Backtestsjsouprun-20261006-181938#2504
W3CDom serializes plain XML with the reserved XML namespace
Our patch was judged better than the merged fix.
In this judgment, Patch A = our agent, Patch B = the merged fix.
Both patches fix the root cause the same way. When the W3C element is created, they map jsoup's internal plain-XML sentinel namespace (Parser.NamespaceXml) to null. Explicit namespaces are kept, and xml:* attributes are left alone. The main difference is how they treat prefixed elements that sit in the sentinel namespace without a declaration (e.g. `<foo:bar>`). - **Patch A** nulls the namespace unless the tag name starts with `xml:`. An undeclared `foo:bar` then goes through the existing imputed-empty-namespace path instead of being bound to the reserved XML namespace, and `xml:baz` correctly keeps the reserved namespace. - **Patch B** only nulls when `tag().prefix()` is empty. An undeclared `foo:bar` therefore still asks for createElementNS(XML-namespace, "foo:bar"). That is invalid per the Namespaces spec, so it either throws and falls back or keeps the reserved namespace on the element. Patch B also depends on `Tag.prefix()`, which is not shown in the diff. **Tests.** Patch A's tests are broader. They cover: - the exact issue example, round-tripped through a namespace-aware parser; - the static `convert` path; - xml:lang preserved alongside explicit default and prefixed namespaces; - undeclared prefixes and `xml:`-prefixed elements. Patch B's tests are solid but narrower: plain XML with xml:lang, and explicit default namespaces including `xmlns=""`. They do not cover the undeclared-prefix edge. **Other differences.** Patch B adds a CHANGES entry and a tidy helper method with @Nullable, which is nice craft but secondary. Overall, Patch A handles more of the edge cases correctly and verifies them more thoroughly.
The judge saw only the two patches, labelled A and B in random order, with issue links and version numbers removed. The reasoning above is shown verbatim.
Solve
18m 44s
Wall-clock time in the container
Full test suite
No new failures
2,018 passed · 0 skipped · 2,018 total
Contamination audit
Clean
No sign the agent saw the real fix
Patch size
82vs 59 merged
Changed lines, ours vs merged
The issue
View on GitHubPlain XML parsed with Parser.xmlParser() should round-trip through the default namespace-aware W3CDom conversion as parseable XML. But for an ordinary no-namespace XML document, W3CDom currently serializes jsoup's internal XML parser namespace as the document's default namespace.
For example, converting <root><child>One</child></root> through new W3CDom().fromJsoup(doc) and W3CDom.asString(..., W3CDom.OutputXml()) emits <root xmlns="http://www.w3.org/XML/1998/namespace">.... That namespace is reserved for the xml prefix and cannot be bound as the default namespace, so a normal XML parser rejects the output.
Fix is to keep explicit source namespaces as W3C DOM namespaces, but map our internal plain-XML sentinel namespace to no namespace when creating unprefixed W3C elements. xml:* attributes such as xml:lang should continue to use the reserved XML namespace.
The two patches
src/main/java/org/jsoup/helper/W3CDom.javaindex b26cb93f..d505961e 100644@@ -4,6 +4,7 @@ import org.jsoup.internal.Normalizer; import org.jsoup.internal.StringUtil; import org.jsoup.nodes.Attribute; import org.jsoup.parser.HtmlTreeBuilder;+import org.jsoup.parser.Parser; import org.jsoup.select.NodeVisitor; import org.jsoup.select.Selector; import org.w3c.dom.Comment;@@ -360,6 +361,9 @@ public class W3CDom { org.jsoup.nodes.Element sourceEl = (org.jsoup.nodes.Element) source; String namespace = namespaceAware ? sourceEl.tag().namespace() : null; String tagName = Normalizer.xmlSafeTagName(sourceEl.tagName());+ // the XML parser's internal default namespace is reserved for the xml: prefix; other elements without a declared namespace are in no namespace+ if (Parser.NamespaceXml.equals(namespace) && !tagName.startsWith("xml:"))+ namespace = null; try { // use an empty namespace if none is present but the tag name has a prefix String imputedNamespace = namespace == null && tagName.contains(":") ? "" : namespace;src/test/java/org/jsoup/helper/W3CDomTest.javaindex b9d1d330..4acdab36 100644@@ -518,6 +518,84 @@ public class W3CDomTest { assertEquals("<?xml version=\"1.0\" encoding=\"UTF-8\"?