dompurify 3.4.8 → 3.4.10

package/dist/purify.js

@@ -1,4 +1,4 @@
-/*! @license DOMPurify 3.4.8 | (c) Cure53 and other contributors | Released under the Apache license 2.0 and Mozilla Public License 2.0 | github.com/cure53/DOMPurify/blob/3.4.8/LICENSE */
+/*! @license DOMPurify 3.4.10 | (c) Cure53 and other contributors | Released under the Apache license 2.0 and Mozilla Public License 2.0 | github.com/cure53/DOMPurify/blob/3.4.10/LICENSE */
 
 (function (global, factory) {
   typeof exports === 'object' && typeof module !== 'undefined' ? module.exports = factory() :
@@ -333,6 +333,13 @@
   );
   const DOCTYPE_NAME = seal(/^html$/i);
   const CUSTOM_ELEMENT = seal(/^[a-z][.\w]*(-[.\w]+)+$/i);
+  // Markup-significant character probes used by _sanitizeElements.
+  // Shared module-level instances are safe despite the sticky /g flags:
+  // unapply() resets lastIndex for RegExp receivers before every call.
+  const ELEMENT_MARKUP_PROBE = seal(/<[/\w!]/g);
+  const COMMENT_MARKUP_PROBE = seal(/<[/\w]/g);
+  const FALLBACK_TAG_CLOSE = seal(/<\/no(script|embed|frames)/i);
+  const SELF_CLOSING_TAG = seal(/\/>/i);
 
   /* eslint-disable @typescript-eslint/indent */
   // https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType
@@ -345,7 +352,7 @@
     // Deprecated
     entityNode: 6,
     // Deprecated
-    progressingInstruction: 7,
+    processingInstruction: 7,
     comment: 8,
     document: 9,
     documentType: 10,
@@ -406,10 +413,25 @@
       uponSanitizeShadowNode: []
     };
   };
+  /**
+   * Resolve a set-valued configuration option: a fresh set built from
+   * cfg[key] when it is an own array property (seeded with a clone of
+   * options.base when given, case-normalized via options.transform),
+   * the fallback set otherwise.
+   *
+   * @param cfg the cloned, prototype-free configuration object
+   * @param key the configuration property to read
+   * @param fallback the set to use when the option is absent or not an array
+   * @param options transform and optional base set to merge into
+   * @returns the resolved set
+   */
+  const _resolveSetOption = function _resolveSetOption(cfg, key, fallback, options) {
+    return objectHasOwnProperty(cfg, key) && arrayIsArray(cfg[key]) ? addToSet(options.base ? clone(options.base) : {}, cfg[key], options.transform) : fallback;
+  };
   function createDOMPurify() {
     let window = arguments.length > 0 && arguments[0] !== undefined ? arguments[0] : getGlobal();
     const DOMPurify = root => createDOMPurify(root);
-    DOMPurify.version = '3.4.8';
+    DOMPurify.version = '3.4.10';
     DOMPurify.removed = [];
     if (!window || !window.document || window.document.nodeType !== NODE_TYPE.document || !window.Element) {
       // Not running in a browser, provide a factory function
@@ -454,23 +476,54 @@
     }
     let trustedTypesPolicy;
     let emptyHTML = '';
+    // The instance's own internal Trusted Types policy. Unlike a caller-supplied
+    // `TRUSTED_TYPES_POLICY`, this is created at most once — Trusted Types throws
+    // on duplicate policy names — and is the only policy allowed to persist
+    // across configurations and survive `clearConfig()`.
+    let defaultTrustedTypesPolicy;
+    let defaultTrustedTypesPolicyResolved = false;
     // Tracks whether we are already inside a call to the configured Trusted Types
-    // policy's `createHTML`. If the supplied `TRUSTED_TYPES_POLICY.createHTML`
+    // policy (`createHTML` or `createScriptURL`). If a supplied policy callback
     // itself calls `DOMPurify.sanitize` (the cause of #1422), `sanitize` would
     // re-enter the policy and recurse until the stack overflows. We detect that
-    // re-entry and throw a clear, actionable error instead.
