id: PYSEC-2026-2614 published: "2026-07-13T15:46:29.614932Z" modified: "2026-07-13T16:04:44.674997Z" aliases: - CVE-2026-49825 - GHSA-4jhm-jv67-739f summary: "`lxml_html_clean.Cleaner` does not strip `javascript:` URLs from namespaced URL attributes" details: "# `lxml_html_clean.Cleaner` does not strip `javascript:` URLs from namespaced URL attributes (`xlink:href`)\n\n**Reporter:** Guillem Lefait · **Date:** 2026-05-10\n**Affected:** `lxml` ≤ 6.1.0 and `lxml_html_clean` ≤ 0.4.4 (latest stable)\n**Confirmed against:** lxml 6.1.0 + lxml_html_clean 0.4.4 on Python 3.13.5, 3.14.4, and 3.15.0a8 (libxml2 2.14.6 / 2.9.14 — bug is in pure-Python sanitizer logic, independent of the libxml2 backend)\n**Root-cause class:** same as CVE-2021-28957 (`formaction` missing from `link_attrs`)\n\n## Summary\n\n`Cleaner` filters URL schemes (`javascript:`, `vbscript:`, …) by walking links via `rewrite_links()`, which delegates to `iterlinks()`, which only yields attributes named in `lxml.html.defs.link_attrs`. That allow-list contains no prefixed names (`xlink:href`) and no `srcset`. As a result, when `Cleaner` is configured with `safe_attrs_only=False` — a documented option for callers that want lenient attribute handling but still expect URL-scheme scrubbing — `` survives sanitization untouched, and any browser that follows the SVG-anchor specification will execute the JavaScript when the rendered link is clicked.\n\n**CWE:** CWE-79 (XSS), with CWE-184 (Incomplete List of Disallowed Inputs) as the underlying defect class.\n\n## Affected components\n\n| Package | Versions tested | File / line |\n|--------------------|------------------------|----------------------------------|\n| `lxml` | 4.9.x, 5.2.1, 6.1.0 | `src/lxml/html/defs.py:20` |\n| `lxml` | \" | `src/lxml/html/__init__.py:485-528` |\n| `lxml_html_clean` | 0.4.0 – 0.4.4 | `lxml_html_clean/clean.py:348,576` |\n\nThe legacy `lxml.html.clean` module — bundled in `lxml < 5.2.0` and still installable on newer versions via the `lxml[html_clean]` extra — shares the same bug.\n\n## Root cause\n\n`defs.link_attrs` is a flat string set; the literal `xlink:href` is absent:\n\n```python\n# lxml/html/defs.py\nlink_attrs = frozenset([\n 'action', 'archive', 'background', 'cite', 'classid',\n 'codebase', 'data', 'href', 'longdesc', 'profile', 'src',\n 'usemap', 'dynsrc', 'lowsrc', 'formaction',\n])\n```\n\n`HtmlMixin.iterlinks()` (`lxml/html/__init__.py:526-528`) only yields attributes whose key is in that set:\n\n```python\nfor attrib in link_attrs:\n if attrib in attribs:\n yield (el, attrib, attribs[attrib], 0)\n```\n\n`Cleaner.__call__` registers the URL-scheme filter via `rewrite_links` (`lxml_html_clean/clean.py:348`), which is a thin wrapper around `iterlinks()`. Because `xlink:href` is never yielded, `_remove_javascript_link` (`clean.py:576`) is never invoked for it.\n\n## Minimal reproducer\n\n```python\nfrom lxml import html\nfrom lxml_html_clean import Cleaner\n\nfor payload in (\n '',\n '

y

',\n):\n tree = html.fromstring(payload)\n Cleaner(safe_attrs_only=False)(tree)\n print(html.tostring(tree).decode())\n print(' iterlinks:', list(html.fromstring(payload).iterlinks()))\n# ← unchanged\n# iterlinks: [] ← link rewriter blind\n#

