Added Google Analytics for telemetry - Review zinggAI#41

navinrathore · Dec 11, 2021 · a900ba8 · a900ba8
1 parent f38e810
commit a900ba8
Show file tree

Hide file tree

Showing 5 changed files with 44 additions and 40 deletions.
diff --git a/client/src/main/java/zingg/client/Client.java b/client/src/main/java/zingg/client/Client.java
@@ -1,6 +1,8 @@
 package zingg.client;
 
 import java.io.Serializable;
+import java.util.Arrays;
+import java.util.stream.Collectors;
 
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
@@ -108,8 +110,8 @@ public static void printAnalyticsBanner(boolean collectMetrics) {
 			LOG.info("*  Please note that Zingg captures a few metrics about application's       *");
 			LOG.info("*  runtime parameters. However, no user's personal data or application     *");
 			LOG.info("*  data is captured. If you want to switch off this feature, please        *");
-			LOG.info("*  set the flag collectMetrics to false in config. For details,            *");
-			LOG.info("*  please refer to the zingg docs (https://docs.zingg.ai/docs/analytics.html)     *");
+			LOG.info("*  set the flag collectMetrics to false in config. For details, please     *");
+			LOG.info("*  refer to the zingg docs (https://docs.zingg.ai/docs/analytics.html)     *");
 			LOG.info("****************************************************************************");
 			LOG.info("");
 		}
@@ -148,8 +150,8 @@ public static void main(String... args) {
 			client = new Client(arguments, options);	
 			client.init();
 			client.execute();
-			Analytics.track(Metric.METRIC_EXEC_TIME, (System.currentTimeMillis() - startTime) / 1000);
-			client.trackMetrics(phase);
+			Analytics.track(Metric.METRIC_EXEC_TIME, (System.currentTimeMillis() - startTime) / 1000, arguments.getCollectMetrics());
+			client.postMetrics(phase, arguments.getCollectMetrics());
 			LOG.warn("Zingg processing has completed");				
 		} 
 		catch(ZinggClientException e) {
@@ -218,20 +220,20 @@ public ClientOptions getOptions() {
 	public void setOptions(ClientOptions options) {
 		this.options = options;
 	}
-
-	private void trackMetrics(String phase) {
-		if (!getArguments().getCollectMetrics()) {
-			return;
-		}
-		Analytics.track(Metric.METRIC_FEATURE_COUNT, getArguments().getFieldDefinition().size());
+
+	private void postMetrics(String phase, boolean collectMetrics) {
+		Analytics.track(Metric.METRIC_FIELDS_COUNT, getArguments().getFieldDefinition().size(), collectMetrics);
+
 		Pipe[] dataPipes = getArguments().getData();
-		for (Pipe p : dataPipes) {
-			Analytics.track(Metric.METRIC_SOURCE_TYPE, p.getFormat().type());
-		}
- 		Pipe[] outputPipes = getArguments().getOutput();
-		for (Pipe p : outputPipes) {
-			Analytics.track(Metric.METRIC_SINK_TYPE, p.getFormat().type());
-		}
-		Analytics.trackEvent(phase);
- 	}
+		String inPipesStr = Arrays.stream(dataPipes).map(p -> p.getFormat().type()).collect(Collectors.toList())
+				.stream().reduce((p1, p2) -> p1 + "," + p2).map(Object::toString).orElse("");
+		Analytics.track(Metric.METRIC_DATA_FORMAT, inPipesStr, collectMetrics);
+
+		Pipe[] outputPipes = getArguments().getOutput();
+		String outPipesStr = Arrays.stream(outputPipes).map(p -> p.getFormat().type()).collect(Collectors.toList())
+				.stream().reduce((p1, p2) -> p1 + "," + p2).map(Object::toString).orElse("");
+		Analytics.track(Metric.METRIC_OUTPUT_FORMAT, outPipesStr, collectMetrics);
+
+		Analytics.postEvent(phase, collectMetrics);
+	}
 }
diff --git a/client/src/main/java/zingg/client/util/Analytics.java b/client/src/main/java/zingg/client/util/Analytics.java
@@ -38,15 +38,20 @@ private static Map<String, String> getMetrics() {
 		return metrics;
 	}
 
-	public static void track(String metricName, String metricValue) {
-		getMetrics().put(metricName, metricValue);
+	public static void track(String metricName, String metricValue, boolean collectMetrics) {
+		if (collectMetrics) {
+			getMetrics().put(metricName, metricValue);
+		}
 	}
 
