gCubeSystem
/
accounting-dashboard-harvester-se-plugin
16
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Added logic to exclude harvesting in some cases 

git-svn-id: https://svn.d4science.research-infrastructures.eu/gcube/trunk/accounting/accounting-dashboard-harvester-se-plugin@169212 82a268e6-3cf1-43bd-a215-b396298e98cf
This commit is contained in:
Luca Frosini 2018-06-14 14:19:25 +00:00
parent 248bb98b88
commit 13c74dc4f9
3 changed files with 132 additions and 124 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
src/main/java/org/gcube/dataharvest/AccountingDataHarvesterPlugin.java
View File

@ -3,6 +3,7 @@ package org.gcube.dataharvest;
import java.io.IOException;
import java.io.InputStream;
import java.util.ArrayList;
import java.util.Calendar;
import java.util.Date;
import java.util.List;
import java.util.Map;
@ -45,6 +46,7 @@ public class AccountingDataHarvesterPlugin extends Plugin<DataHarvestPluginDecla
public static final String GET_VRE_USERS_INPUT_PARAMETER = "getVREUsers";
public static final String DRY_RUN_INPUT_PARAMETER = "dryRun";
public static final String SO_BIG_DATA_VO = "/d4science.research-infrastructures.eu/SoBigData";
public static final String SO_BIG_DATA_CATALOGUE_CONTEXT = "/d4science.research-infrastructures.eu/SoBigData/ResourceCatalogue";
public static final String TAGME_CONTEXT = "/d4science.research-infrastructures.eu/SoBigData/TagMe";
@ -184,9 +186,13 @@ public class AccountingDataHarvesterPlugin extends Plugin<DataHarvestPluginDecla
try {
// Collecting Google Analytics Data for VREs Accesses
List<HarvestedData> harvested = vreAccessesHarvester.getData();
data.addAll(harvested);
if(context.startsWith(SO_BIG_DATA_VO) && start.before(DateUtils.getStartCalendar(2018, Calendar.APRIL, 1).getTime())) {
logger.info("Not Harvesting VREs Accesses for {} from {} to {}", context, DateUtils.format(start), DateUtils.format(end));
} else {
// Collecting Google Analytics Data for VREs Accesses
List<HarvestedData> harvested = vreAccessesHarvester.getData();
data.addAll(harvested);
}
} catch(Exception e) {
logger.error("Error harvesting Social Interactions for {}", context, e);
}
@ -205,7 +211,8 @@ public class AccountingDataHarvesterPlugin extends Plugin<DataHarvestPluginDecla
// Collecting info on VRE users
if(getVREUsers) {
// Harvesting Users only for VREs (not for VO and ROOT which is the sum of the children contexts)
if(scopeBean.is(Type.VRE)) {
// The VREUsers can be only Harvested for the lst month
if(scopeBean.is(Type.VRE) && start.equals(DateUtils.getPreviousPeriod(measureType).getTime())) {
VREUsersHarvester vreUsersHarvester = new VREUsersHarvester(start, end);
List<HarvestedData> harvested = vreUsersHarvester.getData();
data.addAll(harvested);

227
src/main/java/org/gcube/dataharvest/harvester/VREAccessesHarvester.java
View File

@ -69,21 +69,21 @@ public class VREAccessesHarvester extends BasicHarvester {
private static Logger logger = LoggerFactory.getLogger(VREAccessesHarvester.class);
private static final JsonFactory JSON_FACTORY = GsonFactory.getDefaultInstance();
private static final String SERVICE_ENDPOINT_CATEGORY = "OnlineService";
private static final String SERVICE_ENDPOINT_NAME = "BigGAnalyticsReportService";
private static final String AP_VIEWS_PROPERTY = "views";
private static final String AP_CLIENT_PROPERTY = "clientId";
private static final String AP_PRIVATEKEY_PROPERTY = "privateKeyId";
private static final String APPLICATION_NAME = "Analytics Reporting";
private List<VREAccessesReportRow> vreAccesses;
public VREAccessesHarvester(Date start, Date end) throws Exception {
super(start, end);
vreAccesses = getAllAccesses(start, end);
}
@Override
public List<HarvestedData> getData() throws Exception {
try {
@ -93,22 +93,21 @@ public class VREAccessesHarvester extends BasicHarvester {
ScopeBean scopeBean = new ScopeBean(context);
String lowerCasedContext = scopeBean.name().toLowerCase();
String case1=lowerCasedContext+"/";
String case2=lowerCasedContext+"?";
for (VREAccessesReportRow row : vreAccesses) {
String case1 = lowerCasedContext + "/";
String case2 = lowerCasedContext + "?";
for(VREAccessesReportRow row : vreAccesses) {
String pagePath = row.getPagePath();
if (!pagePath.contains("_redirect=/group")) {
if (pagePath.endsWith(lowerCasedContext)) {
System.out.println("match end->"+pagePath);
if(!pagePath.contains("_redirect=/group")) {
if(pagePath.endsWith(lowerCasedContext)) {
logger.trace("Matched endsWith({}) : {}", lowerCasedContext, pagePath);
measure++;
}
else if (pagePath.contains(case1) || pagePath.contains(case2) ) {
System.out.println("match compare->"+pagePath);
} else if(pagePath.contains(case1) || pagePath.contains(case2)) {
logger.trace("Matched contains({}) || contains({}) : {}", case1, case2, pagePath);
measure++;
}
}
}
HarvestedData harvest = new HarvestedData(HarvestedDataKey.ACCESSES, context, measure);
logger.debug(harvest.toString());
data.add(harvest);
@ -117,7 +116,7 @@ public class VREAccessesHarvester extends BasicHarvester {
throw e;
}
}
/**
*
* @return a list of {@link VREAccessesReportRow} objects containing the pagePath and the visit number e.g.
