openaire-usage-stats-api-r5/src/main/java/eu/dnetlib/usagestats/repositories/UsageStatsRepository.java

1455 lines
95 KiB
Java
Raw Normal View History

2020-12-30 14:56:37 +01:00
package eu.dnetlib.usagestats.repositories;
import org.apache.log4j.Logger;
import org.springframework.data.redis.core.HashOperations;
import org.springframework.data.redis.core.RedisTemplate;
import org.springframework.stereotype.Repository;
import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.usagestats.portal.CountryRepositories;
import eu.dnetlib.usagestats.portal.CountryUsageStats;
import eu.dnetlib.usagestats.portal.CountryUsageStatsAll;
import eu.dnetlib.usagestats.portal.MonthlyStats;
import eu.dnetlib.usagestats.portal.MonthlyUsageStats;
import eu.dnetlib.usagestats.portal.RepositoryStats;
import eu.dnetlib.usagestats.portal.TotalStats;
import eu.dnetlib.usagestats.portal.TotalStatsReposViewsDownloads;
import eu.dnetlib.usagestats.portal.UsageStats;
import eu.dnetlib.usagestats.portal.YearlyStats;
import eu.dnetlib.usagestats.sushilite.domain.ItemIdentifier;
import eu.dnetlib.usagestats.sushilite.domain.ItemPerformance;
import eu.dnetlib.usagestats.sushilite.domain.ReportItem;
import org.apache.commons.dbutils.DbUtils;
import javax.sql.DataSource;
import java.security.MessageDigest;
import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.text.SimpleDateFormat;
import java.util.ArrayList;
import java.util.Calendar;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import org.springframework.beans.factory.annotation.Value;
@Repository
public class UsageStatsRepository {
private final DataSource usageStatsDB;
private final HashOperations<String, String, String> jedis;
private final Logger log = Logger.getLogger(this.getClass());
@Value("${prod.statsdb}")
private String statsDB;
@Value("${prod.usagestatsImpalaDB}")
private String usagestatsImpalaDB;
public UsageStatsRepository(DataSource usageStatsDB, RedisTemplate<String, String> redisTemplate) {
this.usageStatsDB = usageStatsDB;
this.jedis = redisTemplate.opsForHash();
}
private static String MD5(String string) throws java.security.NoSuchAlgorithmException {
MessageDigest md = MessageDigest.getInstance("MD5");
md.update(string.getBytes());
byte byteData[] = md.digest();
StringBuilder sb = new StringBuilder();
for (byte aByteData : byteData) {
sb.append(Integer.toString((aByteData & 0xff) + 0x100, 16).substring(1));
}
return sb.toString();
}
private static String toJson(Object o) throws com.fasterxml.jackson.core.JsonProcessingException {
ObjectMapper objectMapper = new ObjectMapper();
return objectMapper.writeValueAsString(o);
}
private static UsageStats fromJson(String string) throws java.io.IOException {
ObjectMapper objectMapper = new ObjectMapper();
return objectMapper.readValue(string, UsageStats.class);
}
/*
private static List<ReportItem> reportItemsFromJson(String string) throws Exception {
ObjectMapper objectMapper = new ObjectMapper();
return objectMapper.readValue(string, objectMapper.getTypeFactory().constructCollectionType(List.class, ReportItem.class));
}
*/
public List<MonthlyUsageStats> executeMontlyUsageStats(String query) {
List<MonthlyUsageStats> montlhyList = new ArrayList<MonthlyUsageStats>();
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
log.info(connection.toString());
st = connection.prepareStatement(query);
log.info(st.toString());
rs = st.executeQuery();
while (rs.next()) {
MonthlyUsageStats monthlyUsageStats = new MonthlyUsageStats();
monthlyUsageStats.addDate(rs.getString(1));
monthlyUsageStats.addDownloads(rs.getString(2));
monthlyUsageStats.addViews(rs.getString(3));
montlhyList.add(monthlyUsageStats);
}
} catch (Exception e) {
System.out.println(e);
}
try {
jedis.put("test", "result", toJson(montlhyList));
jedis.put("test", "persistent", "false");
jedis.put("test", "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
}finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return montlhyList;
}
public TotalStatsReposViewsDownloads executeTotalStatsReposViewsDownloads(String query) {
TotalStatsReposViewsDownloads totalStatsReposViewsDownlads = new TotalStatsReposViewsDownloads();
String total_repos = " ";
String views = " ";
String downloads = " ";
String redis_key = "";
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
log.info(connection.toString());
st = connection.prepareStatement(query);
log.info(st.toString());
rs = st.executeQuery();
redis_key = MD5(st.toString());
while (rs.next()) {
totalStatsReposViewsDownlads.addRepositories(rs.getString(1));
totalStatsReposViewsDownlads.addViews(rs.getString(2));
totalStatsReposViewsDownlads.addDownloads(rs.getString(3));
}
} catch (Exception e) {
System.out.println(e);
}
try {
jedis.put(redis_key, "result", toJson(totalStatsReposViewsDownlads));
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
}finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return totalStatsReposViewsDownlads;
}
public CountryUsageStatsAll executeCountryUsageStats(String query) {
CountryUsageStatsAll countryListAll = new CountryUsageStatsAll();
List<CountryUsageStats> countryList = new ArrayList<CountryUsageStats>();
String date = " ";
String total_repos = " ";
String views = " ";
String downloads = " ";
String redis_key = "redis_key";
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
int total_views = 0;
int total_downloads = 0;
try {
connection = usageStatsDB.getConnection();
log.info(connection.toString());
st = connection.prepareStatement(query);
log.info(st.toString());
rs = st.executeQuery();
redis_key = MD5(st.toString());
while (rs.next()) {
CountryUsageStats countryUsageStats = new CountryUsageStats();
countryUsageStats.addCountry(rs.getString(1));
countryUsageStats.addTotalRepos(rs.getString(2));
countryUsageStats.addViews(rs.getString(3));
countryUsageStats.addDownloads(rs.getString(4));
total_views += Integer.parseInt(rs.getString(3));
total_downloads += Integer.parseInt(rs.getString(4));
countryList.add(countryUsageStats);
}
countryListAll.addViewsAll(Integer.toString(total_views));
countryListAll.addDownloadsAll(Integer.toString(total_downloads));
countryListAll.addCountryUsageStats(countryList);
} catch (Exception e) {
log.info(e);
System.out.println(e);
}
try {
jedis.put(redis_key, "result", toJson(countryListAll));
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return countryListAll;
}
public CountryUsageStats executeCountryUsageStats(String query, String country) {
CountryUsageStats countryUsageStats = new CountryUsageStats();
String total_repos = " ";
String views = " ";
String downloads = " ";
String redis_key = "";
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
