spyros
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Corrections for Irus stats after discussing with Dimitris

This commit is contained in:
Spyros Zoupanos 2020-10-08 18:53:23 +03:00
parent 715bbd487d
commit 8b08f35dfe
4 changed files with 24 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/ExecuteWorkflow.java
View File

@ -57,7 +57,7 @@ public class ExecuteWorkflow {
static boolean sarcDownloadReports;
static boolean sarcProcessStats;
static int sarcNumberOfOpendoarsToDownload;
public static void main(String args[]) throws Exception {
// Sending the logs to the console
@ -140,7 +140,6 @@ public class ExecuteWorkflow {
irusProcessStats = false;
irusNumberOfOpendoarsToDownload = Integer.parseInt(parser.get("irusNumberOfOpendoarsToDownload"));
if (parser.get("sarcCreateTablesEmptyDirs").toLowerCase().equals("true"))
sarcCreateTablesEmptyDirs = true;
else
@ -154,7 +153,7 @@ public class ExecuteWorkflow {
else
sarcProcessStats = false;
sarcNumberOfOpendoarsToDownload = Integer.parseInt(parser.get("sarcNumberOfOpendoarsToDownload"));
UsageStatsExporter usagestatsExport = new UsageStatsExporter();
usagestatsExport.export();
}

22
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/IrusStats.java
View File

@ -169,7 +169,8 @@ public class IrusStats {
logger.info("Inserting to irus_sushilogtmp table");
String insertSushilogtmp = "INSERT INTO " + ConnectDB.getUsageStatsDBSchema() + ".irus_sushilogtmp " +
"SELECT 'IRUS-UK', 'opendoar____::', `ItemIdent`.`Value`, `ItemPerf`.`Period`.`Begin`, " +
"SELECT 'IRUS-UK', CONCAT('opendoar____::', split(split(INPUT__FILE__NAME,'IrusIRReport_')[1],'_')[0]), " +
"`ItemIdent`.`Value`, `ItemPerf`.`Period`.`Begin`, " +
"`ItemPerf`.`Instance`.`MetricType`, `ItemPerf`.`Instance`.`Count` " +
"FROM " + ConnectDB.getUsageStatsDBSchema() + ".irus_sushilogtmp_json " +
"LATERAL VIEW posexplode(ItemIdentifier) ItemIdentifierTable AS seqi, ItemIdent " +
@ -201,13 +202,26 @@ public class IrusStats {
stmt.executeUpdate(insertDStats);
logger.info("Inserted into downloads_stats");
logger.info("Creating sushilog table");
String createSushilog = "CREATE TABLE IF NOT EXISTS " + ConnectDB.getUsageStatsDBSchema()
+ ".sushilog " +
"(`source` string, " +
"`repository_id` string, " +
"`rid` string, " +
"`date` string, " +
"`metric_type` string, " +
"`count` int)";
stmt.executeUpdate(createSushilog);
logger.info("Created sushilog table");
logger.info("Inserting to sushilog table");
String insertToShushilog = "INSERT INTO " + ConnectDB.getUsageStatsDBSchema() + ".sushilog SELECT * FROM " +
ConnectDB.getUsageStatsDBSchema()
+ ".irus_sushilogtmp";
stmt.executeUpdate(insertToShushilog);
logger.info("Inserted to sushilog table");
ConnectDB.getHiveConnection().close();
}
public void getIrusRRReport(String irusUKReportPath) throws Exception {
@ -313,7 +327,7 @@ public class IrusStats {
+ "&ItemIdentifier=&ItemDataType=&hasDOI=&Granularity=Monthly&Callback=";
start.add(Calendar.MONTH, 1);
System.out.println("Downloading file: " + reportUrl);
logger.info("Downloading file: " + reportUrl);
String text = getJson(reportUrl, "", "");
if (text == null) {
continue;
@ -322,7 +336,7 @@ public class IrusStats {
FileSystem fs = FileSystem.get(new Configuration());
String filePath = irusUKReportPath + "/" + "IrusIRReport_" +
opendoar + "_" + simpleDateFormat.format(start.getTime()) + ".json";
System.out.println("Storing to file: " + filePath);
logger.info("Storing to file: " + filePath);
FSDataOutputStream fin = fs.create(new Path(filePath), true);
JSONParser parser = new JSONParser();

6
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/SarcStats.java
View File

@ -71,20 +71,20 @@ public class SarcStats {
public void reCreateLogDirs() throws IOException {
FileSystem dfs = FileSystem.get(new Configuration());
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("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));
}
public void processSarc(String sarcsReportPathArray, String sarcsReportPathNonArray,
String url, String issn) throws Exception {
Statement stmt = ConnectDB.getHiveConnection().createStatement();

2
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/UsageStatsExporter.java
View File

@ -149,7 +149,7 @@ public class UsageStatsExporter {
}
System.exit(0);
SarcStats sarcStats = new SarcStats();
if (ExecuteWorkflow.sarcCreateTablesEmptyDirs) {
sarcStats.reCreateLogDirs();