-    let IN_POLICY_CREATE_HTML = 0;
-    const _createTrustedHTML = function _createTrustedHTML(html) {
-      if (IN_POLICY_CREATE_HTML > 0) {
-        throw typeErrorCreate('The configured TRUSTED_TYPES_POLICY.createHTML must not call ' + 'DOMPurify.sanitize, as that causes infinite recursion. Do not pass ' + 'a policy whose createHTML wraps DOMPurify as TRUSTED_TYPES_POLICY; ' + 'see the "DOMPurify and Trusted Types" section of the README.');
+    // re-entry and throw a clear, actionable error instead. The guard is shared
+    // across both callbacks, because either one re-entering `sanitize` triggers
+    // the same unbounded recursion.
+    let IN_TRUSTED_TYPES_POLICY = 0;
+    const _assertNotInTrustedTypesPolicy = function _assertNotInTrustedTypesPolicy() {
+      if (IN_TRUSTED_TYPES_POLICY > 0) {
+        throw typeErrorCreate('A configured TRUSTED_TYPES_POLICY callback (createHTML or ' + 'createScriptURL) must not call DOMPurify.sanitize, as that causes ' + 'infinite recursion. Do not pass a policy whose callbacks wrap ' + 'DOMPurify as TRUSTED_TYPES_POLICY; see the "DOMPurify and Trusted ' + 'Types" section of the README.');
       }
-      IN_POLICY_CREATE_HTML++;
+    };
+    const _createTrustedHTML = function _createTrustedHTML(html) {
+      _assertNotInTrustedTypesPolicy();
+      IN_TRUSTED_TYPES_POLICY++;
       try {
         return trustedTypesPolicy.createHTML(html);
       } finally {
-        IN_POLICY_CREATE_HTML--;
+        IN_TRUSTED_TYPES_POLICY--;
       }
     };
+    const _createTrustedScriptURL = function _createTrustedScriptURL(scriptUrl) {
+      _assertNotInTrustedTypesPolicy();
+      IN_TRUSTED_TYPES_POLICY++;
+      try {
+        return trustedTypesPolicy.createScriptURL(scriptUrl);
+      } finally {
+        IN_TRUSTED_TYPES_POLICY--;
+      }
+    };
+    // Lazily resolve (and cache) the instance's internal default policy.
+    // Resolution is attempted at most once: a successful `createPolicy` cannot be
+    // repeated (Trusted Types throws on duplicate names), and a failed or
+    // unsupported attempt must not be retried on every parse.
+    const _getDefaultTrustedTypesPolicy = function _getDefaultTrustedTypesPolicy() {
+      if (!defaultTrustedTypesPolicyResolved) {
+        defaultTrustedTypesPolicy = _createTrustedTypesPolicy(trustedTypes, currentScript);
+        defaultTrustedTypesPolicyResolved = true;
+      }
+      return defaultTrustedTypesPolicy;
+    };
     const _document = document,
       implementation = _document.implementation,
       createNodeIterator = _document.createNodeIterator,
@@ -609,7 +662,17 @@
     let USE_PROFILES = {};
     /* Tags to ignore content of when KEEP_CONTENT is true */
     let FORBID_CONTENTS = null;
-    const DEFAULT_FORBID_CONTENTS = addToSet({}, ['annotation-xml', 'audio', 'colgroup', 'desc', 'foreignobject', 'head', 'iframe', 'math', 'mi', 'mn', 'mo', 'ms', 'mtext', 'noembed', 'noframes', 'noscript', 'plaintext', 'script', 'style', 'svg', 'template', 'thead', 'title', 'video', 'xmp']);
+    const DEFAULT_FORBID_CONTENTS = addToSet({}, ['annotation-xml', 'audio', 'colgroup', 'desc', 'foreignobject', 'head', 'iframe', 'math', 'mi', 'mn', 'mo', 'ms', 'mtext', 'noembed', 'noframes', 'noscript', 'plaintext', 'script',
+    // <selectedcontent> mirrors the selected <option>'s subtree, cloned by
+    // the UA (customizable <select>) — including any on* handlers — and the
+    // engine re-mirrors synchronously whenever a removal changes which
+    // option/selectedcontent is current, even inside DOMPurify's inert
+    // DOMParser document. Hoisting its children on removal re-inserts a fresh
+    // mirror target ahead of the walk, which the engine refills, looping
+    // forever (DoS) and amplifying output. Dropping its content on removal
+    // (rather than hoisting) breaks that cascade; the content is a duplicate
+    // of the option, which is sanitized on its own. See campaign-3 F1/F6.
