Sanitize attribute names in fenced_code and legacy_attrs

docs/changelog.md

@@ -16,6 +16,8 @@ See the [Contributing Guide](contributing.md) for details.
 
 ### Fixed
 
+* Sanitize attribute names in the `fenced_code` and `legacy_attrs` extensions so an
+  attribute name cannot break out of its tag and inject markup (#1663).
 * For backwards compatibility in some languages, revert the enforcement of CommonMark
   style punctuation rules (#1657).
 

markdown/extensions/attr_list.py

@@ -88,17 +88,28 @@ def isheader(elem: Element) -> bool:
     return elem.tag in ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']
 
 
+NAME_RE = re.compile(r'[^A-Z_a-z\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02ff'
+                     r'\u0370-\u037d\u037f-\u1fff\u200c-\u200d'
+                     r'\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff'
+                     r'\uf900-\ufdcf\ufdf0-\ufffd'
+                     r'\:\-\.0-9\u00b7\u0300-\u036f\u203f-\u2040]+')
+
+
+def sanitize_name(name: str) -> str:
+    """
+    Sanitize name as 'an XML Name, minus the `:`.'
+    See <https://www.w3.org/TR/REC-xml-names/#NT-NCName>.
+    """
+    return NAME_RE.sub('_', name)
+
+
 class AttrListTreeprocessor(Treeprocessor):
 
     BASE_RE = r'\{\:?[ ]*([^\}\n ][^\n]*)[ ]*\}'
     HEADER_RE = re.compile(r'[ ]+{}[ ]*$'.format(BASE_RE))
     BLOCK_RE = re.compile(r'\n[ ]*{}[ ]*$'.format(BASE_RE))
     INLINE_RE = re.compile(r'^{}'.format(BASE_RE))
-    NAME_RE = re.compile(r'[^A-Z_a-z\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02ff'
-                         r'\u0370-\u037d\u037f-\u1fff\u200c-\u200d'
-                         r'\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff'
-                         r'\uf900-\ufdcf\ufdf0-\ufffd'
-                         r'\:\-\.0-9\u00b7\u0300-\u036f\u203f-\u2040]+')
+    NAME_RE = NAME_RE
 
     def run(self, doc: Element) -> None:
         for elem in doc.iter():
@@ -190,7 +201,7 @@ def sanitize_name(self, name: str) -> str:
         Sanitize name as 'an XML Name, minus the `:`.'
         See <https://www.w3.org/TR/REC-xml-names/#NT-NCName>.
         """
-        return self.NAME_RE.sub('_', name)
+        return sanitize_name(name)
 
 
 class AttrListExtension(Extension):

markdown/extensions/fenced_code.py

@@ -25,7 +25,7 @@
 from . import Extension
 from ..preprocessors import Preprocessor
 from .codehilite import CodeHilite, CodeHiliteExtension, parse_hl_lines
-from .attr_list import get_attrs_and_remainder, AttrListExtension
+from .attr_list import get_attrs_and_remainder, sanitize_name, AttrListExtension
 from ..util import parseBoolValue
 from ..serializers import _escape_attrib_html
 import re
@@ -157,7 +157,8 @@ def run(self, lines: list[str]) -> list[str]:
                         # pairs were defined on the code block, and the `use_pygments` key was not set to `True`. The
                         # `use_pygments` key could be either set to `False` or not defined. It is omitted from output.
                         kv_pairs = ''.join(
-                            f' {k}="{_escape_attrib_html(v)}"' for k, v in config.items() if k != 'use_pygments'
+                            f' {sanitize_name(k)}="{_escape_attrib_html(v)}"'
+                            for k, v in config.items() if k != 'use_pygments'
                         )
                     code = self._escape(m.group('code'))
                     code = f'<pre{id_attr}{class_attr}><code{lang_attr}{kv_pairs}>{code}</code></pre>'

markdown/extensions/legacy_attrs.py

@@ -33,6 +33,7 @@
 import re
 from markdown.treeprocessors import Treeprocessor, isString
 from markdown.extensions import Extension
+from markdown.extensions.attr_list import sanitize_name
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:  # pragma: no cover
@@ -57,7 +58,7 @@ def run(self, doc: etree.Element) -> None:
     def handleAttributes(self, el: etree.Element, txt: str) -> str:
         """ Set attributes and return text without definitions. """
         def attributeCallback(match: re.Match[str]):
-            el.set(match.group(1), match.group(2).replace('\n', ' '))
+            el.set(sanitize_name(match.group(1)), match.group(2).replace('\n', ' '))
         return ATTR_RE.sub(attributeCallback, txt)
 
 

tests/test_syntax/extensions/test_fenced_code.py

@@ -304,6 +304,25 @@ def testFencedLanguageNoCodehiliteWithAttrList(self):
             extensions=['fenced_code', 'attr_list']
         )
 
+    def testFencedCodeAttrNameSanitizedNoCodehiliteWithAttrList(self):
+        # An attribute name is sanitized so it cannot break out of the `code` tag (see #1663).
+        self.assertMarkdownRenders(
+            self.dedent(
+                '''
+                ``` { .python x><svg/onload=alert(1) }
+                # Some python code
+                ```
+                '''
+            ),
+            self.dedent(
+                '''
+                <pre><code class="language-python" x_svg_onload="alert(1)"># Some python code
+                </code></pre>
+                '''
+            ),
+            extensions=['fenced_code', 'attr_list']
+        )
+
     def testFencedLanguagePygmentsDisabledInAttrNoCodehiliteWithAttrList(self):
         self.assertMarkdownRenders(
             self.dedent(

tests/test_syntax/extensions/test_legacy_attrs.py

@@ -65,3 +65,11 @@ def testLegacyAttrs(self):
             """),  # noqa: E501
             extensions=['legacy_attrs']
         )
+
+    def testLegacyAttrsSanitizesName(self):
+        # An attribute name is sanitized so it cannot break out of the element (see #1663).
+        self.assertMarkdownRenders(
+            'text {@x><svg/onload=alert(1)=y} more',
+            '<p x_svg_onload_alert_1_="y">text  more</p>',
+            extensions=['legacy_attrs']
+        )