y

← unchanged\n# iterlinks: [] ← link rewriter blind\n```\n\nBoth SVG and MathML scopes are vulnerable — same allow-list gap, both render anchors that browsers treat as navigable. Other lab-confirmed surviving variants (same scope, different scheme encoding): mixed-case (`JaVaScRiPt:`), HTML-entity (`javascript:`), embedded tab (`java\\tscript:`).\n\n## Impact\n\nA caller that uses `Cleaner` to neutralise untrusted HTML and chooses `safe_attrs_only=False` — typically because the application wants to allow custom data-/aria-/vendor attributes — will silently pass `javascript:` payloads carried on `xlink:href` through to victim renders. Stored XSS in any application that round-trips user-supplied HTML through this configuration. Reach is conditional on the `safe_attrs_only=False` toggle, but that is a documented public option; consumers reasonably expect URL-scheme scrubbing to be independent of attribute allow-listing.\n\n## Suggested fix\n\n**Extend `link_attrs`** to include `xlink:href`. In HTML mode, `lxml.html` keeps prefixed attribute names verbatim — the parsed key is the literal string `xlink:href`, not a Clark-notation form — so the existing allow-list lookup is a plain string match. Same shape as the CVE-2021-28957 fix:\n\n```diff\n # lxml/html/defs.py\n link_attrs = frozenset([\n 'action', 'archive', 'background', 'cite', 'classid',\n 'codebase', 'data', 'href', 'longdesc', 'profile', 'src',\n 'usemap', 'dynsrc', 'lowsrc', 'formaction',\n+ 'xlink:href',\n ])\n```\n\nThis single change closes the reported XSS for both SVG `` and MathML ``. `lxml_html_clean` is the canonical home of the `Cleaner` code (881 lines); `lxml.html.clean` is a 21-line backward-compat shim (`from lxml_html_clean import *`) that picks up the fix automatically once `link_attrs` is updated upstream. Since the upstream change requires lxml maintainer action, see the alternative below if a self-contained patch in `lxml_html_clean` is preferred.\n\n**Alternative (in-package fix, no lxml coordination needed):** add a namespaced-URL-attribute walk inside `Cleaner.__call__` so the URL-scheme filter doesn't depend on `link_attrs`. Sketch:\n\n```python\n# lxml_html_clean/clean.py — supplements rewrite_links() in __call__\n_NS_URL_ATTRS = ('xlink:href',) # extend as needed\n_BAD_SCHEME = re.compile(r'^\\s*(javascript|vbscript|data):', re.I)\n\nfor el in doc.iter():\n for attr in _NS_URL_ATTRS:\n if attr in el.attrib and _BAD_SCHEME.match(el.attrib[attr]):\n del el.attrib[attr]\n```\n\nThis decouples the cleaner from the upstream `link_attrs` set and matches the security-ownership boundary established when the cleaner was extracted in lxml 5.2.0.\n\n**Defense in depth (optional, regardless of which fix path is taken):**\n- Also handle `srcset`: the value is a `url 1x, url 2x, …` descriptor list, so split on commas and validate each candidate URL. Not directly executable in current browsers, but closes the same gap.\n- Also accept Clark-notation forms (`{http://www.w3.org/1999/xlink}href`) so XML-mode callers using `lxml.etree` get the same protection. HTML mode never produces this form, so not needed for the reported bug.\n\n## Severity\n\nCVSS 3.1 base score: **8.2 / High** — `AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:L/A:N` (stored XSS; victim must click the SVG anchor; scope-changed because script executes in the rendering origin). PR:N reflects the common case where untrusted HTML enters the sanitizer from anonymous sources (comments, support tickets); deployments that gate writes behind authentication can score with PR:L (→ 7.6).\n\nSeverity is **CONDITIONAL** on the caller passing `safe_attrs_only=False`. With the class default (`True`), attribute allow-listing strips `xlink:href` before scheme scrubbing runs, and the bug does not fire — verified at HEAD: default-config `Cleaner()()` → ``.\n\n## Prior art / novelty\n\n- **CVE-2021-28957 (lxml 4.6.3)** — same root cause, different attribute (`formaction`). Fix was a one-line extension of `link_attrs`. Direct precedent.\n- **CVE-2022-34473** (Mozilla Sanitizer API) — `xlink:href` URL bypass primitive in a different sanitizer.\n- **Bleach (Mozilla, Python)** explicitly handles the `xlink` namespace; `enshrined/svg-sanitize` (PHP) ships `cleanXlinkHrefs()`; DOMPurify scrubs `xlink:href` via `ALLOWED_URI_REGEXP`.\n- **`nh3`** (the alternative recommended in `lxml_html_clean`'s own README for security-sensitive use) is **not vulnerable** to this primitive — verified 2026-05-10 on `nh3==0.3.5`: with `