Search in sources :

Example 11 with TypeStorage

use of org.apache.rya.indexing.entity.storage.TypeStorage in project incubator-rya by apache.

the class DuplicateDataDetectorIT method testCreateEntityNearDuplicate.

@Test
public void testCreateEntityNearDuplicate() throws EntityStorageException, TypeStorageException, ObjectStorageException {
    // Create the types the Entity uses.
    final TypeStorage typeStorage = new MongoTypeStorage(super.getMongoClient(), RYA_INSTANCE_NAME);
    final Type personType = createPersonType();
    final Type employeeType = createEmployeeType();
    typeStorage.create(personType);
    typeStorage.create(employeeType);
    final Optional<Type> storedPersonType = typeStorage.get(personType.getId());
    final Optional<Type> storedEmployeeType = typeStorage.get(employeeType.getId());
    assertTrue(storedPersonType.isPresent());
    assertTrue(storedEmployeeType.isPresent());
    // Create it.
    final DuplicateDataConfig duplicateDataConfig = new DuplicateDataConfig(// boolean
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // byte
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // date
    new Tolerance(500.0, ToleranceType.DIFFERENCE), // double
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // float
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // integer
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // long
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // short
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // string
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // uri
    new Tolerance(1.0, ToleranceType.DIFFERENCE), new HashMap<String, List<String>>(), true);
    final DuplicateDataDetector duplicateDataDetector = new DuplicateDataDetector(duplicateDataConfig);
    final EntityStorage entityStorage = new MongoEntityStorage(super.getMongoClient(), RYA_INSTANCE_NAME, duplicateDataDetector);
    final Entity bobEntity = createBobEntity();
    entityStorage.create(bobEntity);
    assertTrue(entityStorage.get(bobEntity.getSubject()).isPresent());
    final Builder duplicateBobBuilder = Entity.builder(createBobEntity());
    duplicateBobBuilder.setSubject(createRyaUri("Robert"));
    // Modify a property for each type that is within tolerance
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 41)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(250.76f)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(72.499)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(50001)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 2)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(123456789013L)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 1).minusYears(40))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 1))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(true)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bob.smitch01@gmail.com"))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("124 Fake St. Washington, DC 20024")));
    duplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 556)));
    final Entity duplicateBobEntity = duplicateBobBuilder.build();
    // Try to create another entity that's considered a duplicate.
    // It will NOT be be created.
    boolean hasDuplicate = false;
    try {
        entityStorage.create(duplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        hasDuplicate = true;
    }
    assertTrue(hasDuplicate);
    assertFalse(entityStorage.get(duplicateBobEntity.getSubject()).isPresent());
    final Builder notDuplicateBobBuilder = Entity.builder(createBobEntity());
    notDuplicateBobBuilder.setSubject(createRyaUri("Not Bob"));
    // Modify a property for each type that is within tolerance
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 50)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(300.0f)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(100.0)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(60000)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 5)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(9L)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 10000000L).minusYears(40))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 10000000L))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(false)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bad.email.address@gmail.com"))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("123456789 Fake St. Washington, DC 20024")));
    notDuplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 1000)));
    final Entity notDuplicateBobEntity = notDuplicateBobBuilder.build();
    // It will be created.
    try {
        entityStorage.create(notDuplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        fail();
    }
    assertTrue(entityStorage.get(notDuplicateBobEntity.getSubject()).isPresent());
}
Also used : Entity(org.apache.rya.indexing.entity.model.Entity) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) EntityStorage(org.apache.rya.indexing.entity.storage.EntityStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) ReflectionToStringBuilder(org.apache.commons.lang.builder.ReflectionToStringBuilder) Builder(org.apache.rya.indexing.entity.model.Entity.Builder) URIImpl(org.openrdf.model.impl.URIImpl) DateTime(org.joda.time.DateTime) Date(java.util.Date) MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) TypeStorage(org.apache.rya.indexing.entity.storage.TypeStorage) MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) RyaType(org.apache.rya.api.domain.RyaType) RyaTypeUtils.shortRyaType(org.apache.rya.api.domain.RyaTypeUtils.shortRyaType) RyaTypeUtils.floatRyaType(org.apache.rya.api.domain.RyaTypeUtils.floatRyaType) RyaTypeUtils.uriRyaType(org.apache.rya.api.domain.RyaTypeUtils.uriRyaType) RyaTypeUtils.longRyaType(org.apache.rya.api.domain.RyaTypeUtils.longRyaType) RyaTypeUtils.stringRyaType(org.apache.rya.api.domain.RyaTypeUtils.stringRyaType) RyaTypeUtils.doubleRyaType(org.apache.rya.api.domain.RyaTypeUtils.doubleRyaType) RyaTypeUtils.byteRyaType(org.apache.rya.api.domain.RyaTypeUtils.byteRyaType) RyaTypeUtils.booleanRyaType(org.apache.rya.api.domain.RyaTypeUtils.booleanRyaType) RyaTypeUtils.dateRyaType(org.apache.rya.api.domain.RyaTypeUtils.dateRyaType) Type(org.apache.rya.indexing.entity.model.Type) RyaTypeUtils.intRyaType(org.apache.rya.api.domain.RyaTypeUtils.intRyaType) DuplicateDataConfig(org.apache.rya.indexing.smarturi.duplication.conf.DuplicateDataConfig) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) Property(org.apache.rya.indexing.entity.model.Property) Test(org.junit.Test)

