accounting-dashboard-harves.../src/main/java/org/gcube/dataharvest/harvester/VREAccessesHarvester.java

302 lines
12 KiB
Java

package org.gcube.dataharvest.harvester;
import static org.gcube.resources.discovery.icclient.ICFactory.clientFor;
import static org.gcube.resources.discovery.icclient.ICFactory.queryFor;
import java.io.IOException;
import java.security.GeneralSecurityException;
import java.time.Instant;
import java.time.LocalDate;
import java.time.ZoneId;
import java.time.format.DateTimeFormatter;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import org.gcube.accounting.accounting.summary.access.model.ScopeDescriptor;
import org.gcube.accounting.accounting.summary.access.model.update.AccountingRecord;
import org.gcube.common.encryption.encrypter.StringEncrypter;
import org.gcube.common.resources.gcore.ServiceEndpoint;
import org.gcube.common.resources.gcore.ServiceEndpoint.AccessPoint;
import org.gcube.common.resources.gcore.ServiceEndpoint.Property;
import org.gcube.common.resources.gcore.utils.Group;
import org.gcube.common.scope.api.ScopeProvider;
import org.gcube.common.scope.impl.ScopeBean;
import org.gcube.dataharvest.AccountingDashboardHarvesterPlugin;
import org.gcube.dataharvest.datamodel.AnalyticsReportCredentials;
import org.gcube.dataharvest.datamodel.HarvestedDataKey;
import org.gcube.dataharvest.datamodel.VREAccessesReportRow;
import org.gcube.resources.discovery.client.api.DiscoveryClient;
import org.gcube.resources.discovery.client.queries.api.SimpleQuery;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.google.analytics.data.v1beta.BetaAnalyticsDataClient;
import com.google.analytics.data.v1beta.BetaAnalyticsDataSettings;
import com.google.analytics.data.v1beta.DateRange;
import com.google.analytics.data.v1beta.DateRange.Builder;
import com.google.analytics.data.v1beta.Dimension;
import com.google.analytics.data.v1beta.Metric;
import com.google.analytics.data.v1beta.Row;
import com.google.analytics.data.v1beta.RunReportRequest;
import com.google.analytics.data.v1beta.RunReportResponse;
import com.google.api.client.json.JsonFactory;
import com.google.api.client.json.gson.GsonFactory;
import com.google.api.gax.core.FixedCredentialsProvider;
import com.google.auth.oauth2.ServiceAccountCredentials;
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 = "GA4AnalyticsDataService";
private static final String AP_VIEWS_PROPERTY = "views";
private static final String AP_CLIENT_ID = "client_id";
private static final String AP_PRIVATEKEY_ID_PROPERTY = "private_key_id";
private List<VREAccessesReportRow> vreAccesses;
public VREAccessesHarvester(Date start, Date end) throws Exception {
super(start, end);
vreAccesses = getAllAccesses(start, end);
}
@Override
public List<AccountingRecord> getAccountingRecords() throws Exception {
try {
String context = org.gcube.dataharvest.utils.Utils.getCurrentContext();
ArrayList<AccountingRecord> accountingRecords = new ArrayList<AccountingRecord>();
int measure = 0;
ScopeBean scopeBean = new ScopeBean(context);
String lowerCasedContext = scopeBean.name().toLowerCase();
String case1 = lowerCasedContext + "/";
String case2 = lowerCasedContext + "?";
for(VREAccessesReportRow row : vreAccesses) {
String pagePath = row.getPagePath();
if (!pagePath.contains("_redirect=/group") && !pagePath.contains("workspace")) {
if(pagePath.endsWith(lowerCasedContext)) {
logger.debug("Matched endsWith({}) : {}", lowerCasedContext, pagePath);
measure += row.getVisitNumber();
} else if(pagePath.contains(case1) || pagePath.contains(case2)) {
logger.debug("Matched contains({}) || contains({}) : {}", case1, case2, pagePath);
measure += row.getVisitNumber();
}
}
}
ScopeDescriptor scopeDescriptor = AccountingDashboardHarvesterPlugin.getScopeDescriptor();
AccountingRecord ar = new AccountingRecord(scopeDescriptor, instant, getDimension(HarvestedDataKey.ACCESSES), (long) measure);
logger.debug("{} : {}", ar.getDimension().getId(), ar.getMeasure());
accountingRecords.add(ar);
return accountingRecords;
} catch(Exception e) {
throw e;
}
}
/**
*
* @return a list of {@link VREAccessesReportRow} objects containing the pagePath and the visit number e.g.
