address a corner-case in XMLLiteral creation

Porges · Porges · commit 3e589ad428ae · 2026-04-06T07:28:20.000+12:00
diff --git a/html2rdf/src/host_language/xhtml.rs b/html2rdf/src/host_language/xhtml.rs
@@ -156,19 +156,38 @@ impl super::Element for (&uppsala::Document<'_>, uppsala::NodeId) {
         // NOTE: I assume that ids are stable when cloning
         let el = doc.element_mut(self.1).unwrap();
 
+        // Collect inherited namespace declarations from ancestors of the current
+        // element. We must be careful to not shadow these with a prefix that
+        // only differs by case.
+        let mut inherited_ns: std::collections::HashMap<&str, &str> =
+            std::collections::HashMap::new();
+        for ancestor_id in self.0.ancestors(self.1).into_iter().rev() {
+            if let Some(ancestor_el) = self.0.element(ancestor_id) {
+                for (prefix, uri) in &ancestor_el.namespace_declarations {
+                    inherited_ns.insert(prefix.as_ref(), uri.as_ref());
+                }
+            }
+        }
+
         // copy in all the active mappings as xmlns
         // this is simpler than tracking whether we need to add them as xmlns
         // or as prefixes, and it matches what the test suite does
         let mut prefixes = Vec::new();
         for (prefix, value) in active_mappings.mappings() {
             // note that there's a little tricky case where a prefix redefined a xmlns
             // in which case we need to ensure that _both_ are present on child nodes
-            // so the XML side and the RDFa side remain correct
-            if el
+            // so the XML side and the RDFa side remain correct.
+            // We must also check inherited ancestor namespaces — a lowercased @prefix
+            // colliding with an inherited xmlns would corrupt element name resolution.
+            let conflicts_with_existing = el
                 .namespace_declarations
                 .iter()
                 .any(|(p, v)| p == prefix && v != value)
-            {
+                || inherited_ns
+                    .get(prefix.as_str())
+                    .is_some_and(|v| *v != value);
+
+            if conflicts_with_existing {
                 prefixes.push((prefix, value));
             } else {
                 el.namespace_declarations
@@ -197,7 +216,7 @@ impl super::Element for (&uppsala::Document<'_>, uppsala::NodeId) {
                             if !prefix_attr.is_empty() {
                                 prefix_attr.push(' ');
                             }
-                            _ = write!(prefix_attr, "{prefix}: {value} ");
+                            _ = write!(prefix_attr, "{prefix}: {value}");
                         }
                     }
                     if !prefix_attr.is_empty() {
diff --git a/html2rdf/tests/xml-tests.rs b/html2rdf/tests/xml-tests.rs
@@ -1,3 +1,4 @@
+use insta::assert_snapshot;
 use oxrdf::{Graph, TermRef, vocab::rdf};
 
 use crate::utils::serialize_graph;
@@ -102,7 +103,7 @@ fn element_prefix_xmlns_conflict_parent() {
     insta::assert_snapshot!(serialize_graph(&g, utils::base().as_str()), @r#"
     @base <http://example.test/> .
     @prefix rdf: <//www.w3.org/1999/02/22-rdf-syntax-ns#> .
-    <> <ex:desc> "<k:p xmlns:k=\"http://another.ns/\" prefix=\"k: http://prefix.ns/ \">.<b xmlns=\"http://www.w3.org/1999/xhtml\" property=\"k:foo\">value</b>.</k:p>"^^rdf:XMLLiteral ;
+    <> <ex:desc> "<k:p xmlns:k=\"http://another.ns/\" prefix=\"k: http://prefix.ns/\">.<b xmlns=\"http://www.w3.org/1999/xhtml\" property=\"k:foo\">value</b>.</k:p>"^^rdf:XMLLiteral ;
     	<//prefix.ns/foo> "value" .
     "#);
 }
@@ -181,3 +182,34 @@ fn element_prefix_xmlns_conflict_child() {
     	<//prefix.ns/foo> "value" .
     "#);
 }
+
+/// When @prefix lowercases FOO→foo, and xmlns:foo already exists with a
+/// different URI, we must ensure that it doesn't affect the XML namespaces.
+#[test]
+fn lowered_prefix_shadows_xmlns_in_xmlliteral() {
+    let input = r#"<?xml version="1.0" encoding="UTF-8"?>
+    <html xmlns="http://www.w3.org/1999/xhtml" xmlns:foo="http://original.ns/">
+    <body prefix="FOO: http://rdfa-prefix.ns/">
+    <div property="ex:desc" datatype="rdf:XMLLiteral">
+      <foo:bar>content</foo:bar>
+    </div>
+    </body>
+    </html>"#;
+    let (g, _) = utils::parse_xhtml(input);
+
+    // The XMLLiteral should preserve the original XML namespace for foo:bar,
+    // i.e. xmlns:foo="http://original.ns/", NOT the lowered @prefix value.
+    let xml_literal = g
+        .iter()
+        .find_map(|t| match t.object {
+            oxrdf::TermRef::Literal(l) if l.datatype() == rdf::XML_LITERAL => {
+                Some(l.value().to_string())
+            }
+            _ => None,
+        })
+        .expect("should have XMLLiteral");
+
+    // foo:bar must be in namespace http://original.ns/, not http://rdfa-prefix.ns/
+    assert_snapshot!(xml_literal,
+        @r#"<foo:bar xmlns:foo="http://original.ns/" prefix="foo: http://rdfa-prefix.ns/">content</foo:bar>"#);
+}