Example 12 with TypeStorage

use of org.apache.rya.indexing.entity.storage.TypeStorage in project incubator-rya by apache.

the class MongoEntityIndex2IT method beforeClass.

@Before
public void beforeClass() throws Exception {
    optimizer = new EntityIndexOptimizer();
    optimizer.setConf(conf);
    final TypeStorage typeStorage = optimizer.getTypeStorage();
    typeStorage.create(PERSON_TYPE);
    final Entity entity = Entity.builder().setSubject(new RyaURI("urn:SSN:111-11-1111")).setExplicitType(RYA_PERSON_TYPE).setProperty(RYA_PERSON_TYPE, new Property(new RyaURI("urn:age"), new RyaType("25"))).setProperty(RYA_PERSON_TYPE, new Property(new RyaURI("urn:eye"), new RyaType("blue"))).setProperty(RYA_PERSON_TYPE, new Property(new RyaURI("urn:name"), new RyaType("bob"))).build();
    entityStorage = optimizer.getEntityStorage();
    entityStorage.create(entity);
}
Also used : TypeStorage(org.apache.rya.indexing.entity.storage.TypeStorage) Entity(org.apache.rya.indexing.entity.model.Entity) RyaURI(org.apache.rya.api.domain.RyaURI) EntityIndexOptimizer(org.apache.rya.indexing.entity.EntityIndexOptimizer) RyaType(org.apache.rya.api.domain.RyaType) Property(org.apache.rya.indexing.entity.model.Property) Before(org.junit.Before)

Example 13 with TypeStorage

use of org.apache.rya.indexing.entity.storage.TypeStorage in project incubator-rya by apache.

the class MongoEntityIndexerIT method deleteStatement_deletesType.

@Test
public void deleteStatement_deletesType() throws Exception {
    try (MongoEntityIndexer indexer = new MongoEntityIndexer()) {
        indexer.setConf(conf);
        indexer.init();
        // Load the type into the TypeStorage.
        final TypeStorage types = new MongoTypeStorage(getMongoClient(), conf.getRyaInstanceName());
        types.create(PERSON_TYPE);
        types.create(EMPLOYEE_TYPE);
        // Index a bunch of RyaStatements.
        final RyaURI aliceSSN = new RyaURI("urn:SSN/111-11-1111");
        indexer.storeStatements(Sets.newHashSet(new RyaStatement(aliceSSN, new RyaURI(RDF.TYPE.toString()), new RyaType(XMLSchema.ANYURI, "urn:person")), new RyaStatement(aliceSSN, new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice")), new RyaStatement(aliceSSN, new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30")), new RyaStatement(aliceSSN, new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue"))));
        // Remove the explicit type from Alice.
        indexer.deleteStatement(new RyaStatement(aliceSSN, new RyaURI(RDF.TYPE.toString()), new RyaType(XMLSchema.ANYURI, "urn:person")));
        // Fetch the Entity from storage and ensure it looks correct.
        final EntityStorage entities = new MongoEntityStorage(getMongoClient(), conf.getRyaInstanceName());
        final Entity entity = entities.get(new RyaURI("urn:SSN/111-11-1111")).get();
        final Entity expected = Entity.builder().setSubject(aliceSSN).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue"))).setProperty(new RyaURI("urn:employee"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setVersion(entity.getVersion()).build();
        assertEquals(expected, entity);
    }
}
Also used : MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) TypeStorage(org.apache.rya.indexing.entity.storage.TypeStorage) RyaURI(org.apache.rya.api.domain.RyaURI) Entity(org.apache.rya.indexing.entity.model.Entity) MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) EntityStorage(org.apache.rya.indexing.entity.storage.EntityStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) RyaStatement(org.apache.rya.api.domain.RyaStatement) RyaType(org.apache.rya.api.domain.RyaType) Property(org.apache.rya.indexing.entity.model.Property) Test(org.junit.Test)

