provenance() {}
public Provenance() {
}
public Provenance(String datasourceId, String datasourceName, String nsPrefix) {
this.datasourceId = datasourceId;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.parser.utility;
public class VtdException extends Exception {

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.parser.utility;
import com.ximpleware.AutoPilot;
@ -36,7 +37,8 @@ public class VtdUtilityParser {
final Map<String, String> currentAttributes = new HashMap<>();
if (attributes != null) {
attributes.forEach(
attributes
.forEach(
attributeKey -> {
try {
int attr = vn.getAttrVal(attributeKey);
@ -58,7 +60,8 @@ public class VtdUtilityParser {
ap.selectXPath(xpath);
while (ap.evalXPath() != -1) {
int t = vn.getText();
if (t > -1) results.add(vn.toNormalizedString(t));
if (t > -1)
results.add(vn.toNormalizedString(t));
}
return results;
} catch (Exception e) {
@ -72,7 +75,8 @@ public class VtdUtilityParser {
ap.selectXPath(xpath);
while (ap.evalXPath() != -1) {
int it = nav.getText();
if (it > -1) return nav.toNormalizedString(it);
if (it > -1)
return nav.toNormalizedString(it);
}
return null;
} catch (Exception e) {

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils;
import com.jayway.jsonpath.JsonPath;
@ -59,7 +60,8 @@ public class DHPUtils {
public static String getJPathString(final String jsonPath, final String json) {
try {
Object o = JsonPath.read(json, jsonPath);
if (o instanceof String) return (String) o;
if (o instanceof String)
return (String) o;
if (o instanceof JSONArray && ((JSONArray) o).size() > 0)
return (String) ((JSONArray) o).get(0);
return o.toString();

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils.saxon;
import net.sf.saxon.expr.XPathContext;
@ -9,8 +10,7 @@ import net.sf.saxon.trans.XPathException;
public abstract class AbstractExtensionFunction extends ExtensionFunctionDefinition {
public static String DEFAULT_SAXON_EXT_NS_URI =
"http://www.d-net.research-infrastructures.eu/saxon-extension";
public static String DEFAULT_SAXON_EXT_NS_URI = "http://www.d-net.research-infrastructures.eu/saxon-extension";
public abstract String getName();

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils.saxon;
import java.text.ParseException;
@ -13,7 +14,9 @@ import net.sf.saxon.value.StringValue;
public class ExtractYear extends AbstractExtensionFunction {
private static final String[] dateFormats = {"yyyy-MM-dd", "yyyy/MM/dd"};
private static final String[] dateFormats = {
"yyyy-MM-dd", "yyyy/MM/dd"
};
@Override
public String getName() {
@ -44,7 +47,9 @@ public class ExtractYear extends AbstractExtensionFunction {
@Override
public SequenceType[] getArgumentTypes() {
return new SequenceType[] {SequenceType.OPTIONAL_ITEM};
return new SequenceType[] {
SequenceType.OPTIONAL_ITEM
};
}
@Override

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils.saxon;
import java.text.ParseException;
@ -43,7 +44,9 @@ public class NormalizeDate extends AbstractExtensionFunction {
@Override
public SequenceType[] getArgumentTypes() {
return new SequenceType[] {SequenceType.OPTIONAL_ITEM};
return new SequenceType[] {
SequenceType.OPTIONAL_ITEM
};
}
@Override

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils.saxon;
import net.sf.saxon.expr.XPathContext;
@ -49,7 +50,9 @@ public class PickFirst extends AbstractExtensionFunction {
@Override
public SequenceType[] getArgumentTypes() {
return new SequenceType[] {SequenceType.OPTIONAL_ITEM};
return new SequenceType[] {
SequenceType.OPTIONAL_ITEM
};
}
@Override

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.utils.saxon;
import java.io.StringReader;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.message;
import com.fasterxml.jackson.core.JsonProcessingException;
@ -20,7 +21,8 @@ public class Message {
return jsonMapper.readValue(json, Message.class);
}
public Message() {}
public Message() {
}
public Message(String workflowId, String jobName, MessageType type, Map<String, String> body) {
this.workflowId = workflowId;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.message;
import com.rabbitmq.client.AMQP;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.message;
import com.rabbitmq.client.Channel;

View File

@ -1,6 +1,6 @@
package eu.dnetlib.message;
public enum MessageType {
ONGOING,
REPORT
ONGOING, REPORT
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.scholexplorer.relation;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.scholexplorer.relation;
import com.fasterxml.jackson.databind.ObjectMapper;
@ -9,8 +10,7 @@ public class RelationMapper extends HashMap<String, RelInfo> implements Serializ
public static RelationMapper load() throws Exception {
final String json =
IOUtils.toString(RelationMapper.class.getResourceAsStream("relations.json"));
final String json = IOUtils.toString(RelationMapper.class.getResourceAsStream("relations.json"));
ObjectMapper mapper = new ObjectMapper();
return mapper.readValue(json, RelationMapper.class);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.application;
import static org.junit.jupiter.api.Assertions.assertEquals;
@ -10,12 +11,13 @@ public class ArgumentApplicationParserTest {
@Test
public void testParseParameter() throws Exception {
final String jsonConfiguration =
IOUtils.toString(
final String jsonConfiguration = IOUtils
.toString(
this.getClass().getResourceAsStream("/eu/dnetlib/application/parameters.json"));
assertNotNull(jsonConfiguration);
ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
parser.parseArgument(
parser
.parseArgument(
new String[] {
"-p",
"value0",

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import static org.junit.jupiter.api.Assertions.*;
@ -65,8 +66,9 @@ public class HdfsSupportTest {
// then
assertEquals(2, paths.size());
List<String> expecteds =
Arrays.stream(new String[] {subDir1.toString(), subDir2.toString()})
List<String> expecteds = Arrays.stream(new String[] {
subDir1.toString(), subDir2.toString()
})
.sorted()
.collect(Collectors.toList());
List<String> actuals = paths.stream().sorted().collect(Collectors.toList());

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.common;
import static org.mockito.Mockito.*;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.model.mdstore;
import static org.junit.jupiter.api.Assertions.assertTrue;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.message;
import static org.junit.jupiter.api.Assertions.*;
@ -33,8 +34,7 @@ public class MessageTest {
@Test
public void toStringTest() {
final String expectedJson =
"{\"workflowId\":\"wId\",\"jobName\":\"Collection\",\"type\":\"ONGOING\",\"body\":{\"ExecutionTime\":\"30s\",\"parsedItem\":\"300\"}}";
final String expectedJson = "{\"workflowId\":\"wId\",\"jobName\":\"Collection\",\"type\":\"ONGOING\",\"body\":{\"ExecutionTime\":\"30s\",\"parsedItem\":\"300\"}}";
Message m = new Message();
m.setWorkflowId("wId");
m.setType(MessageType.ONGOING);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.scholexplorer.relation;
import org.junit.jupiter.api.Test;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.action;
import com.fasterxml.jackson.databind.annotation.JsonDeserialize;
@ -11,7 +12,8 @@ public class AtomicAction<T extends Oaf> implements Serializable {
private T payload;
public AtomicAction() {}
public AtomicAction() {
}
public AtomicAction(Class<T> clazz, T payload) {
this.clazz = clazz;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.action;
import com.fasterxml.jackson.core.JsonParser;

View File

@ -1,16 +1,11 @@
package eu.dnetlib.dhp.schema.common;
import eu.dnetlib.dhp.schema.oaf.OafEntity;
/** Actual entity types in the Graph */
public enum EntityType {
publication,
dataset,
otherresearchproduct,
software,
datasource,
organization,
project;
publication, dataset, otherresearchproduct, software, datasource, organization, project;
/**
* Resolves the EntityType, given the relative class name

View File

@ -1,9 +1,7 @@
package eu.dnetlib.dhp.schema.common;
/** Main entity types in the Graph */
public enum MainEntityType {
result,
datasource,
organization,
project
result, datasource, organization, project
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.common;
import eu.dnetlib.dhp.schema.oaf.Qualifier;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.common;
import com.google.common.collect.Maps;
@ -52,7 +53,8 @@ public class ModelSupport {
private static final String schemeTemplate = "dnet:%s_%s_relations";
private ModelSupport() {}
private ModelSupport() {
}
/**
* Checks subclass-superclass relationship.
@ -142,7 +144,8 @@ public class ModelSupport {
}
public static String getScheme(final String sourceType, final String targetType) {
return String.format(
return String
.format(
schemeTemplate,
entityMapping.get(EntityType.valueOf(sourceType)).name(),
entityMapping.get(EntityType.valueOf(targetType)).name());
@ -159,29 +162,31 @@ public class ModelSupport {
private static <T extends Oaf> String idFnForRelation(T t) {
Relation r = (Relation) t;
return Optional.ofNullable(r.getSource())
return Optional
.ofNullable(r.getSource())
.map(
source ->
Optional.ofNullable(r.getTarget())
source -> Optional
.ofNullable(r.getTarget())
.map(
target ->
Optional.ofNullable(r.getRelType())
target -> Optional
.ofNullable(r.getRelType())
.map(
relType ->
Optional.ofNullable(r.getSubRelType())
relType -> Optional
.ofNullable(r.getSubRelType())
.map(
subRelType ->
Optional.ofNullable(r.getRelClass())
subRelType -> Optional
.ofNullable(r.getRelClass())
.map(
relClass ->
String.join(
relClass -> String
.join(
source,
target,
relType,
subRelType,
relClass))
.orElse(
String.join(
String
.join(
source,
target,
relType,

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -68,8 +69,10 @@ public class Author implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
Author author = (Author) o;
return Objects.equals(fullname, author.fullname)
&& Objects.equals(name, author.name)

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -31,9 +32,12 @@ public class Context implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
Context other = (Context) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.util.Objects;
@ -16,9 +17,12 @@ public class Country extends Qualifier {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (!super.equals(o)) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
if (!super.equals(o))
return false;
Country country = (Country) o;
return Objects.equals(dataInfo, country.dataInfo);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -62,8 +63,10 @@ public class DataInfo implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
DataInfo dataInfo = (DataInfo) o;
return Objects.equals(invisible, dataInfo.invisible)
&& Objects.equals(inferred, dataInfo.inferred)
@ -75,7 +78,8 @@ public class DataInfo implements Serializable {
@Override
public int hashCode() {
return Objects.hash(
return Objects
.hash(
invisible, inferred, deletedbyinference, trust, inferenceprovenance, provenanceaction);
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import eu.dnetlib.dhp.schema.common.ModelConstants;
@ -90,8 +91,7 @@ public class Dataset extends Result implements Serializable {
final Dataset d = (Dataset) e;
storagedate =
d.getStoragedate() != null && compareTrust(this, e) < 0 ? d.getStoragedate() : storagedate;
storagedate = d.getStoragedate() != null && compareTrust(this, e) < 0 ? d.getStoragedate() : storagedate;
device = d.getDevice() != null && compareTrust(this, e) < 0 ? d.getDevice() : device;
@ -99,13 +99,11 @@ public class Dataset extends Result implements Serializable {
version = d.getVersion() != null && compareTrust(this, e) < 0 ? d.getVersion() : version;
lastmetadataupdate =
d.getLastmetadataupdate() != null && compareTrust(this, e) < 0
lastmetadataupdate = d.getLastmetadataupdate() != null && compareTrust(this, e) < 0
? d.getLastmetadataupdate()
: lastmetadataupdate;
metadataversionnumber =
d.getMetadataversionnumber() != null && compareTrust(this, e) < 0
metadataversionnumber = d.getMetadataversionnumber() != null && compareTrust(this, e) < 0
? d.getMetadataversionnumber()
: metadataversionnumber;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -372,120 +373,93 @@ public class Datasource extends OafEntity implements Serializable {
Datasource d = (Datasource) e;
datasourcetype =
d.getDatasourcetype() != null && compareTrust(this, e) < 0
datasourcetype = d.getDatasourcetype() != null && compareTrust(this, e) < 0
? d.getDatasourcetype()
: datasourcetype;
openairecompatibility =
d.getOpenairecompatibility() != null && compareTrust(this, e) < 0
openairecompatibility = d.getOpenairecompatibility() != null && compareTrust(this, e) < 0
? d.getOpenairecompatibility()
: openairecompatibility;
officialname =
d.getOfficialname() != null && compareTrust(this, e) < 0
officialname = d.getOfficialname() != null && compareTrust(this, e) < 0
? d.getOfficialname()
: officialname;
englishname =
d.getEnglishname() != null && compareTrust(this, e) < 0 ? d.getEnglishname() : officialname;
websiteurl =
d.getWebsiteurl() != null && compareTrust(this, e) < 0 ? d.getWebsiteurl() : websiteurl;
englishname = d.getEnglishname() != null && compareTrust(this, e) < 0 ? d.getEnglishname() : officialname;
websiteurl = d.getWebsiteurl() != null && compareTrust(this, e) < 0 ? d.getWebsiteurl() : websiteurl;
logourl = d.getLogourl() != null && compareTrust(this, e) < 0 ? d.getLogourl() : getLogourl();
contactemail =
d.getContactemail() != null && compareTrust(this, e) < 0
contactemail = d.getContactemail() != null && compareTrust(this, e) < 0
? d.getContactemail()
: contactemail;
namespaceprefix =
d.getNamespaceprefix() != null && compareTrust(this, e) < 0
namespaceprefix = d.getNamespaceprefix() != null && compareTrust(this, e) < 0
? d.getNamespaceprefix()
: namespaceprefix;
latitude = d.getLatitude() != null && compareTrust(this, e) < 0 ? d.getLatitude() : latitude;
longitude =
d.getLongitude() != null && compareTrust(this, e) < 0 ? d.getLongitude() : longitude;
dateofvalidation =
d.getDateofvalidation() != null && compareTrust(this, e) < 0
longitude = d.getLongitude() != null && compareTrust(this, e) < 0 ? d.getLongitude() : longitude;
dateofvalidation = d.getDateofvalidation() != null && compareTrust(this, e) < 0
? d.getDateofvalidation()
: dateofvalidation;
description =
d.getDescription() != null && compareTrust(this, e) < 0 ? d.getDescription() : description;
description = d.getDescription() != null && compareTrust(this, e) < 0 ? d.getDescription() : description;
subjects = mergeLists(subjects, d.getSubjects());
// opendoar specific fields (od*)
odnumberofitems =
d.getOdnumberofitems() != null && compareTrust(this, e) < 0
odnumberofitems = d.getOdnumberofitems() != null && compareTrust(this, e) < 0
? d.getOdnumberofitems()
: odnumberofitems;
odnumberofitemsdate =
d.getOdnumberofitemsdate() != null && compareTrust(this, e) < 0
odnumberofitemsdate = d.getOdnumberofitemsdate() != null && compareTrust(this, e) < 0
? d.getOdnumberofitemsdate()
: odnumberofitemsdate;
odpolicies =
d.getOdpolicies() != null && compareTrust(this, e) < 0 ? d.getOdpolicies() : odpolicies;
odpolicies = d.getOdpolicies() != null && compareTrust(this, e) < 0 ? d.getOdpolicies() : odpolicies;
odlanguages = mergeLists(odlanguages, d.getOdlanguages());
odcontenttypes = mergeLists(odcontenttypes, d.getOdcontenttypes());
accessinfopackage = mergeLists(accessinfopackage, d.getAccessinfopackage());
// re3data fields
releasestartdate =
d.getReleasestartdate() != null && compareTrust(this, e) < 0
releasestartdate = d.getReleasestartdate() != null && compareTrust(this, e) < 0
? d.getReleasestartdate()
: releasestartdate;
releaseenddate =
d.getReleaseenddate() != null && compareTrust(this, e) < 0
releaseenddate = d.getReleaseenddate() != null && compareTrust(this, e) < 0
? d.getReleaseenddate()
: releaseenddate;
missionstatementurl =
d.getMissionstatementurl() != null && compareTrust(this, e) < 0
missionstatementurl = d.getMissionstatementurl() != null && compareTrust(this, e) < 0
? d.getMissionstatementurl()
: missionstatementurl;
dataprovider =
d.getDataprovider() != null && compareTrust(this, e) < 0
dataprovider = d.getDataprovider() != null && compareTrust(this, e) < 0
? d.getDataprovider()
: dataprovider;
serviceprovider =
d.getServiceprovider() != null && compareTrust(this, e) < 0
serviceprovider = d.getServiceprovider() != null && compareTrust(this, e) < 0
? d.getServiceprovider()
: serviceprovider;
// {open, restricted or closed}
databaseaccesstype =
d.getDatabaseaccesstype() != null && compareTrust(this, e) < 0
databaseaccesstype = d.getDatabaseaccesstype() != null && compareTrust(this, e) < 0
? d.getDatabaseaccesstype()
: databaseaccesstype;
// {open, restricted or closed}
datauploadtype =
d.getDatauploadtype() != null && compareTrust(this, e) < 0
datauploadtype = d.getDatauploadtype() != null && compareTrust(this, e) < 0
? d.getDatauploadtype()
: datauploadtype;
// {feeRequired, registration, other}
databaseaccessrestriction =
d.getDatabaseaccessrestriction() != null && compareTrust(this, e) < 0
databaseaccessrestriction = d.getDatabaseaccessrestriction() != null && compareTrust(this, e) < 0
? d.getDatabaseaccessrestriction()
: databaseaccessrestriction;
// {feeRequired, registration, other}
datauploadrestriction =
d.getDatauploadrestriction() != null && compareTrust(this, e) < 0
datauploadrestriction = d.getDatauploadrestriction() != null && compareTrust(this, e) < 0
? d.getDatauploadrestriction()
: datauploadrestriction;
versioning =
d.getVersioning() != null && compareTrust(this, e) < 0 ? d.getVersioning() : versioning;
citationguidelineurl =
d.getCitationguidelineurl() != null && compareTrust(this, e) < 0
versioning = d.getVersioning() != null && compareTrust(this, e) < 0 ? d.getVersioning() : versioning;
citationguidelineurl = d.getCitationguidelineurl() != null && compareTrust(this, e) < 0
? d.getCitationguidelineurl()
: citationguidelineurl;
// {yes, no, unknown}
qualitymanagementkind =
d.getQualitymanagementkind() != null && compareTrust(this, e) < 0
qualitymanagementkind = d.getQualitymanagementkind() != null && compareTrust(this, e) < 0
? d.getQualitymanagementkind()
: qualitymanagementkind;
pidsystems =
d.getPidsystems() != null && compareTrust(this, e) < 0 ? d.getPidsystems() : pidsystems;
pidsystems = d.getPidsystems() != null && compareTrust(this, e) < 0 ? d.getPidsystems() : pidsystems;
certificates =
d.getCertificates() != null && compareTrust(this, e) < 0
certificates = d.getCertificates() != null && compareTrust(this, e) < 0
? d.getCertificates()
: certificates;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -94,8 +95,10 @@ public class ExternalReference implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
ExternalReference that = (ExternalReference) o;
return Objects.equals(sitename, that.sitename)
&& Objects.equals(label, that.label)
@ -109,7 +112,8 @@ public class ExternalReference implements Serializable {
@Override
public int hashCode() {
return Objects.hash(
return Objects
.hash(
sitename, label, url, description, qualifier, refidentifier, query, dataInfo);
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -57,8 +58,10 @@ public class ExtraInfo implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
ExtraInfo extraInfo = (ExtraInfo) o;
return Objects.equals(name, extraInfo.name)
&& Objects.equals(typology, extraInfo.typology)

