D-Net

package eu.dnetlib.grid.process;

import java.util.List;

import java.util.Map;

import java.util.Objects;

import java.util.stream.Collectors;

import org.springframework.boot.CommandLineRunner;

import org.springframework.boot.SpringApplication;

import org.springframework.boot.autoconfigure.SpringBootApplication;

import com.googlecode.protobuf.format.JsonFormat;

import eu.dnetlib.grid.process.model.GridOrganization;

import eu.dnetlib.grid.process.utils.Datasource;

import eu.dnetlib.grid.process.utils.GridCleaner;

import eu.dnetlib.grid.process.utils.GridUtils;

import eu.dnetlib.miscutils.functional.hash.Hashing;

@SpringBootApplication

public class GridOrganizationsApplication implements CommandLineRunner {

	private static final Datasource gridDs =

			new Datasource("10|openaire____::" + Hashing.md5("grid"), "GRID - Global Research Identifier Database", "grid________");

	public static void main(final String[] args) {

		SpringApplication.run(GridOrganizationsApplication.class, args);

	}

	@Override

	public void run(final String... args) throws Exception {

		if (args.length == 1) {

			processFile(args[0]);

		} else {

			System.err.println();

			System.err.println("Usage: java -jar <JARFILE> <GridJsonFile>");

			System.err.println();

		}

	}

	protected void processFile(final String jsonFile) throws Exception {

		final GridCleaner cleaner = new GridCleaner();

		final Map<String, String> parents = GridUtils.streamOrganizations(jsonFile)

				.filter(Objects::nonNull)

				.filter(o -> o.getStatus().equals("active"))

				.collect(Collectors.toMap(GridOrganization::getId, GridUtils::findShortName));

		final JsonFormat jsonFormat = new JsonFormat();

		GridUtils.streamOrganizations(jsonFile)

				.filter(Objects::nonNull)

				.filter(o -> o.getStatus().equals("active"))

				.map(cleaner)

				.map(org -> GridUtils.toProtos(org, parents, gridDs))

				.flatMap(List::stream)

				.map(jsonFormat::printToString)

				.forEach(System.out::println);

	}

}