+    'selectedcontent', 'style', 'svg', 'template', 'thead', 'title', 'video', 'xmp']);
     /* Tags that are safe for data: URIs */
     let DATA_URI_TAGS = null;
     const DEFAULT_DATA_URI_TAGS = addToSet({}, ['audio', 'video', 'img', 'source', 'image', 'track']);
@@ -625,8 +688,10 @@
     /* Allowed XHTML+XML namespaces */
     let ALLOWED_NAMESPACES = null;
     const DEFAULT_ALLOWED_NAMESPACES = addToSet({}, [MATHML_NAMESPACE, SVG_NAMESPACE, HTML_NAMESPACE], stringToString);
-    let MATHML_TEXT_INTEGRATION_POINTS = addToSet({}, ['mi', 'mo', 'mn', 'ms', 'mtext']);
-    let HTML_INTEGRATION_POINTS = addToSet({}, ['annotation-xml']);
+    const DEFAULT_MATHML_TEXT_INTEGRATION_POINTS = freeze(['mi', 'mo', 'mn', 'ms', 'mtext']);
+    let MATHML_TEXT_INTEGRATION_POINTS = addToSet({}, DEFAULT_MATHML_TEXT_INTEGRATION_POINTS);
+    const DEFAULT_HTML_INTEGRATION_POINTS = freeze(['annotation-xml']);
+    let HTML_INTEGRATION_POINTS = addToSet({}, DEFAULT_HTML_INTEGRATION_POINTS);
     // Certain elements are allowed in both SVG and HTML
     // namespace. We need to specify them explicitly
     // so that they don't get erroneously deleted from
@@ -668,14 +733,32 @@
       // HTML tags and attributes are not case-sensitive, converting to lowercase. Keeping XHTML as is.
       transformCaseFunc = PARSER_MEDIA_TYPE === 'application/xhtml+xml' ? stringToString : stringToLowerCase;
       /* Set configuration parameters */
-      ALLOWED_TAGS = objectHasOwnProperty(cfg, 'ALLOWED_TAGS') && arrayIsArray(cfg.ALLOWED_TAGS) ? addToSet({}, cfg.ALLOWED_TAGS, transformCaseFunc) : DEFAULT_ALLOWED_TAGS;
-      ALLOWED_ATTR = objectHasOwnProperty(cfg, 'ALLOWED_ATTR') && arrayIsArray(cfg.ALLOWED_ATTR) ? addToSet({}, cfg.ALLOWED_ATTR, transformCaseFunc) : DEFAULT_ALLOWED_ATTR;
-      ALLOWED_NAMESPACES = objectHasOwnProperty(cfg, 'ALLOWED_NAMESPACES') && arrayIsArray(cfg.ALLOWED_NAMESPACES) ? addToSet({}, cfg.ALLOWED_NAMESPACES, stringToString) : DEFAULT_ALLOWED_NAMESPACES;
-      URI_SAFE_ATTRIBUTES = objectHasOwnProperty(cfg, 'ADD_URI_SAFE_ATTR') && arrayIsArray(cfg.ADD_URI_SAFE_ATTR) ? addToSet(clone(DEFAULT_URI_SAFE_ATTRIBUTES), cfg.ADD_URI_SAFE_ATTR, transformCaseFunc) : DEFAULT_URI_SAFE_ATTRIBUTES;
-      DATA_URI_TAGS = objectHasOwnProperty(cfg, 'ADD_DATA_URI_TAGS') && arrayIsArray(cfg.ADD_DATA_URI_TAGS) ? addToSet(clone(DEFAULT_DATA_URI_TAGS), cfg.ADD_DATA_URI_TAGS, transformCaseFunc) : DEFAULT_DATA_URI_TAGS;
-      FORBID_CONTENTS = objectHasOwnProperty(cfg, 'FORBID_CONTENTS') && arrayIsArray(cfg.FORBID_CONTENTS) ? addToSet({}, cfg.FORBID_CONTENTS, transformCaseFunc) : DEFAULT_FORBID_CONTENTS;
-      FORBID_TAGS = objectHasOwnProperty(cfg, 'FORBID_TAGS') && arrayIsArray(cfg.FORBID_TAGS) ? addToSet({}, cfg.FORBID_TAGS, transformCaseFunc) : clone({});
-      FORBID_ATTR = objectHasOwnProperty(cfg, 'FORBID_ATTR') && arrayIsArray(cfg.FORBID_ATTR) ? addToSet({}, cfg.FORBID_ATTR, transformCaseFunc) : clone({});
