Apply namespaces when fragment parsing

For #2299

Author: jhy

src/main/java/org/jsoup/parser/XmlTreeBuilder.java

@@ -15,6 +15,7 @@
 import org.jsoup.nodes.Node;
 import org.jsoup.nodes.TextNode;
 import org.jsoup.nodes.XmlDeclaration;
+import org.jsoup.select.Elements;
 import org.jspecify.annotations.Nullable;
 
 import java.io.Reader;
@@ -60,11 +61,23 @@ protected void initialiseParse(Reader input, String baseUri, Parser parser) {
     @Override
     void initialiseParseFragment(@Nullable Element context) {
         super.initialiseParseFragment(context);
-        if (context != null) {
-            TokeniserState textState = context.tag().textState();
-            if (textState != null) tokeniser.transition(textState);
+        if (context == null) return;
+
+        // transition to the tag's text state if available
+        TokeniserState textState = context.tag().textState();
+        if (textState != null) tokeniser.transition(textState);
+
+        // reconstitute the namespace stack by traversing the element and its parents (top down)
+        Elements chain = context.parents();
+        chain.add(0, context);
+        for (int i = chain.size() - 1; i >= 0; i--) {
+            Element el = chain.get(i);
+            HashMap<String, String> namespaces = new HashMap<>(namespacesStack.peek());
+            namespacesStack.push(namespaces);
+            if (el.attributesSize() > 0) {
+                processNamespaces(el.attributes(), namespaces);
+            }
         }
-
     }
 
     Document parse(Reader input, String baseUri) {
@@ -130,14 +143,16 @@ void insertElementFor(Token.StartTag startTag) {
         HashMap<String, String> namespaces = new HashMap<>(namespacesStack.peek());
         namespacesStack.push(namespaces);
 
-        if (startTag.attributes != null) {
-            startTag.attributes.deduplicate(settings);
-            processNamespaces(startTag.attributes, namespaces);
+        Attributes attributes = startTag.attributes;
+        if (attributes != null) {
+            attributes.deduplicate(settings);
+            processNamespaces(attributes, namespaces);
+            applyNamespacesToAttributes(attributes, namespaces);
         }
 
         String ns = resolveNamespace(startTag.tagName, namespaces);
         Tag tag = tagFor(startTag.tagName, startTag.normalName, ns, settings);
-        Element el = new Element(tag, null, settings.normalizeAttributes(startTag.attributes));
+        Element el = new Element(tag, null, settings.normalizeAttributes(attributes));
         currentElement().appendChild(el);
         push(el);
 
@@ -162,6 +177,9 @@ private static void processNamespaces(Attributes attributes, HashMap<String, Str
                 namespaces.put(nsPrefix, value);
             }
         }
+    }
+
+    private static void applyNamespacesToAttributes(Attributes attributes, HashMap<String, String> namespaces) {
         // second pass, apply namespace to attributes. Collects them first then adds (as userData is an attribute)
         Map<String, String> attrPrefix = new HashMap<>();
         for (Attribute attr: attributes) {

src/test/java/org/jsoup/parser/XmlTreeBuilderTest.java

@@ -575,4 +575,31 @@ private static void assertXmlNamespace(Element el) {
         assertEquals("", other.namespace());
     }
 
+    @Test void elementsViaAppendHtmlAreNamespaced() {
+        // tests that when elements / attributes are added via a fragment parse, they inherit the namespace stack, and can still override
+        String xml = "<out xmlns='/out'><bk:book xmlns:bk='/books' xmlns:edi='/edi'><bk:title>Test</bk:title><li edi:foo='bar'></bk:book></out>";
+        Document doc = Jsoup.parse(xml, Parser.xmlParser());
+
+        // insert some parsed xml, inherit bk and edi, and with an inner node override bk
+        Element book = doc.expectFirst("bk|book");
+        book.append("<bk:content edi:foo=qux>Content</bk:content>");
+
+        Element out = doc.expectFirst("out");
+        assertEquals("/out", out.tag().namespace());
+
+        Element content = book.expectFirst("bk|content");
+        assertEquals("bk:content", content.tag().name());
+        assertEquals("/books", content.tag().namespace());
+        assertEquals("/edi", content.attribute("edi:foo").namespace());
+
+        content.append("<data>Data</data><html xmlns='/html' xmlns:bk='/update'><p>Foo</p><bk:news>News</bk:news></html>");
+        // p should be in /html, news in /update
+        Element p = content.expectFirst("p");
+        assertEquals("/html", p.tag().namespace());
+        Element news = content.expectFirst("bk|news");
+        assertEquals("/update", news.tag().namespace());
+        Element data = content.expectFirst("data");
+        assertEquals("/out", data.tag().namespace());
+    }
+
 }