use of org.apache.clerezza.commons.rdf.IRI in project stanbol by apache.
the class EntityLinkerConfig method setTypeMapping.
/**
* @param conceptType the type of the concept or <code>null</code> to
* add the default dc:type mapping. See also {@link #setDefaultDcType(IRI)}
* @param dcType the dc:type for the parsed concept type
* @return the previously mapped dc:type value if an existing mapping
* was updated or <code>null</code> if a new mapping was added.
*/
public IRI setTypeMapping(String conceptType, IRI dcType) {
if (dcType == null) {
throw new IllegalArgumentException("The parsed dc:type URI MUST NOT be NULL!");
}
if (conceptType == null) {
// handle setting of the default dc:type value
IRI oldDefault = getDefaultDcType();
setDefaultDcType(dcType);
return oldDefault;
}
return typeMappings.put(conceptType, dcType);
}
use of org.apache.clerezza.commons.rdf.IRI in project stanbol by apache.
the class LangIdEnhancementEngine method computeEnhancements.
public void computeEnhancements(ContentItem ci) throws EngineException {
Entry<IRI, Blob> contentPart = ContentItemHelper.getBlob(ci, SUPPORTED_MIMTYPES);
if (contentPart == null) {
throw new IllegalStateException("No ContentPart with Mimetype '" + TEXT_PLAIN_MIMETYPE + "' found for ContentItem " + ci.getUri() + ": This is also checked in the canEnhance method! -> This " + "indicated an Bug in the implementation of the " + "EnhancementJobManager!");
}
String text = "";
try {
text = ContentItemHelper.getText(contentPart.getValue());
} catch (IOException e) {
throw new InvalidContentException(this, ci, e);
}
if (text.trim().length() == 0) {
log.info("No text contained in ContentPart {} of ContentItem {}", contentPart.getKey(), ci.getUri());
return;
}
// truncate text to some piece from the middle if probeLength > 0
int checkLength = probeLength;
if (checkLength > 0 && text.length() > checkLength) {
text = text.substring(text.length() / 2 - checkLength / 2, text.length() / 2 + checkLength / 2);
}
LanguageIdentifier languageIdentifier = new LanguageIdentifier(text);
String language = languageIdentifier.getLanguage();
log.info("language identified as " + language);
// add language to metadata
Graph g = ci.getMetadata();
ci.getLock().writeLock().lock();
try {
IRI textEnhancement = EnhancementEngineHelper.createTextEnhancement(ci, this);
g.add(new TripleImpl(textEnhancement, DC_LANGUAGE, new PlainLiteralImpl(language)));
g.add(new TripleImpl(textEnhancement, DC_TYPE, DCTERMS_LINGUISTIC_SYSTEM));
} finally {
ci.getLock().writeLock().unlock();
}
}
use of org.apache.clerezza.commons.rdf.IRI in project stanbol by apache.
the class FstLinkingEngineComponent method applyConfig.
/**
* Called by {@link #activate(ComponentContext)},
* {@link PlainFstLinkingComponnet#activate(ComponentContext)} and
* {@link NamedEntityFstLinkingComponnet#activate(ComponentContext)} to
* apply the parsed {@link ComponentContext#getProperties()}. The
* {@link LinkingModeEnum linking mode} is parsed separately as OSGI does not
* allow to modify the parsed config and sup-classes do need to override
* the linking mode.