+      ALLOWED_TAGS = _resolveSetOption(cfg, 'ALLOWED_TAGS', DEFAULT_ALLOWED_TAGS, {
+        transform: transformCaseFunc
+      });
+      ALLOWED_ATTR = _resolveSetOption(cfg, 'ALLOWED_ATTR', DEFAULT_ALLOWED_ATTR, {
+        transform: transformCaseFunc
+      });
+      ALLOWED_NAMESPACES = _resolveSetOption(cfg, 'ALLOWED_NAMESPACES', DEFAULT_ALLOWED_NAMESPACES, {
+        transform: stringToString
+      });
+      URI_SAFE_ATTRIBUTES = _resolveSetOption(cfg, 'ADD_URI_SAFE_ATTR', DEFAULT_URI_SAFE_ATTRIBUTES, {
+        transform: transformCaseFunc,
+        base: DEFAULT_URI_SAFE_ATTRIBUTES
+      });
+      DATA_URI_TAGS = _resolveSetOption(cfg, 'ADD_DATA_URI_TAGS', DEFAULT_DATA_URI_TAGS, {
+        transform: transformCaseFunc,
+        base: DEFAULT_DATA_URI_TAGS
+      });
+      FORBID_CONTENTS = _resolveSetOption(cfg, 'FORBID_CONTENTS', DEFAULT_FORBID_CONTENTS, {
+        transform: transformCaseFunc
+      });
+      FORBID_TAGS = _resolveSetOption(cfg, 'FORBID_TAGS', clone({}), {
+        transform: transformCaseFunc
+      });
+      FORBID_ATTR = _resolveSetOption(cfg, 'FORBID_ATTR', clone({}), {
+        transform: transformCaseFunc
+      });
       USE_PROFILES = objectHasOwnProperty(cfg, 'USE_PROFILES') ? cfg.USE_PROFILES && typeof cfg.USE_PROFILES === 'object' ? clone(cfg.USE_PROFILES) : cfg.USE_PROFILES : false;
       ALLOW_ARIA_ATTR = cfg.ALLOW_ARIA_ATTR !== false; // Default true
       ALLOW_DATA_ATTR = cfg.ALLOW_DATA_ATTR !== false; // Default true
@@ -694,8 +777,8 @@
       IN_PLACE = cfg.IN_PLACE || false; // Default false
       IS_ALLOWED_URI$1 = isRegex(cfg.ALLOWED_URI_REGEXP) ? cfg.ALLOWED_URI_REGEXP : IS_ALLOWED_URI; // Default regexp
       NAMESPACE = typeof cfg.NAMESPACE === 'string' ? cfg.NAMESPACE : HTML_NAMESPACE; // Default HTML namespace
-      MATHML_TEXT_INTEGRATION_POINTS = objectHasOwnProperty(cfg, 'MATHML_TEXT_INTEGRATION_POINTS') && cfg.MATHML_TEXT_INTEGRATION_POINTS && typeof cfg.MATHML_TEXT_INTEGRATION_POINTS === 'object' ? clone(cfg.MATHML_TEXT_INTEGRATION_POINTS) : addToSet({}, ['mi', 'mo', 'mn', 'ms', 'mtext']); // Default built-in map
-      HTML_INTEGRATION_POINTS = objectHasOwnProperty(cfg, 'HTML_INTEGRATION_POINTS') && cfg.HTML_INTEGRATION_POINTS && typeof cfg.HTML_INTEGRATION_POINTS === 'object' ? clone(cfg.HTML_INTEGRATION_POINTS) : addToSet({}, ['annotation-xml']); // Default built-in map
+      MATHML_TEXT_INTEGRATION_POINTS = objectHasOwnProperty(cfg, 'MATHML_TEXT_INTEGRATION_POINTS') && cfg.MATHML_TEXT_INTEGRATION_POINTS && typeof cfg.MATHML_TEXT_INTEGRATION_POINTS === 'object' ? clone(cfg.MATHML_TEXT_INTEGRATION_POINTS) : addToSet({}, DEFAULT_MATHML_TEXT_INTEGRATION_POINTS); // Default built-in map
+      HTML_INTEGRATION_POINTS = objectHasOwnProperty(cfg, 'HTML_INTEGRATION_POINTS') && cfg.HTML_INTEGRATION_POINTS && typeof cfg.HTML_INTEGRATION_POINTS === 'object' ? clone(cfg.HTML_INTEGRATION_POINTS) : addToSet({}, DEFAULT_HTML_INTEGRATION_POINTS); // Default built-in map
