D-Net

package eu.dnetlib.parthenos.catalogue;

import java.io.BufferedOutputStream;

import java.io.ByteArrayOutputStream;

import java.io.IOException;

import java.net.URISyntaxException;

import java.util.Iterator;

import com.fasterxml.jackson.core.JsonEncoding;

import com.fasterxml.jackson.core.JsonFactory;

import com.fasterxml.jackson.core.JsonGenerator;

import com.google.common.base.Joiner;

import eu.dnetlib.parthenos.jrr.ParthenosRegistryResource;

import eu.dnetlib.parthenos.publisher.ParthenosPublisherException;

import eu.dnetlib.parthenos.rdf.ResourceReader;

import org.apache.commons.lang3.StringUtils;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

import org.apache.jena.rdf.model.Resource;

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

import org.springframework.stereotype.Component;

import static eu.dnetlib.parthenos.CRM.*;

/**

 * Created by Alessia Bardi on 21/11/2017.

 *

 * @author Alessia Bardi

 */

@Component

public class CatalogueRegistrator {

	private static final Log log = LogFactory.getLog(CatalogueRegistrator.class);

	private final String PARTHENOS_BASE_URL = "http://parthenos.d4science.org";

	private final String CATALOGUE_NAME_REFIX = "parthenos_";

	@Autowired

	private ResourceReader resourceReader;

	@Autowired

	private CatalogueAPIClient catalogueAPIClient;

	private String groupTemplate = "{"

			+ "  \"name\":\"%s\","

			+ "  \"id\":\"%s\","

			+ "  \"title\": \"%s\""

			//+ "  \"description\": \"%s\""

			+ "}";

	public String register(final Resource resource, final Resource type) throws IOException, ParthenosPublisherException, URISyntaxException {

		String resURI = resource.getURI();

		log.debug(String.format("Catalogue --> Processing resource : %s with type: %s", resURI, type.getLocalName()));

		//For the catalogue: Must be purely lowercase alphanumeric (ascii) characters and these symbols: -_

		//Replacing all non-alphanumeric characters with empty strings

		String resCatName = getNameForCatalogue(resURI.substring(resURI.lastIndexOf("/") + 1));

		ParthenosRegistryResource prr = catalogueAPIClient.getRegistered(resCatName);

		if(prr != null){

			//TODO: what to do if already registered?

			log.debug(resCatName+ " is already registered");

			return prr.getUuid();

		}

		else {

			//resource not yet registered

			ensureGroups(resource);

			String json;

			switch (type.getLocalName()) {

			case "E7_Activity":

				json = getJsonForActivity(resource, resCatName);

				break;

			case "E39_Actor":

				json = getJsonForActor(resource, resCatName);

				break;

			case "E70_Thing":

				json = getJsonForThing(resource, resCatName);

				break;

			case "E29_Design_or_Procedure":

				json = getJsonForDesignProcedure(resource, resCatName);

				break;

			default:

				throw new IllegalArgumentException(String.format("Type " + type.getLocalName() + " not supported"));

			}

			String uuid = catalogueAPIClient.doRegister(json, resCatName);

			if(StringUtils.isBlank(uuid)){

				log.warn(String.format("%s could not be registered", resURI));

			}

			else {

				log.debug(String.format("%s registered on the catalogue with uuid: %s", resURI, uuid));

			}

			return uuid;

		}

	}

	protected boolean purge(final String resCatName) throws URISyntaxException, IOException {

		return catalogueAPIClient.purgeItem(resCatName);

	}

	/**

	 * The catalogue accepts names which are strings between 2 and 100 characters long, containing only lowercase alphanumeric characters, - and _

	 * @param name the original name

	 * @return a string which is name adapted for the catalogue

	 */

	protected String getNameForCatalogue(final String name){

		String n = name.replaceAll("[^A-Za-z0-9]","_").toLowerCase();

		if(n.length() > 100){

			n = n.substring(0,99);

		}

		if(n.length() < 2){

			n = CATALOGUE_NAME_REFIX+n;

		}

		return n;

	}

	/**

	 * Ensure that providers referred in the Resource are available as "groups" in the registry.

	 * @param res Resource

	 */

	protected void ensureGroups(final Resource res) throws ParthenosPublisherException, IOException, URISyntaxException {

		log.debug("Ensuring groups exist");

		Iterator<String> providerNames = resourceReader.getProviderNames(res);

		while(providerNames.hasNext()){

			String name = providerNames.next();

			if(StringUtils.isNotBlank(name)){

				String groupName = getNameForCatalogue(name);

				if(!catalogueAPIClient.groupExist(groupName)){

					String groupJson = String.format(groupTemplate, groupName, groupName, name);

					catalogueAPIClient.registerGroup(groupJson, groupName);

					log.info("NEW GROUP REGISTERED: "+groupName);

				}

			}

		}

	}

	protected String getJsonForActivity(final Resource res, final String resNameForCatalogue) throws IOException {

		JsonFactory jsonFactory = new JsonFactory();

		final ByteArrayOutputStream out = new ByteArrayOutputStream();

		BufferedOutputStream bos = new BufferedOutputStream(out);

		JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);

		jg.writeStartObject();

		writeCommonFields(jg, res, resNameForCatalogue);

		//TODO: from which linked E39_Actor?