* @param linkingMode the linking mode
* @param properties
* @throws ConfigurationException
*/
protected void applyConfig(LinkingModeEnum linkingMode, Dictionary<String, Object> properties, NamespacePrefixService prefixService) throws ConfigurationException {
// (0) The name for the Enhancement Engine and the basic metadata
Object value = properties.get(PROPERTY_NAME);
if (value == null || value.toString().isEmpty()) {
throw new ConfigurationException(PROPERTY_NAME, "The EnhancementEngine name MUST BE configured!");
} else {
this.engineName = value.toString();
}
log.info(" - engine name: {}", engineName);
engineMetadata = new Hashtable<String, Object>();
engineMetadata.put(PROPERTY_NAME, this.engineName);
value = properties.get(Constants.SERVICE_RANKING);
engineMetadata.put(Constants.SERVICE_RANKING, value == null ? Integer.valueOf(0) : value);
// (0) set the linking mode
this.linkingMode = linkingMode;
log.info(" - linking mode: {}", linkingMode);
// (1) parse the TextProcessing configuration
// TODO: decide if we should use the TextProcessingConfig for this engine
textProcessingConfig = TextProcessingConfig.createInstance(properties);
// change default for EntityLinkerConfig.MIN_FOUND_TOKENS
value = properties.get(EntityLinkerConfig.MIN_FOUND_TOKENS);
entityLinkerConfig = EntityLinkerConfig.createInstance(properties, prefixService);
if (value == null) {
// no MIN_FOUND_TOKENS config present
// manually set the default to the value used by this engine
entityLinkerConfig.setMinFoundTokens(FST_DEFAULT_MIN_FOUND_TOKENS);
}
// (2) parse the configured IndexReference
value = properties.get(SOLR_CORE);
if (value == null) {
throw new ConfigurationException(SOLR_CORE, "Missing required configuration of the SolrCore");
} else {
indexReference = IndexReference.parse(value.toString());
}
value = properties.get(IndexConfiguration.FIELD_ENCODING);
if (value == null) {
throw new ConfigurationException(IndexConfiguration.FIELD_ENCODING, "Missing required configuration of the Solr Field Encoding");
} else {
try {
fieldEncoding = FieldEncodingEnum.valueOf(value.toString().trim());
} catch (IllegalArgumentException e) {
throw new ConfigurationException(IndexConfiguration.FIELD_ENCODING, "The configured " + "FieldEncoding MUST BE a member of " + Arrays.toString(FieldEncodingEnum.values()), e);
}
}
value = properties.get(IndexConfiguration.SKIP_ALT_TOKENS);
if (value instanceof Boolean) {
skipAltTokensConfig = ((Boolean) value);
} else if (value != null) {
skipAltTokensConfig = Boolean.valueOf(value.toString());
}
// else no config -> will use the default
// (4) parse Origin information
value = properties.get(ORIGIN);
if (value instanceof RDFTerm) {
origin = (RDFTerm) origin;
} else if (value instanceof String) {
try {
URI originUri = new URI((String) value);
if (originUri.isAbsolute()) {
origin = new IRI((String) value);
} else {
origin = new PlainLiteralImpl((String) value);
}
} catch (URISyntaxException e) {
origin = new PlainLiteralImpl((String) value);
}
log.info(" - origin: {}", origin);
} else if (value != null) {
log.warn("Values of the {} property MUST BE of type RDFTerm or String " + "(parsed: {} (type:{}))", new Object[] { ORIGIN, value, value.getClass() });
}
// else no ORIGIN information provided
// (5) init the FST configuration
// We can create the default configuration only here, as it depends on the
// name of the solrIndex
String defaultConfig = "*;" + IndexConfiguration.PARAM_FST + "=" + indexReference.getIndex() + ";" + IndexConfiguration.PARAM_FIELD + "=" + IndexConfiguration.DEFAULT_FIELD;
fstConfig = new LanguageConfiguration(IndexConfiguration.FST_CONFIG, new String[] { defaultConfig });
// now set the actual configuration parsed to the engine
value = properties.get(IndexConfiguration.FST_CONFIG);
if (value != null && !StringUtils.isBlank(value.toString())) {
fstConfig.setConfiguration(properties);
}
// else keep the default
value = properties.get(IndexConfiguration.FST_FOLDER);
if (value instanceof String) {
this.fstFolder = ((String) value).trim();
if (this.fstFolder.isEmpty()) {
this.fstFolder = null;
}
} else if (value == null) {
this.fstFolder = null;
} else {
throw new ConfigurationException(IndexConfiguration.FST_FOLDER, "Values MUST BE of type String" + "(found: " + value.getClass().getName() + ")!");
}
// (6) Create the ThreadPool used for the runtime creation of FST models
value = properties.get(FST_THREAD_POOL_SIZE);
int tpSize;
if (value instanceof Number) {
tpSize = ((Number) value).intValue();
} else if (value != null) {
try {
tpSize = Integer.parseInt(value.toString());
} catch (NumberFormatException e) {
throw new ConfigurationException(FST_THREAD_POOL_SIZE, "Unable to parse the integer FST thread pool size from the " + "configured " + value.getClass().getSimpleName() + " '" + value + "'!", e);
}
} else {
tpSize = -1;
}
if (tpSize <= 0) {
// if configured value <= 0 we use the default
tpSize = DEFAULT_FST_THREAD_POOL_SIZE;
}
// build a ThreadFactoryBuilder for low priority daemon threads that
// do use a meaningful name
ThreadFactoryBuilder tfBuilder = new ThreadFactoryBuilder();
// should be stopped if the VM closes
tfBuilder.setDaemon(true);
// low priority
tfBuilder.setPriority(Thread.MIN_PRIORITY);
tfBuilder.setNameFormat(engineName + "-FstRuntimeCreation-thread-%d");
if (fstCreatorService != null && !fstCreatorService.isTerminated()) {
// NOTE: We can not call terminateNow, because to interrupt threads
// here would also close FileChannels used by the SolrCore
// and produce java.nio.channels.ClosedByInterruptException
// exceptions followed by java.nio.channels.ClosedChannelException
// on following calls to affected files of the SolrIndex.
