sbt-assembly merge strategy for aop.xml files

gistfile1.scala

// Create a new MergeStrategy for aop.xml files
val aopMerge: MergeStrategy = new MergeStrategy {
  val name = "aopMerge"
  import scala.xml._
  import scala.xml.dtd._

  def apply(tempDir: File, path: String, files: Seq[File]): Either[String, Seq[(File, String)]] = {
    val dt = DocType("aspectj", PublicID("-//AspectJ//DTD//EN", "http://www.eclipse.org/aspectj/dtd/aspectj.dtd"), Nil)
    val file = MergeStrategy.createMergeTarget(tempDir, path)
    val xmls: Seq[Elem] = files.map(XML.loadFile)
    val aspectsChildren: Seq[Node] = xmls.flatMap(_ \\ "aspectj" \ "aspects" \ "_")
    val weaverChildren: Seq[Node] = xmls.flatMap(_ \\ "aspectj" \ "weaver" \ "_")
    val options: String = xmls.map(x => (x \\ "aspectj" \ "weaver" \ "@options").text).mkString(" ").trim
    val weaverAttr = if (options.isEmpty) Null else new UnprefixedAttribute("options", options, Null)
    val aspects = new Elem(null, "aspects", Null, TopScope, false, aspectsChildren: _*)
    val weaver = new Elem(null, "weaver", weaverAttr, TopScope, false, weaverChildren: _*)
    val aspectj = new Elem(null, "aspectj", Null, TopScope, false, aspects, weaver)
    XML.save(file.toString, aspectj, "UTF-8", xmlDecl = false, dt)
    IO.append(file, IO.Newline.getBytes(IO.defaultCharset))
    Right(Seq(file -> path))
  }
}

// Use defaultMergeStrategy with a case for aop.xml
// I like this better than the inline version mentioned in assembly's README
val customMergeStrategy: String => MergeStrategy = {
  case PathList("META-INF", "aop.xml") =>
    aopMerge
  case s =>
    defaultMergeStrategy(s)
}

// Use the customMergeStrategy in your settings
mergeStrategy in assembly := customMergeStrategy

Thank you @colestanfield. I had a

case "META-INF/aop.xml"                                            => MergeStrategy.first

and it wouldn't work. Your aopMerge strategy works just fine!

Thanks @colestanfield... this works great! I would modify the first line to

val aopMerge = new sbtassembly.MergeStrategy {

so folks don't need to import MergeStrategy.

ok first of all: major kudos to @colestanfield! 'found this thread because i ran into the same issue with aop.xml with Kamon libraries ...this is probably a stupid question: i copied the defaultMergeStrategy from the sbt-assembly README file because of missing reference to this function error...am i supposed to do that or am i supposed to add some import?

I wasn't too thrilled about putting this code in build.sbt...you can't add this to any .sbt file under project folder because sbt don't allow imports (at least for the current 0.13.11 version)...also, with the autoPlugins support...having a separate .sbt file (except for specific plugins that don't support autoplugin trait) is actually discouraged (i.e. you add sbt-assembly plugin in the plugins.sbt).

anyway, so i created a project/AssemblyMergeStrategies.scala and import that in the build.sbt file so i can use the assemblyMergeStrategy in assembly := customMergeStrategy line

BTW mergeStrategy is already deprecated; use assemblyMergeStrategy instead

It worked .. thanks @colestanfield for nice finding

I've tried using this snippet and for some reason sbt assembly suspends while trying to read an XML file from a directory target/streams/$global making us unable to use that - does anyone knows what may be causing this to try and read this directory or why is it created?

Hello @colestanfield
I notice that with sbt 1.4.9 this cause some problem
The error message during the build is
org.xml.sax.SAXParseExceptionpublicId: -//AspectJ//DTD//EN; systemId: http://www.eclipse.org/aspectj/dtd/aspectj.dtd; lineNumber: 1; columnNumber: 2; The markup declarations contained or pointed to by the document type declaration must be well-formed.

With sbt 1.10.11 and sbt-assembly 2.3.1 this can be

lazy val aopMerge = CustomMergeStrategy("aopMerge") { (conflicts: Vector[Dependency]) =>
  import scala.xml.*
  import scala.xml.dtd.*

  val parser = {
    val factory = javax.xml.parsers.SAXParserFactory.newInstance
    factory.setFeature("http://apache.org/xml/features/nonvalidating/load-external-dtd", false)
    factory.newSAXParser
  }
  val xmls: Seq[Elem] = conflicts.map { conflict =>
    Using.resource((dependency: Dependency) => dependency.stream())(conflict) { is: InputStream =>
      XML.loadXML(Source.fromInputStream(is), parser)
      // XML.load(is)
    }
  }

  val dt = DocType("aspectj", PublicID("-//AspectJ//DTD//EN", "https://www.eclipse.org/aspectj/dtd/aspectj.dtd"), Nil)
  val aspectsChildren: Seq[Node] = xmls.flatMap(_ \\ "aspectj" \ "aspects" \ "_")
  val weaverChildren: Seq[Node] = xmls.flatMap(_ \\ "aspectj" \ "weaver" \ "_")
  val options: String = xmls.map(x => (x \\ "aspectj" \ "weaver" \ "@options").text).mkString(" ").trim
  val weaverAttr = if (options.isEmpty) Null else new UnprefixedAttribute("options", options, Null)
  val aspects = new Elem(null, "aspects", Null, TopScope, false, aspectsChildren: _*)
  val weaver = new Elem(null, "weaver", weaverAttr, TopScope, false, weaverChildren: _*)
  val aspectj = new Elem(null, "aspectj", Null, TopScope, false, aspects, weaver)

  val bytes = scala.util.Using.resource(new CharArrayWriter()) { writer =>
    XML.write(writer, aspectj, "UTF-8", xmlDecl = false, dt)
    (writer.toString + IO.Newline).getBytes(IO.defaultCharset)
  }

  Right(Vector(JarEntry(conflicts.head.target, () => new java.io.ByteArrayInputStream(bytes))))
}