       const customElementHandling = objectHasOwnProperty(cfg, 'CUSTOM_ELEMENT_HANDLING') && cfg.CUSTOM_ELEMENT_HANDLING && typeof cfg.CUSTOM_ELEMENT_HANDLING === 'object' ? clone(cfg.CUSTOM_ELEMENT_HANDLING) : create(null);
       CUSTOM_ELEMENT_HANDLING = create(null);
       if (objectHasOwnProperty(customElementHandling, 'tagNameCheck') && isRegexOrFunction(customElementHandling.tagNameCheck)) {
@@ -707,6 +790,7 @@
       if (objectHasOwnProperty(customElementHandling, 'allowCustomizedBuiltInElements') && typeof customElementHandling.allowCustomizedBuiltInElements === 'boolean') {
         CUSTOM_ELEMENT_HANDLING.allowCustomizedBuiltInElements = customElementHandling.allowCustomizedBuiltInElements; // Default undefined
       }
+      seal(CUSTOM_ELEMENT_HANDLING);
       if (SAFE_FOR_TEMPLATES) {
         ALLOW_DATA_ATTR = false;
       }
@@ -790,6 +874,13 @@
         addToSet(ALLOWED_TAGS, ['tbody']);
         delete FORBID_TAGS.tbody;
       }
+      // Re-derive the active Trusted Types policy from this configuration on
+      // every parse. The active policy must never be sticky closure state that
+      // outlives the config that set it: a caller-supplied policy left in place
+      // after `clearConfig()` — or after a later call that supplied none, or
+      // `TRUSTED_TYPES_POLICY: null` — could sign a subsequent "default"
+      // `RETURN_TRUSTED_TYPE` result with a foreign, possibly unsafe policy.
+      // See GHSA-vxr8-fq34-vvx9.
       if (cfg.TRUSTED_TYPES_POLICY) {
         if (typeof cfg.TRUSTED_TYPES_POLICY.createHTML !== 'function') {
           throw typeErrorCreate('TRUSTED_TYPES_POLICY configuration option must provide a "createHTML" hook.');
@@ -797,7 +888,7 @@
         if (typeof cfg.TRUSTED_TYPES_POLICY.createScriptURL !== 'function') {
           throw typeErrorCreate('TRUSTED_TYPES_POLICY configuration option must provide a "createScriptURL" hook.');
         }
-        // Overwrite existing TrustedTypes policy.
+        // A caller-supplied policy applies to this configuration only.
         const previousTrustedTypesPolicy = trustedTypesPolicy;
         trustedTypesPolicy = cfg.TRUSTED_TYPES_POLICY;
         // Sign local variables required by `sanitize`. If the supplied policy's
@@ -810,16 +901,30 @@
           trustedTypesPolicy = previousTrustedTypesPolicy;
           throw error;
         }
+      } else if (cfg.TRUSTED_TYPES_POLICY === null) {
+        // Explicit opt-out for this call: perform no Trusted Types signing and
+        // create nothing (so a strict `trusted-types` CSP that disallows a
+        // `dompurify` policy can still call `sanitize` from inside its own
+        // policy — see #1422). Resetting to `undefined` rather than a sticky
+        // `null` also drops any previously retained caller policy, so it cannot
+        // resurface on a later call, while still allowing the next config-less
+        // call to restore the internal default policy. See GHSA-vxr8-fq34-vvx9.