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -31,9 +32,12 @@ public class Field<T> implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
Field<T> other = (Field<T>) obj;
return getValue().equals(other.getValue());
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import com.fasterxml.jackson.annotation.JsonIgnore;
@ -44,7 +45,8 @@ public class GeoLocation implements Serializable {
public String toComparableString() {
return isBlank()
? ""
: String.format(
: String
.format(
"%s::%s%s",
point != null ? point.toLowerCase() : "",
box != null ? box.toLowerCase() : "",
@ -58,9 +60,12 @@ public class GeoLocation implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
GeoLocation other = (GeoLocation) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -121,7 +122,8 @@ public class Instance implements Serializable {
}
public String toComparableString() {
return String.format(
return String
.format(
"%s::%s::%s::%s",
hostedby != null && hostedby.getKey() != null ? hostedby.getKey().toLowerCase() : "",
accessright != null && accessright.getClassid() != null ? accessright.getClassid() : "",
@ -136,9 +138,12 @@ public class Instance implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
Instance other = (Instance) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -127,8 +128,10 @@ public class Journal implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
Journal journal = (Journal) o;
return Objects.equals(name, journal.name)
&& Objects.equals(issnPrinted, journal.issnPrinted)
@ -146,7 +149,8 @@ public class Journal implements Serializable {
@Override
public int hashCode() {
return Objects.hash(
return Objects
.hash(
name,
issnPrinted,
issnOnline,

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import com.fasterxml.jackson.annotation.JsonIgnore;
@ -39,7 +40,8 @@ public class KeyValue implements Serializable {
public String toComparableString() {
return isBlank()
? ""
: String.format(
: String
.format(
"%s::%s",
key != null ? key.toLowerCase() : "", value != null ? value.toLowerCase() : "");
}
@ -56,9 +58,12 @@ public class KeyValue implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
KeyValue other = (KeyValue) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -17,8 +18,10 @@ public class OAIProvenance implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
OAIProvenance that = (OAIProvenance) o;
return Objects.equals(originDescription, that.originDescription);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -37,11 +38,13 @@ public abstract class Oaf implements Serializable {
}
public void mergeOAFDataInfo(Oaf e) {
if (e.getDataInfo() != null && compareTrust(this, e) < 0) dataInfo = e.getDataInfo();
if (e.getDataInfo() != null && compareTrust(this, e) < 0)
dataInfo = e.getDataInfo();
}
protected String extractTrust(Oaf e) {
if (e == null || e.getDataInfo() == null || e.getDataInfo().getTrust() == null) return "0.0";
if (e == null || e.getDataInfo() == null || e.getDataInfo().getTrust() == null)
return "0.0";
return e.getDataInfo().getTrust();
}
@ -51,8 +54,10 @@ public abstract class Oaf implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
Oaf oaf = (Oaf) o;
return Objects.equals(dataInfo, oaf.dataInfo)
&& Objects.equals(lastupdatetimestamp, oaf.lastupdatetimestamp);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -78,7 +79,8 @@ public abstract class OafEntity extends Oaf implements Serializable {
public void mergeFrom(OafEntity e) {
if (e == null) return;
if (e == null)
return;
originalId = mergeLists(originalId, e.getOriginalId());
@ -100,7 +102,8 @@ public abstract class OafEntity extends Oaf implements Serializable {
protected <T> List<T> mergeLists(final List<T>... lists) {
return Arrays.stream(lists)
return Arrays
.stream(lists)
.filter(Objects::nonNull)
.flatMap(List::stream)
.distinct()
@ -109,9 +112,12 @@ public abstract class OafEntity extends Oaf implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (!super.equals(o)) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
if (!super.equals(o))
return false;
OafEntity oafEntity = (OafEntity) o;
return Objects.equals(id, oafEntity.id);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -175,50 +176,38 @@ public class Organization extends OafEntity implements Serializable {
}
final Organization o = (Organization) e;
legalshortname =
o.getLegalshortname() != null && compareTrust(this, e) < 0
legalshortname = o.getLegalshortname() != null && compareTrust(this, e) < 0
? o.getLegalshortname()
: legalshortname;
legalname =
o.getLegalname() != null && compareTrust(this, e) < 0 ? o.getLegalname() : legalname;
legalname = o.getLegalname() != null && compareTrust(this, e) < 0 ? o.getLegalname() : legalname;
alternativeNames = mergeLists(o.getAlternativeNames(), alternativeNames);
websiteurl =
o.getWebsiteurl() != null && compareTrust(this, e) < 0 ? o.getWebsiteurl() : websiteurl;
websiteurl = o.getWebsiteurl() != null && compareTrust(this, e) < 0 ? o.getWebsiteurl() : websiteurl;
logourl = o.getLogourl() != null && compareTrust(this, e) < 0 ? o.getLogourl() : logourl;
eclegalbody =
o.getEclegalbody() != null && compareTrust(this, e) < 0 ? o.getEclegalbody() : eclegalbody;
eclegalperson =
o.getEclegalperson() != null && compareTrust(this, e) < 0
eclegalbody = o.getEclegalbody() != null && compareTrust(this, e) < 0 ? o.getEclegalbody() : eclegalbody;
eclegalperson = o.getEclegalperson() != null && compareTrust(this, e) < 0
? o.getEclegalperson()
: eclegalperson;
ecnonprofit =
o.getEcnonprofit() != null && compareTrust(this, e) < 0 ? o.getEcnonprofit() : ecnonprofit;
ecresearchorganization =
o.getEcresearchorganization() != null && compareTrust(this, e) < 0
ecnonprofit = o.getEcnonprofit() != null && compareTrust(this, e) < 0 ? o.getEcnonprofit() : ecnonprofit;
ecresearchorganization = o.getEcresearchorganization() != null && compareTrust(this, e) < 0
? o.getEcresearchorganization()
: ecresearchorganization;
echighereducation =
o.getEchighereducation() != null && compareTrust(this, e) < 0
echighereducation = o.getEchighereducation() != null && compareTrust(this, e) < 0
? o.getEchighereducation()
: echighereducation;
ecinternationalorganizationeurinterests =
o.getEcinternationalorganizationeurinterests() != null && compareTrust(this, e) < 0
ecinternationalorganizationeurinterests = o.getEcinternationalorganizationeurinterests() != null
&& compareTrust(this, e) < 0
? o.getEcinternationalorganizationeurinterests()
: ecinternationalorganizationeurinterests;
ecinternationalorganization =
o.getEcinternationalorganization() != null && compareTrust(this, e) < 0
ecinternationalorganization = o.getEcinternationalorganization() != null && compareTrust(this, e) < 0
? o.getEcinternationalorganization()
: ecinternationalorganization;
ecenterprise =
o.getEcenterprise() != null && compareTrust(this, e) < 0
ecenterprise = o.getEcenterprise() != null && compareTrust(this, e) < 0
? o.getEcenterprise()
: ecenterprise;
ecsmevalidated =
o.getEcsmevalidated() != null && compareTrust(this, e) < 0
ecsmevalidated = o.getEcsmevalidated() != null && compareTrust(this, e) < 0
? o.getEcsmevalidated()
: ecsmevalidated;
ecnutscode =
o.getEcnutscode() != null && compareTrust(this, e) < 0 ? o.getEcnutscode() : ecnutscode;
ecnutscode = o.getEcnutscode() != null && compareTrust(this, e) < 0 ? o.getEcnutscode() : ecnutscode;
country = o.getCountry() != null && compareTrust(this, e) < 0 ? o.getCountry() : country;
mergeOAFDataInfo(o);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -67,8 +68,10 @@ public class OriginDescription implements Serializable {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
OriginDescription that = (OriginDescription) o;
return Objects.equals(harvestDate, that.harvestDate)
&& Objects.equals(altered, that.altered)

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import eu.dnetlib.dhp.schema.common.ModelConstants;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -275,63 +276,48 @@ public class Project extends OafEntity implements Serializable {
Project p = (Project) e;
websiteurl =
p.getWebsiteurl() != null && compareTrust(this, e) < 0 ? p.getWebsiteurl() : websiteurl;
websiteurl = p.getWebsiteurl() != null && compareTrust(this, e) < 0 ? p.getWebsiteurl() : websiteurl;
code = p.getCode() != null && compareTrust(this, e) < 0 ? p.getCode() : code;
acronym = p.getAcronym() != null && compareTrust(this, e) < 0 ? p.getAcronym() : acronym;
title = p.getTitle() != null && compareTrust(this, e) < 0 ? p.getTitle() : title;
startdate =
p.getStartdate() != null && compareTrust(this, e) < 0 ? p.getStartdate() : startdate;
startdate = p.getStartdate() != null && compareTrust(this, e) < 0 ? p.getStartdate() : startdate;
enddate = p.getEnddate() != null && compareTrust(this, e) < 0 ? p.getEnddate() : enddate;
callidentifier =
p.getCallidentifier() != null && compareTrust(this, e) < 0
callidentifier = p.getCallidentifier() != null && compareTrust(this, e) < 0
? p.getCallidentifier()
: callidentifier;
keywords = p.getKeywords() != null && compareTrust(this, e) < 0 ? p.getKeywords() : keywords;
duration = p.getDuration() != null && compareTrust(this, e) < 0 ? p.getDuration() : duration;
ecsc39 = p.getEcsc39() != null && compareTrust(this, e) < 0 ? p.getEcsc39() : ecsc39;
oamandatepublications =
p.getOamandatepublications() != null && compareTrust(this, e) < 0
oamandatepublications = p.getOamandatepublications() != null && compareTrust(this, e) < 0
? p.getOamandatepublications()
: oamandatepublications;
ecarticle29_3 =
p.getEcarticle29_3() != null && compareTrust(this, e) < 0
ecarticle29_3 = p.getEcarticle29_3() != null && compareTrust(this, e) < 0
? p.getEcarticle29_3()
: ecarticle29_3;
subjects = mergeLists(subjects, p.getSubjects());
fundingtree = mergeLists(fundingtree, p.getFundingtree());
contracttype =
p.getContracttype() != null && compareTrust(this, e) < 0
contracttype = p.getContracttype() != null && compareTrust(this, e) < 0
? p.getContracttype()
: contracttype;
optional1 =
p.getOptional1() != null && compareTrust(this, e) < 0 ? p.getOptional1() : optional1;
optional2 =
p.getOptional2() != null && compareTrust(this, e) < 0 ? p.getOptional2() : optional2;
jsonextrainfo =
p.getJsonextrainfo() != null && compareTrust(this, e) < 0
optional1 = p.getOptional1() != null && compareTrust(this, e) < 0 ? p.getOptional1() : optional1;
optional2 = p.getOptional2() != null && compareTrust(this, e) < 0 ? p.getOptional2() : optional2;
jsonextrainfo = p.getJsonextrainfo() != null && compareTrust(this, e) < 0
? p.getJsonextrainfo()
: jsonextrainfo;
contactfullname =
p.getContactfullname() != null && compareTrust(this, e) < 0
contactfullname = p.getContactfullname() != null && compareTrust(this, e) < 0
? p.getContactfullname()
: contactfullname;
contactfax =
p.getContactfax() != null && compareTrust(this, e) < 0 ? p.getContactfax() : contactfax;
contactphone =
p.getContactphone() != null && compareTrust(this, e) < 0
contactfax = p.getContactfax() != null && compareTrust(this, e) < 0 ? p.getContactfax() : contactfax;
contactphone = p.getContactphone() != null && compareTrust(this, e) < 0
? p.getContactphone()
: contactphone;
contactemail =
p.getContactemail() != null && compareTrust(this, e) < 0
contactemail = p.getContactemail() != null && compareTrust(this, e) < 0
? p.getContactemail()
: contactemail;
summary = p.getSummary() != null && compareTrust(this, e) < 0 ? p.getSummary() : summary;
currency = p.getCurrency() != null && compareTrust(this, e) < 0 ? p.getCurrency() : currency;
totalcost =
p.getTotalcost() != null && compareTrust(this, e) < 0 ? p.getTotalcost() : totalcost;
fundedamount =
p.getFundedamount() != null && compareTrust(this, e) < 0
totalcost = p.getTotalcost() != null && compareTrust(this, e) < 0 ? p.getTotalcost() : totalcost;
fundedamount = p.getFundedamount() != null && compareTrust(this, e) < 0
? p.getFundedamount()
: fundedamount;
mergeOAFDataInfo(e);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import eu.dnetlib.dhp.schema.common.ModelConstants;
@ -30,7 +31,8 @@ public class Publication extends Result implements Serializable {
Publication p = (Publication) e;
if (p.getJournal() != null && compareTrust(this, e) < 0) journal = p.getJournal();
if (p.getJournal() != null && compareTrust(this, e) < 0)
journal = p.getJournal();
mergeOAFDataInfo(e);
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import com.fasterxml.jackson.annotation.JsonIgnore;
@ -46,7 +47,8 @@ public class Qualifier implements Serializable {
public String toComparableString() {
return isBlank()
? ""
: String.format(
: String
.format(
"%s::%s::%s::%s",
classid != null ? classid : "",
classname != null ? classname : "",
@ -69,9 +71,12 @@ public class Qualifier implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
Qualifier other = (Qualifier) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import static com.google.common.base.Preconditions.checkArgument;
@ -68,11 +69,14 @@ public class Relation extends Oaf {
checkArgument(Objects.equals(getRelClass(), r.getRelClass()), "relClass(es) must be equal");
setCollectedfrom(
Stream.concat(
Optional.ofNullable(getCollectedfrom())
Stream
.concat(
Optional
.ofNullable(getCollectedfrom())
.map(Collection::stream)
.orElse(Stream.empty()),
Optional.ofNullable(r.getCollectedfrom())
Optional
.ofNullable(r.getCollectedfrom())
.map(Collection::stream)
.orElse(Stream.empty()))
.distinct() // relies on KeyValue.equals
@ -81,8 +85,10 @@ public class Relation extends Oaf {
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
if (this == o)
return true;
if (o == null || getClass() != o.getClass())
return false;
Relation relation = (Relation) o;
return relType.equals(relation.relType)
&& subRelType.equals(relation.subRelType)