><html xmlns=\"http://www.w3.org/1999/xhtml\"><head/><body><div xmlns:v-bind=\"undefined\" v-bind:class=\"test\"><span v-bind:style=\"color:red\"/></div></body></html>", xml); } + private static Document reparseNamespaceAware(String xml) throws Exception {+ DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();+ factory.setNamespaceAware(true);+ return factory.newDocumentBuilder().parse(new InputSource(new StringReader(xml)));+ }++ @Test void plainXmlHasNoNamespaceAndRoundTrips() throws Exception {+ org.jsoup.nodes.Document jdoc = Jsoup.parse("<root><child>One</child></root>", "", Parser.xmlParser());+ Document w3cDoc = new W3CDom().fromJsoup(jdoc);++ org.w3c.dom.Element root = w3cDoc.getDocumentElement();+ assertNull(root.getNamespaceURI());+ assertNull(root.getFirstChild().getNamespaceURI());++ String out = W3CDom.asString(w3cDoc, W3CDom.OutputXml());+ assertFalse(out.contains("xmlns"), out);+ assertTrue(out.contains("<root><child>One</child></root>"), out);++ Document reparsed = reparseNamespaceAware(out); // would throw if the reserved xml namespace was bound as default+ assertEquals("root", reparsed.getDocumentElement().getLocalName());+ assertNull(reparsed.getDocumentElement().getNamespaceURI());+ assertEquals("One", reparsed.getDocumentElement().getTextContent());+ }++ @Test void plainXmlViaStaticConvertRoundTrips() throws Exception {+ org.jsoup.nodes.Document jdoc = Jsoup.parse("<root><child>One</child></root>", "", Parser.xmlParser());+ Document w3cDoc = W3CDom.convert(jdoc);+ assertNull(w3cDoc.getDocumentElement().getNamespaceURI());+ String out = W3CDom.asString(w3cDoc, W3CDom.OutputXml());+ assertFalse(out.contains(Parser.NamespaceXml), out);+ reparseNamespaceAware(out);+ }++ @Test void plainXmlKeepsXmlPrefixedAttributesAndExplicitNamespaces() throws Exception {+ String xml = "<root xml:lang=\"en\"><a xmlns=\"urn:a\"><b/></a><x:c xmlns:x=\"urn:x\"/><d/></root>";+ org.jsoup.nodes.Document jdoc = Jsoup.parse(xml, "", Parser.xmlParser());+ Document w3cDoc = new W3CDom().fromJsoup(jdoc);++ org.w3c.dom.Element root = w3cDoc.getDocumentElement();+ assertNull(root.getNamespaceURI());+ assertEquals("en", root.getAttributeNS(Parser.NamespaceXml, "lang"));++ NodeList kids = root.getChildNodes();+ assertEquals("urn:a", kids.item(0).getNamespaceURI());+ assertEquals("urn:a", kids.item(0).getFirstChild().getNamespaceURI());+ assertEquals("urn:x", kids.item(1).getNamespaceURI());+ assertNull(kids.item(2).getNamespaceURI());++ String out = W3CDom.asString(w3cDoc, W3CDom.OutputXml());+ assertFalse(out.contains("xmlns=\"" + Parser.NamespaceXml), out);+ Document reparsed = reparseNamespaceAware(out);+ org.w3c.dom.Element rRoot = reparsed.getDocumentElement();+ assertNull(rRoot.getNamespaceURI());+ assertEquals("en", rRoot.getAttributeNS(Parser.NamespaceXml, "lang"));+ assertEquals("urn:a", rRoot.getChildNodes().item(0).getNamespaceURI());+ assertEquals("urn:x", rRoot.getChildNodes().item(1).getNamespaceURI());+ assertNull(rRoot.getChildNodes().item(2).getNamespaceURI());+ }++ @Test void plainXmlUndeclaredPrefixNotBoundToXmlNamespace() {+ org.jsoup.nodes.Document jdoc = Jsoup.parse("<root><foo:bar>x</foo:bar><xml:baz/></root>", "", Parser.xmlParser());+ Document w3cDoc = new W3CDom().fromJsoup(jdoc);++ NodeList kids = w3cDoc.getDocumentElement().getChildNodes();+ Node fooBar = kids.item(0);+ assertEquals(Node.ELEMENT_NODE, fooBar.getNodeType());+ assertEquals("foo:bar", fooBar.getNodeName());+ assertNotEquals(Parser.NamespaceXml, fooBar.getNamespaceURI());++ Node xmlBaz = kids.item(1); // the xml: prefix keeps its reserved namespace+ assertEquals("xml:baz", xmlBaz.getNodeName());+ assertEquals(Parser.NamespaceXml, xmlBaz.getNamespaceURI());++ String out = W3CDom.asString(w3cDoc, W3CDom.OutputXml());+ assertFalse(out.contains(Parser.NamespaceXml), out);+ assertTrue(out.contains("<root><foo:bar>x</foo:bar><xml:baz/></root>"), out);+ }+ private static Stream<Arguments> parserProvider() { return Stream.of( Arguments.of(Parser.htmlParser()),
