D-Net

package eu.dnetlib.msro.workflows.dedup;

import java.util.List;

import java.util.Map;

import javax.annotation.Resource;

import org.apache.commons.lang.StringUtils;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

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

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

import com.google.common.collect.Iterables;

import com.google.gson.Gson;

import com.googlecode.sarasvati.Arc;

import com.googlecode.sarasvati.NodeToken;

import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;

import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;

import eu.dnetlib.enabling.locators.UniqueServiceLocator;

import eu.dnetlib.miscutils.datetime.DateUtils;

import eu.dnetlib.msro.workflows.dedup.conf.DedupConfigurationOrchestrationLoader;

import eu.dnetlib.msro.workflows.nodes.SimpleJobNode;

public class PrepareDedupIndexJobNode extends SimpleJobNode {

	/**

	 * logger.

	 */

	private static final Log log = LogFactory.getLog(PrepareDedupIndexJobNode.class); // NOPMD by marko on 11/24/08 5:02 PM

	@Resource

	private UniqueServiceLocator serviceLocator;

	@Autowired

	private DedupConfigurationOrchestrationLoader dedupOrchestrationLoader;

	private String rottenRecordsPathParam;

	private String hbaseTable;

	private String dedupConfig;

	@Override

	protected String execute(final NodeToken token) throws Exception {

		log.info("start preparing job");

		final String fields = getFields(env("format", token), env("layout", token));

		token.getEnv().setAttribute("index.fields", fields);

		if (!StringUtils.isBlank(getRottenRecordsPathParam())) {

			token.getEnv().setAttribute(getRottenRecordsPathParam(), "/tmp" + getFileName(token, "rottenrecords"));

		}

		token.getEnv().setAttribute("index.solr.url", getIndexSolrUrlZk());

		token.getEnv().setAttribute("index.solr.collection", getCollectionName(token));

		token.getEnv().setAttribute("index.shutdown.wait.time", getIndexSolrShutdownWait());

		token.getEnv().setAttribute("index.buffer.flush.threshold", getIndexBufferFlushTreshold());

		token.getEnv().setAttribute("index.solr.sim.mode", isFeedingSimulationMode());

		token.getEnv().setAttribute("index.feed.timestamp", DateUtils.now_ISO8601());

		@SuppressWarnings("unchecked")

		final List<Map<String, String>> sets = new Gson().fromJson(token.getEnv().getAttribute("sets"), List.class);

		token.getEnv().setAttribute("actionset", Iterables.getOnlyElement(sets).get("set"));

		token.getEnv().setAttribute("entityType", token.getEnv().getAttribute("entityType"));

		token.getEnv().setAttribute("entityTypeId", token.getEnv().getAttribute("entityTypeId"));

		return Arc.DEFAULT_ARC;

	}

	private String getFields(final String format, final String layout) throws ISLookUpException {

		return isLookup(String

				.format("<FIELDS>{for $x in collection('')//RESOURCE_PROFILE[.//RESOURCE_TYPE/@value = 'MDFormatDSResourceType' and .//NAME='%s']//LAYOUT[@name='%s']/FIELDS/FIELD return $x[string(@path)]}</FIELDS>",

						format, layout));

	}

	public String getIndexSolrUrlZk() throws ISLookUpException {

		return isLookup("for $x in /RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='IndexServiceResourceType'] return $x//PROTOCOL[./@name='solr']/@address/string()");

	}

	public String getIndexSolrShutdownWait() throws ISLookUpException {

		return queryForServiceProperty("solr:feedingShutdownTolerance");

	}

	public String getIndexBufferFlushTreshold() throws ISLookUpException {

		return queryForServiceProperty("solr:feedingBufferFlushThreshold");

	}

	public String isFeedingSimulationMode() throws ISLookUpException {

		return queryForServiceProperty("solr:feedingSimulationMode");

	}

	private String queryForServiceProperty(final String key) throws ISLookUpException {

		return isLookup("for $x in /RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='IndexServiceResourceType'] return $x//SERVICE_PROPERTIES/PROPERTY[./@ key='"

				+ key + "']/@value/string()");

	}

	private String isLookup(final String xquery) throws ISLookUpException {

		log.debug("quering for service property: " + xquery);

		final String res = serviceLocator.getService(ISLookUpService.class).getResourceProfileByQuery(xquery);

		if (StringUtils.isBlank(res)) throw new IllegalStateException("unable to find unique service property, xquery: " + xquery);

		return res;

	}

	private String getFileName(final NodeToken token, final String fileNamePrefix) {

		return "/" + fileNamePrefix + "_" + getHbaseTable() + "_" + token.getEnv().getAttribute("format") + ".seq";

	}

	private String getCollectionName(final NodeToken token) {

		return env("format", token) + "-" + env("layout", token) + "-" + env("interpretation", token);

	}

	private String env(final String s, final NodeToken token) {

		return token.getEnv().getAttribute(s);

	}

	public String getHbaseTable() {

		return hbaseTable;

	}

	@Required

	public void setHbaseTable(final String hbaseTable) {

		this.hbaseTable = hbaseTable;

	}

	public String getRottenRecordsPathParam() {

		return rottenRecordsPathParam;

	}

	public void setRottenRecordsPathParam(final String rottenRecordsPathParam) {

		this.rottenRecordsPathParam = rottenRecordsPathParam;

	}

	public String getDedupConfig() {

		return dedupConfig;

	}

	public void setDedupConfig(final String dedupConfig) {

		this.dedupConfig = dedupConfig;

	}

}