int total_views = 0;
int total_downloads = 0;
try {
connection = usageStatsDB.getConnection();
st = connection.prepareStatement(query);
redis_key = MD5(st.toString());
//st.setString(1, country);
log.info(st.toString());
rs = st.executeQuery();
while (rs.next()) {
countryUsageStats.addCountry(country);
countryUsageStats.addTotalRepos(rs.getString(1));
countryUsageStats.addViews(rs.getString(2));
countryUsageStats.addDownloads(rs.getString(3));
}
} catch (Exception e) {
log.info(e);
System.out.println(e);
}
try {
jedis.put(redis_key, "result", toJson(countryUsageStats));
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return countryUsageStats;
}
public List<CountryRepositories> executeCountryRepositories(String query) {
List<CountryRepositories> countryReposList = new ArrayList<CountryRepositories>();
String country = " ";
String repository = " ";
String redis_key = "";
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
log.info(connection.toString());
st = connection.prepareStatement(query);
log.info(st.toString());
rs = st.executeQuery();
redis_key = MD5(st.toString());
while (rs.next()) {
CountryRepositories countryRepository = new CountryRepositories();
countryRepository.addCountry(rs.getString(1));
countryRepository.addRepository(rs.getString(2));
countryReposList.add(countryRepository);
}
} catch (Exception e) {
System.out.println(e);
}
try {
jedis.put(redis_key, "result", toJson(countryReposList));
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return countryReposList;
}
public List<MonthlyUsageStats> executeMontlyUsageStatsForRepo(String query, String datasourceId) {
List<MonthlyUsageStats> montlhyList = new ArrayList<MonthlyUsageStats>();
String redis_key = "";
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
st = connection.prepareStatement(query);
redis_key = MD5(st.toString());
st.setString(1, datasourceId);
log.info(connection.toString());
rs = st.executeQuery();
while (rs.next()) {
MonthlyUsageStats monthlyUsageStats = new MonthlyUsageStats();
monthlyUsageStats.addDate(rs.getString(1));
monthlyUsageStats.addDownloads(rs.getString(2));
monthlyUsageStats.addViews(rs.getString(3));
montlhyList.add(monthlyUsageStats);
}
} catch (Exception e) {
System.out.println(e);
}
try {
jedis.put(redis_key, "result", toJson(montlhyList));
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
System.out.println(e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return montlhyList;
}
public UsageStats executeUsageStats(String query, List<String> values, String type) {
UsageStats usageStats = new UsageStats();
int total_views = 0;
int total_downloads = 0;
int page_views = 0;
int openaire_downloads = 0;
int openaire_views = 0;
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
st = connection.prepareStatement(query);
int i = 1;
for (String s : values) {
st.setString(i, s);
i++;
}
String redis_key = MD5(st.toString());
String redis_result = jedis.get(redis_key, "result");
if (redis_result != null) {
return fromJson(redis_result);
}
rs = st.executeQuery();
if (type.equals("result")) {
while (rs.next()) {
if (rs.getString(1).equals("views") && rs.getString(4) != null && !rs.getString(4).equals("") && !rs.getString(4).equals("null")) {
usageStats.addViews(new RepositoryStats(rs.getString(3), rs.getString(2), rs.getString(4), rs.getString(5)));
total_views += Integer.parseInt(rs.getString(4));
openaire_views += Integer.parseInt(rs.getString(5));
} else if (rs.getString(1).equals("downloads") && rs.getString(4) != null && !rs.getString(4).equals("") && !rs.getString(4).equals("null")) {
usageStats.addDownloads(new RepositoryStats(rs.getString(3), rs.getString(2), rs.getString(4), "0"));
total_downloads += Integer.parseInt(rs.getString(4));
openaire_downloads += Integer.parseInt(rs.getString(5));
} else if (rs.getString(1).equals("pageviews") && rs.getString(4) != null && !rs.getString(4).equals("") && !rs.getString(4).equals("null")) {
page_views = Integer.parseInt(rs.getString(4));
}
}
usageStats.setTotal_views(Integer.toString(total_views));
usageStats.setTotal_downloads(Integer.toString(total_downloads));
usageStats.setPageViews(Integer.toString(page_views));
usageStats.setTotal_openaire_views(Integer.toString(openaire_views));
usageStats.setTotal_openaire_downloads(Integer.toString(openaire_downloads));
} else if (type.equals("project") || type.equals("datasource")) {
while (rs.next()) {
if (rs.getString(1).equals("views") && rs.getString(2) != null && !rs.getString(2).equals("") && !rs.getString(2).equals("null")) {
total_views += Integer.parseInt(rs.getString(2));
openaire_views += Integer.parseInt(rs.getString(3));
} else if (rs.getString(1).equals("downloads") && rs.getString(2) != null && !rs.getString(2).equals("") && !rs.getString(2).equals("null")) {
total_downloads += Integer.parseInt(rs.getString(2));
openaire_downloads += Integer.parseInt(rs.getString(3));
} else if (rs.getString(1).equals("pageviews") && rs.getString(2) != null && !rs.getString(2).equals("") && !rs.getString(2).equals("null")) {
page_views = Integer.parseInt(rs.getString(2));
}
/*
else if (rs.getString(1).equals("openaire") && rs.getString(2) != null && !rs.getString(2).equals("") && !rs.getString(2).equals("null")) {
openaire = Integer.parseInt(rs.getString(2));
}
*/
}
usageStats.setTotal_views(Integer.toString(total_views));
usageStats.setTotal_downloads(Integer.toString(total_downloads));
usageStats.setPageViews(Integer.toString(page_views));
usageStats.setTotal_openaire_views(Integer.toString(openaire_views));
usageStats.setTotal_openaire_downloads(Integer.toString(openaire_downloads));
}
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "query", st.toString());
//jedis.put(redis_key, "result", toString(usageStats));
jedis.put(redis_key, "result", toJson(usageStats));
jedis.put(redis_key, "fetchMode", "3");
} catch (Exception e) {
log.error("Cannot execute query2 : ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return usageStats;
}
public TotalStats executeTotalStats() {
TotalStats totalStats = null;
try {
String redis_result = jedis.get("total_stats", "result");
if (redis_result != null) {
totalStats = fromJsonTotalStats(redis_result);
} else {
return updateTotalStats();
}
} catch (Exception e) {
log.error("Cannot execute totalStats : ", e);
}
return totalStats;
}
public TotalStats updateTotalStats() {
TotalStats totalStats = new TotalStats();
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
HashMap<Integer, List<MonthlyStats>> monthlyStatsMap = new HashMap<>();
try {
connection = usageStatsDB.getConnection();
