chore | adding the excluded domains for creating backends

hypertrace-trace-enricher/hypertrace-trace-enricher-impl/src/main/java/org/hypertrace/traceenricher/enrichment/enrichers/backend/AbstractBackendEntityEnricher.java

@@ -27,6 +27,7 @@
 import org.hypertrace.entity.data.service.v1.Entity.Builder;
 import org.hypertrace.entity.service.constants.EntityConstants;
 import org.hypertrace.entity.v1.entitytype.EntityType;
+import org.hypertrace.semantic.convention.utils.http.HttpSemanticConventionUtils;
 import org.hypertrace.semantic.convention.utils.span.SpanSemanticConventionUtils;
 import org.hypertrace.traceenricher.enrichedspan.constants.BackendType;
 import org.hypertrace.traceenricher.enrichedspan.constants.EnrichedSpanConstants;
@@ -64,10 +65,12 @@ public abstract class AbstractBackendEntityEnricher extends AbstractTraceEnriche
       EnrichedSpanConstants.getValue(Backend.BACKEND_OPERATION);
   private static final String BACKEND_DESTINATION_ATTR =
       EnrichedSpanConstants.getValue(Backend.BACKEND_DESTINATION);
+  private static final String BACKEND_EXCLUDED_DOMAINS = "backend.excluded.domains";
 
   private EdsClient edsClient;
   private EntityCache entityCache;
   private FqnResolver fqnResolver;
+  private List<String> backendExcludedDomains;
 
   @Override
   public void init(Config enricherConfig, ClientRegistry clientRegistry) {
@@ -76,6 +79,10 @@ public void init(Config enricherConfig, ClientRegistry clientRegistry) {
     this.entityCache = clientRegistry.getEntityCache();
     setup(enricherConfig, clientRegistry);
     this.fqnResolver = getFqnResolver();
+    this.backendExcludedDomains =
+        enricherConfig.hasPath(BACKEND_EXCLUDED_DOMAINS)
+            ? enricherConfig.getStringList(BACKEND_EXCLUDED_DOMAINS)
+            : Collections.emptyList();
   }
 
   public abstract void setup(Config enricherConfig, ClientRegistry clientRegistry);
@@ -122,18 +129,6 @@ protected boolean canResolveBackend(StructuredTraceGraph structuredTraceGraph, E
     return true;
   }
 
-  /**
-   * Method to check if backend uri is valid or not. This will enable any custom logic to be
-   * inserted in the implementing classes.
-   *
-   * @param backendURI backend URI information
-   * @return true if backend uri resolution is allowed
-   */
-  protected boolean isValidBackendUri(String backendURI) {
-    // by default allow the backend uri to proceed
-    return true;
-  }
-
   /** Checks if the candidateEntity is indeed a backend Entity */
   private boolean isValidBackendEntity(
       StructuredTrace trace, Event backendSpan, BackendInfo candidateInfo) {
@@ -323,7 +318,7 @@ public Optional<BackendInfo> resolve(
       }
 
       String backendUri = maybeBackendUri.get();
-      if (!isValidBackendUri(backendUri)) {
+      if (isDomainExcluded(event, backendUri)) {
         return Optional.empty();
       }
 
@@ -349,6 +344,13 @@ public Optional<BackendInfo> resolve(
     return Optional.empty();
   }
 
+  private boolean isDomainExcluded(Event event, String backendUri) {
+    String[] hostAndPort = backendUri.split(COLON);
+    String host = fqnResolver.resolve(hostAndPort[0], event);
+    String domain = HttpSemanticConventionUtils.getPrimaryDomain(host);
+    return backendExcludedDomains.contains(domain);
+  }
+
   private Builder getEntityBuilder(
       StructuredTrace trace, Event event, BackendType type, String backendUri) {
     try {

semantic-convention-utils/src/main/java/org/hypertrace/semantic/convention/utils/http/HttpSemanticConventionUtils.java

@@ -44,6 +44,7 @@
 import com.google.common.base.Splitter;
 import com.google.common.collect.Lists;
 import com.google.common.collect.Sets;
+import com.google.common.net.InternetDomainName;
 import io.micrometer.core.instrument.util.StringUtils;
 import java.net.HttpCookie;
 import java.net.MalformedURLException;
@@ -230,6 +231,15 @@ public static Optional<String> getEnvironmentForSpan(Event event) {
             event, OTelDeploymentSemanticConventions.DEPLOYMENT_ENVIRONMENT.getValue()));
   }
 
+  public static String getPrimaryDomain(String host) {
+    try {
+      return InternetDomainName.from(host).topPrivateDomain().toString();
+    } catch (Exception exception) {
+      LOGGER.error("Error while extracting the primary domain from the host {} ", host, exception);
+      return host;
+    }
+  }
+
   /**
    * OTel mandates one of the following set to be present for http server span - http.scheme,
    * http.host, http.target - http.scheme, http.server_name, net.host.port, http.target -

semantic-convention-utils/src/test/java/org/hypertrace/semantic/convention/utils/http/HttpSemanticConventionUtilsTest.java

@@ -591,4 +591,15 @@ public void testGetPathFromUrl() {
             "/api/v1/gatekeeper/check?url=%2Fpixel%2Factivities%3Fadvertisable%3DTRHRT&method=GET&service=pixel");
     Assertions.assertEquals(path.get(), "/api/v1/gatekeeper/check");
   }
+
+  @Test
+  public void testGetPrimaryDomain() {
+    Assertions.assertEquals(HttpSemanticConventionUtils.getPrimaryDomain("www.xyz.com"), "xyz.com");
+    Assertions.assertEquals(
+        HttpSemanticConventionUtils.getPrimaryDomain("www.abc.xyz.com"), "xyz.com");
+    Assertions.assertEquals(HttpSemanticConventionUtils.getPrimaryDomain("abc.xyz.com"), "xyz.com");
+    Assertions.assertEquals(HttpSemanticConventionUtils.getPrimaryDomain("xyz.com"), "xyz.com");
+
+    Assertions.assertEquals(HttpSemanticConventionUtils.getPrimaryDomain("10.0.0.0"), "10.0.0.0");
+  }
 }