Merge branch 'master' into feature/467-sbom-creation-per-pr

Author: lsulak

agent/src/main/resources/reference.conf

@@ -14,6 +14,11 @@
 # dispatcher to be used (http, console, capture)
 atum.dispatcher.type="http"
 
+# Optional author/createdBy identity used for auditing (partitionings, checkpoints, additional data).
+# If unset, it falls back to System.getProperty("user.name"). Useful when the JVM user is a generic
+# system account (e.g. "yarn") and the application name is more meaningful.
+#atum.author="my-application-name"
+
 # The REST API URI of the atum server
 #atum.dispatcher.http.url=
 

agent/src/main/scala/za/co/absa/atum/agent/AtumAgent.scala

@@ -32,14 +32,26 @@ trait AtumAgent {
   val dispatcher: Dispatcher
 
   /**
-   *  Returns a user under whose security context the JVM is running.
-   *  Its purpose is for auditing in author/createdBy fields.
+   *  The user used for auditing in author/createdBy fields.
    *
-   *  Important: It's not supposed to be used for authorization as it can be spoofed!
+   *  It is resolved once per agent (on first access) and then cached for the agent's lifetime:
+   *    - if the `atum.author` configuration key is set (e.g. in `application.conf` or via the
+   *      `-Datum.author=...` JVM system property), that value is used - this is useful when the JVM
+   *      user is a generic system account (e.g. `yarn`) and the application name is more meaningful;
+   *    - otherwise it falls back to the user under whose security context the JVM is running
+   *      (`System.getProperty("user.name")`), which is platform independent.
+   *
+   *  Overriding this method replaces the resolution strategy entirely (see `AtumAgent.fromConfig`).
    *
-   *  @return Current user.
+   *  Important: It's not supposed to be used for authorization as it can be spoofed!
    */
-  private[agent] def currentUser: String = System.getProperty("user.name") // platform independent
+  private[agent] def currentUser: String = resolvedCurrentUser
+
+  // Cached, resolved-once default author identity. Ensures that agents which do not override
+  // `currentUser` keep a stable audit identity for their whole lifetime, even if the Typesafe
+  // Config caches are later invalidated or backing system properties change. The actual
+  // configuration loading is owned by the companion object (see `AtumAgent.defaultAuthor`).
+  private[this] lazy val resolvedCurrentUser: String = AtumAgent.defaultAuthor
 
   /**
    *  Sends `CheckpointDTO` to the AtumService API
@@ -139,6 +151,9 @@ object AtumAgent extends AtumAgent {
 
   override val dispatcher: Dispatcher = dispatcherFromConfig()
 
+  // `currentUser` is intentionally not overridden here: the trait default already delegates to
+  // `defaultAuthor` (below) and caches the result once, which is exactly what the singleton needs.
+
   private[agent] def dispatcherFromConfig(config: Config = ConfigFactory.load()): Dispatcher = {
     config.getString("atum.dispatcher.type") match {
       case "http" => new HttpDispatcher(config)
@@ -148,7 +163,33 @@ object AtumAgent extends AtumAgent {
     }
   }
 
+  /**
+   *  The default author (createdBy) identity, resolved from the globally loaded configuration.
+   *  Callers cache the result (see the trait's `currentUser`), so this is evaluated once per agent.
+   *
+   *  @return the default author identity for agents that do not override `currentUser`.
+   */
+  private[agent] def defaultAuthor: String = resolveAuthor(ConfigFactory.load())
+
+  /**
+   *  Resolves the author (createdBy) identity used for auditing from the given configuration.
+   *
+   *  If the optional `atum.author` key is present and non-blank, its (trimmed) value is used;
+   *  otherwise it falls back to `System.getProperty("user.name")`.
+   *
+   *  @param config configuration to read the optional `atum.author` key from.
+   *  @return the resolved author identity.
+   */
+  private[agent] def resolveAuthor(config: Config): String = {
+    if (config.hasPath("atum.author") && config.getString("atum.author").trim.nonEmpty) {
+      config.getString("atum.author").trim
+    } else {
+      System.getProperty("user.name") // platform independent
+    }
+  }
+
   def fromConfig(config: Config): AtumAgent = new AtumAgent {
     override val dispatcher: Dispatcher = dispatcherFromConfig(config)
+    override val currentUser: String = resolveAuthor(config)
   }
 }