@ -127,23 +126,23 @@ public class VREAccessesHarvester extends BasicHarvester {
*/
private static List<VREAccessesReportRow> getAllAccesses(Date start, Date end) throws Exception {
DateRange dateRange = getDateRangeForAnalytics(start, end);
System.out.println("getting accesses in this time range: " + dateRange.toPrettyString());
logger.trace("Getting accesses in this time range {}", dateRange.toPrettyString());
AnalyticsReportCredentials credentialsFromD4S = getAuthorisedApplicationInfoFromIs();
AnalyticsReporting service = initializeAnalyticsReporting(credentialsFromD4S);
HashMap<String, GetReportsResponse> responses = getReportResponses(service, credentialsFromD4S.getViewIds(), dateRange);
HashMap<String,GetReportsResponse> responses = getReportResponses(service, credentialsFromD4S.getViewIds(),
dateRange);
List<VREAccessesReportRow> totalAccesses = new ArrayList<>();
for (String view : responses.keySet()) {
for(String view : responses.keySet()) {
List<VREAccessesReportRow> viewReport = parseResponse(view, responses.get(view));
System.out.println("got " + viewReport.size() + " entries from view id= "+view);
logger.trace("Got {} entries from view id={}", viewReport.size(), view);
totalAccesses.addAll(viewReport);
}
System.out.println("Merged in " + totalAccesses.size() + " toal entries from all views");
logger.trace("Merged in {} total entries from all views", totalAccesses.size());
return totalAccesses;
}
/**
* Initializes an Analytics Reporting API V4 service object.
*
@ -151,15 +150,16 @@ public class VREAccessesHarvester extends BasicHarvester {
* @throws IOException
* @throws GeneralSecurityException
*/
private static AnalyticsReporting initializeAnalyticsReporting(AnalyticsReportCredentials cred) throws GeneralSecurityException, IOException {
private static AnalyticsReporting initializeAnalyticsReporting(AnalyticsReportCredentials cred)
throws GeneralSecurityException, IOException {
HttpTransport httpTransport = GoogleNetHttpTransport.newTrustedTransport();
GoogleCredential credential = fromD4SServiceEndpoint(cred).createScoped(AnalyticsReportingScopes.all());
// Construct the Analytics Reporting service object.
return new AnalyticsReporting.Builder(httpTransport, JSON_FACTORY, credential)
.setApplicationName(APPLICATION_NAME).build();
}
}
/**
* Queries the Analytics Reporting API V4.
*
@ -167,32 +167,27 @@ public class VREAccessesHarvester extends BasicHarvester {
* @return GetReportResponse The Analytics Reporting API V4 response.
* @throws IOException
*/
private static HashMap<String, GetReportsResponse> getReportResponses(AnalyticsReporting service, List<String> viewIDs, DateRange dateRange) throws IOException {
HashMap<String, GetReportsResponse> reports = new HashMap<>();
private static HashMap<String,GetReportsResponse> getReportResponses(AnalyticsReporting service,
List<String> viewIDs, DateRange dateRange) throws IOException {
HashMap<String,GetReportsResponse> reports = new HashMap<>();
// Create the Metrics object.