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -233,9 +234,11 @@ public class Result extends OafEntity implements Serializable {
if (r.getBestaccessright() != null && compareTrust(this, r) < 0)
bestaccessright = r.getBestaccessright();
if (r.getResulttype() != null && compareTrust(this, r) < 0) resulttype = r.getResulttype();
if (r.getResulttype() != null && compareTrust(this, r) < 0)
resulttype = r.getResulttype();
if (r.getLanguage() != null && compareTrust(this, r) < 0) language = r.getLanguage();
if (r.getLanguage() != null && compareTrust(this, r) < 0)
language = r.getLanguage();
country = mergeLists(country, r.getCountry());
@ -247,7 +250,8 @@ public class Result extends OafEntity implements Serializable {
description = longestLists(description, r.getDescription());
if (r.getPublisher() != null && compareTrust(this, r) < 0) publisher = r.getPublisher();
if (r.getPublisher() != null && compareTrust(this, r) < 0)
publisher = r.getPublisher();
if (r.getEmbargoenddate() != null && compareTrust(this, r) < 0)
embargoenddate = r.getEmbargoenddate();
@ -260,7 +264,8 @@ public class Result extends OafEntity implements Serializable {
contributor = mergeLists(contributor, r.getContributor());
if (r.getResourcetype() != null) resourcetype = r.getResourcetype();
if (r.getResourcetype() != null)
resourcetype = r.getResourcetype();
coverage = mergeLists(coverage, r.getCoverage());
@ -270,16 +275,17 @@ public class Result extends OafEntity implements Serializable {
}
private List<Field<String>> longestLists(List<Field<String>> a, List<Field<String>> b) {
if (a == null || b == null) return a == null ? b : a;
if (a == null || b == null)
return a == null ? b : a;
if (a.size() == b.size()) {
int msa =
a.stream()
int msa = a
.stream()
.filter(i -> i.getValue() != null)
.map(i -> i.getValue().length())
.max(Comparator.naturalOrder())
.orElse(0);
int msb =
b.stream()
int msb = b
.stream()
.filter(i -> i.getValue() != null)
.map(i -> i.getValue().length())
.max(Comparator.naturalOrder())

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import eu.dnetlib.dhp.schema.common.ModelConstants;
@ -63,13 +64,11 @@ public class Software extends Result implements Serializable {
license = mergeLists(license, s.getLicense());
codeRepositoryUrl =
s.getCodeRepositoryUrl() != null && compareTrust(this, s) < 0
codeRepositoryUrl = s.getCodeRepositoryUrl() != null && compareTrust(this, s) < 0
? s.getCodeRepositoryUrl()
: codeRepositoryUrl;
programmingLanguage =
s.getProgrammingLanguage() != null && compareTrust(this, s) < 0
programmingLanguage = s.getProgrammingLanguage() != null && compareTrust(this, s) < 0
? s.getProgrammingLanguage()
: programmingLanguage;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable;
@ -45,9 +46,12 @@ public class StructuredProperty implements Serializable {
@Override
public boolean equals(Object obj) {
if (this == obj) return true;
if (obj == null) return false;
if (getClass() != obj.getClass()) return false;
if (this == obj)
return true;
if (obj == null)
return false;
if (getClass() != obj.getClass())
return false;
StructuredProperty other = (StructuredProperty) obj;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import eu.dnetlib.dhp.schema.oaf.Dataset;
@ -46,7 +47,8 @@ public class DLIDataset extends Dataset {
DLIDataset p = (DLIDataset) e;
if (StringUtils.isBlank(completionStatus) && StringUtils.isNotBlank(p.completionStatus))
completionStatus = p.completionStatus;
if ("complete".equalsIgnoreCase(p.completionStatus)) completionStatus = "complete";
if ("complete".equalsIgnoreCase(p.completionStatus))
completionStatus = "complete";
dlicollectedfrom = mergeProvenance(dlicollectedfrom, p.getDlicollectedfrom());
}
@ -54,7 +56,8 @@ public class DLIDataset extends Dataset {
final List<ProvenaceInfo> a, final List<ProvenaceInfo> b) {
Map<String, ProvenaceInfo> result = new HashMap<>();
if (a != null)
a.forEach(
a
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())
@ -66,7 +69,8 @@ public class DLIDataset extends Dataset {
result.put(p.getId(), p);
});
if (b != null)
b.forEach(
b
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import eu.dnetlib.dhp.schema.oaf.OafEntity;
@ -44,7 +45,8 @@ public class DLIPublication extends Publication implements Serializable {
DLIPublication p = (DLIPublication) e;
if (StringUtils.isBlank(completionStatus) && StringUtils.isNotBlank(p.completionStatus))
completionStatus = p.completionStatus;
if ("complete".equalsIgnoreCase(p.completionStatus)) completionStatus = "complete";
if ("complete".equalsIgnoreCase(p.completionStatus))
completionStatus = "complete";
dlicollectedfrom = mergeProvenance(dlicollectedfrom, p.getDlicollectedfrom());
}
@ -52,7 +54,8 @@ public class DLIPublication extends Publication implements Serializable {
final List<ProvenaceInfo> a, final List<ProvenaceInfo> b) {
Map<String, ProvenaceInfo> result = new HashMap<>();
if (a != null)
a.forEach(
a
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())
@ -64,7 +67,8 @@ public class DLIPublication extends Publication implements Serializable {
result.put(p.getId(), p);
});
if (b != null)
b.forEach(
b
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import eu.dnetlib.dhp.schema.oaf.Relation;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import eu.dnetlib.dhp.schema.oaf.Oaf;
@ -72,7 +73,8 @@ public class DLIUnknown extends Oaf implements Serializable {
}
public void mergeFrom(DLIUnknown p) {
if ("complete".equalsIgnoreCase(p.completionStatus)) completionStatus = "complete";
if ("complete".equalsIgnoreCase(p.completionStatus))
completionStatus = "complete";
dlicollectedfrom = mergeProvenance(dlicollectedfrom, p.getDlicollectedfrom());
}
@ -80,7 +82,8 @@ public class DLIUnknown extends Oaf implements Serializable {
final List<ProvenaceInfo> a, final List<ProvenaceInfo> b) {
Map<String, ProvenaceInfo> result = new HashMap<>();
if (a != null)
a.forEach(
a
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())
@ -92,7 +95,8 @@ public class DLIUnknown extends Oaf implements Serializable {
result.put(p.getId(), p);
});
if (b != null)
b.forEach(
b
.forEach(
p -> {
if (p != null && StringUtils.isNotBlank(p.getId()) && result.containsKey(p.getId())) {
if ("incomplete".equalsIgnoreCase(result.get(p.getId()).getCompletionStatus())