Example 14 with TypeStorage

use of org.apache.rya.indexing.entity.storage.TypeStorage in project incubator-rya by apache.

the class MongoEntityIndexerIT method addStatement_manyUpdates.

@Test
public void addStatement_manyUpdates() throws Exception {
    try (MongoEntityIndexer indexer = new MongoEntityIndexer()) {
        indexer.setConf(conf);
        indexer.init();
        // Load the types into the TypeStorage.
        final TypeStorage types = new MongoTypeStorage(getMongoClient(), conf.getRyaInstanceName());
        types.create(PERSON_TYPE);
        types.create(EMPLOYEE_TYPE);
        // Index a bunch of RyaStatements.
        final RyaURI aliceSSN = new RyaURI("urn:SSN/111-11-1111");
        indexer.storeStatement(new RyaStatement(aliceSSN, new RyaURI(RDF.TYPE.toString()), new RyaType(XMLSchema.ANYURI, "urn:person")));
        indexer.storeStatement(new RyaStatement(aliceSSN, new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice")));
        indexer.storeStatement(new RyaStatement(aliceSSN, new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30")));
        indexer.storeStatement(new RyaStatement(aliceSSN, new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue")));
        // Fetch the Entity from storage and ensure it looks correct.
        final EntityStorage entities = new MongoEntityStorage(getMongoClient(), conf.getRyaInstanceName());
        final Entity entity = entities.get(new RyaURI("urn:SSN/111-11-1111")).get();
        final Entity expected = Entity.builder().setSubject(aliceSSN).setExplicitType(new RyaURI("urn:person")).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue"))).setProperty(new RyaURI("urn:employee"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setVersion(entity.getVersion()).build();
        assertEquals(expected, entity);
    }
}
Also used : MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) TypeStorage(org.apache.rya.indexing.entity.storage.TypeStorage) RyaURI(org.apache.rya.api.domain.RyaURI) Entity(org.apache.rya.indexing.entity.model.Entity) MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) EntityStorage(org.apache.rya.indexing.entity.storage.EntityStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) RyaStatement(org.apache.rya.api.domain.RyaStatement) RyaType(org.apache.rya.api.domain.RyaType) Property(org.apache.rya.indexing.entity.model.Property) Test(org.junit.Test)

Example 15 with TypeStorage

use of org.apache.rya.indexing.entity.storage.TypeStorage in project incubator-rya by apache.

the class MongoEntityIndexerIT method addStatements.

