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

More control on directory cleanup

This commit is contained in:
Spyros Zoupanos 2020-10-15 17:41:36 +03:00
parent 8f24a6388e
commit a1dc930486
5 changed files with 53 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/ExecuteWorkflow.java
View File

@ -37,6 +37,8 @@ public class ExecuteWorkflow {
static String usageStatsDBSchema;
static String statsDBSchema;
static boolean recreateDbAndTables;
static boolean piwikEmptyDirs;
static boolean downloadPiwikLogs;
static boolean processPiwikLogs;
@ -45,6 +47,7 @@ public class ExecuteWorkflow {
static int numberOfPiwikIdsToDownload;
static int numberOfSiteIdsToDownload;
static boolean laReferenciaEmptyDirs;
static boolean downloadLaReferenciaLogs;
static boolean processLaReferenciaLogs;
@ -98,6 +101,11 @@ public class ExecuteWorkflow {
else
recreateDbAndTables = false;
if (parser.get("piwikEmptyDirs").toLowerCase().equals("true"))
piwikEmptyDirs = true;
else
piwikEmptyDirs = false;
if (parser.get("downloadPiwikLogs").toLowerCase().equals("true"))
downloadPiwikLogs = true;
else
@ -119,6 +127,11 @@ public class ExecuteWorkflow {
numberOfPiwikIdsToDownload = Integer.parseInt(parser.get("numberOfPiwikIdsToDownload"));
numberOfSiteIdsToDownload = Integer.parseInt(parser.get("numberOfSiteIdsToDownload"));
if (parser.get("laReferenciaEmptyDirs").toLowerCase().equals("true"))
laReferenciaEmptyDirs = true;
else
laReferenciaEmptyDirs = false;
if (parser.get("downloadLaReferenciaLogs").toLowerCase().equals("true"))
downloadLaReferenciaLogs = true;
else

16
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/PiwikStatsDB.java
View File

@ -39,6 +39,22 @@ public class PiwikStatsDB {
}
public 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("Creating repoLog directory: " + ExecuteWorkflow.repoLogPath);
dfs.mkdirs(new Path(ExecuteWorkflow.repoLogPath));
logger.info("Creating portalLog directory: " + ExecuteWorkflow.portalLogPath);
dfs.mkdirs(new Path(ExecuteWorkflow.portalLogPath));
}
public void recreateDBAndTables() throws Exception {
this.createDatabase();
this.createTables();

13
dhp-workflows/dhp-usage-stats-update/src/main/java/eu/dnetlib/oa/graph/usagestats/export/UsageStatsExporter.java
View File

@ -64,11 +64,13 @@ public class UsageStatsExporter {
logger.info("Initializing the download logs module");
PiwikDownloadLogs piwd = new PiwikDownloadLogs(ExecuteWorkflow.matomoBaseURL, ExecuteWorkflow.matomoAuthToken);
if (ExecuteWorkflow.piwikEmptyDirs) {
logger.info("Recreating Piwik log directories");
piwikstatsdb.reCreateLogDirs();
}
// Downloading piwik logs (also managing directory creation)
if (ExecuteWorkflow.downloadPiwikLogs) {
logger.info("Recreating log directories");
reCreateLogDirs();
logger.info("Downloading piwik logs");
piwd
.GetOpenAIRELogs(
@ -90,6 +92,11 @@ public class UsageStatsExporter {
LaReferenciaDownloadLogs lrf = new LaReferenciaDownloadLogs(ExecuteWorkflow.lareferenciaBaseURL,
ExecuteWorkflow.lareferenciaAuthToken);
if (ExecuteWorkflow.laReferenciaEmptyDirs) {
logger.info("Recreating LaReferencia log directories");
piwikstatsdb.reCreateLogDirs();
}
if (ExecuteWorkflow.downloadLaReferenciaLogs) {
logger.info("Downloading LaReferencia logs");
lrf.GetLaReferenciaRepos(ExecuteWorkflow.lareferenciaLogPath);

12
dhp-workflows/dhp-usage-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/usagestats/export/usagestats_parameters.json
View File

@ -101,6 +101,12 @@
"paramDescription": "Re-create database and initial tables?",
"paramRequired": true
},
{
"paramName": "pwed",
"paramLongName": "piwikEmptyDirs",
"paramDescription": "Empty piwik directories?",
"paramRequired": true
},
{
"paramName": "ppwl",
"paramLongName": "processPiwikLogs",
@ -137,6 +143,12 @@
"paramDescription": "Limit the number of the downloaded siteids (La Referencia logs) to the first numberOfSiteIdsToDownload",
"paramRequired": true
},
{
"paramName": "lerd",
"paramLongName": "laReferenciaEmptyDirs",
"paramDescription": "Empty LaReferencia directories?",
"paramRequired": true
},
{
"paramName": "plrl",
"paramLongName": "processLaReferenciaLogs",

2
dhp-workflows/dhp-usage-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/usagestats/oozie_app/workflow.xml
View File

@ -59,12 +59,14 @@
<arg>--usageStatsDBSchema</arg><arg>${usageStatsDBSchema}</arg>
<arg>--statsDBSchema</arg><arg>${statsDBSchema}</arg>
<arg>--recreateDbAndTables</arg><arg>${recreateDbAndTables}</arg>
<arg>--piwikEmptyDirs</arg><arg>${piwikEmptyDirs}</arg>
<arg>--downloadPiwikLogs</arg><arg>${downloadPiwikLogs}</arg>
<arg>--processPiwikLogs</arg><arg>${processPiwikLogs}</arg>
<arg>--startingLogPeriod</arg><arg>${startingLogPeriod}</arg>
<arg>--endingLogPeriod</arg><arg>${endingLogPeriod}</arg>
<arg>--numberOfPiwikIdsToDownload</arg><arg>${numberOfPiwikIdsToDownload}</arg>
<arg>--numberOfSiteIdsToDownload</arg><arg>${numberOfSiteIdsToDownload}</arg>
<arg>--laReferenciaEmptyDirs</arg><arg>${laReferenciaEmptyDirs}</arg>
<arg>--downloadLaReferenciaLogs</arg><arg>${downloadLaReferenciaLogs}</arg>
<arg>--processLaReferenciaLogs</arg><arg>${processLaReferenciaLogs}</arg>
<arg>--irusCreateTablesEmptyDirs</arg><arg>${irusCreateTablesEmptyDirs}</arg>