D-Net

package eu.dnetlib.msro.workflows.nodes.objectstore;

import java.io.File;

import java.util.regex.Matcher;

import java.util.regex.Pattern;

import eu.dnetlib.miscutils.functional.xml.DnetXsltFunctions;

import eu.dnetlib.rmi.data.ObjectStoreFile;

import eu.dnetlib.rmi.data.Protocols;

public class ArxivIteratorRepository extends IteratorRepository {

	private final Pattern pattern;

	public ArxivIteratorRepository(final Iterable<File> input, final String repositoryPrefix, final String oaiPrefix) {

		super(input, repositoryPrefix, oaiPrefix);

		this.pattern = Pattern.compile("[0-9]");

	}

	@Override

	String generateNextElement() {

		try {

			final String inputname = this.input.next().getCanonicalPath();

			final String[] values = inputname.split("/");

			if (values.length > 1) {

				final ObjectStoreFile info = new ObjectStoreFile();

				String fileName = values[values.length - 1].replace(".pdf", "");

				final Matcher matcher = this.pattern.matcher(fileName);

				if (matcher.find()) {

					if (matcher.start() != 0) {

						fileName = fileName.substring(0, matcher.start()) + "/" + fileName.substring(matcher.start());

					}

				}

				info.setDownloadedURL("http://arxiv.org/abs/" + fileName);

				info.setAccessProtocol(Protocols.None);

				// String value = "oai:arXiv.org:" + fileName;

				final String value = this.oaiPrefix + fileName;

				final String resultID = this.repositoryPrefix + "::" + DnetXsltFunctions.md5(value);

				info.setObjectID(resultID + "::" + DnetXsltFunctions.md5(info.getDownloadedURL()));

				info.setMimeType("pdf");

				info.setURI(inputname);

				return info.toJSON();

			}

		} catch (final Exception e) {

			return null;

		}

		return null;

	}

}