Metric sessions = new Metric()
.setExpression("ga:pageviews")
.setAlias("pages");
Metric sessions = new Metric().setExpression("ga:pageviews").setAlias("pages");
Dimension pageTitle = new Dimension().setName("ga:pagePath");
for (String view : viewIDs) {
logger.info("Getting data from Google Analytics for viewid: "+ view);
for(String view : viewIDs) {
logger.info("Getting data from Google Analytics for viewid: " + view);
// Create the ReportRequest object.
ReportRequest request = new ReportRequest()
.setViewId(view)
.setDateRanges(Arrays.asList(dateRange))
.setMetrics(Arrays.asList(sessions))
.setDimensions(Arrays.asList(pageTitle));
ReportRequest request = new ReportRequest().setViewId(view).setDateRanges(Arrays.asList(dateRange))
.setMetrics(Arrays.asList(sessions)).setDimensions(Arrays.asList(pageTitle));
ArrayList<ReportRequest> requests = new ArrayList<ReportRequest>();
requests.add(request);
// Create the GetReportsRequest object.
GetReportsRequest getReport = new GetReportsRequest()
.setReportRequests(requests);
GetReportsRequest getReport = new GetReportsRequest().setReportRequests(requests);
// Call the batchGet method.
GetReportsResponse response = service.reports().batchGet(getReport).execute();
reports.put(view, response);
@ -200,48 +195,47 @@ public class VREAccessesHarvester extends BasicHarvester {
// Return the response.
return reports;
}
/**
* Parses and prints the Analytics Reporting API V4 response.
*
* @param response An Analytics Reporting API V4 response.
*/
private static List<VREAccessesReportRow> parseResponse(String viewId, GetReportsResponse response) {
System.out.println("\n*** parsing Response for " + viewId);
logger.trace("\n*** parsing Response for {}", viewId);
List<VREAccessesReportRow> toReturn = new ArrayList<>();
for (Report report: response.getReports()) {
for(Report report : response.getReports()) {
ColumnHeader header = report.getColumnHeader();
List<String> dimensionHeaders = header.getDimensions();
List<MetricHeaderEntry> metricHeaders = header.getMetricHeader().getMetricHeaderEntries();
List<ReportRow> rows = report.getData().getRows();
if (rows == null) {
if(rows == null) {
logger.warn("No data found for " + viewId);
}
else
for (ReportRow row: rows) {
} else
for(ReportRow row : rows) {
List<String> dimensions = row.getDimensions();
List<DateRangeValues> metrics = row.getMetrics();
VREAccessesReportRow var = new VREAccessesReportRow();
boolean validEntry = false;
for (int i = 0; i < dimensionHeaders.size() && i < dimensions.size(); i++) {
//System.out.println(dimensionHeaders.get(i) + ": " + dimensions.get(i));
for(int i = 0; i < dimensionHeaders.size() && i < dimensions.size(); i++) {
//logger.trace("{} : {}", dimensionHeaders.get(i), dimensions.get(i));
String pagePath = dimensions.get(i);
if (pagePath.startsWith("/group") || pagePath.startsWith("/web")) {
if(pagePath.startsWith("/group") || pagePath.startsWith("/web")) {
var.setPagePath(dimensions.get(i));
validEntry = true;
}
}
if (validEntry) {
for (int j = 0; j < metrics.size(); j++) {
if(validEntry) {
for(int j = 0; j < metrics.size(); j++) {
DateRangeValues values = metrics.get(j);
for (int k = 0; k < values.getValues().size() && k < metricHeaders.size(); k++) {
for(int k = 0; k < values.getValues().size() && k < metricHeaders.size(); k++) {
var.setVisitNumber(Integer.parseInt(values.getValues().get(k)));
}
}
}
}
toReturn.add(var);
logger.debug(var.toString());
}
@ -249,49 +243,46 @@ public class VREAccessesHarvester extends BasicHarvester {
}
return toReturn;
}
private static GoogleCredential fromD4SServiceEndpoint(AnalyticsReportCredentials cred) throws IOException {
String clientId = cred.getClientId();
String clientEmail = cred.getClientEmail();
String privateKeyPem = cred.getPrivateKeyPem();
String privateKeyId = cred.getPrivateKeyId();
String tokenUri = cred.getTokenUri();
String projectId = cred.getProjectId();
if (clientId == null || clientEmail == null || privateKeyPem == null
|| privateKeyId == null) {
if(clientId == null || clientEmail == null || privateKeyPem == null || privateKeyId == null) {
throw new IOException("Error reading service account credential from stream, "
+ "expecting 'client_id', 'client_email', 'private_key' and 'private_key_id'.");
}
PrivateKey privateKey = privateKeyFromPkcs8(privateKeyPem);
Collection<String> emptyScopes = Collections.emptyList();
Builder credentialBuilder = new GoogleCredential.Builder()
.setTransport( Utils.getDefaultTransport())
.setJsonFactory(Utils.getDefaultJsonFactory())
.setServiceAccountId(clientEmail)
.setServiceAccountScopes(emptyScopes)
.setServiceAccountPrivateKey(privateKey)
Builder credentialBuilder = new GoogleCredential.Builder().setTransport(Utils.getDefaultTransport())
.setJsonFactory(Utils.getDefaultJsonFactory()).setServiceAccountId(clientEmail)
.setServiceAccountScopes(emptyScopes).setServiceAccountPrivateKey(privateKey)
.setServiceAccountPrivateKeyId(privateKeyId);
if (tokenUri != null) {
if(tokenUri != null) {
credentialBuilder.setTokenServerEncodedUrl(tokenUri);
}
if (projectId != null) {
if(projectId != null) {
credentialBuilder.setServiceAccountProjectId(projectId);
}
// Don't do a refresh at this point, as it will always fail before the scopes are added.