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import java.io.Serializable;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.action;
import static org.junit.jupiter.api.Assertions.*;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.common;
import static org.junit.jupiter.api.Assertions.assertFalse;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.oaf;
import static org.junit.jupiter.api.Assertions.*;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.schema.scholexplorer;
import com.fasterxml.jackson.core.JsonProcessingException;
@ -22,13 +23,17 @@ public class DLItest {
a1.setCompletionStatus("complete");
DLIPublication a = new DLIPublication();
a.setPid(
Arrays.asList(
a
.setPid(
Arrays
.asList(
createSP("10.11", "doi", "dnet:pid_types"),
createSP("123456", "pdb", "dnet:pid_types")));
a.setTitle(Collections.singletonList(createSP("A Title", "title", "dnetTitle")));
a.setDlicollectedfrom(
Arrays.asList(
a
.setDlicollectedfrom(
Arrays
.asList(
createCollectedFrom("dct", "datacite", "complete"),
createCollectedFrom("dct", "datacite", "incomplete")));
a.setCompletionStatus("incomplete");
@ -42,8 +47,7 @@ public class DLItest {
@Test
public void testDeserialization() throws IOException {
final String json =
"{\"dataInfo\":{\"invisible\":false,\"inferred\":null,\"deletedbyinference\":false,\"trust\":\"0.9\",\"inferenceprovenance\":null,\"provenanceaction\":null},\"lastupdatetimestamp\":null,\"id\":\"60|bd9352547098929a394655ad1a44a479\",\"originalId\":[\"bd9352547098929a394655ad1a44a479\"],\"collectedfrom\":[{\"key\":\"dli_________::datacite\",\"value\":\"Datasets in Datacite\",\"dataInfo\":null,\"blank\":false}],\"pid\":[{\"value\":\"10.7925/DRS1.DUCHAS_5078760\",\"qualifier\":{\"classid\":\"doi\",\"classname\":\"doi\",\"schemeid\":\"dnet:pid_types\",\"schemename\":\"dnet:pid_types\",\"blank\":false},\"dataInfo\":null}],\"dateofcollection\":\"2020-01-09T08:29:31.885Z\",\"dateoftransformation\":null,\"extraInfo\":null,\"oaiprovenance\":null,\"author\":[{\"fullname\":\"Cathail, S. Ó\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Donnell, Breda Mc\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Ireland. Department of Arts, Culture, and the Gaeltacht\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"University College Dublin\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"National Folklore Foundation\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Cathail, S. Ó\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Donnell, Breda Mc\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null}],\"resulttype\":null,\"language\":null,\"country\":null,\"subject\":[{\"value\":\"Recreation\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null},{\"value\":\"Entertainments and recreational activities\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null},{\"value\":\"Siamsaíocht agus caitheamh aimsire\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null}],\"title\":[{\"value\":\"Games We Play\",\"qualifier\":null,\"dataInfo\":null}],\"relevantdate\":[{\"value\":\"1938-09-28\",\"qualifier\":{\"classid\":\"date\",\"classname\":\"date\",\"schemeid\":\"dnet::date\",\"schemename\":\"dnet::date\",\"blank\":false},\"dataInfo\":null}],\"description\":[{\"value\":\"Story collected by Breda Mc Donnell, a student at Tenure school (Tinure, Co. Louth) (no informant identified).\",\"dataInfo\":null}],\"dateofacceptance\":null,\"publisher\":{\"value\":\"University College Dublin\",\"dataInfo\":null},\"embargoenddate\":null,\"source\":null,\"fulltext\":null,\"format\":null,\"contributor\":null,\"resourcetype\":null,\"coverage\":null,\"refereed\":null,\"context\":null,\"processingchargeamount\":null,\"processingchargecurrency\":null,\"externalReference\":null,\"instance\":[],\"storagedate\":null,\"device\":null,\"size\":null,\"version\":null,\"lastmetadataupdate\":null,\"metadataversionnumber\":null,\"geolocation\":null,\"dlicollectedfrom\":[{\"id\":\"dli_________::datacite\",\"name\":\"Datasets in Datacite\",\"completionStatus\":\"complete\",\"collectionMode\":\"resolved\"}],\"completionStatus\":\"complete\"}";
final String json = "{\"dataInfo\":{\"invisible\":false,\"inferred\":null,\"deletedbyinference\":false,\"trust\":\"0.9\",\"inferenceprovenance\":null,\"provenanceaction\":null},\"lastupdatetimestamp\":null,\"id\":\"60|bd9352547098929a394655ad1a44a479\",\"originalId\":[\"bd9352547098929a394655ad1a44a479\"],\"collectedfrom\":[{\"key\":\"dli_________::datacite\",\"value\":\"Datasets in Datacite\",\"dataInfo\":null,\"blank\":false}],\"pid\":[{\"value\":\"10.7925/DRS1.DUCHAS_5078760\",\"qualifier\":{\"classid\":\"doi\",\"classname\":\"doi\",\"schemeid\":\"dnet:pid_types\",\"schemename\":\"dnet:pid_types\",\"blank\":false},\"dataInfo\":null}],\"dateofcollection\":\"2020-01-09T08:29:31.885Z\",\"dateoftransformation\":null,\"extraInfo\":null,\"oaiprovenance\":null,\"author\":[{\"fullname\":\"Cathail, S. Ó\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Donnell, Breda Mc\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Ireland. Department of Arts, Culture, and the Gaeltacht\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"University College Dublin\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"National Folklore Foundation\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Cathail, S. Ó\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null},{\"fullname\":\"Donnell, Breda Mc\",\"name\":null,\"surname\":null,\"rank\":null,\"pid\":null,\"affiliation\":null}],\"resulttype\":null,\"language\":null,\"country\":null,\"subject\":[{\"value\":\"Recreation\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null},{\"value\":\"Entertainments and recreational activities\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null},{\"value\":\"Siamsaíocht agus caitheamh aimsire\",\"qualifier\":{\"classid\":\"dnet:subject\",\"classname\":\"dnet:subject\",\"schemeid\":\"unknown\",\"schemename\":\"unknown\",\"blank\":false},\"dataInfo\":null}],\"title\":[{\"value\":\"Games We Play\",\"qualifier\":null,\"dataInfo\":null}],\"relevantdate\":[{\"value\":\"1938-09-28\",\"qualifier\":{\"classid\":\"date\",\"classname\":\"date\",\"schemeid\":\"dnet::date\",\"schemename\":\"dnet::date\",\"blank\":false},\"dataInfo\":null}],\"description\":[{\"value\":\"Story collected by Breda Mc Donnell, a student at Tenure school (Tinure, Co. Louth) (no informant identified).\",\"dataInfo\":null}],\"dateofacceptance\":null,\"publisher\":{\"value\":\"University College Dublin\",\"dataInfo\":null},\"embargoenddate\":null,\"source\":null,\"fulltext\":null,\"format\":null,\"contributor\":null,\"resourcetype\":null,\"coverage\":null,\"refereed\":null,\"context\":null,\"processingchargeamount\":null,\"processingchargecurrency\":null,\"externalReference\":null,\"instance\":[],\"storagedate\":null,\"device\":null,\"size\":null,\"version\":null,\"lastmetadataupdate\":null,\"metadataversionnumber\":null,\"geolocation\":null,\"dlicollectedfrom\":[{\"id\":\"dli_________::datacite\",\"name\":\"Datasets in Datacite\",\"completionStatus\":\"complete\",\"collectionMode\":\"resolved\"}],\"completionStatus\":\"complete\"}";
ObjectMapper mapper = new ObjectMapper();
mapper.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager;
import com.google.common.base.Splitter;
@ -24,8 +25,7 @@ import org.slf4j.LoggerFactory;
public class ISClient implements Serializable {
private static final Logger log =
LoggerFactory.getLogger(PartitionActionSetsByPayloadTypeJob.class);
private static final Logger log = LoggerFactory.getLogger(PartitionActionSetsByPayloadTypeJob.class);
private static final String INPUT_ACTION_SET_ID_SEPARATOR = ",";
@ -37,14 +37,16 @@ public class ISClient implements Serializable {
public List<String> getLatestRawsetPaths(String setIds) {
List<String> ids =
Lists.newArrayList(
Splitter.on(INPUT_ACTION_SET_ID_SEPARATOR)
List<String> ids = Lists
.newArrayList(
Splitter
.on(INPUT_ACTION_SET_ID_SEPARATOR)
.omitEmptyStrings()
.trimResults()
.split(setIds));
return ids.stream()
return ids
.stream()
.map(id -> getSet(isLookup, id))
.map(as -> as.getPathToLatest())
.collect(Collectors.toCollection(ArrayList::new));
@ -52,8 +54,7 @@ public class ISClient implements Serializable {
private ActionManagerSet getSet(ISLookUpService isLookup, final String setId) {
final String q =
"for $x in collection('/db/DRIVER/ActionManagerSetDSResources/ActionManagerSetDSResourceType') "
final String q = "for $x in collection('/db/DRIVER/ActionManagerSetDSResources/ActionManagerSetDSResourceType') "
+ "where $x//SET/@id = '"
+ setId
+ "' return $x";
@ -78,7 +79,8 @@ public class ISClient implements Serializable {
set.setId(doc.valueOf("//SET/@id").trim());
set.setName(doc.valueOf("//SET").trim());
set.setImpact(ImpactTypes.valueOf(doc.valueOf("//IMPACT").trim()));
set.setLatest(
set
.setLatest(
doc.valueOf("//RAW_SETS/LATEST/@id"),
doc.valueOf("//RAW_SETS/LATEST/@creationDate"),
doc.valueOf("//RAW_SETS/LATEST/@lastUpdate"));
@ -87,7 +89,8 @@ public class ISClient implements Serializable {
if (expiredNodes != null) {
for (int i = 0; i < expiredNodes.size(); i++) {
Element ex = (Element) expiredNodes.get(i);
set.addExpired(
set
.addExpired(
ex.attributeValue("id"),
ex.attributeValue("creationDate"),
ex.attributeValue("lastUpdate"));
@ -114,8 +117,7 @@ public class ISClient implements Serializable {
private String queryServiceProperty(ISLookUpService isLookup, final String propertyName)
throws ActionManagerException {
final String q =
"for $x in /RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='ActionManagerServiceResourceType'] return $x//SERVICE_PROPERTIES/PROPERTY[./@ key='"
final String q = "for $x in /RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='ActionManagerServiceResourceType'] return $x//SERVICE_PROPERTIES/PROPERTY[./@ key='"
+ propertyName
+ "']/@value/string()";
log.debug("quering for service property: " + q);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.migration;
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;
@ -8,38 +9,58 @@ public class LicenseComparator implements Comparator<Qualifier> {
@Override
public int compare(Qualifier left, Qualifier right) {
if (left == null && right == null) return 0;
if (left == null) return 1;
if (right == null) return -1;
if (left == null && right == null)
return 0;
if (left == null)
return 1;
if (right == null)
return -1;
String lClass = left.getClassid();
String rClass = right.getClassid();
if (lClass.equals(rClass)) return 0;
if (lClass.equals(rClass))
return 0;
if (lClass.equals("OPEN SOURCE")) return -1;
if (rClass.equals("OPEN SOURCE")) return 1;
if (lClass.equals("OPEN SOURCE"))
return -1;
if (rClass.equals("OPEN SOURCE"))
return 1;
if (lClass.equals("OPEN")) return -1;
if (rClass.equals("OPEN")) return 1;
if (lClass.equals("OPEN"))
return -1;
if (rClass.equals("OPEN"))
return 1;
if (lClass.equals("6MONTHS")) return -1;
if (rClass.equals("6MONTHS")) return 1;
if (lClass.equals("6MONTHS"))
return -1;
if (rClass.equals("6MONTHS"))
return 1;
if (lClass.equals("12MONTHS")) return -1;
if (rClass.equals("12MONTHS")) return 1;
if (lClass.equals("12MONTHS"))
return -1;
if (rClass.equals("12MONTHS"))
return 1;
if (lClass.equals("EMBARGO")) return -1;
if (rClass.equals("EMBARGO")) return 1;
if (lClass.equals("EMBARGO"))
return -1;
if (rClass.equals("EMBARGO"))
return 1;
if (lClass.equals("RESTRICTED")) return -1;
if (rClass.equals("RESTRICTED")) return 1;
if (lClass.equals("RESTRICTED"))
return -1;
if (rClass.equals("RESTRICTED"))
return 1;
if (lClass.equals("CLOSED")) return -1;
if (rClass.equals("CLOSED")) return 1;
if (lClass.equals("CLOSED"))
return -1;
if (rClass.equals("CLOSED"))
return 1;
if (lClass.equals("UNKNOWN")) return -1;
if (rClass.equals("UNKNOWN")) return 1;
if (lClass.equals("UNKNOWN"))
return -1;
if (rClass.equals("UNKNOWN"))
return 1;
// Else (but unlikely), lexicographical ordering will do.
return lClass.compareTo(rClass);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.migration;
import com.google.common.base.Splitter;
@ -34,10 +35,11 @@ public class MigrateActionSet {
private static final String RAWSET_PREFIX = "rawset_";
public static void main(String[] args) throws Exception {
final ArgumentApplicationParser parser =
new ArgumentApplicationParser(
IOUtils.toString(
MigrateActionSet.class.getResourceAsStream(
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils
.toString(
MigrateActionSet.class
.getResourceAsStream(
"/eu/dnetlib/dhp/actionmanager/migration/migrate_actionsets_parameters.json")));
parser.parseArgument(args);
@ -68,8 +70,7 @@ public class MigrateActionSet {
Configuration conf = getConfiguration(distcp_task_timeout, distcp_memory_mb, distcp_num_maps);
FileSystem targetFS = FileSystem.get(conf);
Configuration sourceConf =
getConfiguration(distcp_task_timeout, distcp_memory_mb, distcp_num_maps);
Configuration sourceConf = getConfiguration(distcp_task_timeout, distcp_memory_mb, distcp_num_maps);
sourceConf.set(FileSystem.FS_DEFAULT_NAME_KEY, sourceNN);
FileSystem sourceFS = FileSystem.get(sourceConf);
@ -78,7 +79,8 @@ public class MigrateActionSet {
List<Path> targetPaths = new ArrayList<>();
final List<Path> sourcePaths = getSourcePaths(sourceNN, isLookUp);
log.info(
log
.info(
"paths to process:\n{}",
sourcePaths.stream().map(p -> p.toString()).collect(Collectors.joining("\n")));
for (Path source : sourcePaths) {
@ -87,8 +89,7 @@ public class MigrateActionSet {
log.warn("skipping unexisting path: {}", source);
} else {
LinkedList<String> pathQ =
Lists.newLinkedList(Splitter.on(SEPARATOR).split(source.toUri().getPath()));
LinkedList<String> pathQ = Lists.newLinkedList(Splitter.on(SEPARATOR).split(source.toUri().getPath()));
final String rawSet = pathQ.pollLast();
log.info("got RAWSET: {}", rawSet);
@ -97,8 +98,8 @@ public class MigrateActionSet {
final String actionSetDirectory = pathQ.pollLast();
final Path targetPath =
new Path(targetNN + workDir + SEPARATOR + actionSetDirectory + SEPARATOR + rawSet);
final Path targetPath = new Path(
targetNN + workDir + SEPARATOR + actionSetDirectory + SEPARATOR + rawSet);
log.info("using TARGET PATH: {}", targetPath);
@ -115,7 +116,8 @@ public class MigrateActionSet {
}
}
props.setProperty(
props
.setProperty(
TARGET_PATHS, targetPaths.stream().map(p -> p.toString()).collect(Collectors.joining(",")));
File file = new File(System.getProperty("oozie.action.output.properties"));
@ -140,8 +142,8 @@ public class MigrateActionSet {
op.preserve(DistCpOptions.FileAttribute.REPLICATION);
op.preserve(DistCpOptions.FileAttribute.CHECKSUMTYPE);
int res =
ToolRunner.run(
int res = ToolRunner
.run(
new DistCp(conf, op),
new String[] {
"-Dmapred.task.timeout=" + distcp_task_timeout,
@ -171,8 +173,7 @@ public class MigrateActionSet {
private List<Path> getSourcePaths(String sourceNN, ISLookUpService isLookUp)
throws ISLookUpException {
String XQUERY =
"distinct-values(\n"
String XQUERY = "distinct-values(\n"
+ "let $basePath := collection('/db/DRIVER/ServiceResources/ActionManagerServiceResourceType')//SERVICE_PROPERTIES/PROPERTY[@key = 'basePath']/@value/string()\n"
+ "for $x in collection('/db/DRIVER/ActionManagerSetDSResources/ActionManagerSetDSResourceType') \n"
+ "let $setDir := $x//SET/@directory/string()\n"
@ -180,7 +181,9 @@ public class MigrateActionSet {
+ "return concat($basePath, '/', $setDir, '/', $rawSet))";
log.info(String.format("running xquery:\n%s", XQUERY));
return isLookUp.quickSearchProfile(XQUERY).stream()
return isLookUp
.quickSearchProfile(XQUERY)
.stream()
.map(p -> sourceNN + p)
.map(Path::new)
.collect(Collectors.toList());

