D-Net

package eu.dnetlib.data.collector.plugins.oai;

import java.util.ArrayList;

import java.util.Iterator;

import java.util.List;

import com.google.common.base.Splitter;

import com.google.common.collect.Iterators;

import com.google.common.collect.Lists;

import eu.dnetlib.rmi.data.CollectorServiceException;

import eu.dnetlib.rmi.data.InterfaceDescriptor;

import eu.dnetlib.rmi.data.plugin.AbstractCollectorPlugin;

import org.springframework.beans.factory.annotation.Required;

public class OaiCollectorPlugin extends AbstractCollectorPlugin {

	private static final String FORMAT_PARAM = "format";

	private static final String OAI_SET_PARAM = "set";

	private OaiIteratorFactory oaiIteratorFactory;

	@Override

	public Iterable<String> collect(final InterfaceDescriptor interfaceDescriptor, final String fromDate, final String untilDate)

			throws CollectorServiceException {

		final String baseUrl = interfaceDescriptor.getBaseUrl();

		final String mdFormat = interfaceDescriptor.getParams().get(FORMAT_PARAM);

		final String setParam = interfaceDescriptor.getParams().get(OAI_SET_PARAM);

		final List<String> sets = new ArrayList<>();

		if (setParam != null) {

			sets.addAll(Lists.newArrayList(Splitter.on(",").omitEmptyStrings().trimResults().split(setParam)));

		}

		if (sets.isEmpty()) {

			// If no set is defined, ALL the sets must be harvested

			sets.add("");

		}

		if (baseUrl == null || baseUrl.isEmpty()) { throw new CollectorServiceException("Param 'baseurl' is null or empty"); }

		if (mdFormat == null || mdFormat.isEmpty()) { throw new CollectorServiceException("Param 'mdFormat' is null or empty"); }

        if (fromDate != null && !fromDate.matches("\\d{4}-\\d{2}-\\d{2}(T\\d{2}:\\d{2}:\\d{2}Z)?+")) {

            throw new CollectorServiceException("Invalid date (YYYY-MM-DD): " + fromDate);

        }

        if (untilDate != null && !untilDate.matches("\\d{4}-\\d{2}-\\d{2}(T\\d{2}:\\d{2}:\\d{2}Z)?+")) {

            throw new CollectorServiceException("Invalid date (YYYY-MM-DD): " + untilDate);

        }

		return () -> {

			final Iterator<Iterator<String>> iterator =

					sets.stream().map(set -> oaiIteratorFactory.newIterator(baseUrl, mdFormat, set, fromDate, untilDate)).iterator();

			return Iterators.concat(iterator);

		};

	}

	public OaiIteratorFactory getOaiIteratorFactory() {

		return oaiIteratorFactory;

	}

	@Required

	public void setOaiIteratorFactory(final OaiIteratorFactory oaiIteratorFactory) {

		this.oaiIteratorFactory = oaiIteratorFactory;

	}

}