		//for project we can take it from the maintainer team: is this correct also for other Things (like Services and Software?)

		jg.writeStringField("maintainer", Joiner.on(", ").join(resourceReader.getMaintainerUrls(res)));

		//TODO: it should be better to identify email contacts rather than generic contact labels of maintainer

		//jg.writeStringField("maintainer_email", Joiner.on(", ").join(resourceReader.getMaintainerContacts(res)));

		jg.writeArrayFieldStart("extras");

		addExtra(jg, "system:type", E7_Activity.getLocalName());

		//specific class

		addExtra(jg, "instance of", resourceReader.findSpecificType(res, E7_Activity).getLocalName());

		if (res.getURI().startsWith(PARTHENOS_BASE_URL))

			addExtra(jg, "Parthenos URL", res.getURI());

		//for Services

		addExtra(jg, "competence", resourceReader.getCompetence(res));

		addExtra(jg, "provided by", Joiner.on(", ").join(resourceReader.getProviderNames(res)));

		addExtra(jg, "declared begin/end of operation", Joiner.on(", ").join(resourceReader.getDeclarativeTimes(res)));

		//TODO: where to get it?

		addExtra(jg, "last confirmation", "");

		//TODO: where to get it?

		addExtra(jg, "date of registration", "");

		addExtra(jg, "availability", resourceReader.getAvailability(res));

		//TODO: where to get it?

		addExtra(jg, "condition of use", resourceReader.getConditionOfUse(res));

		addExtra(jg, "contact points", Joiner.on(", ").join(resourceReader.getProviderContactPoints(res)));

		addExtra(jg, "activity type", Joiner.on(", ").join(resourceReader.getActivityTypes(res)));

		addExtra(jg, "hosts", Joiner.on(", ").join(resourceReader.getHostedStuff(res)));

		addExtra(jg, "online access point", Joiner.on(", ").join(resourceReader.getAccessPoints(res)));

		addExtra(jg, "protocol", Joiner.on(", ").join(resourceReader.getProtocols(res)));

		addExtra(jg,"curates", Joiner.on(", ").join(resourceReader.getCuratedObjects(res)));

		//TODO: where to get it?

		addExtra(jg, "runs on request", "");

		addExtra(jg, "delivers on request", Joiner.on(", ").join(resourceReader.getDeliversOnRequest(res)));

		addExtra(jg, "uses curation plan", Joiner.on(", ").join(resourceReader.getCurationPlans(res)));

		//for Projects

		addExtra(jg, "offers", Joiner.on(", ").join(resourceReader.getOfferedServiceUrls(res)));

		addExtra(jg, "started", Joiner.on(", ").join(resourceReader.getStartTimes(res)));

		jg.writeEndArray(); //end extras

		jg.writeEndObject();

		jg.close();

		return out.toString("UTF-8");

	}

	protected String getJsonForActor(final Resource res, final String resNameForCatalogue) throws IOException {

		JsonFactory jsonFactory = new JsonFactory();

		final ByteArrayOutputStream out = new ByteArrayOutputStream();

		BufferedOutputStream bos = new BufferedOutputStream(out);

		JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);

		jg.writeStartObject();

		writeCommonFields(jg, res, resNameForCatalogue);

		jg.writeArrayFieldStart("extras");

		addExtra(jg, "system:type", E39_Actor.getLocalName());

		//specific class

		addExtra(jg, "instance of", resourceReader.findSpecificType(res, E39_Actor).getLocalName());

		if (res.getURI().startsWith(PARTHENOS_BASE_URL))

			addExtra(jg, "Parthenos URL", res.getURI());

		addExtra(jg, "has member", Joiner.on(", ").join(resourceReader.getMemberUrls(res)));

		addExtra(jg, "is member of", Joiner.on(", ").join(resourceReader.isMemberOf(res)));

		addExtra(jg, "has contact point", Joiner.on(", ").join(resourceReader.getResourceDirectContactPoints(res)));

		addExtra(jg, "provides", Joiner.on(", ").join(resourceReader.getProvidedServiceUrls(res)));

		jg.writeEndArray();

		jg.writeEndObject();

		jg.close();

		return out.toString("UTF-8");

	}

	protected String getJsonForThing(final Resource res, final String resNameForCatalogue) throws IOException {

		JsonFactory jsonFactory = new JsonFactory();

		final ByteArrayOutputStream out = new ByteArrayOutputStream();

		BufferedOutputStream bos = new BufferedOutputStream(out);

		JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);

		jg.writeStartObject();

		writeCommonFields(jg, res, resNameForCatalogue);

		jg.writeArrayFieldStart("extras");

		addExtra(jg, "system:type", E70_Thing.getLocalName());

		//specific class

		addExtra(jg, "instance of", resourceReader.findSpecificType(res, E70_Thing).getLocalName());

		if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {

			addExtra(jg, "Parthenos URL", res.getURI());

		}

		//TODO: things include digital objects, software, datasets, schema. Guess we should know what to add here.

		addExtra(jg, "is part of", Joiner.on(", ").join(resourceReader.getIsPartOfUrls(res)));

		addExtra(jg, "has part", Joiner.on(", ").join(resourceReader.getHasPartUrls(res)));

		addExtra(jg, "curated by", Joiner.on(", ").join(resourceReader.getCuratorUrls(res)));

		addExtra(jg, "curation plan", Joiner.on(", ").join(resourceReader.getCurationPlans(res)));

		addExtra(jg, "hosted by", Joiner.on(", ").join(resourceReader.getHostedBys(res)));

		//TODO where to get the encoding types? Should we through the Creation event?

		//addExtra(jg, "encoding type", Joiner.on(", ").join());

		//TODO where to get the schema/formats? Should we through the Creation event?

		//addExtra(jg, "encoding type", Joiner.on(", ").join());

		//TODO where to get the creator? Should we through the Creation event?

		//addExtra(jg, "creator", Joiner.on(", ").join());

		addExtra(jg, "subject", Joiner.on(", ").join(resourceReader.getSubjects(res)));

		addExtra(jg, "temporal coverage", Joiner.on(", ").join(resourceReader.getTemporalCoverages(res)));

		addExtra(jg, "spatial coverage", Joiner.on(", ").join(resourceReader.getSpatialCoverages(res)));

		jg.writeEndArray();

		jg.writeEndObject();

		jg.close();

		return out.toString("UTF-8");

	}

	protected String getJsonForDesignProcedure(final Resource res, final String resNameForCatalogue) throws IOException {

		JsonFactory jsonFactory = new JsonFactory();

		final ByteArrayOutputStream out = new ByteArrayOutputStream();

		BufferedOutputStream bos = new BufferedOutputStream(out);

		JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);

		jg.writeStartObject();

		writeCommonFields(jg, res, resNameForCatalogue);

		jg.writeArrayFieldStart("extras");

		addExtra(jg, "system:type", E29_Design_or_Procedure.getLocalName());

		//specific class

		addExtra(jg, "instance of", resourceReader.findSpecificType(res, E29_Design_or_Procedure).getLocalName());

		if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {

			addExtra(jg, "Parthenos URL", res.getURI());

		}

		//TODO: add additional metadata for E29_Design_or_Procedure, if any

		jg.writeEndArray();

		jg.writeEndObject();

		jg.close();

		return out.toString("UTF-8");

	}

	protected void addExtra(final JsonGenerator jg, final String key, final String value) throws IOException {

		jg.writeStartObject();

		jg.writeStringField("key", key);

		jg.writeStringField("value", value);

		jg.writeEndObject();

	}

	protected void writeCommonFields(final JsonGenerator jg, final Resource res, final String resNameForCatalogue) throws IOException {

		//end of URI

		jg.writeStringField("name", resNameForCatalogue);

		//default license

		jg.writeStringField("license_id", "notspecified");

		String title = resourceReader.getTitle(res);

		if (StringUtils.isBlank(title))

			title = resNameForCatalogue;

		jg.writeStringField("title", title);

		//description

		jg.writeStringField("notes", resourceReader.getDescription(res));

		//the names of all superclasses of the entity

		jg.writeArrayFieldStart("tags");

		Iterator<String> classNames = resourceReader.getRDFClassNames(res);

		while (classNames.hasNext()) {

			jg.writeStartObject();

			jg.writeStringField("name", classNames.next());

			jg.writeEndObject();

		}

		jg.writeEndArray();

		//RI from which the entity has been collected, the source from which the RI collected the entity (if available).

		//TODO: other Actors to add as catalogue group?

		Iterator<String> providers = resourceReader.getProviderNames(res);

		jg.writeArrayFieldStart("groups");

		while(providers.hasNext()){

			String provider = providers.next();

			jg.writeStartObject();

			jg.writeStringField("name", provider);

			jg.writeEndObject();

		}

		jg.writeEndArray();

	}

	public ResourceReader getResourceReader() {

		return resourceReader;

	}

	public void setResourceReader(final ResourceReader resourceReader) {

		this.resourceReader = resourceReader;

	}

	public CatalogueAPIClient getCatalogueAPIClient() {

		return catalogueAPIClient;

	}

	public void setCatalogueAPIClient(final CatalogueAPIClient catalogueAPIClient) {

		this.catalogueAPIClient = catalogueAPIClient;

	}

}