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.migration;
import static eu.dnetlib.data.proto.KindProtos.Kind.entity;
@ -46,7 +47,8 @@ public class ProtoConverter implements Serializable {
rel.setRelType(r.getRelType().toString());
rel.setSubRelType(r.getSubRelType().toString());
rel.setRelClass(r.getRelClass());
rel.setCollectedfrom(
rel
.setCollectedfrom(
r.getCollectedfromCount() > 0
? r.getCollectedfromList().stream().map(kv -> mapKV(kv)).collect(Collectors.toList())
: null);
@ -97,14 +99,16 @@ public class ProtoConverter implements Serializable {
}
private static Organization convertOrganization(OafProtos.Oaf oaf) {
final OrganizationProtos.Organization.Metadata m =
oaf.getEntity().getOrganization().getMetadata();
final OrganizationProtos.Organization.Metadata m = oaf.getEntity().getOrganization().getMetadata();
final Organization org = setOaf(new Organization(), oaf);
setEntity(org, oaf);
org.setLegalshortname(mapStringField(m.getLegalshortname()));
org.setLegalname(mapStringField(m.getLegalname()));
org.setAlternativeNames(
m.getAlternativeNamesList().stream()
org
.setAlternativeNames(
m
.getAlternativeNamesList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
org.setWebsiteurl(mapStringField(m.getWebsiteurl()));
@ -114,7 +118,8 @@ public class ProtoConverter implements Serializable {
org.setEcnonprofit(mapStringField(m.getEcnonprofit()));
org.setEcresearchorganization(mapStringField(m.getEcresearchorganization()));
org.setEchighereducation(mapStringField(m.getEchighereducation()));
org.setEcinternationalorganizationeurinterests(
org
.setEcinternationalorganizationeurinterests(
mapStringField(m.getEcinternationalorganizationeurinterests()));
org.setEcinternationalorganization(mapStringField(m.getEcinternationalorganization()));
org.setEcenterprise(mapStringField(m.getEcenterprise()));
@ -129,8 +134,11 @@ public class ProtoConverter implements Serializable {
final DatasourceProtos.Datasource.Metadata m = oaf.getEntity().getDatasource().getMetadata();
final Datasource datasource = setOaf(new Datasource(), oaf);
setEntity(datasource, oaf);
datasource.setAccessinfopackage(
m.getAccessinfopackageList().stream()
datasource
.setAccessinfopackage(
m
.getAccessinfopackageList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
datasource.setCertificates(mapStringField(m.getCertificates()));
@ -151,12 +159,18 @@ public class ProtoConverter implements Serializable {
datasource.setLogourl(mapStringField(m.getLogourl()));
datasource.setMissionstatementurl(mapStringField(m.getMissionstatementurl()));
datasource.setNamespaceprefix(mapStringField(m.getNamespaceprefix()));
datasource.setOdcontenttypes(
m.getOdcontenttypesList().stream()
datasource
.setOdcontenttypes(
m
.getOdcontenttypesList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
datasource.setOdlanguages(
m.getOdlanguagesList().stream()
datasource
.setOdlanguages(
m
.getOdlanguagesList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
datasource.setOdnumberofitems(mapStringField(m.getOdnumberofitems()));
@ -165,14 +179,18 @@ public class ProtoConverter implements Serializable {
datasource.setOfficialname(mapStringField(m.getOfficialname()));
datasource.setOpenairecompatibility(mapQualifier(m.getOpenairecompatibility()));
datasource.setPidsystems(mapStringField(m.getPidsystems()));
datasource.setPolicies(
datasource
.setPolicies(
m.getPoliciesList().stream().map(ProtoConverter::mapKV).collect(Collectors.toList()));
datasource.setQualitymanagementkind(mapStringField(m.getQualitymanagementkind()));
datasource.setReleaseenddate(mapStringField(m.getReleaseenddate()));
datasource.setServiceprovider(mapBoolField(m.getServiceprovider()));
datasource.setReleasestartdate(mapStringField(m.getReleasestartdate()));
datasource.setSubjects(
m.getSubjectsList().stream()
datasource
.setSubjects(
m
.getSubjectsList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
datasource.setVersioning(mapBoolField(m.getVersioning()));
@ -204,13 +222,17 @@ public class ProtoConverter implements Serializable {
project.setFundedamount(m.getFundedamount());
project.setTotalcost(m.getTotalcost());
project.setKeywords(mapStringField(m.getKeywords()));
project.setSubjects(
m.getSubjectsList().stream()
project
.setSubjects(
m
.getSubjectsList()
.stream()
.map(sp -> mapStructuredProperty(sp))
.collect(Collectors.toList()));
project.setTitle(mapStringField(m.getTitle()));
project.setWebsiteurl(mapStringField(m.getWebsiteurl()));
project.setFundingtree(
project
.setFundingtree(
m.getFundingtreeList().stream().map(f -> mapStringField(f)).collect(Collectors.toList()));
project.setJsonextrainfo(mapStringField(m.getJsonextrainfo()));
project.setSummary(mapStringField(m.getSummary()));
@ -242,12 +264,18 @@ public class ProtoConverter implements Serializable {
setEntity(software, oaf);
setResult(software, oaf);
software.setDocumentationUrl(
m.getDocumentationUrlList().stream()
software
.setDocumentationUrl(
m
.getDocumentationUrlList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
software.setLicense(
m.getLicenseList().stream()
software
.setLicense(
m
.getLicenseList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
software.setCodeRepositoryUrl(mapStringField(m.getCodeRepositoryUrl()));
@ -260,15 +288,22 @@ public class ProtoConverter implements Serializable {
OtherResearchProduct otherResearchProducts = setOaf(new OtherResearchProduct(), oaf);
setEntity(otherResearchProducts, oaf);
setResult(otherResearchProducts, oaf);
otherResearchProducts.setContactperson(
m.getContactpersonList().stream()
otherResearchProducts
.setContactperson(
m
.getContactpersonList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
otherResearchProducts.setContactgroup(
m.getContactgroupList().stream()
otherResearchProducts
.setContactgroup(
m
.getContactgroupList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
otherResearchProducts.setTool(
otherResearchProducts
.setTool(
m.getToolList().stream().map(ProtoConverter::mapStringField).collect(Collectors.toList()));
return otherResearchProducts;
@ -296,8 +331,11 @@ public class ProtoConverter implements Serializable {
dataset.setVersion(mapStringField(m.getVersion()));
dataset.setLastmetadataupdate(mapStringField(m.getLastmetadataupdate()));
dataset.setMetadataversionnumber(mapStringField(m.getMetadataversionnumber()));
dataset.setGeolocation(
m.getGeolocationList().stream()
dataset
.setGeolocation(
m
.getGeolocationList()
.stream()
.map(ProtoConverter::mapGeolocation)
.collect(Collectors.toList()));
return dataset;
@ -314,16 +352,23 @@ public class ProtoConverter implements Serializable {
final OafProtos.OafEntity e = oaf.getEntity();
entity.setId(e.getId());
entity.setOriginalId(e.getOriginalIdList());
entity.setCollectedfrom(
entity
.setCollectedfrom(
e.getCollectedfromList().stream().map(ProtoConverter::mapKV).collect(Collectors.toList()));
entity.setPid(
e.getPidList().stream()
entity
.setPid(
e
.getPidList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
entity.setDateofcollection(e.getDateofcollection());
entity.setDateoftransformation(e.getDateoftransformation());
entity.setExtraInfo(
e.getExtraInfoList().stream()
entity
.setExtraInfo(
e
.getExtraInfoList()
.stream()
.map(ProtoConverter::mapExtraInfo)
.collect(Collectors.toList()));
return entity;
@ -332,55 +377,87 @@ public class ProtoConverter implements Serializable {
public static <T extends Result> T setResult(T entity, OafProtos.Oaf oaf) {
// setting Entity fields
final ResultProtos.Result.Metadata m = oaf.getEntity().getResult().getMetadata();
entity.setAuthor(
entity
.setAuthor(
m.getAuthorList().stream().map(ProtoConverter::mapAuthor).collect(Collectors.toList()));
entity.setResulttype(mapQualifier(m.getResulttype()));
entity.setLanguage(mapQualifier(m.getLanguage()));
entity.setCountry(
m.getCountryList().stream()
entity
.setCountry(
m
.getCountryList()
.stream()
.map(ProtoConverter::mapQualifierAsCountry)
.collect(Collectors.toList()));
entity.setSubject(
m.getSubjectList().stream()
entity
.setSubject(
m
.getSubjectList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
entity.setTitle(
m.getTitleList().stream()
entity
.setTitle(
m
.getTitleList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
entity.setRelevantdate(
m.getRelevantdateList().stream()
entity
.setRelevantdate(
m
.getRelevantdateList()
.stream()
.map(ProtoConverter::mapStructuredProperty)
.collect(Collectors.toList()));
entity.setDescription(
m.getDescriptionList().stream()
entity
.setDescription(
m
.getDescriptionList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setDateofacceptance(mapStringField(m.getDateofacceptance()));
entity.setPublisher(mapStringField(m.getPublisher()));
entity.setEmbargoenddate(mapStringField(m.getEmbargoenddate()));
entity.setSource(
m.getSourceList().stream()
entity
.setSource(
m
.getSourceList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setFulltext(
m.getFulltextList().stream()
entity
.setFulltext(
m
.getFulltextList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setFormat(
m.getFormatList().stream()
entity
.setFormat(
m
.getFormatList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setContributor(
m.getContributorList().stream()
entity
.setContributor(
m
.getContributorList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setResourcetype(mapQualifier(m.getResourcetype()));
entity.setCoverage(
m.getCoverageList().stream()
entity
.setCoverage(
m
.getCoverageList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
entity.setContext(
entity
.setContext(
m.getContextList().stream().map(ProtoConverter::mapContext).collect(Collectors.toList()));
entity.setBestaccessright(getBestAccessRights(oaf.getEntity().getResult().getInstanceList()));
@ -390,8 +467,10 @@ public class ProtoConverter implements Serializable {
private static Qualifier getBestAccessRights(List<ResultProtos.Result.Instance> instanceList) {
if (instanceList != null) {
final Optional<FieldTypeProtos.Qualifier> min =
instanceList.stream().map(i -> i.getAccessright()).min(new LicenseComparator());
final Optional<FieldTypeProtos.Qualifier> min = instanceList
.stream()
.map(i -> i.getAccessright())
.min(new LicenseComparator());
final Qualifier rights = min.isPresent() ? mapQualifier(min.get()) : new Qualifier();
@ -418,8 +497,11 @@ public class ProtoConverter implements Serializable {
final Context entity = new Context();
entity.setId(context.getId());
entity.setDataInfo(
context.getDataInfoList().stream()
entity
.setDataInfo(
context
.getDataInfoList()
.stream()
.map(ProtoConverter::mapDataInfo)
.collect(Collectors.toList()));
return entity;
@ -543,8 +625,11 @@ public class ProtoConverter implements Serializable {
entity.setName(author.getName());
entity.setSurname(author.getSurname());
entity.setRank(author.getRank());
entity.setPid(
author.getPidList().stream()
entity
.setPid(
author
.getPidList()
.stream()
.map(
kv -> {
final StructuredProperty sp = new StructuredProperty();
@ -556,8 +641,11 @@ public class ProtoConverter implements Serializable {
return sp;
})
.collect(Collectors.toList()));
entity.setAffiliation(
author.getAffiliationList().stream()
entity
.setAffiliation(
author
.getAffiliationList()
.stream()
.map(ProtoConverter::mapStringField)
.collect(Collectors.toList()));
return entity;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.migration;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
@ -40,15 +41,16 @@ public class TransformActions implements Serializable {
private static final String SEPARATOR = "/";
public static void main(String[] args) throws Exception {
final ArgumentApplicationParser parser =
new ArgumentApplicationParser(
IOUtils.toString(
MigrateActionSet.class.getResourceAsStream(
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils
.toString(
MigrateActionSet.class
.getResourceAsStream(
"/eu/dnetlib/dhp/actionmanager/migration/transform_actionsets_parameters.json")));
parser.parseArgument(args);
Boolean isSparkSessionManaged =
Optional.ofNullable(parser.get("isSparkSessionManaged"))
Boolean isSparkSessionManaged = Optional
.ofNullable(parser.get("isSparkSessionManaged"))
.map(Boolean::valueOf)
.orElse(Boolean.TRUE);
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
@ -83,8 +85,7 @@ public class TransformActions implements Serializable {
final String rawset = pathQ.pollLast();
final String actionSetDirectory = pathQ.pollLast();
final Path targetDirectory =
new Path(targetBaseDir + SEPARATOR + actionSetDirectory + SEPARATOR + rawset);
final Path targetDirectory = new Path(targetBaseDir + SEPARATOR + actionSetDirectory + SEPARATOR + rawset);
if (fs.exists(targetDirectory)) {
log.info("found target directory '{}", targetDirectory);
@ -94,7 +95,8 @@ public class TransformActions implements Serializable {
log.info("transforming actions from '{}' to '{}'", sourcePath, targetDirectory);
sc.sequenceFile(sourcePath, Text.class, Text.class)
sc
.sequenceFile(sourcePath, Text.class, Text.class)
.map(a -> eu.dnetlib.actionmanager.actions.AtomicAction.fromJSON(a._2().toString()))
.map(TransformActions::doTransform)
.filter(Objects::nonNull)
@ -129,8 +131,12 @@ public class TransformActions implements Serializable {
case project:
return new AtomicAction<>(Project.class, (Project) oaf);
case result:
final String resulttypeid =
proto_oaf.getEntity().getResult().getMetadata().getResulttype().getClassid();
final String resulttypeid = proto_oaf
.getEntity()
.getResult()
.getMetadata()
.getResulttype()
.getClassid();
switch (resulttypeid) {
case "publication":
return new AtomicAction<>(Publication.class, (Publication) oaf);
@ -157,8 +163,7 @@ public class TransformActions implements Serializable {
private static String getTargetBaseDir(String isLookupUrl) throws ISLookUpException {
ISLookUpService isLookUp = ISLookupClientFactory.getLookUpService(isLookupUrl);
String XQUERY =
"collection('/db/DRIVER/ServiceResources/ActionManagerServiceResourceType')//SERVICE_PROPERTIES/PROPERTY[@key = 'basePath']/@value/string()";
String XQUERY = "collection('/db/DRIVER/ServiceResources/ActionManagerServiceResourceType')//SERVICE_PROPERTIES/PROPERTY[@key = 'basePath']/@value/string()";
return isLookUp.getResourceProfileByQuery(XQUERY);
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.partition;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
@ -23,20 +24,22 @@ import org.slf4j.LoggerFactory;
/** Partitions given set of action sets by payload type. */
public class PartitionActionSetsByPayloadTypeJob {
private static final Logger logger =
LoggerFactory.getLogger(PartitionActionSetsByPayloadTypeJob.class);
private static final Logger logger = LoggerFactory.getLogger(PartitionActionSetsByPayloadTypeJob.class);
private static final StructType KV_SCHEMA =
StructType$.MODULE$.apply(
Arrays.asList(
private static final StructType KV_SCHEMA = StructType$.MODULE$
.apply(
Arrays
.asList(
StructField$.MODULE$.apply("key", DataTypes.StringType, false, Metadata.empty()),
StructField$.MODULE$.apply("value", DataTypes.StringType, false, Metadata.empty())));
private static final StructType ATOMIC_ACTION_SCHEMA =
StructType$.MODULE$.apply(
Arrays.asList(
private static final StructType ATOMIC_ACTION_SCHEMA = StructType$.MODULE$
.apply(
Arrays
.asList(
StructField$.MODULE$.apply("clazz", DataTypes.StringType, false, Metadata.empty()),
StructField$.MODULE$.apply(
StructField$.MODULE$
.apply(
"payload", DataTypes.StringType, false, Metadata.empty())));
private ISClient isClient;
@ -45,18 +48,20 @@ public class PartitionActionSetsByPayloadTypeJob {
this.isClient = new ISClient(isLookupUrl);
}
public PartitionActionSetsByPayloadTypeJob() {}
public PartitionActionSetsByPayloadTypeJob() {
}
public static void main(String[] args) throws Exception {
String jsonConfiguration =
IOUtils.toString(
PromoteActionPayloadForGraphTableJob.class.getResourceAsStream(
String jsonConfiguration = IOUtils
.toString(
PromoteActionPayloadForGraphTableJob.class
.getResourceAsStream(
"/eu/dnetlib/dhp/actionmanager/partition/partition_action_sets_by_payload_type_input_parameters.json"));
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
parser.parseArgument(args);
Boolean isSparkSessionManaged =
Optional.ofNullable(parser.get("isSparkSessionManaged"))
Boolean isSparkSessionManaged = Optional
.ofNullable(parser.get("isSparkSessionManaged"))
.map(Boolean::valueOf)
.orElse(Boolean.TRUE);
logger.info("isSparkSessionManaged: {}", isSparkSessionManaged);
@ -97,7 +102,8 @@ public class PartitionActionSetsByPayloadTypeJob {
private static void readAndWriteActionSetsFromPaths(
SparkSession spark, List<String> inputActionSetPaths, String outputPath) {
inputActionSetPaths.stream()
inputActionSetPaths
.stream()
.filter(path -> HdfsSupport.exists(path, spark.sparkContext().hadoopConfiguration()))
.forEach(
inputActionSetPath -> {
@ -111,8 +117,8 @@ public class PartitionActionSetsByPayloadTypeJob {
JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<Row> rdd =
sc.sequenceFile(path, Text.class, Text.class)
JavaRDD<Row> rdd = sc
.sequenceFile(path, Text.class, Text.class)
.map(x -> RowFactory.create(x._1().toString(), x._2().toString()));
return spark

