WX-744 Optionally rewrite blob paths to appear as local paths

build.sbt

@@ -245,6 +245,7 @@ lazy val tesBackend = (project in backendRoot / "tes")
   .dependsOn(sfsBackend)
   .dependsOn(ftpFileSystem)
   .dependsOn(drsFileSystem)
+  .dependsOn(azureBlobFileSystem)
   .dependsOn(backend % "test->test")
   .dependsOn(common % "test->test")
 

supportedBackends/tes/src/main/scala/cromwell/backend/impl/tes/TesAsyncBackendJobExecutionActor.scala

@@ -80,6 +80,14 @@ class TesAsyncBackendJobExecutionActor(override val standardParams: StandardAsyn
 
   private val tesEndpoint = workflowDescriptor.workflowOptions.getOrElse("endpoint", tesConfiguration.endpointURL)
 
+  // Temporary support for configuring the format we use to send BlobPaths to TES.
+  // Added 10/2022 as a workaround for the CromwellOnAzure TES server expecting
+  // blob containers to be mounted via blobfuse rather than addressed natively.
+  private val transformBlobToLocalPaths: Boolean =
+    configurationDescriptor.backendConfig
+      .getAs[Boolean]("transform-blob-to-local-path")
+      .getOrElse(false)
+
   override lazy val jobTag: String = jobDescriptor.key.tag
 
   private val outputMode = validate {
@@ -148,7 +156,7 @@ class TesAsyncBackendJobExecutionActor(override val standardParams: StandardAsyn
         mode)
     })
 
-    tesTask.map(TesTask.makeTask)
+    tesTask.map(TesTask.makeTask(_, transformBlobToLocalPaths))
   }
 
   def writeScriptFile(): Future[Unit] = {

supportedBackends/tes/src/main/scala/cromwell/backend/impl/tes/TesTask.scala

@@ -258,21 +258,37 @@ object TesTask {
     )
   }
 
-  def makeTask(tesTask: TesTask): Task = {
+  def makeTask(tesTask: TesTask, transformBlobToLocalPath: Boolean = false): Task = {
+    val inputs = if (transformBlobToLocalPath) transformInputs(tesTask.inputs) else tesTask.inputs
+    val outputs = if (transformBlobToLocalPath) transformOutputs(tesTask.outputs) else tesTask.outputs
     Task(
       id = None,
       state = None,
       name = Option(tesTask.name),
       description = Option(tesTask.description),
-      inputs = Option(tesTask.inputs),
-      outputs = Option(tesTask.outputs),
+      inputs = Option(inputs),
+      outputs = Option(outputs),
       resources = Option(tesTask.resources),
       executors = tesTask.executors,
       volumes = None,
       tags = Option(tesTask.jobDescriptor.workflowDescriptor.customLabels.asMap),
       logs = None
     )
   }
+
+  def transformInputs(inputs: Seq[Input]): Seq[Input] = inputs.map(i =>
+    i.copy(url=i.url.map(transformBlobString))
+  )
+
+  def transformOutputs(outputs: Seq[Output]): Seq[Output] = outputs.map(i =>
+    i.copy(url=i.url.map(transformBlobString))
+  )
+
+  val blobSegment = ".blob.core.windows.net"
+  def transformBlobString(s: String): String = if (s.contains(blobSegment)) {
+    s.replaceFirst("https:/", "").replaceFirst(blobSegment, "")
+  } else s
+
 }
 
 // Field requirements in classes below based off GA4GH schema

supportedBackends/tes/src/test/scala/cromwell/backend/impl/tes/TesTaskSpec.scala

@@ -88,4 +88,28 @@ class TesTaskSpec
 
     task.tags shouldBe Option(Map("foo" -> "bar"))
   }
+
+  it should "apply the correct transformation to a blob path" in {
+    val orig = "https://coaexternalstorage.blob.core.windows.net/cromwell/mydir/myfile.txt"
+    TesTask.transformBlobString(orig) shouldBe "/coaexternalstorage/cromwell/mydir/myfile.txt"
+  }
+
+  it should "not transform a non-blob path" in {
+    val orig = "https://some-bogus-url.test/cromwell/mydir/myfile.txt"
+    TesTask.transformBlobString(orig) shouldBe orig
+  }
+
+  it should "transform inputs" in {
+    val baseInput = Input(
+      Option("name"),
+      Option("descr"),
+      Option("https://coaexternalstorage.blob.core.windows.net/cromwell/mydir/myfile.txt"),
+      "path",
+      Option("type"),
+      Option("content"),
+    )
+    val inputs = Option(Seq(baseInput))
+    val outcome = inputs.map(TesTask.transformInputs)
+    outcome.get.head.url.get shouldBe "/coaexternalstorage/cromwell/mydir/myfile.txt"
+  }
 }