return credentialBuilder.build();
}
private static PrivateKey privateKeyFromPkcs8(String privateKeyPem) throws IOException {
Reader reader = new StringReader(privateKeyPem);
Section section = PemReader.readFirstSectionAndClose(reader, "PRIVATE KEY");
if (section == null) {
if(section == null) {
throw new IOException("Invalid PKCS8 data.");
}
byte[] bytes = section.getBase64DecodedBytes();
@ -301,78 +292,82 @@ public class VREAccessesHarvester extends BasicHarvester {
KeyFactory keyFactory = SecurityUtils.getRsaKeyFactory();
PrivateKey privateKey = keyFactory.generatePrivate(keySpec);
return privateKey;
} catch (NoSuchAlgorithmException exception) {
} catch(NoSuchAlgorithmException exception) {
unexpectedException = exception;
} catch (InvalidKeySpecException exception) {
} catch(InvalidKeySpecException exception) {
unexpectedException = exception;
}
throw new IOException("Unexpected exception reading PKCS data", unexpectedException);
}
private static List<ServiceEndpoint> getAnalyticsReportingConfigurationFromIS(String infrastructureScope) throws Exception {
private static List<ServiceEndpoint> getAnalyticsReportingConfigurationFromIS(String infrastructureScope)
throws Exception {
String scope = infrastructureScope;
String currScope = ScopeProvider.instance.get();
String currScope = ScopeProvider.instance.get();
ScopeProvider.instance.set(scope);
SimpleQuery query = queryFor(ServiceEndpoint.class);
query.addCondition("$resource/Profile/Category/text() eq '"+ SERVICE_ENDPOINT_CATEGORY +"'");
query.addCondition("$resource/Profile/Name/text() eq '"+ SERVICE_ENDPOINT_NAME +"'");
query.addCondition("$resource/Profile/Category/text() eq '" + SERVICE_ENDPOINT_CATEGORY + "'");
query.addCondition("$resource/Profile/Name/text() eq '" + SERVICE_ENDPOINT_NAME + "'");
DiscoveryClient<ServiceEndpoint> client = clientFor(ServiceEndpoint.class);
List<ServiceEndpoint> toReturn = client.submit(query);
ScopeProvider.instance.set(currScope);
return toReturn;
}
}
/**
* l
* @throws Exception
*/
private static AnalyticsReportCredentials getAuthorisedApplicationInfoFromIs() throws Exception {
AnalyticsReportCredentials reportCredentials = new AnalyticsReportCredentials();
String context = org.gcube.dataharvest.utils.Utils.getCurrentContext();
try {
List<ServiceEndpoint> list = getAnalyticsReportingConfigurationFromIS(context);
if (list.size() > 1) {
logger.error("Too many Service Endpoints having name " + SERVICE_ENDPOINT_NAME +" in this scope having Category " + SERVICE_ENDPOINT_CATEGORY);
}
else if (list.size() == 0){
logger.warn("There is no Service Endpoint having name " + SERVICE_ENDPOINT_NAME +" and Category " + SERVICE_ENDPOINT_CATEGORY + " in this context: " + context);
}
else {
for (ServiceEndpoint res : list) {
if(list.size() > 1) {
logger.error("Too many Service Endpoints having name " + SERVICE_ENDPOINT_NAME
+ " in this scope having Category " + SERVICE_ENDPOINT_CATEGORY);
} else if(list.size() == 0) {
logger.warn("There is no Service Endpoint having name " + SERVICE_ENDPOINT_NAME + " and Category "
+ SERVICE_ENDPOINT_CATEGORY + " in this context: " + context);
} else {
for(ServiceEndpoint res : list) {
reportCredentials.setTokenUri(res.profile().runtime().hostedOn());
Group<AccessPoint> apGroup = res.profile().accessPoints();
Group<AccessPoint> apGroup = res.profile().accessPoints();
