view src/de/mpiwg/anteater/persons/PersonFinderController.java @ 0:036535fcd179

anteater
author jdamerow
date Fri, 14 Sep 2012 10:30:43 +0200
parents
children
line wrap: on
line source

package de.mpiwg.anteater.persons;

import java.io.File;
import java.util.ArrayList;
import java.util.List;

import de.mpiwg.anteater.AnteaterConfiguration;
import de.mpiwg.anteater.persons.impl.StanfordNLPPersonFinder;
import de.mpiwg.anteater.text.TextInformation;
import de.mpiwg.anteater.text.TextPart;
import de.mpiwg.anteater.text.TextType;
import de.mpiwg.anteater.xml.IPersonFinderParser;
import de.mpiwg.anteater.xml.impl.AnalysisXMLManager;
import de.mpiwg.anteater.xml.impl.StanfordNERXMLParser;

public class PersonFinderController {
	
	public final static String COMPONENT_NAME = PersonFinderController.class.getSimpleName();
	
	private AnteaterConfiguration configuration;

	public PersonFinderController(AnteaterConfiguration configuration) {
		this.configuration = configuration;
	}
	
	public List<PersonsExtraction> findPersonsInXML(TextInformation info) {
		List<PersonsExtraction> results = new ArrayList<PersonsExtraction>();
		List<String> summaryAnalysisResults = new ArrayList<String>();
		List<String> supplinfAnalysisResults = new ArrayList<String>();
		
		// check if there are already stored results
		AnalysisXMLManager analysisManager = null;
		if (configuration.getAnalysisPath() != null && !configuration.getAnalysisPath().isEmpty()) {
			File file = new File(info.getFilepath());
			
			analysisManager = new AnalysisXMLManager(configuration.getAnalysisPath() + File.separator +  file.getName());
			
			configuration.getLogger().logMessageWithoutNewLine(COMPONENT_NAME, "Check analysis file for persons in summaries...");
			summaryAnalysisResults = analysisManager.getSummaryPersonsResult();
			configuration.getLogger().logMessage("found " + summaryAnalysisResults.size() + " result(s).");
		
			configuration.getLogger().logMessageWithoutNewLine(COMPONENT_NAME, "Check analysis file for persons in supplementary information...");
			supplinfAnalysisResults = analysisManager.getSupplementaryInfoPersonsResult();
			configuration.getLogger().logMessage("found " + supplinfAnalysisResults.size() + " result(s).");
		}
		
		IPersonFinder personFinder = new StanfordNLPPersonFinder(configuration.getLogger());
		
		// if there are no results for summaries, ask place finding service.
		if (summaryAnalysisResults.size() == 0) {
			configuration.getLogger().logMessage(COMPONENT_NAME, "No results found for summaries, so will ask Stanford NLP NER Parser.");
						
			for (TextPart sum : info.getSummaries()) {
				String sumResult = personFinder.findPersons(sum.getText());
				if (sumResult != null) {
					summaryAnalysisResults.add(sumResult);
					
					// if there is an analysis folder, add result to analysis file
					if (analysisManager != null)
						analysisManager.addSummaryPersonsResult(sumResult);
				}
			}
		}
		
		// if there are no results for supplementary information, ask GNRD name fining service
		if (supplinfAnalysisResults.size() == 0) {
			configuration.getLogger().logMessage(COMPONENT_NAME, "No results found for supplementary information, so will ask Stanford NLP NER Parser.");
			
			for (TextPart sInf : info.getSupplInfos()) {
				String supinfResult = personFinder.findPersons(sInf.getText());
				if (supinfResult != null) {
					supplinfAnalysisResults.add(supinfResult);
					
					// if there is an analysis folder, add result to analysis file
					if (analysisManager != null)
						analysisManager.addSupplInfPersonsResult(supinfResult);
				}
			}
		}
		
		// create objects
		configuration.getLogger().logMessage(COMPONENT_NAME, "Creating analysis results...");
		int idx = 0;
		for (String summaryResult : summaryAnalysisResults) {
			IPersonFinderParser parser = new StanfordNERXMLParser(summaryResult);
			
			List<APerson> persons = parser.parsePersons();
			PersonsExtraction result = new PersonsExtraction();
			result.setType(TextType.TYPE_SUMMARY);
			result.setPerson(persons);
			result.setTextIdx(idx);
			
			results.add(result);
			idx++;
		}
		
		idx = 0;
		for (String suplinfResult : supplinfAnalysisResults) {
			IPersonFinderParser parser = new StanfordNERXMLParser(suplinfResult);
			
			List<APerson> persons = parser.parsePersons();
			PersonsExtraction result = new PersonsExtraction();
			result.setType(TextType.TYPE_SUPLINF);
			result.setPerson(persons);
			result.setTextIdx(idx);
			
			results.add(result);
			idx++;
		}
		
		return results;
	}
}