//st = connection.prepareStatement("SELECT count(distinct d.repository_id) AS repository, count(distinct d.result_id) AS items, sum(d.count) AS downloads, sum(v.count) AS views from public.downloads_stats d FULL OUTER JOIN public.views_stats v ON d.source=v.source AND d.repository_id=v.repository_id AND d.result_id=v.result_id AND d.date=v.date;");
st = connection.prepareStatement("SELECT ndv(distinct repository_id) AS repository, ndv(distinct result_id) AS items, sum(downloads) AS downloads, sum(views) AS views FROM "+usagestatsImpalaDB+".usage_stats;");
rs = st.executeQuery();
rs.next();
totalStats.setRepositories(rs.getInt(1));
totalStats.setItems(rs.getInt(2));
totalStats.setDownloads(rs.getInt(3));
totalStats.setViews(rs.getInt(4));
rs.close();
st.close();
//st = connection.prepareStatement("select coalesce(d.date,v.date) as month, count(distinct d.repository_id) as repository, count(distinct d.result_id) as items, sum(d.count) as downloads, sum(v.count) as views from public.downloads_stats d FULL OUTER JOIN public.views_stats v ON d.source=v.source AND d.repository_id=v.repository_id AND d.result_id=v.result_id AND d.date=v.date group by month order by month;");
st = connection.prepareStatement("SELECT `date`, ndv(distinct repository_id) AS repository, ndv(distinct result_id) AS items, sum(downloads) AS downloads, sum(views) AS views FROM "+usagestatsImpalaDB+".usage_stats GROUP BY `date` ORDER BY `date`;");
rs = st.executeQuery();
while (rs.next()) {
int year = Integer.parseInt(rs.getString(1).substring(0, 4));
int month = Integer.parseInt(rs.getString(1).substring(5));
MonthlyStats monthlyStats = new MonthlyStats();
monthlyStats.setMonth(month);
monthlyStats.setRepositories(rs.getInt(2));
monthlyStats.setItems(rs.getInt(3));
monthlyStats.setDownloads(rs.getInt(4));
monthlyStats.setViews(rs.getInt(5));
if (monthlyStatsMap.get(year) != null) {
monthlyStatsMap.get(year).add(monthlyStats);
} else {
List<MonthlyStats> newList = new ArrayList<>();
newList.add(monthlyStats);
monthlyStatsMap.put(year, newList);
}
}
rs.close();
st.close();
//st = connection.prepareStatement("SELECT COALESCE(SUBSTRING(d.date FROM 1 FOR 4), SUBSTRING(v.date FROM 1 FOR 4)) AS year, COUNT(DISTINCT d.repository_id) AS repository, COUNT(DISTINCT d.result_id) AS items, SUM(d.count) AS downloads, SUM(v.count) AS views FROM public.downloads_stats d FULL OUTER JOIN public.views_stats v ON d.source=v.source AND d.repository_id=v.repository_id AND d.result_id=v.result_id AND d.date=v.date GROUP BY year ORDER BY year;");
st = connection.prepareStatement("SELECT SUBSTR(`date`,1,4) AS year, ndv(DISTINCT repository_id) AS repository, \n" +
"ndv(DISTINCT result_id) AS items, SUM(downloads) AS downloads, SUM(views) AS views \n" +
"FROM "+usagestatsImpalaDB+".usage_stats GROUP BY year ORDER BY year;");
rs = st.executeQuery();
List<YearlyStats> yearlyStatsList = new ArrayList<>();
while (rs.next()) {
YearlyStats yearlyStats = new YearlyStats();
yearlyStats.setYear(rs.getInt(1));
yearlyStats.setRepositories(rs.getInt(2));
yearlyStats.setItems(rs.getInt(3));
yearlyStats.setDownloads(rs.getInt(4));
yearlyStats.setViews(rs.getInt(5));
yearlyStats.setMonthlyStats(monthlyStatsMap.get(rs.getInt(1)));
yearlyStatsList.add(yearlyStats);
}
totalStats.setYearlyStats(yearlyStatsList);
jedis.put("total_stats", "result", toJson(totalStats));
jedis.put("total_stats", "persistent", "false");
} catch (Exception e) {
log.error("Cannot execute totalStats : ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return totalStats;
}
private static TotalStats fromJsonTotalStats(String string) throws java.io.IOException {
ObjectMapper objectMapper = new ObjectMapper();
return objectMapper.readValue(string, TotalStats.class);
}
public String executeRepoId(String repositoryIdentifier, String report) {
PreparedStatement st = null;
Connection connection = null;
ResultSet rs = null;
log.info("database "+statsDB);
try {
connection = usageStatsDB.getConnection();
String[] split = repositoryIdentifier.split(":");
String openaire_id = "-1";
switch (split[0].toLowerCase()) {
case "openaire":
if (!report.equals("jr1")) {
st = connection.prepareStatement("select id from "+statsDB+".datasource where id=?");
st.setString(1, repositoryIdentifier.replaceFirst(split[0] + ":", ""));
} else {
st = connection.prepareStatement("select id from "+statsDB+".datasource where id=? AND (type='Journal' OR type='Journal Aggregator/Publisher')");
st.setString(1, repositoryIdentifier.replaceFirst(split[0] + ":", ""));
}
rs = st.executeQuery();
while (rs.next()) {
openaire_id = rs.getString(1);
}
return openaire_id;
case "opendoar":
if (!report.equals("jr1")) {
st = connection.prepareStatement("select id from "+statsDB+".datasource_oids where oid=?");
st.setString(1, "opendoar____::" + repositoryIdentifier.replaceFirst(split[0] + ":", ""));
} else {
st = connection.prepareStatement("select distinct d.id from "+statsDB+".datasource d, "+statsDB+".datasource_oids di where di.oid=? and d.id=di.id and (type='Journal' OR type='Journal Aggregator/Publisher')");
st.setString(1, "opendoar____::" + repositoryIdentifier.replaceFirst(split[0] + ":", ""));
}
rs = st.executeQuery();
while (rs.next()) {
openaire_id = rs.getString(1);
}
return openaire_id;
case "issn":
st = connection.prepareStatement("select distinct d.id from "+statsDB+".datasource d, "+statsDB+".datasource_oids di, "+statsDB+".datasource_results dr where d.id=dr.id and di.oid like ? and d.id=di.id and (type='Journal' OR type='Journal Aggregator/Publisher')");
st.setString(1, "%" + repositoryIdentifier.replaceFirst(split[0] + ":", "") + "%");
rs = st.executeQuery();
while (rs.next()) {
openaire_id = rs.getString(1);
}
return openaire_id;
default:
return "-1";
}
} catch (Exception e) {
log.error("Repository id failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
return "-1";
}
public void executeItem(List<ReportItem> reportItems, String itemIdentifier, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
String[] split = itemIdentifier.split(":");
switch (split[0].toLowerCase()) {
case "oid":
executeOid(reportItems, itemIdentifier.replaceFirst(split[0] + ":", ""), repositoryIdentifier, itemDataType, beginDate, endDate, granularity);
break;
case "doi":
executeDoi(reportItems, itemIdentifier.replaceFirst(split[0] + ":", ""), repositoryIdentifier, itemDataType, beginDate, endDate, granularity);
break;
case "openaire":
executeOpenaire(reportItems, itemIdentifier.replaceFirst(split[0] + ":", ""), repositoryIdentifier, itemDataType, beginDate, endDate, granularity);
break;
default:
}
}
private void executeOid(List<ReportItem> reportItems, String oid, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
