dnet-hadoop/dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/UsageStatsExporter.java

package eu.dnetlib.oa.graph.usagestats.export;

import java.io.IOException;
import java.sql.ResultSet;
import java.sql.Statement;

import javax.sound.midi.SysexMessage;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

/**
 * Main class for downloading and processing Usage statistics
 * 
 * @author D. Pierrakos, S. Zoupanos
 */
public class UsageStatsExporter {

	public UsageStatsExporter() {

	}

	private static final Logger logger = LoggerFactory.getLogger(UsageStatsExporter.class);

	public void runImpalaQuery() throws Exception {
		Statement stmt = ConnectDB.getImpalaConnection().createStatement();
		ConnectDB.getImpalaConnection().setAutoCommit(false);

		logger.info("Executing Impala query");
		Statement statement = ConnectDB.getImpalaConnection().createStatement();

		ResultSet rs = statement
			.executeQuery(
//				"CREATE TABLE usagestats_20200913.spyros_tmp5 AS\n" +
//					"SELECT s.source, d.id AS repository_id, ro.id as result_id, s.count, '0'  \n" +
//					"FROM usagestats_20200913.sarc_sushilogtmp2 s,  \n" +
//					"openaire_prod_stats_shadow_20200821.datasource_oids d,  \n" +
//					"openaire_prod_stats_shadow_20200821.datasource_results dr,  \n" +
//					"openaire_prod_stats_shadow_20200821.result_pids ro  \n" +
//					"WHERE d.oid LIKE CONCAT('%', s.repository, '%') AND dr.id=d.id AND dr.result=ro.id  \n" +
//					"AND s.rid=ro.pid AND ro.type='doi' AND metric_type='ft_total' AND s.source='SARC-OJS' ");

				"CREATE TABLE usagestats_20200913.spyros_tmp6 AS\n" +
					"SELECT *   \n" +
					"FROM usagestats_20200913.sarc_sushilogtmp2");

		stmt.close();
	}

	private void reCreateLogDirs() throws IllegalArgumentException, IOException {
		FileSystem dfs = FileSystem.get(new Configuration());

		logger.info("Deleting repoLog directory: " + ExecuteWorkflow.repoLogPath);
		dfs.delete(new Path(ExecuteWorkflow.repoLogPath), true);

		logger.info("Deleting portalLog directory: " + ExecuteWorkflow.portalLogPath);
		dfs.delete(new Path(ExecuteWorkflow.portalLogPath), true);

		logger.info("Deleting irusUKReport directory: " + ExecuteWorkflow.irusUKReportPath);
		dfs.delete(new Path(ExecuteWorkflow.irusUKReportPath), true);

		logger.info("Deleting sarcsReport (Array) directory: " + ExecuteWorkflow.sarcsReportPathArray);
		dfs.delete(new Path(ExecuteWorkflow.sarcsReportPathArray), true);

		logger.info("Deleting sarcsReport (NonArray) directory: " + ExecuteWorkflow.sarcsReportPathNonArray);
		dfs.delete(new Path(ExecuteWorkflow.sarcsReportPathNonArray), true);

		logger.info("Deleting lareferenciaLog directory: " + ExecuteWorkflow.lareferenciaLogPath);
		dfs.delete(new Path(ExecuteWorkflow.lareferenciaLogPath), true);

		logger.info("Creating repoLog directory: " + ExecuteWorkflow.repoLogPath);
		dfs.mkdirs(new Path(ExecuteWorkflow.repoLogPath));

		logger.info("Creating portalLog directory: " + ExecuteWorkflow.portalLogPath);
		dfs.mkdirs(new Path(ExecuteWorkflow.portalLogPath));

		logger.info("Creating irusUKReport directory: " + ExecuteWorkflow.irusUKReportPath);
		dfs.mkdirs(new Path(ExecuteWorkflow.irusUKReportPath));

		logger.info("Creating sarcsReport (Array) directory: " + ExecuteWorkflow.sarcsReportPathArray);
		dfs.mkdirs(new Path(ExecuteWorkflow.sarcsReportPathArray));

		logger.info("Creating sarcsReport (NonArray) directory: " + ExecuteWorkflow.sarcsReportPathNonArray);
		dfs.mkdirs(new Path(ExecuteWorkflow.sarcsReportPathNonArray));

		logger.info("Creating lareferenciaLog directory: " + ExecuteWorkflow.lareferenciaLogPath);
		dfs.mkdirs(new Path(ExecuteWorkflow.lareferenciaLogPath));
	}

	public void export() throws Exception {

		logger.info("Initialising DB properties");
		ConnectDB.init();

//		runImpalaQuery();

		PiwikStatsDB piwikstatsdb = new PiwikStatsDB(ExecuteWorkflow.repoLogPath, ExecuteWorkflow.portalLogPath);

		logger.info("Re-creating database and tables");
		if (ExecuteWorkflow.recreateDbAndTables)
			piwikstatsdb.recreateDBAndTables();
		;

		logger.info("Initializing the download logs module");
		PiwikDownloadLogs piwd = new PiwikDownloadLogs(ExecuteWorkflow.matomoBaseURL, ExecuteWorkflow.matomoAuthToken);

		// Downloading piwik logs (also managing directory creation)
		if (ExecuteWorkflow.downloadLogs) {
			logger.info("Recreating log directories");
			reCreateLogDirs();

			logger.info("Downloading piwik logs");
			piwd
				.GetOpenAIRELogs(
					ExecuteWorkflow.repoLogPath,
					ExecuteWorkflow.portalLogPath, ExecuteWorkflow.portalMatomoID);
		}
		logger.info("Downloaded piwik logs");

		// Create DB tables, insert/update statistics
		String cRobotsUrl = "https://raw.githubusercontent.com/atmire/COUNTER-Robots/master/COUNTER_Robots_list.json";
		piwikstatsdb.setCounterRobotsURL(cRobotsUrl);

		if (ExecuteWorkflow.processPiwikLogs) {
			logger.info("Processing logs");
			piwikstatsdb.processLogs();
		}

		logger.info("Creating LaReferencia tables");
		LaReferenciaDownloadLogs lrf = new LaReferenciaDownloadLogs(ExecuteWorkflow.lareferenciaBaseURL,
			ExecuteWorkflow.lareferenciaAuthToken);
		logger.info("Downloading LaReferencia logs");
		lrf.GetLaReferenciaRepos(ExecuteWorkflow.lareferenciaLogPath);
		logger.info("Downloaded LaReferencia logs");
		LaReferenciaStats lastats = new LaReferenciaStats(ExecuteWorkflow.lareferenciaLogPath);
		logger.info("Processing LaReferencia logs");
//		lastats.processLogs();
//        log.info("LaReferencia logs done");

		System.exit(0);

//		IrusStats irusstats = new IrusStats(irusUKBaseURL);
//		irusstats.getIrusRRReport(irusUKReportPath);

//		irusstats.processIrusStats();
//		log.info("irus done");

//		SarcStats sarcStats = new SarcStats();
//		sarcStats.getAndProcessSarc(sarcsReportPathArray, sarcsReportPathNonArray);
//		sarcStats.finalizeSarcStats();
//		log.info("sarc done");

//		// finalize usagestats
		piwikstatsdb.finalizeStats();
//		log.info("finalized stats");
	}
}