fix: apply upstream patch for in-context parsing (#3116)

flavorjones · web-flow · commit a26536d7a41f · 2024-01-30T10:42:07.000-05:00
**What problem is this PR intended to solve?** Apply upstream fix for #3112 - upstream bug report https://gitlab.gnome.org/GNOME/libxml2/-/issues/672 - upstream fix https://gitlab.gnome.org/GNOME/libxml2/-/commit/95f2a17440568694a6df6a326c5b411e77597be2 Fixes #3112 **Have you included adequate test coverage?** Yes **Does this change affect the behavior of either the C or the Java implementations?** Only affects the C implementation.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -15,6 +15,7 @@ Nokogiri follows [Semantic Versioning](https://semver.org/), please see the [REA
 
 * [CRuby] `XML::Reader` defaults the encoding to UTF-8 if it's not specified in either the document or as a method parameter. Previously non-ASCII characters were serialized as NCRs in this case. [#2891] (@flavorjones)
 * [CRuby] Restored support for compilation by GCC versions earlier than 4.6, which was broken in v1.15.0 (540e9aee). [#3090] (@adfoster-r7)
+* [CRuby] Patched upstream libxml2 to allow parsing HTML5 in the context of a namespaced node (e.g., foreign content like MathML). [#3112, #3116] (@flavorjones)
 
 
 ## v1.16.0 / 2023-12-27
diff --git a/patches/libxml2/0012-parser-Fix-crash-in-xmlParseInNodeContext-with-HTML.patch b/patches/libxml2/0012-parser-Fix-crash-in-xmlParseInNodeContext-with-HTML.patch
@@ -0,0 +1,33 @@
+From 95f2a17440568694a6df6a326c5b411e77597be2 Mon Sep 17 00:00:00 2001
+From: Nick Wellnhofer <wellnhofer@aevum.de>
+Date: Tue, 30 Jan 2024 13:25:17 +0100
+Subject: [PATCH] parser: Fix crash in xmlParseInNodeContext with HTML
+ documents
+
+Ignore namespaces if we have an HTML document with namespaces added
+manually.
+
+Fixes #672.
+---
+ parser.c | 4 +++-
+ 1 file changed, 3 insertions(+), 1 deletion(-)
+
+diff --git a/parser.c b/parser.c
+index 1038d71b..f7842ed1 100644
+--- a/parser.c
++++ b/parser.c
+@@ -12415,8 +12415,10 @@ xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen,
+     }
+     xmlAddChild(node, fake);
+ 
+-    if (node->type == XML_ELEMENT_NODE) {
++    if (node->type == XML_ELEMENT_NODE)
+ 	nodePush(ctxt, node);
++
++    if ((ctxt->html == 0) && (node->type == XML_ELEMENT_NODE)) {
+ 	/*
+ 	 * initialize the SAX2 namespaces stack
+ 	 */
+-- 
+2.42.0
+
diff --git a/test/html5/test_api.rb b/test/html5/test_api.rb
@@ -238,6 +238,22 @@ def test_node_wrap_uses_parent_node_as_parsing_context_node
     assert_equal("select", el.parent.parent.name)
   end
 
+  def test_parse_in_context_of_foreign_namespace
+    if Nokogiri.uses_libxml?("~> 2.12.0")
+      skip_unless_libxml2_patch("0012-parser-Fix-crash-in-xmlParseInNodeContext-with-HTML.patch")
+    end
+
+    # https://github.com/sparklemotion/nokogiri/issues/3112
+    # https://gitlab.gnome.org/GNOME/libxml2/-/issues/672
+    doc = Nokogiri::HTML5::Document.parse("<html><body><math>")
+    math = doc.at_css("math")
+
+    nodes = math.parse("mrow") # segfaults in libxml 2.12 before 95f2a174
+
+    assert_kind_of(Nokogiri::XML::NodeSet, nodes)
+    assert_equal(1, nodes.length)
+  end
+
   describe Nokogiri::HTML5::Document do
     describe "#fragment" do
       it "parses text nodes in a `body` context" do