//st = connection.prepareStatement("SELECT DISTINCT roid.id FROM public.result_oids roid, public.downloads_stats s WHERE s.result_id=roid.id AND roid.orid=? UNION SELECT DISTINCT roid.id FROM public.result_oids roid, public.views_stats s WHERE s.result_id=roid.id AND roid.orid=?");
st = connection.prepareStatement("SELECT DISTINCT roid.id FROM "+statsDB+".result_oids roid, "+usagestatsImpalaDB+".usage_stats us WHERE us.result_id=roid.id AND roid.oid=?");
st.setString(1, oid);
//st.setString(2, oid);
rs = st.executeQuery();
while (rs.next()) {
executeOpenaire(reportItems, rs.getString(1), repositoryIdentifier, itemDataType, beginDate, endDate, granularity);
}
connection.close();
} catch (Exception e) {
log.error("Oid to OpenAIRE id failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
private void executeDoi(List<ReportItem> reportItems, String doi, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
//st = connection.prepareStatement("SELECT DISTINCT poid.id FROM public.result_pids poid, public.downloads_stats s WHERE s.result_id=poid.id AND poid.type='doi' AND poid.pid=? UNION SELECT DISTINCT poid.id FROM public.result_pids poid, public.views_stats s WHERE s.result_id=poid.id AND poid.type='doi' AND poid.pid=?");
st = connection.prepareStatement("SELECT DISTINCT poid.id FROM "+statsDB+".result_pids poid, "+usageStatsDB+".usage_stats us WHERE us.result_id=poid.id AND poid.type='Digital Object Identifier' AND poid.pid=?");
st.setString(1, doi);
//st.setString(2, doi);
rs = st.executeQuery();
while (rs.next()) {
executeOpenaire(reportItems, rs.getString(1), repositoryIdentifier, itemDataType, beginDate, endDate, granularity);
}
} catch (Exception e) {
log.error("Doi to OpenAIRE id failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
private void executeOpenaire(List<ReportItem> reportItems, String openaire, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
SimpleDateFormat report_dateFormat = new SimpleDateFormat("yyyy-MM-dd");
SimpleDateFormat postgresFormat = new SimpleDateFormat("yyyy/MM");
String beginDateStr = postgresFormat.format(beginDate);
String endDateStr = postgresFormat.format(endDate);
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
/*
Calendar startCalendar = Calendar.getInstance();
startCalendar.setTime(beginDate);
Calendar endCalendar = Calendar.getInstance();
endCalendar.setTime(endDate);
int diffYear = endCalendar.get(Calendar.YEAR) - startCalendar.get(Calendar.YEAR);
int diffMonth = diffYear * 12 + endCalendar.get(Calendar.MONTH) - startCalendar.get(Calendar.MONTH);
*/
try {
connection = usageStatsDB.getConnection();
if (repositoryIdentifier.equals("")) {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids WHERE pids.id=? AND type='doi' GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.ddate;");
//st = connection.prepareStatement("SELECT res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.date, oids.orid, res.downloads, res.views FROM (SELECT us.repository_id, us.result_id, us.date, us.downloads, us.views FROM usage_stats us WHERE us.date>=? AND us.date<=? AND us.result_id=?) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids WHERE pids.id=? AND type='doi' GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.date;");
st = connection.prepareStatement("SELECT distinct res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads, res.views "
+ "FROM (SELECT us.repository_id, us.result_id, us.`date`, us.downloads, us.views FROM "+usagestatsImpalaDB+".usage_stats us "
+ "WHERE us.`date`>=? AND us.`date`<=? AND us.result_id=?) AS res "
+ "JOIN "+statsDB+".result r ON res.result_id=r.id "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN (select id, group_concat(type,',') as type FROM "+statsDB+".result_classifications where id=? "
+ "GROUP by id) rc ON rc.id=r.id "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid FROM "+statsDB+".result_pids pids "
+ "WHERE pids.id=? AND type='Digital Object Identifier' GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT id, group_concat(oids.oid, '#!#') AS oid FROM "+statsDB+".result_oids oids "
+ "WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, openaire);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, openaire);
st.setString(4, openaire);
st.setString(5, openaire);
st.setString(6, openaire);
} else {
//st = connection.prepareStatement("SELECT res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id AND rc.type=? LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids WHERE pids.id=? AND type='doi' GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.ddate;");
st = connection.prepareStatement("SELECT distinct res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads, "
+ "res.views FROM (SELECT us.repository_id, us.result_id, us.`date`, us.downloads, us.views FROM "+usagestatsImpalaDB+".usage_stats us "
+ "WHERE us.`date`>=? AND us.`date`<=? AND us.result_id=?) AS res "
+ "JOIN "+statsDB+".result r ON res.result_id=r.id "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".result_classifications rc ON rc.id=r.id AND rc.type=? "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid "
+ "FROM "+statsDB+".result_pids pids WHERE pids.id=? "
+ "AND type='Digital Object Identifier' GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT oids.id, group_concat(oids.oid, '#!#') AS oid "
+ "FROM "+statsDB+".result_oids oids WHERE oids.id=? "
+ "GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, openaire);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, openaire);
st.setString(4, itemDataType);
st.setString(5, openaire);
st.setString(6, openaire);
}
} else {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids WHERE pids.id=? AND type='doi' GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.ddate;");
st = connection.prepareStatement("SELECT distinct res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads, res.views "
+ "FROM (SELECT us.repository_id, us.result_id, us.`date`, us.downloads, us.views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? AND us.result_id=? AND us.repository_id=?) AS res "
+ "JOIN "+statsDB+".result r ON res.result_id=r.id JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN (select id, group_concat(type,',') as type from "+statsDB+".result_classifications where id=? group by id) rc ON rc.id=r.id "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid FROM "+statsDB+".result_pids pids "
+ "WHERE pids.id=? AND type='Digital Object Identifier' GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT oids.id, group_concat(oids.oid, '#!#') AS oid FROM "+statsDB+".result_oids oids "
+ "WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, openaire);