-	public static void track(String metricName, double metricValue) {
-		track(metricName, String.valueOf(metricValue));
+	public static void track(String metricName, double metricValue, boolean collectMetrics) {
+		track(metricName, String.valueOf(metricValue), collectMetrics);
 	}
 
-	public static void trackEvent(String phase) {
+	public static void postEvent(String phase, boolean collectMetrics) {
+		if (collectMetrics == false) {
+			return;
+		}
 		ObjectMapper mapper = new ObjectMapper();
 		ObjectNode rootNode = mapper.createObjectNode();
 
@@ -90,13 +95,13 @@ private static void sendEvents(String param) {
 		try {
 			URL url = uri.toURL();
    			String response = executePostRequest(url.toString(), param);
-   		} catch (IOException e) {
+    		} catch (IOException e) {
 			e.printStackTrace();
 		}
 		LOG.debug("Event tracked.");
 	}
 
-	public static String executePostRequest(String targetURL, String urlParameters) {
+	private static String executePostRequest(String targetURL, String urlParameters) {
 		HttpURLConnection connection = null;
 		try {
 			//Create connection

diff --git a/client/src/main/java/zingg/client/util/Metric.java b/client/src/main/java/zingg/client/util/Metric.java
@@ -4,13 +4,13 @@
 import org.apache.spark.sql.Row;
 
 public class Metric {
-    public static final String METRIC_SOURCE_TYPE = "source_type";
-    public static final String METRIC_SINK_TYPE = "sink_type";
-    public static final String METRIC_FEATURE_COUNT = "no_features";
-    public static final String METRIC_EXEC_TIME = "exec_time";
-    public static final String METRIC_POSITIVE_COUNT = "positive_count";
-    public static final String METRIC_NEGATIVE_COUNT = "negative_count";
-    public static final String METRIC_TOTAL_COUNT = "total_count";
+    public static final String METRIC_DATA_FORMAT = "dataFormat";
+    public static final String METRIC_OUTPUT_FORMAT = "outputFormat";
+    public static final String METRIC_FIELDS_COUNT = "numFields";
+    public static final String METRIC_EXEC_TIME = "executionTime";
+    public static final String METRIC_TRAINING_POSITIVES = "trainingDataPositives";
+    public static final String METRIC_TRAINING_NEGATIVES = "trainingDataNegatives";
+    public static final String METRIC_DATA_COUNT = "dataCount";
 
     public static final long timeout = 1200L;
     public static final double confidence = 0.95; // default value

diff --git a/core/src/main/java/zingg/Matcher.java b/core/src/main/java/zingg/Matcher.java
@@ -104,9 +104,8 @@ public void execute() throws ZinggClientException {
 			//testData = dropDuplicates(testData);
 			double count = testData.count();
 			LOG.info("Read " + count);
-			if (args.getCollectMetrics()) {
-				Analytics.track(Metric.METRIC_TOTAL_COUNT, count);
-			}
+			Analytics.track(Metric.METRIC_DATA_COUNT, count, args.getCollectMetrics());
+
 			Dataset<Row> blocked = getBlocked(testData);
 			LOG.info("Blocked ");
 			/*blocked = blocked.cache();

diff --git a/core/src/main/java/zingg/Trainer.java b/core/src/main/java/zingg/Trainer.java
@@ -55,10 +55,8 @@ public void execute() throws ZinggClientException {
 			Model model = ModelUtil.createModel(positives, negatives, new Model(this.featurers), spark);
 			model.save(args.getModel());
 			LOG.info("Learnt similarity rules and saved output at " + args.getZinggDir());
-			if (args.getCollectMetrics()) {
-				Analytics.track(Metric.METRIC_POSITIVE_COUNT, Metric.approxCount(positives));
-				Analytics.track(Metric.METRIC_NEGATIVE_COUNT, Metric.approxCount(negatives));
-			}
+			Analytics.track(Metric.METRIC_TRAINING_POSITIVES, Metric.approxCount(positives), args.getCollectMetrics());
+			Analytics.track(Metric.METRIC_TRAINING_NEGATIVES, Metric.approxCount(negatives), args.getCollectMetrics());
 			LOG.info("Finished Learning phase");			
 		} catch (Exception e) {
 			e.printStackTrace();