+        trustedTypesPolicy = undefined;
+        emptyHTML = '';
       } else {
-        // Uninitialized policy, attempt to initialize the internal dompurify policy.
-        if (trustedTypesPolicy === undefined && cfg.TRUSTED_TYPES_POLICY !== null) {
-          trustedTypesPolicy = _createTrustedTypesPolicy(trustedTypes, currentScript);
-        }
-        // If creating the internal policy succeeded sign internal variables.
-        // Note: a falsy `trustedTypesPolicy` (null when policy creation failed or
-        // was skipped via `TRUSTED_TYPES_POLICY: null`, or undefined when no
-        // policy has been initialized yet) must be excluded here, otherwise we
-        // would call `.createHTML` on a non-policy and throw. See #1422.
+        // No policy supplied: keep the currently active policy if one is set — a
+        // previously supplied policy is intentionally sticky across config-less
+        // calls — otherwise fall back to the instance's own internal policy,
+        // created at most once. (A policy supplied for a *single* call still
+        // lingers by design; what must not linger is a policy whose configuration
+        // has been torn down via `clearConfig()`, which restores the default.)
+        if (trustedTypesPolicy === undefined) {
+          trustedTypesPolicy = _getDefaultTrustedTypesPolicy();
+        }
+        // Sign internal variables only when a policy is active. A falsy policy
+        // (Trusted Types unsupported, creation failed, or an explicit opt-out)
+        // leaves `emptyHTML` as a plain string, so we never call `.createHTML` on
+        // a non-policy and throw. See #1422.
         if (trustedTypesPolicy && typeof emptyHTML === 'string') {
           emptyHTML = _createTrustedHTML('');
         }
@@ -851,6 +956,77 @@
      * correctly. */
     const ALL_SVG_TAGS = addToSet({}, [...svg$1, ...svgFilters, ...svgDisallowed]);
     const ALL_MATHML_TAGS = addToSet({}, [...mathMl$1, ...mathMlDisallowed]);
+    /**
+     * Namespace rules for an element in the SVG namespace.
+     *
+     * @param tagName the element's lowercase tag name
+     * @param parent the (possibly simulated) parent node
+     * @param parentTagName the parent's lowercase tag name
+     * @returns true if a spec-compliant parser could produce this element
+     */
+    const _checkSvgNamespace = function _checkSvgNamespace(tagName, parent, parentTagName) {
+      // The only way to switch from HTML namespace to SVG
+      // is via <svg>. If it happens via any other tag, then
+      // it should be killed.
+      if (parent.namespaceURI === HTML_NAMESPACE) {
+        return tagName === 'svg';
+      }
+      // The only way to switch from MathML to SVG is via <svg>
+      // if the parent is either <annotation-xml> or a MathML
+      // text integration point.
+      if (parent.namespaceURI === MATHML_NAMESPACE) {
+        return tagName === 'svg' && (parentTagName === 'annotation-xml' || MATHML_TEXT_INTEGRATION_POINTS[parentTagName]);
+      }
+      // We only allow elements that are defined in SVG
+      // spec. All others are disallowed in SVG namespace.
+      return Boolean(ALL_SVG_TAGS[tagName]);
+    };
+    /**
+     * Namespace rules for an element in the MathML namespace.
+     *
+     * @param tagName the element's lowercase tag name
+     * @param parent the (possibly simulated) parent node
+     * @param parentTagName the parent's lowercase tag name
+     * @returns true if a spec-compliant parser could produce this element
+     */
+    const _checkMathMlNamespace = function _checkMathMlNamespace(tagName, parent, parentTagName) {
+      // The only way to switch from HTML namespace to MathML
+      // is via <math>. If it happens via any other tag, then
+      // it should be killed.
