D-Net

package eu.dnetlib.msro.openaireplus.workflows.nodes.contexts;

import java.io.IOException;

import java.util.*;

import com.google.common.collect.Maps;

import com.ximpleware.AutoPilot;

import com.ximpleware.VTDException;

import com.ximpleware.VTDGen;

import com.ximpleware.VTDNav;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

import org.dom4j.Document;

import org.dom4j.DocumentHelper;

import org.dom4j.Element;

import org.joda.time.DateTime;

import static eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions.md5;

public class BuildH2020FETTaxonomy {

	private static final Log log = LogFactory.getLog(BuildH2020FETTaxonomy.class);

	private Iterator<String> projects;

	private Map<String, Map<String, Map<String, List<Info>>>> dic;

	public BuildH2020FETTaxonomy() {

		dic = Maps.newHashMap();

	}

	public void setIterator(Iterator<String> projects) {

		this.projects = projects;

	}

	public String parseProjects() throws Exception {

		while (projects.hasNext())

			parse(projects.next());

		return getTaxonomy();

	}

	public String getTaxonomy() throws Exception {

		Document taxonomy = DocumentHelper.createDocument();

		Element root1 = taxonomy.addElement("RESOURCE_PROFILE");

		createHeader(root1);

		Element body = root1.addElement("BODY");

		Element configurationElement = body.addElement("CONFIGURATION");

		createContext(configurationElement);

		body.addElement("STATUS");

		body.addElement("SECURITY_PARAMETERS");

		return taxonomy.getDocument().asXML();

	}

	private void createContext(final Element configurationElement) {

		Element context = configurationElement.addElement("context");

		context.addAttribute("type", "community");

		context.addAttribute("id", "fet-h2020");

		context.addAttribute("label", "FET H2020");

		for (Map.Entry<String, Map<String, Map<String, List<Info>>>> entry : dic.entrySet()) {

			String type = entry.getKey();

			Element cat = context.addElement("category");

			String catid = "fet-h2020::" + type.toLowerCase();

			cat.addAttribute("id", catid);

			cat.addAttribute("label", "FET " + type);

			cat.addAttribute("claim", "false");

			//int count = 0;

			for (Map.Entry<String, Map<String, List<Info>>> entryst : dic.get(type).entrySet()) {

				String subtype = entryst.getKey();

				for (Map.Entry<String, List<Info>> entryp : dic.get(type).get(subtype).entrySet()) {

					String opt = entryp.getKey();

			//		count++;

					Element subcat = cat.addElement("concept");

					String proj_md5 = md5(opt).substring(0,12);

					//String subcatid = catid + "::" + count;

					String subcatid = catid + "::" + proj_md5;

					subcat.addAttribute("id", subcatid);

					Element pa = subcat.addElement("param");

					pa.addAttribute("name", "CALL_ID");

					List<Info> projectRows = dic.get(type).get(subtype).get(opt);

					if (!projectRows.isEmpty()) {

						for (Info row : projectRows) {

							createProjectConcept(row, subcat, subcatid);

						}

						subcat.addAttribute("label", projectRows.get(0).getCallname());

						subcat.addAttribute("claim", "false");

						pa.setText(projectRows.get(0).getCallID());

					} else throw new RuntimeException("The lis of projects for " + subcatid + " is empty. This exception should never be thrown.");

				}

			}

		}

	}

	private void createProjectConcept(Info row, Element father, String prefix) {

		String code = row.getCode();

		String acronym = row.getAcronym();

		String title = row.getTitle();

		String projcallid = row.getProjcallid();

		Element concept = father.addElement("concept");

		concept.addAttribute("id", prefix + "::" + code);

		concept.addAttribute("label", title);

		concept.addAttribute("claim", "true");

		Element p = concept.addElement("param");

		p.addAttribute("name", "CD_PROJ_ID");

		p.setText(code);

		p = concept.addElement("param");

		p.addAttribute("name", "CD_CALL_ID");

		p.setText(projcallid);

		p = concept.addElement("param");

		p.addAttribute("name", "CD_PROJECT_NUMBER");

		p.setText(code);

		p = concept.addElement("param");

		p.addAttribute("name", "CD_ACRONYM");

		p.setText(acronym);

		p = concept.addElement("param");

		p.addAttribute("name", "CD_FRAMEWORK");

		p.setText("H2020");

		p = concept.addElement("param");

		p.addAttribute("name", "funder");

		p.setText("EC");

	}

	private void createHeader(final Element profile) {

		Element header = profile.addElement("HEADER");

