package oboannotator.annotator;

import java.util.regex.Matcher;
import java.util.regex.Pattern;

import oboannotator.typesys.MedlineField;

import org.apache.uima.analysis_component.JCasAnnotator_ImplBase;
import org.apache.uima.jcas.JCas;


/**
 * Example annotator that detects room numbers using Java 1.4 regular
 * expressions.
 */
public class MedlineFieldAnnotator extends JCasAnnotator_ImplBase {
	private Pattern uiPattern = Pattern.compile("(^UI|PMID)\\s*\\-(.*?)^([A-Z][A-Z]+\\s*\\-)", Pattern.DOTALL | Pattern.MULTILINE);
	private Pattern tiPattern = Pattern.compile("(^TI)\\s*\\-(.*?)^([A-Z][A-Z]+\\s*\\-)", Pattern.DOTALL | Pattern.MULTILINE);
	private Pattern abPattern = Pattern.compile("(^AB)\\s*\\-(.*?)^([A-Z][A-Z]+\\s*\\-)", Pattern.DOTALL | Pattern.MULTILINE);

	
	private Pattern ontologyPattern = Pattern.compile("([A-Z]*?)_(\\d+)");
	
	
	
	/**
	 * @see JCasAnnotator_ImplBase#process(JCas)
	 */
	public void process(JCas aJCas) {
		// get document text
		String docText = aJCas.getDocumentText();
		// search for Yorktown room numbers

		
		Matcher ontologyMatcher = ontologyPattern.matcher(docText);
		while (ontologyMatcher.find()) {
			// found one - create annotation
			MedlineField annotation = new MedlineField(aJCas);
			annotation.setBegin(ontologyMatcher.start());
			annotation.setEnd(ontologyMatcher.end());
			annotation.setName(ontologyMatcher.group(1));
			annotation.setText(ontologyMatcher.group(2));
			//System.out.println("Name: " + ontologyMatcher.group(1) + "\n Text: "+ ontologyMatcher.group(2));
			annotation.addToIndexes();
		}			
		
		
		Matcher matcher = uiPattern.matcher(docText);
		while (matcher.find()) {
			// found one - create annotation
			MedlineField annotation = new MedlineField(aJCas);
			annotation.setBegin(matcher.start());
			annotation.setEnd(matcher.end());
			annotation.setName(matcher.group(1));
			annotation.setText(matcher.group(2));
			//System.out.println("Name: " + annotation.getName() + "\n Text: "+ annotation.getText());
			annotation.addToIndexes();
		}		
		
		
		Matcher matcher2 = abPattern.matcher(docText);
		while (matcher2.find()) {
			// found one - create annotation
			MedlineField annotation = new MedlineField(aJCas);
			annotation.setBegin(matcher2.start());
			annotation.setEnd(matcher2.end());
			annotation.setName(matcher2.group(1));
			annotation.setText(matcher2.group(2));
			//System.out.println("Name: [" + annotation.getName() + "]\n Text: <"
				//	+ annotation.getText() + ">") ;
			annotation.addToIndexes();
		}

		Matcher matcher3 = tiPattern.matcher(docText);
		while (matcher3.find()) {
			// found one - create annotation
			MedlineField annotation = new MedlineField(aJCas);
			annotation.setBegin(matcher3.start());
			annotation.setEnd(matcher3.end());
			annotation.setName(matcher3.group(1));
			annotation.setText(matcher3.group(2));
			//System.out.println("Name: " + annotation.getName() + "\n Text: "
			//		+ annotation.getText());
			annotation.addToIndexes();
		}
	}
}