st.setString(4, repositoryIdentifier);
//st.setString(5, beginDateStr);
//st.setString(6, endDateStr);
//st.setString(7, openaire);
//st.setString(8, repositoryIdentifier);
st.setString(5, openaire);
st.setString(6, openaire);
st.setString(7, openaire);
} else {
//st = connection.prepareStatement("SELECT res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.result_id=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id AND rc.type=? LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids WHERE pids.id=? AND type='doi' GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids WHERE oids.id=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.ddate;");
st = connection.prepareStatement("SELECT distinct res.repository_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads,res.views "
+ "FROM (SELECT us.repository_id, us.result_id, us.`date`, us.downloads, us.views FROM "+usagestatsImpalaDB+".usage_stats us "
+ "WHERE us.`date`>=? AND us.`date`<=? AND us.result_id=? AND us.repository_id=?) AS res "
+ "JOIN "+statsDB+".result r ON res.result_id=r.id JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".result_classifications rc ON rc.id=r.id AND rc.type=?' "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid FROM "+statsDB+".result_pids pids "
+ "WHERE pids.id=? AND type='Digital Object Identifier' GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT oids.id, group_concat(oids.oid, '#!#') AS oid "
+ "FROM "+statsDB+".result_oids oids WHERE oids.id=? "
+ "GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.repository_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, openaire);
st.setString(4, repositoryIdentifier);
//st.setString(5, beginDateStr);
//st.setString(6, endDateStr);
//st.setString(7, openaire);
//st.setString(8, repositoryIdentifier);
st.setString(5, itemDataType);
st.setString(6, openaire);
st.setString(7, openaire);
}
}
rs = st.executeQuery();
String repository = "";
String lastDate = "";
ReportItem reportItem = null;
int ft_total = 0;
int abstr = 0;
if (granularity.equalsIgnoreCase("totals")) {
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
repository = rs.getString(1);
reportItem = new ReportItem(rs.getString(3), rs.getString(7), rs.getString(5), rs.getString(2));
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", openaire));
reportItem.addIdentifier(new ItemIdentifier("URLs", rs.getString(4)));
if (rs.getString(9) != null && !rs.getString(9).equals("")) {
if (rs.getString(9).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9).substring(0, rs.getString(9).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9)));
}
}
if (rs.getString(6) != null && !rs.getString(6).equals("")) {
if (rs.getString(6).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6).substring(0, rs.getString(6).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6)));
}
}
ft_total = 0;
abstr = 0;
}
ft_total += rs.getInt(10);
abstr += rs.getInt(11);
}
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
} else if (granularity.equalsIgnoreCase("monthly")) {
Calendar endCal = Calendar.getInstance();
endCal.setTime(postgresFormat.parse(endDateStr));
endCal.add(Calendar.MONTH, 1);
Date endDateForZeros = endCal.getTime();
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
repository = rs.getString(1);
lastDate = beginDateStr;
reportItem = new ReportItem(rs.getString(3), rs.getString(7), rs.getString(5), rs.getString(2));
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", openaire));
reportItem.addIdentifier(new ItemIdentifier("URLs", rs.getString(4)));
if (rs.getString(9) != null && !rs.getString(9).equals("")) {
if (rs.getString(9).contains("#!#")) {
String allOAIs = rs.getString(9);
String[] oaiArray = allOAIs.split("#!#");
for (int i = 0; i < oaiArray.length; i++) {
reportItem.addIdentifier(new ItemIdentifier("OAI", oaiArray[i]));
}
//reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9).substring(0, rs.getString(9).indexOf("#!#"))));
//reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9).substring(0, rs.getString(9).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9)));
}
}
if (rs.getString(6) != null && !rs.getString(6).equals("")) {
if (rs.getString(6).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6).substring(0, rs.getString(6).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6)));
}
}
}
fillWithZeros(postgresFormat.parse(lastDate), postgresFormat.parse(rs.getString(8)), reportItem);
Calendar endC = Calendar.getInstance();
endC.setTime(postgresFormat.parse(rs.getString(8)));
endC.set(Calendar.DATE, endC.getActualMaximum(Calendar.DATE));
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(postgresFormat.parse(rs.getString(8))), report_dateFormat.format(endC.getTime()), rs.getString(10), rs.getString(11)));
}
endC.setTime(postgresFormat.parse(rs.getString(8)));
endC.add(Calendar.MONTH, 1);
lastDate = postgresFormat.format(endC.getTime());
}
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
}
} catch (Exception e) {
log.error("Single Item Report failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
public void executeRepo(List<ReportItem> reportItems, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
SimpleDateFormat report_dateFormat = new SimpleDateFormat("yyyy-MM-dd");
SimpleDateFormat postgresFormat = new SimpleDateFormat("yyyy/MM");
String beginDateStr = postgresFormat.format(beginDate);
String endDateStr = postgresFormat.format(endDate);
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
if (repositoryIdentifier.equals("")) {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE dois.orid LIKE 'opendoar%' ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) AS downloads, sum(us.views) AS views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? GROUP BY us.source, us.repository_id, us.`date`) "
+ "AS res JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id WHERE dois.oid "
+ "LIKE 'opendoar%' ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
//st.setString(3, beginDateStr);
//st.setString(4, endDateStr);
} else {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s, result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE dois.orid LIKE 'opendoar%' ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) AS downloads, sum(us.views) AS views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us, "+statsDB+".result_classifications rc "
+ "WHERE rc.id=us.result_id AND us.`date`>=? AND us.`date`<=? AND rc.type=? "
+ "GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id WHERE dois.oid LIKE 'opendoar%' "
+ "ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, itemDataType);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, itemDataType);
