From 7a8f49d6739c687288d262cdb8fcea588dcd382e Mon Sep 17 00:00:00 2001
From: AC <ac@fabtek.cn>
Date: Sat, 10 Oct 2026 15:18:02 +0800
Subject: [PATCH 1/2] Use ASCII case folding for HTML element and attribute
 names

Signed-off-by: AC <ac@fabtek.cn>
---
 cssselect/xpath.py      |  4 ++--
 tests/test_cssselect.py | 26 ++++++++++++++++++++++++++
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/cssselect/xpath.py b/cssselect/xpath.py
index 57b9884..377fc28 100644
--- a/cssselect/xpath.py
+++ b/cssselect/xpath.py
@@ -420,7 +420,7 @@ def xpath_attrib(self, selector: Attrib) -> XPathExpr:
             getattr(self, f"xpath_attrib_{operator}"),
         )
         if self.lower_case_attribute_names:
-            name = selector.attrib.lower()
+            name = ascii_lower(selector.attrib)
         else:
             name = selector.attrib
         if selector.namespace == "*":
@@ -470,7 +470,7 @@ def xpath_element(self, selector: Element) -> XPathExpr:
         else:
             safe = bool(is_safe_name(element))
             if self.lower_case_element_names:
-                element = element.lower()
+                element = ascii_lower(element)
             if selector.namespace == "*":
                 # Namespace wildcard, e.g. "*|div": any namespace,
                 # including none. XPath 1.0 has no "*:name" name test
diff --git a/tests/test_cssselect.py b/tests/test_cssselect.py
index d5d9ec7..1af8424 100644
--- a/tests/test_cssselect.py
+++ b/tests/test_cssselect.py
@@ -1361,6 +1361,32 @@ def argument_types(css: str) -> list[str]:
             css = f"::pseudo_element({argument_string})"
             assert argument_types(css) == argument_list
 
+    def test_html_non_ascii_element_names(self) -> None:
+        document = html.fromstring(
+            '<div><x-Ä id="upper"></x-Ä><x-ä id="lower"></x-ä></div>'
+        )
+        translator = HTMLTranslator()
+        for selector in ("x-Ä", "X-Ä", "*|X-Ä"):
+            elements = typing.cast(
+                "list[etree._Element]",
+                document.xpath(translator.css_to_xpath(selector)),
+            )
+            assert [element.get("id") for element in elements] == ["upper"]
+        assert len(document.xpath(translator.css_to_xpath("X-ä"))) == 1
+
+    def test_html_non_ascii_attribute_names(self) -> None:
+        document = html.fromstring(
+            '<div><p id="upper" DATA-Ä="yes"></p><p id="lower" data-ä="yes"></p></div>'
+        )
+        translator = HTMLTranslator()
+        for selector in ("[DATA-Ä]", '[DATA-Ä="yes"]', "[*|DATA-Ä]"):
+            elements = typing.cast(
+                "list[etree._Element]",
+                document.xpath(translator.css_to_xpath(selector)),
+            )
+            assert [element.get("id") for element in elements] == ["upper"]
+        assert len(document.xpath(translator.css_to_xpath('[DATA-ä="yes"]'))) == 1
+
     def test_select(self) -> None:
         document = etree.fromstring(HTML_IDS)
         sort_key = {el: count for count, el in enumerate(document.iter())}.__getitem__

From fa9244eff12ec30ada5460644b1c451c4212de26 Mon Sep 17 00:00:00 2001
From: AC <ac@fabtek.cn>
Date: Tue, 13 Oct 2026 19:20:25 +0800
Subject: [PATCH 2/2] Simplify HTML name regression tests to assert XPath
 output

Signed-off-by: AC <ac@fabtek.cn>
---
 tests/test_cssselect.py | 41 ++++++++++++++++++++---------------------
 1 file changed, 20 insertions(+), 21 deletions(-)

diff --git a/tests/test_cssselect.py b/tests/test_cssselect.py
index 1af8424..4a98eee 100644
--- a/tests/test_cssselect.py
+++ b/tests/test_cssselect.py
@@ -1362,30 +1362,29 @@ def argument_types(css: str) -> list[str]:
             assert argument_types(css) == argument_list
 
     def test_html_non_ascii_element_names(self) -> None:
-        document = html.fromstring(
-            '<div><x-Ä id="upper"></x-Ä><x-ä id="lower"></x-ä></div>'
-        )
-        translator = HTMLTranslator()
-        for selector in ("x-Ä", "X-Ä", "*|X-Ä"):
-            elements = typing.cast(
-                "list[etree._Element]",
-                document.xpath(translator.css_to_xpath(selector)),
-            )
-            assert [element.get("id") for element in elements] == ["upper"]
-        assert len(document.xpath(translator.css_to_xpath("X-ä"))) == 1
+        xpath = HTMLTranslator().css_to_xpath
+        assert xpath("x-Ä") == "descendant-or-self::*[name() = 'x-Ä']"
+        assert xpath("X-Ä") == "descendant-or-self::*[name() = 'x-Ä']"
+        assert xpath("*|X-Ä") == "descendant-or-self::*[local-name() = 'x-Ä']"
+        assert xpath("X-ä") == "descendant-or-self::*[name() = 'x-ä']"
 
     def test_html_non_ascii_attribute_names(self) -> None:
-        document = html.fromstring(
-            '<div><p id="upper" DATA-Ä="yes"></p><p id="lower" data-ä="yes"></p></div>'
+        xpath = HTMLTranslator().css_to_xpath
+        assert (
+            xpath("[DATA-Ä]")
+            == "descendant-or-self::*[attribute::*[name() = 'data-Ä']]"
+        )
+        assert (
+            xpath('[DATA-Ä="yes"]')
+            == "descendant-or-self::*[attribute::*[name() = 'data-Ä'] = 'yes']"
+        )
+        assert xpath("[*|DATA-Ä]") == (
+            "descendant-or-self::*[attribute::*[local-name() = 'data-Ä']]"
+        )
+        assert (
+            xpath('[DATA-ä="yes"]')
+            == "descendant-or-self::*[attribute::*[name() = 'data-ä'] = 'yes']"
         )
-        translator = HTMLTranslator()
-        for selector in ("[DATA-Ä]", '[DATA-Ä="yes"]', "[*|DATA-Ä]"):
-            elements = typing.cast(
-                "list[etree._Element]",
-                document.xpath(translator.css_to_xpath(selector)),
-            )
-            assert [element.get("id") for element in elements] == ["upper"]
-        assert len(document.xpath(translator.css_to_xpath('[DATA-ä="yes"]'))) == 1
 
     def test_select(self) -> None:
         document = etree.fromstring(HTML_IDS)