XML Processing for ZIO


Maven Central

This library provides non-blocking parsers, writers and filters for handling streaming XML in the zio Scala framework, specifically as ZStream. Parsing is done by wrapping the Aalto XML parser. Writing uses the standard Java XMLOutputFactory mechanism (writing to a byte array which is known not to block).

Currently, ZIO 2.0+ is targeted.


A stream of XML is modeled as a ZStream[Any, XMLStreamException, XmlEvent], where XmlEvent is a sealed trait that closely follows Java's own XmlEvent structure. A notable exception is that StartDocument and EndDocument as absent, since start- and end of a document is already indicated by stream semantics themselves.


The XmlParser object provides a ZPipeline that can turn bytes into XML events:

object XmlParser {
  def parser(ignoreInvalidChars: Boolean = false): ZPipeline[Any, XMLStreamException, Byte, XmlEvent]

If you have a ZStream[Any, Nothing, Byte], you can feed that into the pipeline as follows:

val myStream: ZStream[Any, Nothing, Byte] = ???
val events: ZStream[Any, XMLStreamException, XmlEvent] = myStream >>> XmlParser.parser()


Several ways are available to turn XML events back into bytes or DOM-like data structures.

Writing to a document tree

Two ZPipeline variants exist that emit a document tree after a tag (and children) has been written:

object XmlWriter {
  def collectNode(): ZPipeline[Any, XMLStreamException, XmlEvent, scala.xml.Node]
  def collectElement(): ZPipeline[Any, XMLStreamException, XmlEvent, org.w3c.dom.Element]

The former emits a Scala XML Node, the latter emits a DOM Element. Use the variant that matches other libraries you're working with.

Writing to bytes

You can also just write XML back to bytes, using another ZPipeline in XmlWriter.

object XmlWriter {
  def writeDocument(charset: Charset = StandardCharsets.UTF_8): ZPipeline[Any, XMLStreamException, XmlEvent, Byte]
  def writeFragment(charset: Charset = StandardCharsets.UTF_8): ZPipeline[Any, XMLStreamException, XmlEvent, Byte]

Two variants are available. You'll pick one depending on whether you plan to write a single document (writeDocument) or potentially multiple root nodes as an XML fragment (writeFragment).


In addition to parsing and writing, a few filters are presented that have proven useful as glue logic. See their ScalaDoc for details. Combined with XmlWriter.collectNode, they can be used to gather up pieces of a large XML stream for piece-meal further processing.

object XmlFilter {
  /** Filters subtrees of nodes residing in the XML document at the direct ancestors given in [path].  The
    * subtrees will have the last element of [path] as their parent. Higher ancestors are filtered out.  For
    * example, filterSubtree("foo" :: "bar" :: Nil), given <xml><foo><bar>1</bar><hello/><bar>2</bar></xml>,
    * will emit events for <bar>1</bar><bar>2</bar>.
  def filterSubtree(path: Seq[String]): ZPipeline[Any, Nothing, XmlEvent, XmlEvent]

  /** Filters subtrees of nodes in the XML with the given name, at any path.  The subtrees will have [tagName]
    * as their parent (ancestors are filtered out).
  def filterTag(tagName: String): ZPipeline[Any, Nothing, XmlEvent, XmlEvent]

  /** Removes nodes with the given name, and all of their children, from the stream. The node may occur at any
    * level. The rest of the stream is passed through unchanged. */
  def filterTagNot(tagName: String): ZPipeline[Any, Nothing, XmlEvent, XmlEvent]


When writing XML that should be easily readable by humans, it can be convenient to add indentation to make the nesting of XML elements easier to follow. A ZPipeline is provided that will re-indent an XML stream on the fly.

object XmlIndenter {
  /** Indents a stream of XML parse events (removing any previous indentation first) */
  def indent(amount: Int = 2): ZPipeline[Any, Nothing, XmlEvent, XmlEvent]