}
} else {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE dois.orid LIKE 'opendoar%' ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) AS downloads, sum(us.views) AS views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? AND us.repository_id=? "
+ "GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE dois.oid LIKE 'opendoar%' ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, repositoryIdentifier);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, repositoryIdentifier);
} else {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE dois.orid LIKE 'opendoar%' ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) AS downloads, sum(us.views) AS views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us, "+statsDB+".result_classifications rc "
+ "WHERE rc.id=us.result_id AND us.`date`>=? AND us.`date`<=? AND rc.type=? "
+ "AND us.repository_id=? GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE dois.oid LIKE 'opendoar%' ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, itemDataType);
st.setString(4, repositoryIdentifier);
//st.setString(5, beginDateStr);
//st.setString(6, endDateStr);
//st.setString(7, itemDataType);
//st.setString(8, repositoryIdentifier);
}
}
//log.error("RR STATEMENT: " + st);
/*
String redis_key = MD5(st.toString());
if (jedis.hasKey(redis_key, "result")) {
reportItems.addAll(reportItemsFromJson((String) jedis.get(redis_key, "result")));
st.close();
connection.close();
return;
}
*/
rs = st.executeQuery();
String repository = "";
String lastDate = "";
ReportItem reportItem = null;
/*
Calendar startCalendar = Calendar.getInstance();
startCalendar.setTime(beginDate);
Calendar endCalendar = Calendar.getInstance();
endCalendar.setTime(endDate);
*/
int ft_total = 0;
int abstr = 0;
if (granularity.equalsIgnoreCase("totals")) {
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
repository = rs.getString(1);
reportItem = new ReportItem(null, rs.getString(2), "Platform", null);
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("OpenDOAR", rs.getString(4).substring(rs.getString(4).lastIndexOf(":") + 1)));
reportItem.addIdentifier(new ItemIdentifier("URL", rs.getString(3)));
ft_total = 0;
abstr = 0;
}
ft_total += rs.getInt(6);
abstr += rs.getInt(7);
}
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
} else if (granularity.equalsIgnoreCase("monthly")) {
Calendar endCal = Calendar.getInstance();
endCal.setTime(postgresFormat.parse(endDateStr));
endCal.add(Calendar.MONTH, 1);
Date endDateForZeros = endCal.getTime();
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
repository = rs.getString(1);
lastDate = beginDateStr;
reportItem = new ReportItem(null, rs.getString(2), "Platform", null);
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("OpenDOAR", rs.getString(4).substring(rs.getString(4).lastIndexOf(":") + 1)));
reportItem.addIdentifier(new ItemIdentifier("URL", rs.getString(3)));
}
fillWithZeros(postgresFormat.parse(lastDate), postgresFormat.parse(rs.getString(5)), reportItem);
Calendar endC = Calendar.getInstance();
endC.setTime(postgresFormat.parse(rs.getString(5)));
endC.set(Calendar.DATE, endC.getActualMaximum(Calendar.DATE));
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(postgresFormat.parse(rs.getString(5))), report_dateFormat.format(endC.getTime()), rs.getString(6), rs.getString(7)));
}
endC.setTime(postgresFormat.parse(rs.getString(5)));
endC.add(Calendar.MONTH, 1);
lastDate = postgresFormat.format(endC.getTime());
}
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
}
/*
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "query", st.toString());
jedis.put(redis_key, "result", toJson(reportItems));
jedis.put(redis_key, "fetchMode", "3");
*/
rs.close();
st.close();
connection.close();
} catch (Exception e) {
log.error("Repository Report failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
public void executeJournal(List<ReportItem> reportItems, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
SimpleDateFormat report_dateFormat = new SimpleDateFormat("yyyy-MM-dd");
SimpleDateFormat postgresFormat = new SimpleDateFormat("yyyy/MM");
String beginDateStr = postgresFormat.format(beginDate);
String endDateStr = postgresFormat.format(endDate);
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
if (repositoryIdentifier.equals("")) {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
//st.setString(3, beginDateStr);
//st.setString(4, endDateStr);
} else {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us, "+statsDB+".result_classifications rc "
+ "WHERE rc.id=us.result_id AND us.`date`>=? AND us.`date`<=? AND rc.type=? "
+ "GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, itemDataType);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, itemDataType);
}
} else {
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? AND us.repository_id=? "
+ "GROUP BY us.source, us.repository_id, us.`date`) AS res JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, repositoryIdentifier);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, repositoryIdentifier);
} else {
//st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.orid, res.ddate, res.downloads, res.views FROM (SELECT coalesce(ds.source, vs.source), coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.downloads_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS ds FULL OUTER JOIN (SELECT s.source, s.repository_id, s.date, sum(s.count) FROM public.views_stats s, public.result_classifications rc WHERE rc.id=s.result_id AND s.date>=? AND s.date<=? AND rc.type=? AND s.repository_id=? GROUP BY s.source, s.repository_id, s.date) AS vs ON ds.source=vs.source AND ds.repository_id=vs.repository_id AND ds.date=vs.date) AS res JOIN public.datasource d ON d.id=res.repository_id JOIN public.datasource_oids dois ON d.id=dois.id WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.ddate ASC;");
st = connection.prepareStatement("SELECT d.id, d.name, d.websiteurl, dois.oid, res.`date`, res.downloads, res.views "
+ "FROM (SELECT us.source, us.repository_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us, "+statsDB+".result_classifications rc WHERE rc.id=us.result_id "
+ "AND us.`date`>=?' AND us.`date`<=? AND rc.type=? AND us.repository_id=? "
+ "GROUP BY us.source, us.repository_id, us.`date`) AS res "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".datasource_oids dois ON d.id=dois.id "
+ "WHERE (d.type='Journal' OR d.type='Journal Aggregator/Publisher') ORDER BY d.id, d.name, res.`date` ASC;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, itemDataType);
