private void addJCas(List<JCas> jCases, User user, SourceDocument sourceDocument) {
try {
// Check if there is an annotation document entry in the database. If there is none,
// create one.
AnnotationDocument annotationDocument =
repository.createOrGetAnnotationDocument(sourceDocument, user);
// Read the CAS
JCas jcas = repository.readAnnotationCas(annotationDocument);
// Update the annotation document CAS
repository.upgradeCas(jcas.getCas(), annotationDocument);
// After creating an new CAS or upgrading the CAS, we need to save it
repository.writeAnnotationCas(
jcas.getCas().getJCas(), annotationDocument.getDocument(), user);
jCases.add(jcas);
} catch (UIMAException e) {
error(e.getMessage() + " : " + ExceptionUtils.getRootCauseMessage(e));
} catch (IOException e) {
error(e.getMessage());
}
}
public static void main(String[] arguments) {
final CommandLineParser parser = new PosixParser();
final Options opts = new Options();
CommandLine cmd = null;
Pipeline.addLogHelpAndInputOptions(opts);
Pipeline.addTikaOptions(opts);
Pipeline.addOutputOptions(opts);
Pipeline.addJdbcResourceOptions(
opts, DEFAULT_JDBC_DRIVER, DEFAULT_DB_PROVIDER, DEFAULT_DATABASE);
// entity annotator options setup
opts.addOption("Q", "query-file", true, "file with SQL SELECT queries");
OptionBuilder.withLongOpt("query");
OptionBuilder.withArgName("SELECT");
OptionBuilder.hasArgs();
OptionBuilder.withDescription("one or more SQL SELECT queries");
opts.addOption(OptionBuilder.create('q'));
opts.addOption(
"m", "entity-map", true, "name of the entity map file [" + DEFAULT_MAPPING_FILE + "]");
opts.addOption(
"n", "namespace", true, "namespace of the entity annotations [" + DEFAULT_NAMESPACE + "]");
try {
cmd = parser.parse(opts, arguments);
} catch (final ParseException e) {
System.err.println(e.getLocalizedMessage());
System.exit(1); // == exit ==
}
final Logger l =
Pipeline.loggingSetup(cmd, opts, "txtfnnl entities [options] <directory|files...>\n");
// output options
XmiWriter.Builder writer =
Pipeline.configureWriter(cmd, XmiWriter.configure(Pipeline.ensureOutputDirectory(cmd)));
// DB resource
ExternalResourceDescription jdbcResource = null;
try {
jdbcResource =
Pipeline.getJdbcConnectionResource(
cmd, l, DEFAULT_JDBC_DRIVER, DEFAULT_DB_PROVIDER, DEFAULT_DATABASE);
} catch (final ClassNotFoundException e) {
System.err.println("JDBC resoruce setup failed:");
System.err.println(e.toString());
System.exit(1); // == EXIT ==
} catch (ResourceInitializationException e) {
System.err.println("JDBC resoruce setup failed:");
System.err.println(e.toString());
System.exit(1); // == EXIT ==
}
/* BEGIN entity annotator */
final String queryFileName = cmd.getOptionValue('Q');
final String entityMapPath = cmd.getOptionValue('m', DEFAULT_MAPPING_FILE);
final String namespace = cmd.getOptionValue('n', DEFAULT_NAMESPACE);
String[] queries = cmd.getOptionValues('q');
File entityMap; // m
if (queryFileName != null) {
final File queryFile = new File(queryFileName);
if (!queryFile.isFile() || !queryFile.canRead()) {
System.err.print("cannot read query file ");
System.err.println(queryFile);
System.exit(1); // == EXIT ==
}
String[] fileQueries = null;
try {
fileQueries =
IOUtils.read(new FileInputStream(queryFile), Pipeline.inputEncoding(cmd)).split("\n");
} catch (final Exception e) {
System.err.print("cannot read query file ");
System.err.print(queryFile);
System.err.print(":");
System.err.println(e.getLocalizedMessage());
System.exit(1); // == EXIT ==
}
if (queries == null || queries.length == 0) {
queries = fileQueries;
} else {
final String[] tmp = new String[queries.length + fileQueries.length];
System.arraycopy(queries, 0, tmp, 0, queries.length);
System.arraycopy(fileQueries, 0, tmp, queries.length, fileQueries.length);
queries = tmp;
}
}
entityMap = new File(entityMapPath);
if (!entityMap.isFile() || !entityMap.canRead()) {
System.err.print("cannot read entity map file ");
System.err.println(entityMapPath);
System.exit(1); // == EXIT ==
}
if (queries == null || queries.length == 0) {
queries = DEFAULT_SQL_QUERIES;
}
/* END entity annotator */
try {
final Pipeline pipeline = new Pipeline(2); // tika and known entity annotator
KnownEntityAnnotator.Builder builder =
KnownEntityAnnotator.configure(namespace, queries, entityMap, jdbcResource);
pipeline.setReader(cmd);
pipeline.configureTika(cmd);
pipeline.set(1, Pipeline.multiviewEngine(builder.create()));
pipeline.setConsumer(Pipeline.textEngine(writer.create()));
pipeline.run();
pipeline.destroy();
} catch (final UIMAException e) {
l.severe(e.toString());
System.err.println(e.getLocalizedMessage());
System.exit(1); // == EXIT ==
} catch (final IOException e) {
l.severe(e.toString());
System.err.println(e.getLocalizedMessage());
System.exit(1); // == EXIT ==
}
System.exit(0);
}