WX-1137 Replace 4-byte unicode chars in PAPI event descriptions

supportedBackends/google/pipelines/v2beta/src/main/scala/cromwell/backend/google/pipelines/v2beta/PipelinesUtilityConversions.scala

@@ -60,8 +60,16 @@ trait PipelinesUtilityConversions {
       Option(event.getDescription) collect { case d if d.startsWith(s"$k pulling") => "Pulling" + d.substring(s"$k pulling".length)}
     } headOption
 
+    // See WX-1137. ContainerStoppedEvent descriptions contain the stderr, which may include 4-byte unicode
+    // characters (typically emoji). Some databases have trouble storing these; replace them with the standard
+    // "unknown character" unicode symbol.
+    val name = Option(event.getContainerStopped) match {
+      case Some(_) => cleanUtf8mb4(event.getDescription)
+      case _ => event.getDescription
+    }
+
     ExecutionEvent(
-      name = event.getDescription,
+      name = name,
       offsetDateTime = OffsetDateTime.parse(event.getTimestamp),
       grouping = groupingFromAction.orElse(groupingFromPull)
     )
@@ -91,4 +99,10 @@ object PipelinesUtilityConversions {
       }
     }
   }
+
+  lazy val utf8mb4Regex = "[\\x{10000}-\\x{FFFFF}]"
+  lazy val utf8mb3Replacement = "\uFFFD"  // This is the standard char for replacing invalid/unknown unicode chars
+  def cleanUtf8mb4(in: String): String = {
+    in.replaceAll(utf8mb4Regex, utf8mb3Replacement)
+  }
 }

supportedBackends/google/pipelines/v2beta/src/test/scala/cromwell/backend/google/pipelines/v2beta/PipelinesUtilityConversionsSpec.scala

@@ -0,0 +1,26 @@
+package cromwell.backend.google.pipelines.v2beta
+
+import org.scalatest.flatspec.AnyFlatSpec
+import org.scalatest.matchers.should.Matchers
+
+class PipelinesUtilityConversionsSpec extends AnyFlatSpec with Matchers {
+  behavior of "PipelinesUtilityConversions"
+
+  it should "not modify strings that contain only ascii characters" in {
+    val input = "hi there!?"
+    PipelinesUtilityConversions.cleanUtf8mb4(input) shouldBe input
+  }
+
+  it should "not modify strings with 3-byte unicode characters" in {
+    val input = "Here is my non-ascii character: \u1234 Do you like it?"
+    PipelinesUtilityConversions.cleanUtf8mb4(input) shouldBe input
+  }
+
+  it should "replace 4-byte unicode characters" in {
+    val cry = new String(Character.toChars(Integer.parseInt("1F62D", 16)))
+    val barf = new String(Character.toChars(Integer.parseInt("1F92E", 16)))
+    val input = s"When I try to put an emoji in the database it $barf and then I $cry"
+    val cleaned = "When I try to put an emoji in the database it \uFFFD and then I \uFFFD"
+    PipelinesUtilityConversions.cleanUtf8mb4(input) shouldBe cleaned
+  }
+}