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static eu.dnetlib.dhp.schema.common.ModelSupport.isSubClass;
@ -11,17 +12,17 @@ import java.util.function.BiFunction;
/** OAF model merging support. */
public class MergeAndGet {
private MergeAndGet() {}
private MergeAndGet() {
}
/**
* Strategy for merging OAF model objects.
*
* <p>MERGE_FROM_AND_GET: use OAF 'mergeFrom' method SELECT_NEWER_AND_GET: use last update
* timestamp to return newer instance
* <p>
* MERGE_FROM_AND_GET: use OAF 'mergeFrom' method SELECT_NEWER_AND_GET: use last update timestamp to return newer
* instance
*/
public enum Strategy {
MERGE_FROM_AND_GET,
SELECT_NEWER_AND_GET
MERGE_FROM_AND_GET, SELECT_NEWER_AND_GET
}
/**
@ -32,8 +33,8 @@ public class MergeAndGet {
* @param <A> Action payload type
* @return BiFunction to be used to merge OAF objects
*/
public static <G extends Oaf, A extends Oaf>
SerializableSupplier<BiFunction<G, A, G>> functionFor(Strategy strategy) {
public static <G extends Oaf, A extends Oaf> SerializableSupplier<BiFunction<G, A, G>> functionFor(
Strategy strategy) {
switch (strategy) {
case MERGE_FROM_AND_GET:
return () -> MergeAndGet::mergeFromAndGet;
@ -54,7 +55,8 @@ public class MergeAndGet {
return x;
}
throw new RuntimeException(
String.format(
String
.format(
"MERGE_FROM_AND_GET incompatible types: %s, %s",
x.getClass().getCanonicalName(), y.getClass().getCanonicalName()));
}
@ -70,12 +72,14 @@ public class MergeAndGet {
return x;
} else if (isSubClass(x, y) && x.getLastupdatetimestamp() < y.getLastupdatetimestamp()) {
throw new RuntimeException(
String.format(
String
.format(
"SELECT_NEWER_AND_GET cannot return right type when it is not the same as left type: %s, %s",
x.getClass().getCanonicalName(), y.getClass().getCanonicalName()));
}
throw new RuntimeException(
String.format(
String
.format(
"SELECT_NEWER_AND_GET cannot be used when left is not subtype of right: %s, %s",
x.getClass().getCanonicalName(), y.getClass().getCanonicalName()));
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
@ -25,21 +26,21 @@ import org.slf4j.LoggerFactory;
/** Applies a given action payload file to graph table of compatible type. */
public class PromoteActionPayloadForGraphTableJob {
private static final Logger logger =
LoggerFactory.getLogger(PromoteActionPayloadForGraphTableJob.class);
private static final Logger logger = LoggerFactory.getLogger(PromoteActionPayloadForGraphTableJob.class);
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
public static void main(String[] args) throws Exception {
String jsonConfiguration =
IOUtils.toString(
PromoteActionPayloadForGraphTableJob.class.getResourceAsStream(
String jsonConfiguration = IOUtils
.toString(
PromoteActionPayloadForGraphTableJob.class
.getResourceAsStream(
"/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json"));
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
parser.parseArgument(args);
Boolean isSparkSessionManaged =
Optional.ofNullable(parser.get("isSparkSessionManaged"))
Boolean isSparkSessionManaged = Optional
.ofNullable(parser.get("isSparkSessionManaged"))
.map(Boolean::valueOf)
.orElse(Boolean.TRUE);
logger.info("isSparkSessionManaged: {}", isSparkSessionManaged);
@ -59,13 +60,11 @@ public class PromoteActionPayloadForGraphTableJob {
String outputGraphTablePath = parser.get("outputGraphTablePath");
logger.info("outputGraphTablePath: {}", outputGraphTablePath);
MergeAndGet.Strategy strategy =
MergeAndGet.Strategy.valueOf(parser.get("mergeAndGetStrategy").toUpperCase());
MergeAndGet.Strategy strategy = MergeAndGet.Strategy.valueOf(parser.get("mergeAndGetStrategy").toUpperCase());
logger.info("strategy: {}", strategy);
Class<? extends Oaf> rowClazz = (Class<? extends Oaf>) Class.forName(graphTableClassName);
Class<? extends Oaf> actionPayloadClazz =
(Class<? extends Oaf>) Class.forName(actionPayloadClassName);
Class<? extends Oaf> actionPayloadClazz = (Class<? extends Oaf>) Class.forName(actionPayloadClassName);
throwIfGraphTableClassIsNotSubClassOfActionPayloadClass(rowClazz, actionPayloadClazz);
@ -92,8 +91,8 @@ public class PromoteActionPayloadForGraphTableJob {
private static void throwIfGraphTableClassIsNotSubClassOfActionPayloadClass(
Class<? extends Oaf> rowClazz, Class<? extends Oaf> actionPayloadClazz) {
if (!isSubClass(rowClazz, actionPayloadClazz)) {
String msg =
String.format(
String msg = String
.format(
"graph table class is not a subclass of action payload class: graph=%s, action=%s",
rowClazz.getCanonicalName(), actionPayloadClazz.getCanonicalName());
throw new RuntimeException(msg);
@ -113,11 +112,9 @@ public class PromoteActionPayloadForGraphTableJob {
Class<G> rowClazz,
Class<A> actionPayloadClazz) {
Dataset<G> rowDS = readGraphTable(spark, inputGraphTablePath, rowClazz);
Dataset<A> actionPayloadDS =
readActionPayload(spark, inputActionPayloadPath, actionPayloadClazz);
Dataset<A> actionPayloadDS = readActionPayload(spark, inputActionPayloadPath, actionPayloadClazz);
Dataset<G> result =
promoteActionPayloadForGraphTable(
Dataset<G> result = promoteActionPayloadForGraphTable(
rowDS, actionPayloadDS, strategy, rowClazz, actionPayloadClazz)
.map((MapFunction<G, G>) value -> value, Encoders.bean(rowClazz));
@ -147,9 +144,8 @@ public class PromoteActionPayloadForGraphTableJob {
.read()
.parquet(path)
.map(
(MapFunction<Row, A>)
value ->
OBJECT_MAPPER.readValue(value.<String>getAs("payload"), actionPayloadClazz),
(MapFunction<Row, A>) value -> OBJECT_MAPPER
.readValue(value.<String> getAs("payload"), actionPayloadClazz),
Encoders.bean(actionPayloadClazz));
}
@ -159,22 +155,21 @@ public class PromoteActionPayloadForGraphTableJob {
MergeAndGet.Strategy strategy,
Class<G> rowClazz,
Class<A> actionPayloadClazz) {
logger.info(
logger
.info(
"Promoting action payload for graph table: payload={}, table={}",
actionPayloadClazz.getSimpleName(),
rowClazz.getSimpleName());
SerializableSupplier<Function<G, String>> rowIdFn = ModelSupport::idFn;
SerializableSupplier<Function<A, String>> actionPayloadIdFn = ModelSupport::idFn;
SerializableSupplier<BiFunction<G, A, G>> mergeRowWithActionPayloadAndGetFn =
MergeAndGet.functionFor(strategy);
SerializableSupplier<BiFunction<G, A, G>> mergeRowWithActionPayloadAndGetFn = MergeAndGet.functionFor(strategy);
SerializableSupplier<BiFunction<G, G, G>> mergeRowsAndGetFn = MergeAndGet.functionFor(strategy);
SerializableSupplier<G> zeroFn = zeroFn(rowClazz);
SerializableSupplier<Function<G, Boolean>> isNotZeroFn =
PromoteActionPayloadForGraphTableJob::isNotZeroFnUsingIdOrSource;
SerializableSupplier<Function<G, Boolean>> isNotZeroFn = PromoteActionPayloadForGraphTableJob::isNotZeroFnUsingIdOrSource;
Dataset<G> joinedAndMerged =
PromoteActionPayloadFunctions.joinGraphTableWithActionPayloadAndMerge(
Dataset<G> joinedAndMerged = PromoteActionPayloadFunctions
.joinGraphTableWithActionPayloadAndMerge(
rowDS,
actionPayloadDS,
rowIdFn,
@ -183,7 +178,8 @@ public class PromoteActionPayloadForGraphTableJob {
rowClazz,
actionPayloadClazz);
return PromoteActionPayloadFunctions.groupGraphTableByIdAndMerge(
return PromoteActionPayloadFunctions
.groupGraphTableByIdAndMerge(
joinedAndMerged, rowIdFn, mergeRowsAndGetFn, zeroFn, isNotZeroFn, rowClazz);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static eu.dnetlib.dhp.schema.common.ModelSupport.isSubClass;
@ -20,11 +21,11 @@ import scala.Tuple2;
/** Promote action payload functions. */
public class PromoteActionPayloadFunctions {
private PromoteActionPayloadFunctions() {}
private PromoteActionPayloadFunctions() {
}
/**
* Joins dataset representing graph table with dataset representing action payload using supplied
* functions.
* Joins dataset representing graph table with dataset representing action payload using supplied functions.
*
* @param rowDS Dataset representing graph table
* @param actionPayloadDS Dataset representing action payload
@ -51,8 +52,8 @@ public class PromoteActionPayloadFunctions {
}
Dataset<Tuple2<String, G>> rowWithIdDS = mapToTupleWithId(rowDS, rowIdFn, rowClazz);
Dataset<Tuple2<String, A>> actionPayloadWithIdDS =
mapToTupleWithId(actionPayloadDS, actionPayloadIdFn, actionPayloadClazz);
Dataset<Tuple2<String, A>> actionPayloadWithIdDS = mapToTupleWithId(
actionPayloadDS, actionPayloadIdFn, actionPayloadClazz);
return rowWithIdDS
.joinWith(
@ -60,21 +61,17 @@ public class PromoteActionPayloadFunctions {
rowWithIdDS.col("_1").equalTo(actionPayloadWithIdDS.col("_1")),
"full_outer")
.map(
(MapFunction<Tuple2<Tuple2<String, G>, Tuple2<String, A>>, G>)
value -> {
(MapFunction<Tuple2<Tuple2<String, G>, Tuple2<String, A>>, G>) value -> {
Optional<G> rowOpt = Optional.ofNullable(value._1()).map(Tuple2::_2);
Optional<A> actionPayloadOpt = Optional.ofNullable(value._2()).map(Tuple2::_2);
return rowOpt
.map(
row ->
actionPayloadOpt
row -> actionPayloadOpt
.map(
actionPayload ->
mergeAndGetFn.get().apply(row, actionPayload))
actionPayload -> mergeAndGetFn.get().apply(row, actionPayload))
.orElse(row))
.orElseGet(
() ->
actionPayloadOpt
() -> actionPayloadOpt
.filter(
actionPayload -> actionPayload.getClass().equals(rowClazz))
.map(rowClazz::cast)
@ -86,7 +83,8 @@ public class PromoteActionPayloadFunctions {
private static <T extends Oaf> Dataset<Tuple2<String, T>> mapToTupleWithId(
Dataset<T> ds, SerializableSupplier<Function<T, String>> idFn, Class<T> clazz) {
return ds.map(
return ds
.map(
(MapFunction<T, Tuple2<String, T>>) value -> new Tuple2<>(idFn.get().apply(value), value),
Encoders.tuple(Encoders.STRING(), Encoders.kryo(clazz)));
}
@ -110,8 +108,7 @@ public class PromoteActionPayloadFunctions {
SerializableSupplier<G> zeroFn,
SerializableSupplier<Function<G, Boolean>> isNotZeroFn,
Class<G> rowClazz) {
TypedColumn<G, G> aggregator =
new TableAggregator<>(zeroFn, mergeAndGetFn, isNotZeroFn, rowClazz).toColumn();
TypedColumn<G, G> aggregator = new TableAggregator<>(zeroFn, mergeAndGetFn, isNotZeroFn, rowClazz).toColumn();
return rowDS
.groupByKey((MapFunction<G, String>) x -> rowIdFn.get().apply(x), Encoders.STRING())
.agg(aggregator)

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.partition;
import static eu.dnetlib.dhp.common.ThrowingSupport.rethrowAsRuntimeException;
@ -37,19 +38,20 @@ import scala.collection.mutable.Seq;
@ExtendWith(MockitoExtension.class)
public class PartitionActionSetsByPayloadTypeJobTest {
private static final ClassLoader cl =
PartitionActionSetsByPayloadTypeJobTest.class.getClassLoader();
private static final ClassLoader cl = PartitionActionSetsByPayloadTypeJobTest.class.getClassLoader();
private static Configuration configuration;
private static SparkSession spark;
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
private static final StructType ATOMIC_ACTION_SCHEMA =
StructType$.MODULE$.apply(
Arrays.asList(
private static final StructType ATOMIC_ACTION_SCHEMA = StructType$.MODULE$
.apply(
Arrays
.asList(
StructField$.MODULE$.apply("clazz", DataTypes.StringType, false, Metadata.empty()),
StructField$.MODULE$.apply(
StructField$.MODULE$
.apply(
"payload", DataTypes.StringType, false, Metadata.empty())));
@BeforeAll
@ -71,7 +73,8 @@ public class PartitionActionSetsByPayloadTypeJobTest {
@Nested
class Main {
@Mock private ISClient isClient;
@Mock
private ISClient isClient;
@Test
public void shouldPartitionActionSetsByPayloadType(@TempDir Path workingDir) throws Exception {
@ -84,12 +87,14 @@ public class PartitionActionSetsByPayloadTypeJobTest {
List<String> inputActionSetsPaths = resolveInputActionSetPaths(inputActionSetsBaseDir);
// when
Mockito.when(isClient.getLatestRawsetPaths(Mockito.anyString()))
Mockito
.when(isClient.getLatestRawsetPaths(Mockito.anyString()))
.thenReturn(inputActionSetsPaths);
PartitionActionSetsByPayloadTypeJob job = new PartitionActionSetsByPayloadTypeJob();
job.setIsClient(isClient);
job.run(
job
.run(
Boolean.FALSE,
"", // it can be empty we're mocking the response from isClient
// to
@ -114,7 +119,8 @@ public class PartitionActionSetsByPayloadTypeJobTest {
private List<String> resolveInputActionSetPaths(Path inputActionSetsBaseDir) throws IOException {
Path inputActionSetJsonDumpsDir = getInputActionSetJsonDumpsDir();
return Files.list(inputActionSetJsonDumpsDir)
return Files
.list(inputActionSetJsonDumpsDir)
.map(
path -> {
String inputActionSetId = path.getFileName().toString();
@ -128,29 +134,31 @@ public class PartitionActionSetsByPayloadTypeJobTest {
Path inputActionSetJsonDumpsDir = getInputActionSetJsonDumpsDir();
Map<String, List<String>> oafsByType = new HashMap<>();
Files.list(inputActionSetJsonDumpsDir)
Files
.list(inputActionSetJsonDumpsDir)
.forEach(
inputActionSetJsonDumpFile -> {
String inputActionSetId = inputActionSetJsonDumpFile.getFileName().toString();
Path inputActionSetDir = inputActionSetsDir.resolve(inputActionSetId);
Dataset<String> actionDS =
readActionsFromJsonDump(inputActionSetJsonDumpFile.toString()).cache();
Dataset<String> actionDS = readActionsFromJsonDump(inputActionSetJsonDumpFile.toString()).cache();
writeActionsAsJobInput(actionDS, inputActionSetId, inputActionSetDir.toString());
Map<String, List<String>> actionSetOafsByType =
actionDS
Map<String, List<String>> actionSetOafsByType = actionDS
.withColumn("atomic_action", from_json(col("value"), ATOMIC_ACTION_SCHEMA))
.select(expr("atomic_action.*")).groupBy(col("clazz"))
.agg(collect_list(col("payload")).as("payload_list")).collectAsList().stream()
.select(expr("atomic_action.*"))
.groupBy(col("clazz"))
.agg(collect_list(col("payload")).as("payload_list"))
.collectAsList()
.stream()
.map(
row ->
new AbstractMap.SimpleEntry<>(
row -> new AbstractMap.SimpleEntry<>(
row.<String> getAs("clazz"),
mutableSeqAsJavaList(row.<Seq<String>> getAs("payload_list"))))
.collect(
Collectors.toMap(
Collectors
.toMap(
AbstractMap.SimpleEntry::getKey, AbstractMap.SimpleEntry::getValue));
actionSetOafsByType
@ -172,8 +180,10 @@ public class PartitionActionSetsByPayloadTypeJobTest {
}
private static Path getInputActionSetJsonDumpsDir() {
return Paths.get(
Objects.requireNonNull(cl.getResource("eu/dnetlib/dhp/actionmanager/partition/input/"))
return Paths
.get(
Objects
.requireNonNull(cl.getResource("eu/dnetlib/dhp/actionmanager/partition/input/"))
.getFile());
}
@ -195,12 +205,12 @@ public class PartitionActionSetsByPayloadTypeJobTest {
Path outputDatasetDir = outputDir.resolve(String.format("clazz=%s", clazz.getCanonicalName()));
Files.exists(outputDatasetDir);
List<T> actuals =
readActionPayloadFromJobOutput(outputDatasetDir.toString(), clazz).collectAsList();
List<T> actuals = readActionPayloadFromJobOutput(outputDatasetDir.toString(), clazz).collectAsList();
actuals.sort(Comparator.comparingInt(Object::hashCode));
List<T> expecteds =
oafsByClassName.get(clazz.getCanonicalName()).stream()
List<T> expecteds = oafsByClassName
.get(clazz.getCanonicalName())
.stream()
.map(json -> mapToOaf(json, clazz))
.sorted(Comparator.comparingInt(Object::hashCode))
.collect(Collectors.toList());
@ -214,15 +224,15 @@ public class PartitionActionSetsByPayloadTypeJobTest {
.read()
.parquet(path)
.map(
(MapFunction<Row, T>)
value -> OBJECT_MAPPER.readValue(value.<String>getAs("payload"), clazz),
(MapFunction<Row, T>) value -> OBJECT_MAPPER.readValue(value.<String> getAs("payload"), clazz),
Encoders.bean(clazz));
}
private static <T extends Oaf> T mapToOaf(String json, Class<T> clazz) {
return rethrowAsRuntimeException(
() -> OBJECT_MAPPER.readValue(json, clazz),
String.format(
String
.format(
"failed to map json to class: json=%s, class=%s", json, clazz.getCanonicalName()));
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static eu.dnetlib.dhp.actionmanager.promote.MergeAndGet.Strategy;
@ -126,8 +127,10 @@ public class MergeAndGetTest {
@Test
public void shouldThrowForOafEntityAndOafEntityButNotSubclasses() {
// given
class OafEntitySub1 extends OafEntity {}
class OafEntitySub2 extends OafEntity {}
class OafEntitySub1 extends OafEntity {
}
class OafEntitySub2 extends OafEntity {
}
OafEntitySub1 a = mock(OafEntitySub1.class);
OafEntitySub2 b = mock(OafEntitySub2.class);
@ -166,8 +169,7 @@ public class MergeAndGetTest {
Relation b = mock(Relation.class);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
assertThrows(RuntimeException.class, () -> fn.get().apply(a, b));
@ -180,8 +182,7 @@ public class MergeAndGetTest {
OafEntity b = mock(OafEntity.class);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
assertThrows(RuntimeException.class, () -> fn.get().apply(a, b));
@ -194,8 +195,7 @@ public class MergeAndGetTest {
Result b = mock(Result.class);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
assertThrows(RuntimeException.class, () -> fn.get().apply(a, b));
@ -212,8 +212,7 @@ public class MergeAndGetTest {
b.setLastupdatetimestamp(2L);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
assertThrows(RuntimeException.class, () -> fn.get().apply(a, b));
@ -228,8 +227,7 @@ public class MergeAndGetTest {
when(b.getLastupdatetimestamp()).thenReturn(2L);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
Oaf x = fn.get().apply(a, b);
@ -246,8 +244,7 @@ public class MergeAndGetTest {
when(b.getLastupdatetimestamp()).thenReturn(1L);
// when
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn =
functionFor(Strategy.SELECT_NEWER_AND_GET);
SerializableSupplier<BiFunction<Oaf, Oaf, Oaf>> fn = functionFor(Strategy.SELECT_NEWER_AND_GET);
// then
Oaf x = fn.get().apply(a, b);