		Element rs = header.addElement("RESOURCE_IDENTIFIER");

		rs.addAttribute("value", "");

		Element rt = header.addElement("RESOURCE_TYPE");

		rt.addAttribute("value", "ContextDSResourceType");

		Element rk = header.addElement("RESOURCE_KIND");

		rk.addAttribute("value", "ContextDSResources");

		Element ru = header.addElement("RESOURCE_URI");

		ru.addAttribute("value", "");

		Element daoc = header.addElement("DATE_OF_CREATION");

		daoc.addAttribute("value", DateTime.now().toString());

	}

	private boolean parse(String project) throws VTDException, IOException {

		boolean fet = false;

		String callid = "";

		String projectid = "";

		VTDGen vg = new VTDGen();

		vg.setDoc(project.getBytes("UTF-8"));

		vg.parse(false);

		VTDNav vn = vg.getNav();

		AutoPilot ap = new AutoPilot(vn);

		ap.selectXPath("//metadata/ROWS/ROW[@table=\"projects\"]");

		while (ap.evalXPath() != -1) {

			AutoPilot ap1 = new AutoPilot(vn);

			ap1.selectXPath("./FIELD[@name=\"call_identifier\"]");

			if (ap1.evalXPath() != -1)

				callid = vn.toNormalizedString(vn.getText());

			ap1.clearVariableExprs();

			vn.toElement(VTDNav.PARENT);

			ap1.selectXPath("./FIELD[@name=\"optional1\"]");

			if (ap1.evalXPath() != -1)

				projectid = vn.toNormalizedString(vn.getText());

			if (callid.contains("FET")) {

				String type;

				if (callid.contains("OPEN"))

					type = "OPEN";

				else if (callid.contains("PROACT"))

					type = "PROACT";

				else if (callid.contains("HPC"))

					type = "HPC";

				else

					type = "FLAG";

				vn.toElement(VTDNav.ROOT);

				fet = true;

				insert(type, callid, projectid, new Info(vn));

			}

		}

		return fet;

	}

	private void insert(String type, String callid, String projcallid, Info row) {

		Map<String, Map<String, List<Info>>> dopen;

		Map<String, List<Info>> entry;

		List<Info> projects;

		if (dic.containsKey(type)) {

			dopen = dic.get(type);

			if (dopen.containsKey(callid)) {

				entry = dopen.get(callid);

				if (entry.containsKey(projcallid)) {

					entry.get(projcallid).add(row);

				} else {

					projects = new ArrayList<>();

					projects.add(row);

					entry.put(projcallid, projects);

				}

			} else {

				projects = new ArrayList<>();

				entry = new HashMap<>();

				projects.add(row);

				entry.put(projcallid, projects);

				dopen.put(callid, entry);

			}

		} else {

			projects = new ArrayList<>();

			entry = new HashMap<>();

			dopen = new HashMap<>();

			projects.add(row);

			entry.put(projcallid, projects);

			dopen.put(callid, entry);

			dic.put(type, dopen);

		}

	}

	public void parseFETProject(int projects_number) throws VTDException, IOException {

		int parsed = 0;

		while (projects.hasNext() && parsed < projects_number) {

			if (parse(projects.next()))

				parsed++;

		}

		log.debug(parsed);

	}

	private class Info {

		VTDNav vn;

		private String code, acronym, title, call_identifier, projcallid, callname;

		public String getCode() {

			return code;

		}

		public String getAcronym() {

			return acronym;

		}

		public String getTitle() {

			return title;

		}

		public String getCallID() {

			return call_identifier;

		}

		public String getProjcallid() {

			return projcallid;

		}

		public String getCallname() {

			return callname;

		}

		public Info(VTDNav vn) {

			this.vn = vn;

			parse();

		}

		private void parse() {

			try {

				AutoPilot ap = new AutoPilot(vn);

				ap.selectXPath("//FIELD");

				while (ap.evalXPath() != -1) {

					String aname = vn.toNormalizedString(vn.getAttrVal("name"));

					String text = "";

					if (vn.getText() != -1)

						text = vn.toNormalizedString(vn.getText());

					if (aname.equals("code"))

						code = text;

					else if (aname.equals("acronym"))

						acronym = text;

					else if (aname.equals("title"))

						title = text;

					else if (aname.equals("call_identifier"))

						call_identifier = text;

					else if (aname.equals("optional1"))

						projcallid = text;

					else if (aname.equals("optional2"))

						callname = text;

				}

			} catch (Exception e) {

			}

		}

	}

}