st.setString(4, repositoryIdentifier);
//st.setString(5, beginDateStr);
//st.setString(6, endDateStr);
//st.setString(7, itemDataType);
//st.setString(8, repositoryIdentifier);
}
}
//log.error("RR STATEMENT: " + st);
/*
String redis_key = MD5(st.toString());
if (jedis.hasKey(redis_key, "result")) {
reportItems.addAll(reportItemsFromJson((String) jedis.get(redis_key, "result")));
st.close();
connection.close();
return;
}
*/
rs = st.executeQuery();
String repository = "";
String lastDate = "";
ReportItem reportItem = null;
/*
Calendar startCalendar = Calendar.getInstance();
startCalendar.setTime(beginDate);
Calendar endCalendar = Calendar.getInstance();
endCalendar.setTime(endDate);
*/
int ft_total = 0;
int abstr = 0;
if (granularity.equalsIgnoreCase("totals")) {
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
repository = rs.getString(1);
reportItem = new ReportItem(null, rs.getString(2), "Platform", null);
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("ISSN", rs.getString(4).substring(rs.getString(4).lastIndexOf(":") + 1)));
if (rs.getString(3) != null) {
reportItem.addIdentifier(new ItemIdentifier("URL", rs.getString(3)));
}
ft_total = 0;
abstr = 0;
}
ft_total += rs.getInt(6);
abstr += rs.getInt(7);
}
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
} else if (granularity.equalsIgnoreCase("monthly")) {
Calendar endCal = Calendar.getInstance();
endCal.setTime(postgresFormat.parse(endDateStr));
endCal.add(Calendar.MONTH, 1);
Date endDateForZeros = endCal.getTime();
while (rs.next()) {
if (!rs.getString(1).equals(repository)) {
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
repository = rs.getString(1);
lastDate = beginDateStr;
reportItem = new ReportItem(null, rs.getString(2), "Platform", null);
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("ISSN", rs.getString(4).substring(rs.getString(4).lastIndexOf(":") + 1)));
if (rs.getString(3) != null) {
reportItem.addIdentifier(new ItemIdentifier("URL", rs.getString(3)));
}
}
fillWithZeros(postgresFormat.parse(lastDate), postgresFormat.parse(rs.getString(5)), reportItem);
Calendar endC = Calendar.getInstance();
endC.setTime(postgresFormat.parse(rs.getString(5)));
endC.set(Calendar.DATE, endC.getActualMaximum(Calendar.DATE));
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(postgresFormat.parse(rs.getString(5))), report_dateFormat.format(endC.getTime()), rs.getString(6), rs.getString(7)));
}
endC.setTime(postgresFormat.parse(rs.getString(5)));
endC.add(Calendar.MONTH, 1);
lastDate = postgresFormat.format(endC.getTime());
}
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
}
/*
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "query", st.toString());
jedis.put(redis_key, "result", toJson(reportItems));
jedis.put(redis_key, "fetchMode", "3");
*/
rs.close();
st.close();
connection.close();
} catch (Exception e) {
log.error("Repository Report failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
public void executeBatchItems(List<ReportItem> reportItems, String repositoryIdentifier, String itemDataType, Date beginDate, Date endDate, String granularity) {
SimpleDateFormat report_dateFormat = new SimpleDateFormat("yyyy-MM-dd");
SimpleDateFormat postgresFormat = new SimpleDateFormat("yyyy/MM");
String beginDateStr = postgresFormat.format(beginDate);
String endDateStr = postgresFormat.format(endDate);
Connection connection = null;
PreparedStatement st = null;
ResultSet rs = null;
try {
connection = usageStatsDB.getConnection();
if (itemDataType.equals("")) {
//st = connection.prepareStatement("SELECT res.result_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids, public.result_datasources rd WHERE rd.id=pids.id AND type='doi' AND rd.datasource=? GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids, public.result_datasources rd WHERE rd.id=oids.id AND rd.datasource=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.result_id, res.ddate;");
//st = connection.prepareStatement("SELECT res.result_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.date, oids.orid, res.downloads, res.views FROM (SELECT us.repository_id, us.result_id, us.date, sum(us.downloads) as downloads, sum(us.views) as views FROM public.usage_stats us WHERE us.date>=? AND us.date<=? AND us.repository_id=? GROUP BY us.repository_id, us.result_id, us.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids, public.result_datasources rd WHERE rd.id=pids.id AND type='doi' AND rd.datasource=? GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids, public.result_datasources rd WHERE rd.id=oids.id AND rd.datasource=? GROUP BY oids.id) AS oids ON oids.id=r.id ORDER BY res.result_id, res.date;");
st = connection.prepareStatement("SELECT distinct res.result_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads, res.views "
+ "FROM (SELECT us.repository_id, us.result_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views FROM "+usagestatsImpalaDB+".usage_stats us "
+ "WHERE us.`date`>=? AND us.`date`<=? AND us.repository_id=? "
+ "GROUP BY us.repository_id, us.result_id, us.`date`) AS res JOIN "+statsDB+".result r ON res.result_id=r.id "
+ "JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".result_classifications rc ON rc.id=r.id "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid "
+ "FROM "+statsDB+".result_pids pids, "+statsDB+".result_datasources rd "
+ "WHERE rd.id=pids.id AND type='Digital Object Identifier' AND rd.datasource=? "
+ "GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT oids.id, group_concat(oids.oid, '#!#') AS oid "
+ "FROM "+statsDB+".result_oids oids, "+statsDB+".result_datasources rd "
+ "WHERE rd.id=oids.id AND rd.datasource=? GROUP BY oids.id) "
+ "AS oids ON oids.id=r.id ORDER BY res.result_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, repositoryIdentifier);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, repositoryIdentifier);
st.setString(4, repositoryIdentifier);
st.setString(5, repositoryIdentifier);
} else {
