Examples with LanguageConfiguration org.apache.stanbol.enhancer.nlp.utils.LanguageConfiguration used on opensource projects

Example 1 with LanguageConfiguration

use of org.apache.stanbol.enhancer.nlp.utils.LanguageConfiguration 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);
    }
}

Example 2 with LanguageConfiguration

use of org.apache.stanbol.enhancer.nlp.utils.LanguageConfiguration in project stanbol by apache.

the class FstLinkingEngineTest method setup.

@BeforeClass
public static void setup() throws Exception {
    // get the working directory
    // use property substitution to test this feature!
    String prefix = System.getProperty("basedir") == null ? "." : "${basedir}";
    String solrServerDir = prefix + TEST_INDEX_REL_PATH;
    log.info("Test Solr Server Directory: {}", solrServerDir);
    System.setProperty(ManagedSolrServer.MANAGED_SOLR_DIR_PROPERTY, solrServerDir);
    SolrYardConfig config = new SolrYardConfig(TEST_YARD_ID, TEST_SOLR_CORE_NAME);
    config.setAllowInitialisation(false);
    //the dbpedia default data
    config.setIndexConfigurationName(TEST_SOLR_CORE_CONFIGURATION);
    //init from datafile provider
    config.setAllowInitialisation(true);
    config.setName("DBpedia.org default data");
    config.setDescription("Data used for FstLinkingEngie tests");
    // create the Yard used for the tests
    IndexReference solrIndexRef = IndexReference.parse(config.getSolrServerLocation());
    SolrServer server = StandaloneEmbeddedSolrServerProvider.getInstance().getSolrServer(solrIndexRef, config.getIndexConfigurationName());
    Assert.assertNotNull("Unable to initialise SolrServer for testing", server);
    core = ((EmbeddedSolrServer) server).getCoreContainer().getCore(solrIndexRef.getIndex());
    Assert.assertNotNull("Unable to get SolrCore '" + config.getIndexConfigurationName() + "' from SolrServer " + server, core);
    yard = new SolrYard(server, config, null);
    //setup the index configuration
    LanguageConfiguration langConf = new LanguageConfiguration("not.used", new String[] { "en;field=dbpedia-ont:surfaceForm;generate=true" });
    fstConfig = new IndexConfiguration(langConf, core, FieldEncodingEnum.SolrYard, "");
    fstConfig.setExecutorService(Executors.newFixedThreadPool(1));
    fstConfig.setTypeField("rdf:type");
    fstConfig.setRankingField("entityhub:entityRank");
    //fstConfig.setEntityCacheManager(new FastLRUCacheManager(2048));
    fstConfig.setOrigin(new PlainLiteralImpl(TEST_ORIGIN));
    //activate the FST config
    //activate this configuration
    fstConfig.activate();
    //validate that the index contains the expected entities
    validateTestIndex();
    //now create the FST models
    List<Future<?>> creationTasks = new ArrayList<Future<?>>();
    for (CorpusInfo corpus : fstConfig.getCorpora()) {
        Assert.assertTrue("Failure in UnitTest - all FST models need to be generate=true", corpus.allowCreation);
        if (!corpus.isFstFile()) {
            //create a task on the FST corpus creation service
            creationTasks.add(fstConfig.getExecutorService().submit(new CorpusCreationTask(fstConfig, corpus)));
        }
    }
    //typical hardware
    for (Future<?> future : creationTasks) {
        try {
            future.get(FST_CREATION_WAIT_TIME, TimeUnit.SECONDS);
        } catch (TimeoutException e) {
        // we assert on future.isDone instead
        }
        Assert.assertTrue("FST Model creation not finished after " + FST_CREATION_WAIT_TIME + "seconds", future.isDone());
    }
}

Example 3 with LanguageConfiguration

use of org.apache.stanbol.enhancer.nlp.utils.LanguageConfiguration in project stanbol by apache.

the class MainLabelTokenizer method activate.

@Activate
protected void activate(ComponentContext ctx) {
    final BundleContext bundleContext = ctx.getBundleContext();
    final String managerServicePid = (String) ctx.getProperties().get(Constants.SERVICE_PID);
    labelTokenizerTracker = new ServiceTracker(bundleContext, LabelTokenizer.class.getName(), new ServiceTrackerCustomizer() {

        @Override
        public Object addingService(ServiceReference reference) {
            if (managerServicePid.equals(reference.getProperty(Constants.SERVICE_PID))) {
                //do not track this manager!
                return null;
            }
            LanguageConfiguration langConf = new LanguageConfiguration(SUPPORTED_LANUAGES, DEFAULT_LANG_CONF);
            try {
                langConf.setConfiguration(reference);
            } catch (ConfigurationException e) {
                log.error("Unable to track ServiceReference {} becuase of invalid LanguageConfiguration(" + SUPPORTED_LANUAGES + "=" + reference.getProperty(SUPPORTED_LANUAGES) + ")!", e);
                return null;
            }
            Object service = bundleContext.getService(reference);
            if (service != null) {
                ref2LangConfig.put(reference, langConf);
                langTokenizers.clear();
            }
            return service;
        }

        @Override
        public void modifiedService(ServiceReference reference, Object service) {
            if (managerServicePid.equals(reference.getProperty(Constants.SERVICE_PID))) {
                //ignore this service!
                return;
            }
            LanguageConfiguration langConf = new LanguageConfiguration(SUPPORTED_LANUAGES, DEFAULT_LANG_CONF);
            try {
                langConf.setConfiguration(reference);
                ref2LangConfig.put(reference, langConf);
                langTokenizers.clear();
            } catch (ConfigurationException e) {
                log.error("Unable to track ServiceReference {} becuase of invalid LanguageConfiguration(" + SUPPORTED_LANUAGES + "=" + reference.getProperty(SUPPORTED_LANUAGES) + ")!", e);
                if (ref2LangConfig.remove(reference) != null) {
                    langTokenizers.clear();
                }
            }
        }

        @Override
        public void removedService(ServiceReference reference, Object service) {
            if (managerServicePid.equals(reference.getProperty(Constants.SERVICE_PID))) {
                //ignore this service
                return;
            }
            bundleContext.ungetService(reference);
            if (ref2LangConfig.remove(reference) != null) {
                langTokenizers.clear();
            }
        }
    });
    labelTokenizerTracker.open();
}

Example 4 with LanguageConfiguration

use of org.apache.stanbol.enhancer.nlp.utils.LanguageConfiguration in project stanbol by apache.

the class MainLabelTokenizer method initTokenizers.

@SuppressWarnings("unchecked")
private List<ServiceReference> initTokenizers(String language) {
    List<ServiceReference> tokenizers = new ArrayList<ServiceReference>();
    if (labelTokenizerTracker.getServiceReferences() != null) {
        for (ServiceReference ref : labelTokenizerTracker.getServiceReferences()) {
            LanguageConfiguration langConf = ref2LangConfig.get(ref);
            if (langConf != null && langConf.isLanguage(language)) {
                tokenizers.add(ref);
            }
        }
    }
    if (tokenizers.size() > 1) {
        Collections.sort(tokenizers);
    }
    this.langTokenizers.put(language, tokenizers);
    return tokenizers;
}