/**
* Method that initializes all vital prerequisites, including POS Tagger
*
* @param language Language to be processed with this copy of HeidelTime
* @param typeToProcess Domain type to be processed
* @param outputType Output type
* @param configPath Path to the configuration file for HeidelTimeStandalone
* @param posTagger POS Tagger to use for preprocessing
* @param doIntervalTagging Whether or not to invoke the IntervalTagger
*/
public void initialize(
Language language,
DocumentType typeToProcess,
OutputType outputType,
String configPath,
POSTagger posTagger,
Boolean doIntervalTagging) {
logger.log(
Level.INFO, "HeidelTimeStandalone initialized with language " + this.language.getName());
// set the POS tagger
this.posTagger = posTagger;
// set doIntervalTagging flag
this.doIntervalTagging = doIntervalTagging;
// read in configuration in case it's not yet initialized
if (!Config.isInitialized()) {
if (configPath == null) readConfigFile(CLISwitch.CONFIGFILE.getValue().toString());
else readConfigFile(configPath);
}
try {
heidelTime = new HeidelTime();
heidelTime.initialize(
new UimaContextImpl(language, typeToProcess, CLISwitch.VERBOSITY2.getIsActive()));
logger.log(Level.INFO, "HeidelTime initialized");
} catch (Exception e) {
e.printStackTrace();
logger.log(Level.WARNING, "HeidelTime could not be initialized");
}
// Initialize JCas factory -------------
logger.log(Level.FINE, "Initializing JCas factory...");
try {
TypeSystemDescription[] descriptions =
new TypeSystemDescription[] {
UIMAFramework.getXMLParser()
.parseTypeSystemDescription(
new XMLInputSource(
this.getClass()
.getClassLoader()
.getResource(Config.get(Config.TYPESYSTEMHOME))))
};
jcasFactory = new JCasFactoryImpl(descriptions);
logger.log(Level.INFO, "JCas factory initialized");
} catch (Exception e) {
e.printStackTrace();
logger.log(Level.WARNING, "JCas factory could not be initialized");
}
}
/**
* Processes document with HeidelTime
*
* @param document
* @param documentCreationTime Date when document was created - especially important if document
* is of type {@link DocumentType#NEWS}
* @return Annotated document
* @throws DocumentCreationTimeMissingException If document creation time is missing when
* processing a document of type {@link DocumentType#NEWS}
*/
public String process(String document, Date documentCreationTime, ResultFormatter resultFormatter)
throws DocumentCreationTimeMissingException {
logger.log(Level.INFO, "Processing started");
// Generate jcas object ----------
logger.log(Level.FINE, "Generate CAS object");
JCas jcas = null;
try {
jcas = jcasFactory.createJCas();
jcas.setDocumentText(document);
logger.log(Level.FINE, "CAS object generated");
} catch (Exception e) {
e.printStackTrace();
logger.log(Level.WARNING, "Cas object could not be generated");
}
// Process jcas object -----------
try {
logger.log(Level.FINER, "Establishing preconditions...");
provideDocumentCreationTime(jcas, documentCreationTime);
establishHeidelTimePreconditions(jcas);
logger.log(Level.FINER, "Preconditions established");
heidelTime.process(jcas);
logger.log(Level.INFO, "Processing finished");
} catch (Exception e) {
e.printStackTrace();
logger.log(Level.WARNING, "Processing aborted due to errors");
}
// process interval tagging ---
if (doIntervalTagging) runIntervalTagger(jcas);
// Process results ---------------
logger.log(Level.FINE, "Formatting result...");
// PrintAnnotations.printAnnotations(jcas.getCas(), System.out);
String result = null;
try {
result = resultFormatter.format(jcas);
logger.log(Level.INFO, "Result formatted");
} catch (Exception e) {
e.printStackTrace();
logger.log(Level.WARNING, "Result could not be formatted");
}
return result;
}