//st = connection.prepareStatement("SELECT res.result_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.ddate, oids.orid, res.downloads, res.views FROM (SELECT coalesce(ds.repository_id, vs.repository_id) AS repository_id, coalesce(ds.result_id, vs.result_id) AS result_id, coalesce(ds.date, vs.date) AS ddate, coalesce(ds.sum, 0) AS downloads, coalesce(vs.sum,0) AS views FROM (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.downloads_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS ds FULL OUTER JOIN (SELECT s.repository_id, s.result_id, s.date, sum(s.count) FROM public.views_stats s WHERE s.date>=? AND s.date<=? AND s.repository_id=? GROUP BY s.repository_id, s.result_id, s.date) AS vs ON ds.result_id=vs.result_id AND ds.date=vs.date) AS res JOIN public.result r ON res.result_id=r.id JOIN public.datasource d ON d.id=res.repository_id JOIN public.result_classifications rc ON rc.id=r.id LEFT JOIN (SELECT pids.id, string_agg(pids.pid, '#!#') AS pid FROM public.result_pids pids, result_datasources rd WHERE rd.id=pids.id AND type='doi' AND rd.datasource=? GROUP BY pids.id) AS pids ON pids.id=r.id LEFT JOIN (SELECT oids.id, string_agg(oids.orid, '#!#') AS orid FROM public.result_oids oids, public.result_datasources rd WHERE rd.id=oids.id AND rd.datasource=? GROUP BY oids.id) AS oids ON oids.id=r.id WHERE rc.type=? ORDER BY res.result_id, res.ddate;");
st = connection.prepareStatement("SELECT distinct res.result_id, r.title, r.publisher, r.source, rc.type, pids.pid, d.name, res.`date`, oids.oid, res.downloads, res.views "
+ "FROM (SELECT us.repository_id, us.result_id, us.`date`, sum(us.downloads) as downloads, sum(us.views) as views "
+ "FROM "+usagestatsImpalaDB+".usage_stats us WHERE us.`date`>=? AND us.`date`<=? AND us.repository_id=? "
+ "GROUP BY us.repository_id, us.result_id, us.`date`) AS res "
+ "JOIN "+statsDB+".result r ON res.result_id=r.id JOIN "+statsDB+".datasource d ON d.id=res.repository_id "
+ "JOIN "+statsDB+".result_classifications rc ON rc.id=r.id "
+ "LEFT JOIN (SELECT pids.id, group_concat(pids.pid, '#!#') AS pid "
+ "FROM "+statsDB+".result_pids pids, "+statsDB+".result_datasources rd "
+ "WHERE rd.id=pids.id AND type='Digital Object Identifier' "
+ "AND rd.datasource=? GROUP BY pids.id) AS pids ON pids.id=r.id "
+ "LEFT JOIN (SELECT oids.id, group_concat(oids.oid, '#!#') AS oid "
+ "FROM "+statsDB+".result_oids oids, "+statsDB+".result_datasources rd "
+ "WHERE rd.id=oids.id AND rd.datasource=? GROUP BY oids.id) AS oids ON oids.id=r.id "
+ "WHERE rc.type=? ORDER BY res.result_id, res.`date`;");
st.setString(1, beginDateStr);
st.setString(2, endDateStr);
st.setString(3, repositoryIdentifier);
//st.setString(4, beginDateStr);
//st.setString(5, endDateStr);
//st.setString(6, repositoryIdentifier);
st.setString(4, repositoryIdentifier);
st.setString(5, repositoryIdentifier);
st.setString(6, itemDataType);
}
//log.error("IR STATEMENT: " + st);
/*
String redis_key = MD5(st.toString());
if (jedis.hasKey(redis_key, "result")) {
reportItems.addAll(reportItemsFromJson((String) jedis.get(redis_key, "result")));
st.close();
connection.close();
return;
}
*/
rs = st.executeQuery();
String result = "";
String lastDate = "";
ReportItem reportItem = null;
int ft_total = 0;
int abstr = 0;
if (granularity.equalsIgnoreCase("totals")) {
while (rs.next()) {
if (!rs.getString(1).equals(result)) {
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
result = rs.getString(1);
reportItem = new ReportItem(rs.getString(3), rs.getString(7), rs.getString(5), rs.getString(2));
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("URLs", rs.getString(4)));
if (rs.getString(9) != null && !rs.getString(9).equals("")) {
if (rs.getString(9).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9).substring(0, rs.getString(9).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9)));
}
}
if (rs.getString(6) != null && !rs.getString(6).equals("")) {
if (rs.getString(6).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6).substring(0, rs.getString(6).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6)));
}
}
ft_total = 0;
abstr = 0;
}
ft_total += rs.getInt(10);
abstr += rs.getInt(11);
}
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(beginDate), report_dateFormat.format(endDate), Integer.toString(ft_total), Integer.toString(abstr)));
reportItems.add(reportItem);
}
} else if (granularity.equalsIgnoreCase("monthly")) {
Calendar endCal = Calendar.getInstance();
endCal.setTime(postgresFormat.parse(endDateStr));
endCal.add(Calendar.MONTH, 1);
Date endDateForZeros = endCal.getTime();
while (rs.next()) {
if (!rs.getString(1).equals(result)) {
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
result = rs.getString(1);
lastDate = beginDateStr;
reportItem = new ReportItem(rs.getString(3), rs.getString(7), rs.getString(5), rs.getString(2));
reportItem.addIdentifier(new ItemIdentifier("OpenAIRE", rs.getString(1)));
reportItem.addIdentifier(new ItemIdentifier("URLs", rs.getString(4)));
if (rs.getString(9) != null && !rs.getString(9).equals("")) {
if (rs.getString(9).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9).substring(0, rs.getString(9).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("OAI", rs.getString(9)));
}
}
if (rs.getString(6) != null && !rs.getString(6).equals("")) {
if (rs.getString(6).contains("#!#")) {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6).substring(0, rs.getString(6).indexOf("#!#"))));
} else {
reportItem.addIdentifier(new ItemIdentifier("DOI", rs.getString(6)));
}
}
}
fillWithZeros(postgresFormat.parse(lastDate), postgresFormat.parse(rs.getString(8)), reportItem);
Calendar endC = Calendar.getInstance();
endC.setTime(postgresFormat.parse(rs.getString(8)));
endC.set(Calendar.DATE, endC.getActualMaximum(Calendar.DATE));
if (reportItem != null) {
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(postgresFormat.parse(rs.getString(8))), report_dateFormat.format(endC.getTime()), rs.getString(10), rs.getString(11)));
}
endC.setTime(postgresFormat.parse(rs.getString(8)));
endC.add(Calendar.MONTH, 1);
lastDate = postgresFormat.format(endC.getTime());
}
if (reportItem != null) {
fillWithZeros(postgresFormat.parse(lastDate), endDateForZeros, reportItem);
reportItems.add(reportItem);
}
}
/*
jedis.put(redis_key, "persistent", "false");
jedis.put(redis_key, "query", st.toString());
jedis.put(redis_key, "result", toJson(reportItems));
jedis.put(redis_key, "fetchMode", "3");
*/
} catch (Exception e) {
log.error("Batch Item Report failed: ", e);
} finally {
DbUtils.closeQuietly(rs);
DbUtils.closeQuietly(st);
DbUtils.closeQuietly(connection);
}
}
private void fillWithZeros(Date from, Date to, ReportItem reportItem) {
SimpleDateFormat report_dateFormat = new SimpleDateFormat("yyyy-MM-dd");
Calendar fromCalendar = Calendar.getInstance();
fromCalendar.setTime(from);
Calendar toCalendar = Calendar.getInstance();
toCalendar.setTime(to);
while (from.before(to)) {
Calendar temp_c = Calendar.getInstance();
temp_c.setTime(from);
temp_c.set(Calendar.DAY_OF_MONTH, temp_c.getActualMaximum(Calendar.DAY_OF_MONTH));
Date temp_endDate = temp_c.getTime();
reportItem.addPerformance(new ItemPerformance(report_dateFormat.format(from), report_dateFormat.format(temp_endDate), "0", "0"));
fromCalendar.add(Calendar.MONTH, 1);
from = fromCalendar.getTime();
}
}
}