@Test
public void addStatements() throws Exception {
    try (MongoEntityIndexer indexer = new MongoEntityIndexer()) {
        indexer.setConf(conf);
        indexer.init();
        // Load the types into the TypeStorage.
        final TypeStorage types = new MongoTypeStorage(getMongoClient(), conf.getRyaInstanceName());
        types.create(PERSON_TYPE);
        types.create(EMPLOYEE_TYPE);
        // Index a bunch of RyaStatements.
        final RyaURI aliceSSN = new RyaURI("urn:SSN/111-11-1111");
        final RyaURI bobSSN = new RyaURI("urn:SSN/222-22-2222");
        indexer.storeStatements(Sets.newHashSet(new RyaStatement(aliceSSN, new RyaURI(RDF.TYPE.toString()), new RyaType(XMLSchema.ANYURI, "urn:person")), new RyaStatement(aliceSSN, new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice")), new RyaStatement(aliceSSN, new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30")), new RyaStatement(aliceSSN, new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue")), new RyaStatement(bobSSN, new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Bob")), new RyaStatement(bobSSN, new RyaURI("urn:hoursPerWeek"), new RyaType(XMLSchema.INT, "40")), new RyaStatement(bobSSN, new RyaURI(RDF.TYPE.toString()), new RyaType(XMLSchema.ANYURI, "urn:employee"))));
        // Fetch the Entity from storage and ensure it looks correct.
        final EntityStorage entities = new MongoEntityStorage(getMongoClient(), conf.getRyaInstanceName());
        final Entity alice = entities.get(aliceSSN).get();
        final Entity bob = entities.get(bobSSN).get();
        final Set<Entity> storedEntities = Sets.newHashSet(alice, bob);
        final Entity expectedAlice = Entity.builder().setSubject(aliceSSN).setExplicitType(new RyaURI("urn:person")).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:age"), new RyaType(XMLSchema.INT, "30"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:eye"), new RyaType(XMLSchema.STRING, "blue"))).setProperty(new RyaURI("urn:employee"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Alice"))).setVersion(alice.getVersion()).build();
        final Entity expectedBob = Entity.builder().setSubject(bobSSN).setExplicitType(new RyaURI("urn:employee")).setProperty(new RyaURI("urn:employee"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Bob"))).setProperty(new RyaURI("urn:employee"), new Property(new RyaURI("urn:hoursPerWeek"), new RyaType(XMLSchema.INT, "40"))).setProperty(new RyaURI("urn:person"), new Property(new RyaURI("urn:name"), new RyaType(XMLSchema.STRING, "Bob"))).setVersion(bob.getVersion()).build();
        final Set<Entity> expected = Sets.newHashSet(expectedAlice, expectedBob);
        assertEquals(expected, storedEntities);
    }
}
Also used : MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) TypeStorage(org.apache.rya.indexing.entity.storage.TypeStorage) RyaURI(org.apache.rya.api.domain.RyaURI) Entity(org.apache.rya.indexing.entity.model.Entity) MongoTypeStorage(org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) EntityStorage(org.apache.rya.indexing.entity.storage.EntityStorage) MongoEntityStorage(org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage) RyaStatement(org.apache.rya.api.domain.RyaStatement) RyaType(org.apache.rya.api.domain.RyaType) Property(org.apache.rya.indexing.entity.model.Property) Test(org.junit.Test)

Aggregations

TypeStorage (org.apache.rya.indexing.entity.storage.TypeStorage)18 RyaURI (org.apache.rya.api.domain.RyaURI)16 Test (org.junit.Test)14 Entity (org.apache.rya.indexing.entity.model.Entity)11 RyaType (org.apache.rya.api.domain.RyaType)10 EntityStorage (org.apache.rya.indexing.entity.storage.EntityStorage)10 Property (org.apache.rya.indexing.entity.model.Property)9 Type (org.apache.rya.indexing.entity.model.Type)9 MongoEntityStorage (org.apache.rya.indexing.entity.storage.mongo.MongoEntityStorage)8 MongoTypeStorage (org.apache.rya.indexing.entity.storage.mongo.MongoTypeStorage)8 RyaStatement (org.apache.rya.api.domain.RyaStatement)7 ImmutableList (com.google.common.collect.ImmutableList)2 IOException (java.io.IOException)2 Date (java.util.Date)2 List (java.util.List)2 ReflectionToStringBuilder (org.apache.commons.lang.builder.ReflectionToStringBuilder)2 RyaTypeUtils.booleanRyaType (org.apache.rya.api.domain.RyaTypeUtils.booleanRyaType)2 RyaTypeUtils.byteRyaType (org.apache.rya.api.domain.RyaTypeUtils.byteRyaType)2 RyaTypeUtils.dateRyaType (org.apache.rya.api.domain.RyaTypeUtils.dateRyaType)2 RyaTypeUtils.doubleRyaType (org.apache.rya.api.domain.RyaTypeUtils.doubleRyaType)2