AccessPoint[] accessPoints = (AccessPoint[]) apGroup.toArray(new AccessPoint[apGroup.size()]);
AccessPoint found = accessPoints[0];
reportCredentials.setClientEmail(found.address());
reportCredentials.setProjectId(found.username());
reportCredentials.setPrivateKeyPem(StringEncrypter.getEncrypter().decrypt(found.password()));
for (Property prop : found.properties()) {
if (prop.name().compareTo(AP_VIEWS_PROPERTY) == 0) {
for(Property prop : found.properties()) {
if(prop.name().compareTo(AP_VIEWS_PROPERTY) == 0) {
String decryptedValue = StringEncrypter.getEncrypter().decrypt(prop.value());
String[] views = decryptedValue.split(";");
reportCredentials.setViewIds(Arrays.asList(views));
}
if (prop.name().compareTo(AP_CLIENT_PROPERTY) == 0) {
}
if(prop.name().compareTo(AP_CLIENT_PROPERTY) == 0) {
String decryptedValue = StringEncrypter.getEncrypter().decrypt(prop.value());
reportCredentials.setClientId(decryptedValue);
}
if (prop.name().compareTo(AP_PRIVATEKEY_PROPERTY) == 0) {
}
if(prop.name().compareTo(AP_PRIVATEKEY_PROPERTY) == 0) {
String decryptedValue = StringEncrypter.getEncrypter().decrypt(prop.value());
reportCredentials.setPrivateKeyId(decryptedValue);
}
}
}
}
}
}
} catch (Exception e) {
} catch(Exception e) {
e.printStackTrace();
return null;
}
return reportCredentials;
}
private static LocalDate asLocalDate(Date date) {
return Instant.ofEpochMilli(date.getTime()).atZone(ZoneId.systemDefault()).toLocalDate();
}
private static DateRange getDateRangeForAnalytics(Date start, Date end) {
DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd"); //required by Analytics
String startDate = asLocalDate(start).format(formatter);
@ -382,5 +377,5 @@ public class VREAccessesHarvester extends BasicHarvester {
dateRange.setEndDate(endDate);
return dateRange;
}
}

14
src/test/java/org/gcube/dataharvest/AccountingDataHarvesterPluginTest.java
View File

@ -1,6 +1,7 @@
package org.gcube.dataharvest;
import java.util.ArrayList;
import java.util.Calendar;
import java.util.Date;
import java.util.HashMap;
import java.util.LinkedHashMap;
@ -85,10 +86,11 @@ public class AccountingDataHarvesterPluginTest extends ContextTest {
@Test
public void testScopeBean() throws Exception {
org.gcube.dataharvest.utils.Utils.setContext(ROOT);
SortedSet<String> contexts = getContexts();
for(String context : contexts) {
ScopeBean scopeBean = new ScopeBean(context);
logger.debug("Name {}, FullName {}", scopeBean.name(), scopeBean.toString());
logger.debug("FullName {} - Name {}", scopeBean.toString(), scopeBean.name());
}
}
@ -147,9 +149,13 @@ public class AccountingDataHarvesterPluginTest extends ContextTest {
}
try {
// Collecting Google Analytics Data for VREs Accesses
List<HarvestedData> harvested = vreAccessesHarvester.getData();
data.addAll(harvested);
if(context.startsWith(AccountingDataHarvesterPlugin.SO_BIG_DATA_VO) && start.before(DateUtils.getStartCalendar(2018, Calendar.APRIL, 1).getTime())) {
logger.info("Not Harvesting VREs Accesses for {} from {} to {}", context, DateUtils.format(start), DateUtils.format(end));
} else {
// Collecting Google Analytics Data for VREs Accesses
List<HarvestedData> harvested = vreAccessesHarvester.getData();
data.addAll(harvested);
}
} catch(Exception e) {
logger.error("Error harvesting Social Interactions for {}", context, e);
}