CHANGES.mdindex aa3a1545..2a654efe 100644@@ -10,6 +10,7 @@ ### Bug Fixes * Fixed HTML parsing of mixed-case RCDATA end tags after tag-shaped text. For example, `<title><p>Foo</TiTLE>` and `<textarea><img src=x></TeXtArEa>` now keep the tag-shaped content as text instead of promoting it to markup. [#2503](https://github.com/jhy/jsoup/issues/2503)+* Fixed `W3CDom` XML conversion so plain XML elements don't serialize with the reserved XML namespace as the default namespace. Explicit XML namespaces and `xml:*` attributes are still preserved. [#2504](https://github.com/jhy/jsoup/issues/2504) ### Build Changes * Cleaned up the Maven build for the multi-release JAR so Java 8 and Java 11+ sources compile as separate source sets. This avoids spurious Java 8 compiler warnings from newer-language overlay sources, keeps long-running parser checks behind an explicit profile, and preserves the same published artifacts and runtime behavior.src/main/java/org/jsoup/helper/W3CDom.javaindex b26cb93f..f4d2150f 100644@@ -4,6 +4,7 @@ import org.jsoup.internal.Normalizer; import org.jsoup.internal.StringUtil; import org.jsoup.nodes.Attribute; import org.jsoup.parser.HtmlTreeBuilder;+import org.jsoup.parser.Parser; import org.jsoup.select.NodeVisitor; import org.jsoup.select.Selector; import org.w3c.dom.Comment;@@ -358,7 +359,7 @@ public class W3CDom { public void head(org.jsoup.nodes.Node source, int depth) { if (source instanceof org.jsoup.nodes.Element) { org.jsoup.nodes.Element sourceEl = (org.jsoup.nodes.Element) source;- String namespace = namespaceAware ? sourceEl.tag().namespace() : null;+ @Nullable String namespace = namespaceAware ? w3cNamespace(sourceEl) : null; String tagName = Normalizer.xmlSafeTagName(sourceEl.tagName()); try { // use an empty namespace if none is present but the tag name has a prefix@@ -390,6 +391,14 @@ public class W3CDom { } } + private static @Nullable String w3cNamespace(org.jsoup.nodes.Element sourceEl) {+ // In W3C DOM, plain XML elements have no namespace; XML namespace is reserved for the {@code xml} prefix+ String namespace = sourceEl.tag().namespace();+ if (Parser.NamespaceXml.equals(namespace) && sourceEl.tag().prefix().isEmpty())+ return null;+ return namespace;+ }+ private void append(Node append, org.jsoup.nodes.Node source) { append.setUserData(SourceProperty, source, null); dest.appendChild(append);src/test/java/org/jsoup/helper/W3CDomTest.javaindex b9d1d330..c689db55 100644@@ -146,6 +146,53 @@ public class W3CDomTest { assertEquals("img", img.getNodeName()); } + @Test+ public void plainXmlElementsDoNotUseReservedXmlNamespace() {+ String input = "<root xml:lang=\"en\"><child>One</child></root>";+ org.jsoup.nodes.Document jsoupDoc = Jsoup.parse(input, "", Parser.xmlParser());++ Document w3Doc = new W3CDom().fromJsoup(jsoupDoc);+ org.w3c.dom.Element root = w3Doc.getDocumentElement();+ org.w3c.dom.Element child = (org.w3c.dom.Element) root.getFirstChild();++ assertNull(root.getNamespaceURI());+ assertNull(child.getNamespaceURI());+ assertEquals("en", root.getAttributeNS(Parser.NamespaceXml, "lang"));++ String xml = W3CDom.asString(w3Doc, W3CDom.OutputXml());+ assertFalse(xml.contains("xmlns=\"" + Parser.NamespaceXml + "\""), xml);++ Document roundTrip = parseXml(xml, true);+ org.w3c.dom.Element roundRoot = roundTrip.getDocumentElement();+ assertNull(roundRoot.getNamespaceURI());+ assertEquals("en", roundRoot.getAttributeNS(Parser.NamespaceXml, "lang"));+ }++ @Test+ public void explicitXmlNamespacesRoundTripThroughW3cDom() {+ String input = "<root xmlns=\"urn:example\"><child>One</child><plain xmlns=\"\">Two</plain></root>";+ org.jsoup.nodes.Document jsoupDoc = Jsoup.parse(input, "", Parser.xmlParser());++ Document w3Doc = new W3CDom().fromJsoup(jsoupDoc);+ org.w3c.dom.Element root = w3Doc.getDocumentElement();+ org.w3c.dom.Element child = (org.w3c.dom.Element) root.getFirstChild();+ org.w3c.dom.Element plain = (org.w3c.dom.Element) child.getNextSibling();++ assertEquals("urn:example", root.getNamespaceURI());+ assertEquals("urn:example", child.getNamespaceURI());+ assertNull(plain.getNamespaceURI());++ String xml = W3CDom.asString(w3Doc, W3CDom.OutputXml());+ Document roundTrip = parseXml(xml, true);+ org.w3c.dom.Element roundRoot = roundTrip.getDocumentElement();+ NodeList plainEls = roundRoot.getElementsByTagName("plain");++ assertEquals("urn:example", roundRoot.getNamespaceURI(), xml);+ assertEquals(1, roundRoot.getElementsByTagNameNS("urn:example", "child").getLength(), xml);+ assertEquals(1, plainEls.getLength(), xml);+ assertNull(plainEls.item(0).getNamespaceURI(), xml);+ }+ @Test public void handlesInvalidAttributeNames() { String html = "<html><head></head><body style=\"color: red\" \" name\"></body></html>";