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static org.junit.jupiter.api.Assertions.*;
@ -27,8 +28,7 @@ import org.junit.jupiter.params.provider.Arguments;
import org.junit.jupiter.params.provider.MethodSource;
public class PromoteActionPayloadForGraphTableJobTest {
private static final ClassLoader cl =
PromoteActionPayloadForGraphTableJobTest.class.getClassLoader();
private static final ClassLoader cl = PromoteActionPayloadForGraphTableJobTest.class.getClassLoader();
private static SparkSession spark;
@ -52,8 +52,7 @@ public class PromoteActionPayloadForGraphTableJobTest {
@BeforeEach
public void beforeEach() throws IOException {
workingDir =
Files.createTempDirectory(PromoteActionPayloadForGraphTableJobTest.class.getSimpleName());
workingDir = Files.createTempDirectory(PromoteActionPayloadForGraphTableJobTest.class.getSimpleName());
inputDir = workingDir.resolve("input");
inputGraphRootDir = inputDir.resolve("graph");
inputActionPayloadRootDir = inputDir.resolve("action_payload");
@ -81,11 +80,10 @@ public class PromoteActionPayloadForGraphTableJobTest {
Class<OafEntity> actionPayloadClazz = OafEntity.class;
// when
RuntimeException exception =
assertThrows(
RuntimeException exception = assertThrows(
RuntimeException.class,
() ->
PromoteActionPayloadForGraphTableJob.main(
() -> PromoteActionPayloadForGraphTableJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
@ -104,16 +102,15 @@ public class PromoteActionPayloadForGraphTableJobTest {
}));
// then
String msg =
String.format(
String msg = String
.format(
"graph table class is not a subclass of action payload class: graph=%s, action=%s",
rowClazz.getCanonicalName(), actionPayloadClazz.getCanonicalName());
assertTrue(exception.getMessage().contains(msg));
}
@ParameterizedTest(name = "strategy: {0}, graph table: {1}, action payload: {2}")
@MethodSource(
"eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJobTest#promoteJobTestParams")
@MethodSource("eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJobTest#promoteJobTestParams")
public void shouldPromoteActionPayloadForGraphTable(
MergeAndGet.Strategy strategy,
Class<? extends Oaf> rowClazz,
@ -121,13 +118,12 @@ public class PromoteActionPayloadForGraphTableJobTest {
throws Exception {
// given
Path inputGraphTableDir = createGraphTable(inputGraphRootDir, rowClazz);
Path inputActionPayloadDir =
createActionPayload(inputActionPayloadRootDir, rowClazz, actionPayloadClazz);
Path outputGraphTableDir =
outputDir.resolve("graph").resolve(rowClazz.getSimpleName().toLowerCase());
Path inputActionPayloadDir = createActionPayload(inputActionPayloadRootDir, rowClazz, actionPayloadClazz);
Path outputGraphTableDir = outputDir.resolve("graph").resolve(rowClazz.getSimpleName().toLowerCase());
// when
PromoteActionPayloadForGraphTableJob.main(
PromoteActionPayloadForGraphTableJob
.main(
new String[] {
"-isSparkSessionManaged",
Boolean.FALSE.toString(),
@ -148,20 +144,21 @@ public class PromoteActionPayloadForGraphTableJobTest {
// then
assertTrue(Files.exists(outputGraphTableDir));
List<? extends Oaf> actualOutputRows =
readGraphTableFromJobOutput(outputGraphTableDir.toString(), rowClazz).collectAsList()
List<? extends Oaf> actualOutputRows = readGraphTableFromJobOutput(outputGraphTableDir.toString(), rowClazz)
.collectAsList()
.stream()
.sorted(Comparator.comparingInt(Object::hashCode))
.collect(Collectors.toList());
String expectedOutputGraphTableJsonDumpPath =
resultFileLocation(strategy, rowClazz, actionPayloadClazz);
Path expectedOutputGraphTableJsonDumpFile =
Paths.get(
Objects.requireNonNull(cl.getResource(expectedOutputGraphTableJsonDumpPath))
String expectedOutputGraphTableJsonDumpPath = resultFileLocation(strategy, rowClazz, actionPayloadClazz);
Path expectedOutputGraphTableJsonDumpFile = Paths
.get(
Objects
.requireNonNull(cl.getResource(expectedOutputGraphTableJsonDumpPath))
.getFile());
List<? extends Oaf> expectedOutputRows =
readGraphTableFromJsonDump(expectedOutputGraphTableJsonDumpFile.toString(), rowClazz)
.collectAsList().stream()
List<? extends Oaf> expectedOutputRows = readGraphTableFromJsonDump(
expectedOutputGraphTableJsonDumpFile.toString(), rowClazz)
.collectAsList()
.stream()
.sorted(Comparator.comparingInt(Object::hashCode))
.collect(Collectors.toList());
assertIterableEquals(expectedOutputRows, actualOutputRows);
@ -169,7 +166,8 @@ public class PromoteActionPayloadForGraphTableJobTest {
}
public static Stream<Arguments> promoteJobTestParams() {
return Stream.of(
return Stream
.of(
arguments(
MergeAndGet.Strategy.MERGE_FROM_AND_GET,
eu.dnetlib.dhp.schema.oaf.Dataset.class,
@ -196,8 +194,8 @@ public class PromoteActionPayloadForGraphTableJobTest {
private static <G extends Oaf> Path createGraphTable(Path inputGraphRootDir, Class<G> rowClazz) {
String inputGraphTableJsonDumpPath = inputGraphTableJsonDumpLocation(rowClazz);
Path inputGraphTableJsonDumpFile =
Paths.get(Objects.requireNonNull(cl.getResource(inputGraphTableJsonDumpPath)).getFile());
Path inputGraphTableJsonDumpFile = Paths
.get(Objects.requireNonNull(cl.getResource(inputGraphTableJsonDumpPath)).getFile());
Dataset<G> rowDS = readGraphTableFromJsonDump(inputGraphTableJsonDumpFile.toString(), rowClazz);
String inputGraphTableName = rowClazz.getSimpleName().toLowerCase();
Path inputGraphTableDir = inputGraphRootDir.resolve(inputGraphTableName);
@ -206,7 +204,8 @@ public class PromoteActionPayloadForGraphTableJobTest {
}
private static String inputGraphTableJsonDumpLocation(Class<? extends Oaf> rowClazz) {
return String.format(
return String
.format(
"%s/%s.json",
"eu/dnetlib/dhp/actionmanager/promote/input/graph", rowClazz.getSimpleName().toLowerCase());
}
@ -227,14 +226,12 @@ public class PromoteActionPayloadForGraphTableJobTest {
private static <G extends Oaf, A extends Oaf> Path createActionPayload(
Path inputActionPayloadRootDir, Class<G> rowClazz, Class<A> actionPayloadClazz) {
String inputActionPayloadJsonDumpPath =
inputActionPayloadJsonDumpLocation(rowClazz, actionPayloadClazz);
Path inputActionPayloadJsonDumpFile =
Paths.get(Objects.requireNonNull(cl.getResource(inputActionPayloadJsonDumpPath)).getFile());
Dataset<String> actionPayloadDS =
readActionPayloadFromJsonDump(inputActionPayloadJsonDumpFile.toString());
Path inputActionPayloadDir =
inputActionPayloadRootDir.resolve(actionPayloadClazz.getSimpleName().toLowerCase());
String inputActionPayloadJsonDumpPath = inputActionPayloadJsonDumpLocation(rowClazz, actionPayloadClazz);
Path inputActionPayloadJsonDumpFile = Paths
.get(Objects.requireNonNull(cl.getResource(inputActionPayloadJsonDumpPath)).getFile());
Dataset<String> actionPayloadDS = readActionPayloadFromJsonDump(inputActionPayloadJsonDumpFile.toString());
Path inputActionPayloadDir = inputActionPayloadRootDir
.resolve(actionPayloadClazz.getSimpleName().toLowerCase());
writeActionPayloadAsJobInput(actionPayloadDS, inputActionPayloadDir.toString());
return inputActionPayloadDir;
}
@ -242,7 +239,8 @@ public class PromoteActionPayloadForGraphTableJobTest {
private static String inputActionPayloadJsonDumpLocation(
Class<? extends Oaf> rowClazz, Class<? extends Oaf> actionPayloadClazz) {
return String.format(
return String
.format(
"eu/dnetlib/dhp/actionmanager/promote/input/action_payload/%s_table/%s.json",
rowClazz.getSimpleName().toLowerCase(), actionPayloadClazz.getSimpleName().toLowerCase());
}
@ -269,7 +267,8 @@ public class PromoteActionPayloadForGraphTableJobTest {
MergeAndGet.Strategy strategy,
Class<? extends Oaf> rowClazz,
Class<? extends Oaf> actionPayloadClazz) {
return String.format(
return String
.format(
"eu/dnetlib/dhp/actionmanager/promote/output/graph/%s/%s/%s_action_payload/result.json",
strategy.name().toLowerCase(),
rowClazz.getSimpleName().toLowerCase(),

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.actionmanager.promote;
import static org.junit.jupiter.api.Assertions.assertEquals;
@ -43,13 +44,14 @@ public class PromoteActionPayloadFunctionsTest {
@Test
public void shouldThrowWhenTableTypeIsNotSubtypeOfActionPayloadType() {
// given
class OafImpl extends Oaf {}
class OafImpl extends Oaf {
}
// when
assertThrows(
RuntimeException.class,
() ->
PromoteActionPayloadFunctions.joinGraphTableWithActionPayloadAndMerge(
() -> PromoteActionPayloadFunctions
.joinGraphTableWithActionPayloadAndMerge(
null, null, null, null, null, OafImplSubSub.class, OafImpl.class));
}
@ -61,17 +63,16 @@ public class PromoteActionPayloadFunctionsTest {
String id2 = "id2";
String id3 = "id3";
String id4 = "id4";
List<OafImplSubSub> rowData =
Arrays.asList(
List<OafImplSubSub> rowData = Arrays
.asList(
createOafImplSubSub(id0),
createOafImplSubSub(id1),
createOafImplSubSub(id2),
createOafImplSubSub(id3));
Dataset<OafImplSubSub> rowDS =
spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
Dataset<OafImplSubSub> rowDS = spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
List<OafImplSubSub> actionPayloadData =
Arrays.asList(
List<OafImplSubSub> actionPayloadData = Arrays
.asList(
createOafImplSubSub(id1),
createOafImplSubSub(id2),
createOafImplSubSub(id2),
@ -82,22 +83,20 @@ public class PromoteActionPayloadFunctionsTest {
createOafImplSubSub(id4),
createOafImplSubSub(id4),
createOafImplSubSub(id4));
Dataset<OafImplSubSub> actionPayloadDS =
spark.createDataset(actionPayloadData, Encoders.bean(OafImplSubSub.class));
Dataset<OafImplSubSub> actionPayloadDS = spark
.createDataset(actionPayloadData, Encoders.bean(OafImplSubSub.class));
SerializableSupplier<Function<OafImplSubSub, String>> rowIdFn = () -> OafImplRoot::getId;
SerializableSupplier<Function<OafImplSubSub, String>> actionPayloadIdFn =
() -> OafImplRoot::getId;
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSubSub, OafImplSubSub>> mergeAndGetFn =
() ->
(x, y) -> {
SerializableSupplier<Function<OafImplSubSub, String>> actionPayloadIdFn = () -> OafImplRoot::getId;
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSubSub, OafImplSubSub>> mergeAndGetFn = () -> (x,
y) -> {
x.merge(y);
return x;
};
// when
List<OafImplSubSub> results =
PromoteActionPayloadFunctions.joinGraphTableWithActionPayloadAndMerge(
List<OafImplSubSub> results = PromoteActionPayloadFunctions
.joinGraphTableWithActionPayloadAndMerge(
rowDS,
actionPayloadDS,
rowIdFn,
@ -115,7 +114,8 @@ public class PromoteActionPayloadFunctionsTest {
assertEquals(3, results.stream().filter(x -> x.getId().equals(id3)).count());
assertEquals(4, results.stream().filter(x -> x.getId().equals(id4)).count());
results.forEach(
results
.forEach(
result -> {
switch (result.getId()) {
case "id0":
@ -143,17 +143,16 @@ public class PromoteActionPayloadFunctionsTest {
String id2 = "id2";
String id3 = "id3";
String id4 = "id4";
List<OafImplSubSub> rowData =
Arrays.asList(
List<OafImplSubSub> rowData = Arrays
.asList(
createOafImplSubSub(id0),
createOafImplSubSub(id1),
createOafImplSubSub(id2),
createOafImplSubSub(id3));
Dataset<OafImplSubSub> rowDS =
spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
Dataset<OafImplSubSub> rowDS = spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
List<OafImplSub> actionPayloadData =
Arrays.asList(
List<OafImplSub> actionPayloadData = Arrays
.asList(
createOafImplSub(id1),
createOafImplSub(id2),
createOafImplSub(id2),
@ -164,22 +163,19 @@ public class PromoteActionPayloadFunctionsTest {
createOafImplSub(id4),
createOafImplSub(id4),
createOafImplSub(id4));
Dataset<OafImplSub> actionPayloadDS =
spark.createDataset(actionPayloadData, Encoders.bean(OafImplSub.class));
Dataset<OafImplSub> actionPayloadDS = spark
.createDataset(actionPayloadData, Encoders.bean(OafImplSub.class));
SerializableSupplier<Function<OafImplSubSub, String>> rowIdFn = () -> OafImplRoot::getId;
SerializableSupplier<Function<OafImplSub, String>> actionPayloadIdFn =
() -> OafImplRoot::getId;
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSub, OafImplSubSub>> mergeAndGetFn =
() ->
(x, y) -> {
SerializableSupplier<Function<OafImplSub, String>> actionPayloadIdFn = () -> OafImplRoot::getId;
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSub, OafImplSubSub>> mergeAndGetFn = () -> (x, y) -> {
x.merge(y);
return x;
};
// when
List<OafImplSubSub> results =
PromoteActionPayloadFunctions.joinGraphTableWithActionPayloadAndMerge(
List<OafImplSubSub> results = PromoteActionPayloadFunctions
.joinGraphTableWithActionPayloadAndMerge(
rowDS,
actionPayloadDS,
rowIdFn,
@ -197,7 +193,8 @@ public class PromoteActionPayloadFunctionsTest {
assertEquals(3, results.stream().filter(x -> x.getId().equals(id3)).count());
assertEquals(0, results.stream().filter(x -> x.getId().equals(id4)).count());
results.forEach(
results
.forEach(
result -> {
switch (result.getId()) {
case "id0":
@ -224,31 +221,28 @@ public class PromoteActionPayloadFunctionsTest {
String id1 = "id1";
String id2 = "id2";
String id3 = "id3";
List<OafImplSubSub> rowData =
Arrays.asList(
List<OafImplSubSub> rowData = Arrays
.asList(
createOafImplSubSub(id1),
createOafImplSubSub(id2),
createOafImplSubSub(id2),
createOafImplSubSub(id3),
createOafImplSubSub(id3),
createOafImplSubSub(id3));
Dataset<OafImplSubSub> rowDS =
spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
Dataset<OafImplSubSub> rowDS = spark.createDataset(rowData, Encoders.bean(OafImplSubSub.class));
SerializableSupplier<Function<OafImplSubSub, String>> rowIdFn = () -> OafImplRoot::getId;
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSubSub, OafImplSubSub>> mergeAndGetFn =
() ->
(x, y) -> {
SerializableSupplier<BiFunction<OafImplSubSub, OafImplSubSub, OafImplSubSub>> mergeAndGetFn = () -> (x,
y) -> {
x.merge(y);
return x;
};
SerializableSupplier<OafImplSubSub> zeroFn = OafImplSubSub::new;
SerializableSupplier<Function<OafImplSubSub, Boolean>> isNotZeroFn =
() -> x -> Objects.nonNull(x.getId());
SerializableSupplier<Function<OafImplSubSub, Boolean>> isNotZeroFn = () -> x -> Objects.nonNull(x.getId());
// when
List<OafImplSubSub> results =
PromoteActionPayloadFunctions.groupGraphTableByIdAndMerge(
List<OafImplSubSub> results = PromoteActionPayloadFunctions
.groupGraphTableByIdAndMerge(
rowDS, rowIdFn, mergeAndGetFn, zeroFn, isNotZeroFn, OafImplSubSub.class)
.collectAsList();
@ -258,7 +252,8 @@ public class PromoteActionPayloadFunctionsTest {
assertEquals(1, results.stream().filter(x -> x.getId().equals(id2)).count());
assertEquals(1, results.stream().filter(x -> x.getId().equals(id3)).count());
results.forEach(
results
.forEach(
result -> {
switch (result.getId()) {
case "id1":

