Allow document type definitions to be included when parsing XML for b…

…ody matching Fixes: issue ITV#237
danielbeddoe123 · Nov 18, 2022 · 143fe43 · 143fe43
1 parent 211abd3
commit 143fe43
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 1 deletion.
diff --git a/scalapact-shared/src/main/scala/com/itv/scalapact/shared/matchir/MatchIr.scala b/scalapact-shared/src/main/scala/com/itv/scalapact/shared/matchir/MatchIr.scala
@@ -8,10 +8,15 @@ import com.itv.scalapact.shared.utils.Helpers.{
   safeStringToDouble
 }
 
+import javax.xml.parsers.SAXParserFactory
 import scala.util.Try
-import scala.xml.{Elem, Node, XML}
+import scala.xml.factory.XMLLoader
+import scala.xml.{Elem, Node}
 
 object MatchIr {
+
+  private val XML: XMLLoader[Elem] = createXMLParser()
+
   def fromXmlString(xmlString: String): Option[IrNode] =
     Try(XML.loadString(xmlString)).toOption.map(fromXml)
 
@@ -21,6 +26,13 @@ object MatchIr {
   def fromJSON(fromJson: String => Option[IrNode])(jsonString: String): Option[IrNode] =
     fromJson(jsonString)
 
+  private def createXMLParser(): XMLLoader[Elem] = {
+    val sAXParserFactory = SAXParserFactory.newInstance()
+    sAXParserFactory.setFeature("http://apache.org/xml/features/disallow-doctype-decl", false);
+    val saxParser = sAXParserFactory.newSAXParser()
+    xml.XML.withSAXParser(saxParser)
+  }
+
   private def convertAttributes(attributes: Map[String, String], pathToParent: IrNodePath): IrNodeAttributes =
     attributes.toList.reverse
       .map {

diff --git a/tests-with-deps/src/test/scala/com/itv/scalapact/shared/matchir/MatchIrSpec.scala b/tests-with-deps/src/test/scala/com/itv/scalapact/shared/matchir/MatchIrSpec.scala
@@ -97,6 +97,28 @@ class MatchIrSpec extends AnyFunSpec with Matchers {
 
     }
 
+    it("should be able to convert xml with a document type definition") {
+      val xml: String =
+        """<?xml version="1.0" encoding="UTF-8"?><!DOCTYPE animals [<!ELEMENT animals (alligator)*><!ELEMENT alligator (#PCDATA)><!ATTLIST alligator name CDATA #REQUIRED>]><animals><alligator name="Mary"/></animals>""".stripMargin
+
+      val ir: IrNode =
+        IrNode(
+          "animals",
+          IrNode("alligator")
+            .withAttributes(
+              IrNodeAttributes(
+                Map(
+                  "name" -> IrNodeAttribute(IrStringNode("Mary"), IrNodePathEmpty <~ "animals" <~ "alligator" <@ "name")
+                )
+              )
+            )
+            .withPath(IrNodePathEmpty <~ "animals" <~ "alligator")
+            .markAsXml
+        ).withPath(IrNodePathEmpty <~ "animals").markAsXml
+
+      check(MatchIr.fromXmlString(xml).get =<>= ir)
+    }
+
     it("should be able to convert one node with content") {
 
       val xml: String = <fish>haddock</fish>.toString()