dnet-applications/apps/dnet-is-application/src/main/java/eu/dnetlib/is/importer/OldProfilesImporter.java

package eu.dnetlib.is.importer;

import java.io.StringReader;
import java.util.Date;

import javax.transaction.Transactional;

import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.Node;
import org.dom4j.io.SAXReader;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.http.MediaType;
import org.springframework.stereotype.Component;

import eu.dnetlib.is.resource.model.SimpleResource;
import eu.dnetlib.is.resource.repository.SimpleResourceRepository;
import eu.dnetlib.is.vocabulary.model.Synonym;
import eu.dnetlib.is.vocabulary.model.Vocabulary;
import eu.dnetlib.is.vocabulary.model.VocabularyTerm;
import eu.dnetlib.is.vocabulary.repository.VocabularyRepository;
import eu.dnetlib.is.vocabulary.repository.VocabularyTermRepository;

@Component
public class OldProfilesImporter {

	@Autowired
	private SimpleResourceRepository simpleResourceRepository;

	@Autowired
	private VocabularyRepository vocabularyRepository;

	@Autowired
	private VocabularyTermRepository vocabularyTermRepository;

	@Transactional
	public SimpleResource importSimpleResource(final String xml) throws Exception {
		final SAXReader reader = new SAXReader();
		final Document doc = reader.read(new StringReader(xml));

		final String id = StringUtils.substringBefore(doc.valueOf("//RESOURCE_IDENTIFIER/@value"), "_");
		final Date now = new Date();

		final SimpleResource res = new SimpleResource();
		res.setId(id);
		res.setCreationDate(now);
		res.setModificationDate(now);
		res.setContentType(MediaType.APPLICATION_XML_VALUE);

		String resContent;
		switch (doc.valueOf("//RESOURCE_TYPE/@value")) {
		case "CleanerDSResourceType":
			res.setType("cleaning_rule");
			res.setName(doc.valueOf("//CLEANER_NAME"));
			res.setDescription(doc.valueOf("//CLEANER_DESCRIPTION"));
			resContent = doc.selectSingleNode("//CLEANER_RULES").asXML();
			break;
		case "TransformationRuleDSResourceType":
			res.setType("transformation_rule");
			res.setName(doc.valueOf("//SCRIPT/TITLE"));
			res.setDescription("");
			resContent = doc.selectSingleNode("//SCRIPT/CODE").asXML();
			break;
		case "HadoopJobConfigurationDSResourceType":
			res.setType("hadoop_job_configuration");
			res.setName(doc.valueOf("//HADOOP_JOB/@name"));
			res.setDescription(doc.valueOf("//HADOOP_JOB/DESCRIPTION"));
			resContent = doc.selectSingleNode("//HADOOP_JOB").asXML();
			break;
		default:
			throw new Exception("Invalid resource type: " + doc.valueOf("//RESOURCE_TYPE/@value"));
		}

		simpleResourceRepository.save(res);
		simpleResourceRepository.setContentById(id, resContent);

		return res;
	}

	@Transactional
	public Vocabulary importVocabulary(final String xml) throws Exception {
		final SAXReader reader = new SAXReader();
		final Document doc = reader.read(new StringReader(xml));

		final Vocabulary voc = new Vocabulary();
		final String vocId = doc.valueOf("//VOCABULARY_NAME/@code");
		final String vocName = doc.valueOf("//VOCABULARY_NAME");
		final String vocDesc = doc.valueOf("//VOCABULARY_DESCRIPTION");

		voc.setId(vocId);
		voc.setName(vocName);
		voc.setDescription(vocDesc);

		vocabularyRepository.save(voc);

		for (final Node n : doc.selectNodes("//TERM")) {
			final VocabularyTerm term = new VocabularyTerm();
			term.setVocabulary(vocId);
			term.setCode(n.valueOf("@code"));
			term.setName(n.valueOf("@english_name"));
			term.setEncoding(n.valueOf("@encoding"));
			term.setSynonyms(n.selectNodes(".//SYNONYM")
				.stream()
				.map(ns -> new Synonym(ns.valueOf("@term"), ns.valueOf("@encoding")))
				.sorted()
				.distinct()
				.toArray(Synonym[]::new));

			vocabularyTermRepository.save(term);
		}

		return voc;
	}

}