// Because of that we just log a warning and let uncompleted tasks
// complete!
log.warn("some items in a previouse FST Runtime Creation Threadpool have " + "still not finished!");
}
fstCreatorService = Executors.newFixedThreadPool(tpSize, tfBuilder.build());
// (7) Parse the EntityCache config
int entityCacheSize;
value = properties.get(ENTITY_CACHE_SIZE);
if (value instanceof Number) {
entityCacheSize = ((Number) value).intValue();
} else if (value != null) {
try {
entityCacheSize = Integer.parseInt(value.toString());
} catch (NumberFormatException e) {
throw new ConfigurationException(ENTITY_CACHE_SIZE, "Unable to parse the integer EntityCacheSize from the " + "configured " + value.getClass().getSimpleName() + " '" + value + "'!", e);
}
} else {
entityCacheSize = -1;
}
if (entityCacheSize == 0) {
log.info(" ... EntityCache deactivated");
this.entityCacheSize = entityCacheSize;
} else {
this.entityCacheSize = entityCacheSize < 0 ? DEFAULT_ENTITY_CACHE_SIZE : entityCacheSize;
log.info(" ... EntityCache enabled (size: {})", this.entityCacheSize);
}
// (8) parse the Entity type field
value = properties.get(IndexConfiguration.SOLR_TYPE_FIELD);
if (value == null || StringUtils.isBlank(value.toString())) {
solrTypeField = null;
} else {
solrTypeField = value.toString().trim();
}
// (9) parse the Entity Ranking field
value = properties.get(IndexConfiguration.SOLR_RANKING_FIELD);
if (value == null) {
solrRankingField = null;
} else {
solrRankingField = value.toString().trim();
}
// (10) parse the NamedEntity type mappings (if linkingMode = NER)
if (linkingMode == LinkingModeEnum.NER) {
nerTypeMappings = new HashMap<String, Set<String>>();
value = properties.get(NAMED_ENTITY_TYPE_MAPPINGS);
if (value instanceof String[]) {
// support array
value = Arrays.asList((String[]) value);
} else if (value instanceof String) {
// single value
value = Collections.singleton(value);
}
if (value instanceof Collection<?>) {
// and collection
log.info(" - process Named Entity Type Mappings (used by LinkingMode: {})", linkingMode);
configs: for (Object o : (Iterable<?>) value) {
if (o != null) {
StringBuilder usage = new StringBuilder("useage: ");
usage.append("'{namedEntity-tag-or-uri} > {entityType-1}[,{entityType-n}]'");
String[] config = o.toString().split(">");
String namedEntityType = config[0].trim();
if (namedEntityType.isEmpty()) {
log.warn("Invalid Type Mapping Config '{}': Missing namedEntityType ({}) -> ignore this config", o, usage);
continue configs;
}
if (NamespaceMappingUtils.getPrefix(namedEntityType) != null) {
namedEntityType = NamespaceMappingUtils.getConfiguredUri(prefixService, NAMED_ENTITY_TYPE_MAPPINGS, namedEntityType);
}
if (config.length < 2 || config[1].isEmpty()) {
log.warn("Invalid Type Mapping Config '{}': Missing dc:type URI '{}' ({}) -> ignore this config", o, usage);
continue configs;
}
String entityTypes = config[1].trim();
if (config.length > 2) {
log.warn("Configuration after 2nd '>' gets ignored. Will use mapping '{} > {}' from config {}", new Object[] { namedEntityType, entityTypes, o });
}
Set<String> types = nerTypeMappings.get(namedEntityType);
if (types == null) {
// add new element to the mapping
types = new HashSet<String>();
nerTypeMappings.put(namedEntityType, types);
}
for (String entityType : entityTypes.split(";")) {
entityType = entityType.trim();
if (!entityType.isEmpty()) {
String typeUri;
if ("*".equals(entityType)) {
// null is used as wildcard
typeUri = null;
} else {
typeUri = NamespaceMappingUtils.getConfiguredUri(prefixService, NAMED_ENTITY_TYPE_MAPPINGS, entityType);
}
log.info(" - add {} > {}", namedEntityType, typeUri);
types.add(typeUri);
}
// else ignore empty mapping
}
}
}
} else {
// no mappings defined ... set wildcard mapping
log.info(" - No Named Entity type mappings configured. Will use wildcard mappings");
nerTypeMappings = Collections.singletonMap(null, Collections.<String>singleton(null));
}
}
// (11) start tracking the SolrCore
try {
solrServerTracker = new RegisteredSolrServerTracker(bundleContext, indexReference, null) {
@Override
public void removedService(ServiceReference reference, Object service) {
log.info(" ... SolrCore for {} was removed!", reference);
// try to get an other serviceReference from the tracker
if (reference.equals(FstLinkingEngineComponent.this.solrServerReference)) {
updateEngineRegistration(solrServerTracker.getServiceReference(), null);
} else {
log.info(" - removed SolrCore was not used for FST linking");
}
super.removedService(reference, service);
}
@Override
public void modifiedService(ServiceReference reference, Object service) {
log.info(" ... SolrCore for {} was updated!", indexReference);
updateEngineRegistration(solrServerTracker.getServiceReference(), null);
super.modifiedService(reference, service);
}
@Override
public SolrServer addingService(ServiceReference reference) {
SolrServer server = super.addingService(reference);
if (solrCore != null) {
log.info("Multiple SolrCores for name {}! Will update engine " + "with the newly added {}!", new Object[] { solrCore.getName(), indexReference, reference });
}
updateEngineRegistration(reference, server);
return server;
}
};
} catch (InvalidSyntaxException e) {
throw new ConfigurationException(SOLR_CORE, "parsed SolrCore name '" + value.toString() + "' is invalid (expected: '[{server-name}:]{indexname}'");
}
try {
solrServerTracker.open();
} catch (RuntimeException e) {
// FIX for STANBOL-1416 (see https://issues.apache.org/jira/browse/STANBOL-1416)
// If an available SolrCore can not be correctly initialized we will
// get the exception here. In this case we want this component to be
// activated and waiting for further service events. Because of that
// we catch here the exception.
