/**
* Requests an annotation from the indicated annotation tool (running on the local node) for the
* indicated document record. Stores the result in this object for later output through the
* writeOutputFromLastAnnotate() method.
*
* @param record The document record on which we will run the annotation tool
* @param toolToRun The type of annotation that we should get for the record
* @postcondition lastAnnotatedRecord contains all of the record parameter's previous annotations,
* along with the new one
*/
public void annotateSingleDoc(Record record, AnnotationMode toolToRun)
throws ServiceUnavailableException, TException, AnnotationFailedException,
ServiceSecurityException {
// Ask the Curator to perform the annotation
lastAnnotatedRecord = annotate(record, toolToRun);
if (!RecordTools.hasAnnotation(lastAnnotatedRecord, toolToRun)) {
throw new AnnotationFailedException("Annotation was unsuccessful!");
}
}
/**
* Writes the results of the last call to annotate() to the specified directory in HDFS. This is
* equivalent to serializing the results of a call to annotate() to the directory.
*
* @param outputDir The directory to which the results of the last call to annotate() should be
* written. Each serialized document should be named with the document's hash.
*/
public void writeOutputFromLastAnnotate(Path outputDir) throws TException, IOException {
Path fileLoc = getLocForSerializedForm(lastAnnotatedRecord, outputDir);
try {
if (!transport.isOpen()) {
transport.open();
}
serializer.serialize(lastAnnotatedRecord, fileLoc, hdfs);
Record reconstructed = serializer.deserialize(fileLoc, hdfs);
if (!RecordTools.hasAnnotations(reconstructed)) {
throw new IOException(
"Reconstructed record has no annotations, but original has the following: "
+ RecordTools.getContents(lastAnnotatedRecord));
}
} finally {
if (transport.isOpen()) {
transport.close();
}
}
}
