Mercurial > hg > anteater
view src/de/mpiwg/anteater/species/common/CommonNameFindController.java @ 3:ae96e4bc7fb2
save found species to analysis files
author | jdamerow |
---|---|
date | Mon, 22 Oct 2012 14:21:14 -0700 |
parents | 1c2b4f5e2c05 |
children | dcc35f89dce3 |
line wrap: on
line source
package de.mpiwg.anteater.species.common; import java.io.File; import java.util.ArrayList; import java.util.List; import de.mpiwg.anteater.AnteaterConfiguration; import de.mpiwg.anteater.species.common.impl.LinnaeusNameFinder; import de.mpiwg.anteater.species.scientific.IScientificNamesFinder; import de.mpiwg.anteater.species.scientific.ScientificNamesExtraction; import de.mpiwg.anteater.species.scientific.impl.GNRDNameFinder; import de.mpiwg.anteater.text.TextInformation; import de.mpiwg.anteater.text.TextPart; import de.mpiwg.anteater.xml.impl.AnalysisXMLManager; public class CommonNameFindController { public final static String COMPONENT_NAME = CommonNameFindController.class.getSimpleName(); private AnteaterConfiguration configuration; public CommonNameFindController(AnteaterConfiguration configuration) { this.configuration = configuration; } public List<CommonNamesExtraction> findCommonNamesInXML(TextInformation info) { List<CommonNamesExtraction> results = new ArrayList<CommonNamesExtraction>(); 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 scientific names in summaries..."); summaryAnalysisResults = analysisManager.getSummaryCommonNamesResults(); configuration.getLogger().logMessage("found " + summaryAnalysisResults.size() + " result(s)."); configuration.getLogger().logMessageWithoutNewLine(COMPONENT_NAME, "Check analysis file for scientific names in supplementary information..."); supplinfAnalysisResults = analysisManager.getSupplementaryInfoCommonNamesResults(); configuration.getLogger().logMessage("found " + supplinfAnalysisResults.size() + " result(s)."); } ICommonNameFinder nameFinder = new LinnaeusNameFinder(configuration.getLogger()); // if there are no results for summaries, ask GNRD name finding service. if (summaryAnalysisResults.size() == 0) { configuration.getLogger().logMessage(COMPONENT_NAME, "No results found for summaries, so will ask LinnaeusNameFinder."); for (TextPart sum : info.getSummaries()) { String sumResult = nameFinder.findCommonNames(sum.getText()); if (sumResult != null) { summaryAnalysisResults.add(sumResult); // if there is an analysis folder, add result to analysis file if (analysisManager != null) analysisManager.addSummaryCommonNamesResult(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 LinnaeusNameFinder."); for (TextPart sInf : info.getSupplInfos()) { String supinfResult = nameFinder.findCommonNames(sInf.getText()); if (supinfResult != null) { supplinfAnalysisResults.add(supinfResult); // if there is an analysis folder, add result to analysis file if (analysisManager != null) analysisManager.addSupplInfCommonNamesResult(supinfResult); } } } return null; } }