log.debug("Error while processing existing SolrCore Service during " + "opening SolrServiceTracker ... waiting for further service" + "Events", e);
}
}
use of org.apache.clerezza.commons.rdf.IRI in project stanbol by apache.
the class KeywordLinkingEngineTest method testEngine.
/**
* This tests if the Enhancements created by the Engine confirm to the
* rules defined for the Stanbol Enhancement Structure.
* @throws IOException
* @throws EngineException
*/
@Test
public void testEngine() throws IOException, EngineException {
EntityLinkerConfig linkerConfig = new EntityLinkerConfig();
linkerConfig.setRedirectProcessingMode(RedirectProcessingMode.FOLLOW);
KeywordLinkingEngine engine = KeywordLinkingEngine.createInstance(openNLP, searcher, new TextAnalyzerConfig(), linkerConfig);
engine.referencedSiteName = TEST_REFERENCED_SITE_NAME;
ContentItem ci = ciFactory.createContentItem(new StringSource(TEST_TEXT));
// tells the engine that this is an English text
ci.getMetadata().add(new TripleImpl(ci.getUri(), DC_LANGUAGE, new PlainLiteralImpl("en")));
// compute the enhancements
engine.computeEnhancements(ci);
// validate the enhancement results
Map<IRI, RDFTerm> expectedValues = new HashMap<IRI, RDFTerm>();
expectedValues.put(ENHANCER_EXTRACTED_FROM, ci.getUri());
expectedValues.put(DC_CREATOR, LiteralFactory.getInstance().createTypedLiteral(engine.getClass().getName()));
// adding null as expected for confidence makes it a required property
expectedValues.put(Properties.ENHANCER_CONFIDENCE, null);
// validate create fise:TextAnnotations
int numTextAnnotations = validateAllTextAnnotations(ci.getMetadata(), TEST_TEXT, expectedValues);
assertEquals("Four fise:TextAnnotations are expected by this Test", 4, numTextAnnotations);
// validate create fise:EntityAnnotations
int numEntityAnnotations = validateAllEntityAnnotations(ci, expectedValues);
assertEquals("Five fise:EntityAnnotations are expected by this Test", 5, numEntityAnnotations);
}
use of org.apache.clerezza.commons.rdf.IRI in project stanbol by apache.
the class ClerezzaRDFUtils method urifyBlankNodes.
public static void urifyBlankNodes(Graph model) {
HashMap<BlankNode, IRI> blankNodeMap = new HashMap<BlankNode, IRI>();
Graph remove = new SimpleGraph();
Graph add = new SimpleGraph();
for (Triple t : model) {
BlankNodeOrIRI subj = t.getSubject();
RDFTerm obj = t.getObject();
IRI pred = t.getPredicate();
boolean match = false;
if (subj instanceof BlankNode) {
match = true;
IRI ru = blankNodeMap.get(subj);
if (ru == null) {
ru = createRandomUri();
blankNodeMap.put((BlankNode) subj, ru);
}
subj = ru;
}
if (obj instanceof BlankNode) {
match = true;
IRI ru = blankNodeMap.get(obj);
if (ru == null) {
ru = createRandomUri();
blankNodeMap.put((BlankNode) obj, ru);
}
obj = ru;
}
if (match) {
remove.add(t);
add.add(new TripleImpl(subj, pred, obj));
}
}
model.removeAll(remove);
model.addAll(add);
}
Aggregations