agent/src/test/scala/za/co/absa/atum/agent/AtumAgentUnitTests.scala

@@ -194,6 +194,84 @@ class AtumAgentUnitTests extends AnyFunSuiteLike {
     assert(agentA.dispatcher ne agentB.dispatcher)
   }
 
+  test("currentUser falls back to the JVM user when atum.author is not set") {
+    val agent = AtumAgent.fromConfig(configOf(Map(
+      "atum.dispatcher.type" -> "capture",
+      "atum.dispatcher.capture.capture-limit" -> 10
+    )))
+
+    assert(agent.currentUser == System.getProperty("user.name"))
+  }
+
+  test("currentUser uses atum.author from config when set (trimmed)") {
+    val agent = AtumAgent.fromConfig(configOf(Map(
+      "atum.dispatcher.type" -> "capture",
+      "atum.dispatcher.capture.capture-limit" -> 10,
+      "atum.author" -> "  my-application  "
+    )))
+
+    assert(agent.currentUser == "my-application")
+  }
+
+  test("currentUser falls back to the JVM user when atum.author is blank") {
+    val agent = AtumAgent.fromConfig(configOf(Map(
+      "atum.dispatcher.type" -> "capture",
+      "atum.dispatcher.capture.capture-limit" -> 10,
+      "atum.author" -> "   "
+    )))
+
+    assert(agent.currentUser == System.getProperty("user.name"))
+  }
+
+  test("currentUser is resolved independently per config-backed agent") {
+    val agentA = AtumAgent.fromConfig(configOf(Map(
+      "atum.dispatcher.type" -> "capture",
+      "atum.dispatcher.capture.capture-limit" -> 10,
+      "atum.author" -> "alice"
+    )))
+    val agentB = AtumAgent.fromConfig(configOf(Map(
+      "atum.dispatcher.type" -> "capture",
+      "atum.dispatcher.capture.capture-limit" -> 10
+    )))
+
+    assert(agentA.currentUser == "alice")
+    assert(agentB.currentUser == System.getProperty("user.name"))
+  }
+
+  test("currentUser is resolved once and cached for agents that do not override it") {
+    val originalAuthor = Option(System.getProperty("atum.author"))
+    try {
+      System.setProperty("atum.author", "first-app")
+      ConfigFactory.invalidateCaches()
+
+      // a custom agent that does NOT override currentUser -> relies on the cached trait default
+      val agent = new AtumAgent {
+        override val dispatcher: CapturingDispatcher =
+          AtumAgent.dispatcherFromConfig(configOf(Map(
+            "atum.dispatcher.type" -> "capture",
+            "atum.dispatcher.capture.capture-limit" -> 10
+          ))).asInstanceOf[CapturingDispatcher]
+      }
+
+      // first access resolves and caches the identity
+      assert(agent.currentUser == "first-app")
+
+      // change the backing system property and invalidate Typesafe caches:
+      // a recomputing `def` would observe the new value here
+      System.setProperty("atum.author", "second-app")
+      ConfigFactory.invalidateCaches()
+
+      // the resolved-once value must remain stable for the agent's lifetime
+      assert(agent.currentUser == "first-app")
+    } finally {
+      originalAuthor match {
+        case Some(value) => System.setProperty("atum.author", value)
+        case None        => System.clearProperty("atum.author")
+      }
+      ConfigFactory.invalidateCaches()
+    }
+  }
+
   private def configOf(configValues: Map[String, Any]): Config = {
     val emptyConfig = ConfigFactory.empty()
     configValues.foldLeft(emptyConfig) { case (acc, (configKey, value)) =>