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection;
import com.fasterxml.jackson.databind.ObjectMapper;
@ -40,20 +41,22 @@ public class GenerateNativeStoreSparkJob {
final LongAccumulator totalItems,
final LongAccumulator invalidRecords) {
if (totalItems != null) totalItems.add(1);
if (totalItems != null)
totalItems.add(1);
try {
SAXReader reader = new SAXReader();
Document document =
reader.read(new ByteArrayInputStream(input.getBytes(StandardCharsets.UTF_8)));
Document document = reader.read(new ByteArrayInputStream(input.getBytes(StandardCharsets.UTF_8)));
Node node = document.selectSingleNode(xpath);
final String originalIdentifier = node.getText();
if (StringUtils.isBlank(originalIdentifier)) {
if (invalidRecords != null) invalidRecords.add(1);
if (invalidRecords != null)
invalidRecords.add(1);
return null;
}
return new MetadataRecord(originalIdentifier, encoding, provenance, input, dateOfCollection);
} catch (Throwable e) {
if (invalidRecords != null) invalidRecords.add(1);
if (invalidRecords != null)
invalidRecords.add(1);
e.printStackTrace();
return null;
}
@ -61,18 +64,19 @@ public class GenerateNativeStoreSparkJob {
public static void main(String[] args) throws Exception {
final ArgumentApplicationParser parser =
new ArgumentApplicationParser(
IOUtils.toString(
GenerateNativeStoreSparkJob.class.getResourceAsStream(
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils
.toString(
GenerateNativeStoreSparkJob.class
.getResourceAsStream(
"/eu/dnetlib/dhp/collection/collection_input_parameters.json")));
parser.parseArgument(args);
final ObjectMapper jsonMapper = new ObjectMapper();
final Provenance provenance = jsonMapper.readValue(parser.get("provenance"), Provenance.class);
final long dateOfCollection = new Long(parser.get("dateOfCollection"));
final SparkSession spark =
SparkSession.builder()
final SparkSession spark = SparkSession
.builder()
.appName("GenerateNativeStoreSparkJob")
.master(parser.get("master"))
.getOrCreate();
@ -80,20 +84,18 @@ public class GenerateNativeStoreSparkJob {
final Map<String, String> ongoingMap = new HashMap<>();
final Map<String, String> reportMap = new HashMap<>();
final boolean test =
parser.get("isTest") == null ? false : Boolean.valueOf(parser.get("isTest"));
final boolean test = parser.get("isTest") == null ? false : Boolean.valueOf(parser.get("isTest"));
final JavaSparkContext sc = new JavaSparkContext(spark.sparkContext());
final JavaPairRDD<IntWritable, Text> inputRDD =
sc.sequenceFile(parser.get("input"), IntWritable.class, Text.class);
final JavaPairRDD<IntWritable, Text> inputRDD = sc
.sequenceFile(parser.get("input"), IntWritable.class, Text.class);
final LongAccumulator totalItems = sc.sc().longAccumulator("TotalItems");
final LongAccumulator invalidRecords = sc.sc().longAccumulator("InvalidRecords");
final MessageManager manager =
new MessageManager(
final MessageManager manager = new MessageManager(
parser.get("rabbitHost"),
parser.get("rabbitUser"),
parser.get("rabbitPassword"),
@ -101,11 +103,9 @@ public class GenerateNativeStoreSparkJob {
false,
null);
final JavaRDD<MetadataRecord> mappeRDD =
inputRDD
final JavaRDD<MetadataRecord> mappeRDD = inputRDD
.map(
item ->
parseRecord(
item -> parseRecord(
item._2().toString(),
parser.get("xpath"),
parser.get("encoding"),
@ -118,7 +118,8 @@ public class GenerateNativeStoreSparkJob {
ongoingMap.put("ongoing", "0");
if (!test) {
manager.sendMessage(
manager
.sendMessage(
new Message(
parser.get("workflowId"), "DataFrameCreation", MessageType.ONGOING, ongoingMap),
parser.get("rabbitOngoingQueue"),
@ -132,7 +133,8 @@ public class GenerateNativeStoreSparkJob {
mdStoreRecords.add(mdstore.count());
ongoingMap.put("ongoing", "" + totalItems.value());
if (!test) {
manager.sendMessage(
manager
.sendMessage(
new Message(
parser.get("workflowId"), "DataFrameCreation", MessageType.ONGOING, ongoingMap),
parser.get("rabbitOngoingQueue"),
@ -144,7 +146,8 @@ public class GenerateNativeStoreSparkJob {
reportMap.put("invalidRecords", "" + invalidRecords.value());
reportMap.put("mdStoreSize", "" + mdStoreRecords.value());
if (!test) {
manager.sendMessage(
manager
.sendMessage(
new Message(parser.get("workflowId"), "Collection", MessageType.REPORT, reportMap),
parser.get("rabbitReportQueue"),
true,

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.plugin;
import eu.dnetlib.collector.worker.model.ApiDescriptor;

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.plugin.oai;
import com.google.common.base.Splitter;
@ -33,7 +34,8 @@ public class OaiCollectorPlugin implements CollectorPlugin {
final List<String> sets = new ArrayList<>();
if (setParam != null) {
sets.addAll(
sets
.addAll(
Lists.newArrayList(Splitter.on(",").omitEmptyStrings().trimResults().split(setParam)));
}
if (sets.isEmpty()) {
@ -57,15 +59,15 @@ public class OaiCollectorPlugin implements CollectorPlugin {
throw new DnetCollectorException("Invalid date (YYYY-MM-DD): " + untilDate);
}
final Iterator<Iterator<String>> iters =
sets.stream()
final Iterator<Iterator<String>> iters = sets
.stream()
.map(
set ->
getOaiIteratorFactory()
set -> getOaiIteratorFactory()
.newIterator(baseUrl, mdFormat, set, fromDate, untilDate))
.iterator();
return StreamSupport.stream(
return StreamSupport
.stream(
Spliterators.spliteratorUnknownSize(Iterators.concat(iters), Spliterator.ORDERED), false);
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.plugin.oai;
import eu.dnetlib.dhp.collection.worker.DnetCollectorException;
@ -86,12 +87,12 @@ public class OaiIterator implements Iterator<String> {
}
@Override
public void remove() {}
public void remove() {
}
private String firstPage() throws DnetCollectorException {
try {
String url =
baseUrl + "?verb=ListRecords&metadataPrefix=" + URLEncoder.encode(mdFormat, "UTF-8");
String url = baseUrl + "?verb=ListRecords&metadataPrefix=" + URLEncoder.encode(mdFormat, "UTF-8");
if (set != null && !set.isEmpty()) {
url += "&set=" + URLEncoder.encode(set, "UTF-8");
}
@ -154,8 +155,7 @@ public class OaiIterator implements Iterator<String> {
}
}
final Node errorNode =
doc.selectSingleNode("/*[local-name()='OAI-PMH']/*[local-name()='error']");
final Node errorNode = doc.selectSingleNode("/*[local-name()='OAI-PMH']/*[local-name()='error']");
if (errorNode != null) {
final String code = errorNode.valueOf("@code");
if ("noRecordsMatch".equalsIgnoreCase(code.trim())) {
@ -166,8 +166,7 @@ public class OaiIterator implements Iterator<String> {
}
}
for (final Object o :
doc.selectNodes("//*[local-name()='ListRecords']/*[local-name()='record']")) {
for (final Object o : doc.selectNodes("//*[local-name()='ListRecords']/*[local-name()='record']")) {
queue.add(((Node) o).asXML());
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.plugin.oai;
import eu.dnetlib.dhp.collection.worker.utils.HttpConnector;
@ -17,7 +18,8 @@ public class OaiIteratorFactory {
}
private HttpConnector getHttpConnector() {
if (httpConnector == null) httpConnector = new HttpConnector();
if (httpConnector == null)
httpConnector = new HttpConnector();
return httpConnector;
}
}

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.worker;
public class DnetCollectorException extends Exception {

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.worker;
import com.fasterxml.jackson.databind.ObjectMapper;
@ -45,8 +46,7 @@ public class DnetCollectorWorker {
public void collect() throws DnetCollectorException {
try {
final ObjectMapper jsonMapper = new ObjectMapper();
final ApiDescriptor api =
jsonMapper.readValue(argumentParser.get("apidescriptor"), ApiDescriptor.class);
final ApiDescriptor api = jsonMapper.readValue(argumentParser.get("apidescriptor"), ApiDescriptor.class);
final CollectorPlugin plugin = collectorPluginFactory.getPluginByProtocol(api.getProtocol());
@ -71,8 +71,8 @@ public class DnetCollectorWorker {
final Map<String, String> ongoingMap = new HashMap<>();
final Map<String, String> reportMap = new HashMap<>();
final AtomicInteger counter = new AtomicInteger(0);
try (SequenceFile.Writer writer =
SequenceFile.createWriter(
try (SequenceFile.Writer writer = SequenceFile
.createWriter(
conf,
SequenceFile.Writer.file(hdfswritepath),
SequenceFile.Writer.keyClass(IntWritable.class),
@ -88,9 +88,11 @@ public class DnetCollectorWorker {
if (counter.get() % 10 == 0) {
try {
ongoingMap.put("ongoing", "" + counter.get());
log.debug(
log
.debug(
"Sending message: "
+ manager.sendMessage(
+ manager
.sendMessage(
new Message(
argumentParser.get("workflowId"),
"Collection",
@ -111,14 +113,16 @@ public class DnetCollectorWorker {
});
}
ongoingMap.put("ongoing", "" + counter.get());
manager.sendMessage(
manager
.sendMessage(
new Message(
argumentParser.get("workflowId"), "Collection", MessageType.ONGOING, ongoingMap),
argumentParser.get("rabbitOngoingQueue"),
true,
false);
reportMap.put("collected", "" + counter.get());
manager.sendMessage(
manager
.sendMessage(
new Message(
argumentParser.get("workflowId"), "Collection", MessageType.REPORT, reportMap),
argumentParser.get("rabbitOngoingQueue"),

View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.worker;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
@ -8,9 +9,9 @@ import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
/**
* DnetCollectortWorkerApplication is the main class responsible to start the Dnet Collection into
* HDFS. This module will be executed on the hadoop cluster and taking in input some parameters that
* tells it which is the right collector plugin to use and where store the data into HDFS path
* DnetCollectortWorkerApplication is the main class responsible to start the Dnet Collection into HDFS. This module
* will be executed on the hadoop cluster and taking in input some parameters that tells it which is the right collector
* plugin to use and where store the data into HDFS path
*
* @author Sandro La Bruzzo
*/
@ -25,24 +26,23 @@ public class DnetCollectorWorkerApplication {
/** @param args */
public static void main(final String[] args) throws Exception {
argumentParser =
new ArgumentApplicationParser(
IOUtils.toString(
DnetCollectorWorker.class.getResourceAsStream(
argumentParser = new ArgumentApplicationParser(
IOUtils
.toString(
DnetCollectorWorker.class
.getResourceAsStream(
"/eu/dnetlib/collector/worker/collector_parameter.json")));
argumentParser.parseArgument(args);
log.info("hdfsPath =" + argumentParser.get("hdfsPath"));
log.info("json = " + argumentParser.get("apidescriptor"));
final MessageManager manager =
new MessageManager(
final MessageManager manager = new MessageManager(
argumentParser.get("rabbitHost"),
argumentParser.get("rabbitUser"),
argumentParser.get("rabbitPassword"),
false,
false,
null);
final DnetCollectorWorker worker =
new DnetCollectorWorker(collectorPluginFactory, argumentParser, manager);
final DnetCollectorWorker worker = new DnetCollectorWorker(collectorPluginFactory, argumentParser, manager);
worker.collect();
}
}

Some files were not shown because too many files have changed in this diff Show More