* VREAccessesReportRow [pagePath=/group/agroclimaticmodelling/add-new-users, visitNumber=1]
* VREAccessesReportRow [pagePath=/group/agroclimaticmodelling/administration, visitNumber=2]
* VREAccessesReportRow [pagePath=/group/agroclimaticmodelling/agroclimaticmodelling, visitNumber=39]
*/
private static List<VREAccessesReportRow> getAllAccesses(Date start, Date end) throws Exception {
Builder dateRangeBuilder = getDateRangeBuilderForAnalytics(start, end);
logger.debug("Getting accesses in this time range {}", dateRangeBuilder.toString());
AnalyticsReportCredentials credentialsFromD4S = getAuthorisedApplicationInfoFromIs();
logger.debug("Getting credentials credentialsFromD4S");
BetaAnalyticsDataSettings serviceSettings = initializeAnalyticsReporting(credentialsFromD4S);
logger.debug("initializeAnalyticsReporting service settings");
HashMap<String,List<RunReportResponse>> responses = getReportResponses(serviceSettings, credentialsFromD4S.getViewIds(), dateRangeBuilder);
List<VREAccessesReportRow> totalAccesses = new ArrayList<>();
for(String view : responses.keySet()) {
List<VREAccessesReportRow> viewReport = parseResponse(view, responses.get(view));
logger.debug("Got {} entries from view id={}", viewReport.size(), view);
totalAccesses.addAll(viewReport);
}
logger.debug("Merged in {} total entries from all views", totalAccesses.size());
return totalAccesses;
}
/**
* Initializes an Google Analytics Data API service object.
*
* @return An authorized Google Analytics Data API
* @throws IOException
* @throws GeneralSecurityException
*/
private static BetaAnalyticsDataSettings initializeAnalyticsReporting(AnalyticsReportCredentials cred) throws IOException {
return BetaAnalyticsDataSettings.newBuilder()
.setCredentialsProvider(FixedCredentialsProvider.create(
ServiceAccountCredentials.fromPkcs8(cred.getClientId(), cred.getClientEmail(), cred.getPrivateKeyPem(), cred.getPrivateKeyId(), null)))
.build();
}
/**
* Queries Analytics Data API service
*
* @param service Analytics Data API service service settings.
* @return Row Analytics Data API service
* @throws IOException
*/
private static HashMap<String,List<RunReportResponse>> getReportResponses(BetaAnalyticsDataSettings betaAnalyticsDataSettings,
List<String> viewIDs, Builder dateRangeBuilder) throws IOException {
HashMap<String,List<RunReportResponse>> reports = new HashMap<>();
try (BetaAnalyticsDataClient analyticsData = BetaAnalyticsDataClient.create(betaAnalyticsDataSettings)) {
for(String propertyId : viewIDs) {
List<RunReportResponse> gReportResponses = new ArrayList<>();
logger.debug("Getting data from Analytics Data API for propertyId: " + propertyId);
RunReportRequest request =
RunReportRequest.newBuilder()
.setProperty("properties/" + propertyId)
.addDimensions(Dimension.newBuilder().setName("pagePath"))
.addMetrics(Metric.newBuilder().setName("screenPageViews"))
.addDateRanges(dateRangeBuilder)
.build();
// Make the request.
RunReportResponse response = analyticsData.runReport(request);
gReportResponses.add(response);
// Iterate through every row of the API response.
// for (Row row : response.getRowsList()) {
// System.out.printf(
// "%s, %s%n", row.getDimensionValues(0).getValue(), row.getMetricValues(0).getValue());
// }
reports.put(propertyId, gReportResponses);
}
}
return reports;
}
/**
* Parses and prints the Analytics Data API service respose
*
* @param response An Analytics Data API service response.
*/
private static List<VREAccessesReportRow> parseResponse(String viewId, List<RunReportResponse> responses) {
logger.debug("parsing Response for propertyID=" + viewId);
List<VREAccessesReportRow> toReturn = new ArrayList<>();
for (RunReportResponse response : responses) {
for (Row row: response.getRowsList()) {
String dimension = row.getDimensionValues(0).getValue();
String metric = row.getMetricValues(0).getValue();
VREAccessesReportRow var = new VREAccessesReportRow();
boolean validEntry = false;
String pagePath = dimension;
if (pagePath.startsWith("/group") || pagePath.startsWith("/web")) {
var.setPagePath(dimension);
validEntry = true;
}
if (validEntry) {
var.setVisitNumber(Integer.parseInt(metric));
toReturn.add(var);
}
}
}
return toReturn;
}
private static List<ServiceEndpoint> getAnalyticsReportingConfigurationFromIS(String infrastructureScope)
throws Exception {
String scope = infrastructureScope;
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 + "'");
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) {
Group<AccessPoint> apGroup = res.profile().accessPoints();
AccessPoint[] accessPoints = (AccessPoint[]) apGroup.toArray(new AccessPoint[apGroup.size()]);
AccessPoint found = accessPoints[0];
reportCredentials.setClientEmail(found.username());
String decryptedPrivateKey = StringEncrypter.getEncrypter().decrypt(found.password());
reportCredentials.setPrivateKeyPem(decryptedPrivateKey.trim());
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_ID) == 0) {
String decryptedValue = StringEncrypter.getEncrypter().decrypt(prop.value());
reportCredentials.setClientId(decryptedValue);
}
if(prop.name().compareTo(AP_PRIVATEKEY_ID_PROPERTY) == 0) {
String decryptedValue = StringEncrypter.getEncrypter().decrypt(prop.value());
reportCredentials.setPrivateKeyId(decryptedValue);
}
}
}
}
} 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 Builder getDateRangeBuilderForAnalytics(Date start, Date end) {
DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd"); //required by Analytics
String startDate = asLocalDate(start).format(formatter);
String endDate = asLocalDate(end).format(formatter);
Builder dateRangeBuilder = DateRange.newBuilder().setStartDate(startDate).setEndDate(endDate);
return dateRangeBuilder;
}
}