+      if (parent.namespaceURI === HTML_NAMESPACE) {
+        return tagName === 'math';
+      }
+      // The only way to switch from SVG to MathML is via
+      // <math> and HTML integration points
+      if (parent.namespaceURI === SVG_NAMESPACE) {
+        return tagName === 'math' && HTML_INTEGRATION_POINTS[parentTagName];
+      }
+      // We only allow elements that are defined in MathML
+      // spec. All others are disallowed in MathML namespace.
+      return Boolean(ALL_MATHML_TAGS[tagName]);
+    };
+    /**
+     * Namespace rules for an element in the HTML namespace.
+     *
+     * @param tagName the element's lowercase tag name
+     * @param parent the (possibly simulated) parent node
+     * @param parentTagName the parent's lowercase tag name
+     * @returns true if a spec-compliant parser could produce this element
+     */
+    const _checkHtmlNamespace = function _checkHtmlNamespace(tagName, parent, parentTagName) {
+      // The only way to switch from SVG to HTML is via
+      // HTML integration points, and from MathML to HTML
+      // is via MathML text integration points
+      if (parent.namespaceURI === SVG_NAMESPACE && !HTML_INTEGRATION_POINTS[parentTagName]) {
+        return false;
+      }
+      if (parent.namespaceURI === MATHML_NAMESPACE && !MATHML_TEXT_INTEGRATION_POINTS[parentTagName]) {
+        return false;
+      }
+      // We disallow tags that are specific for MathML
+      // or SVG and should never appear in HTML namespace
+      return !ALL_MATHML_TAGS[tagName] && (COMMON_SVG_AND_HTML_ELEMENTS[tagName] || !ALL_SVG_TAGS[tagName]);
+    };
     /**
      * @param element a DOM element whose namespace is being checked
      * @returns Return false if the element has a
@@ -873,51 +1049,13 @@
         return false;
       }
       if (element.namespaceURI === SVG_NAMESPACE) {
-        // The only way to switch from HTML namespace to SVG
-        // is via <svg>. If it happens via any other tag, then
-        // it should be killed.
-        if (parent.namespaceURI === HTML_NAMESPACE) {
-          return tagName === 'svg';
-        }
-        // The only way to switch from MathML to SVG is via`
-        // svg if parent is either <annotation-xml> or MathML
-        // text integration points.
-        if (parent.namespaceURI === MATHML_NAMESPACE) {
-          return tagName === 'svg' && (parentTagName === 'annotation-xml' || MATHML_TEXT_INTEGRATION_POINTS[parentTagName]);
-        }
-        // We only allow elements that are defined in SVG
-        // spec. All others are disallowed in SVG namespace.
-        return Boolean(ALL_SVG_TAGS[tagName]);
+        return _checkSvgNamespace(tagName, parent, parentTagName);
       }
       if (element.namespaceURI === MATHML_NAMESPACE) {
-        // The only way to switch from HTML namespace to MathML
-        // is via <math>. If it happens via any other tag, then
-        // it should be killed.
-        if (parent.namespaceURI === HTML_NAMESPACE) {
-          return tagName === 'math';
-        }
-        // The only way to switch from SVG to MathML is via
-        // <math> and HTML integration points
-        if (parent.namespaceURI === SVG_NAMESPACE) {
-          return tagName === 'math' && HTML_INTEGRATION_POINTS[parentTagName];
-        }
-        // We only allow elements that are defined in MathML
-        // spec. All others are disallowed in MathML namespace.
-        return Boolean(ALL_MATHML_TAGS[tagName]);
+        return _checkMathMlNamespace(tagName, parent, parentTagName);
       }
       if (element.namespaceURI === HTML_NAMESPACE) {
-        // The only way to switch from SVG to HTML is via
-        // HTML integration points, and from MathML to HTML
-        // is via MathML text integration points
-        if (parent.namespaceURI === SVG_NAMESPACE && !HTML_INTEGRATION_POINTS[parentTagName]) {
-          return false;
-        }
-        if (parent.namespaceURI === MATHML_NAMESPACE && !MATHML_TEXT_INTEGRATION_POINTS[parentTagName]) {
-          return false;
-        }
-        // We disallow tags that are specific for MathML
-        // or SVG and should never appear in HTML namespace
-        return !ALL_MATHML_TAGS[tagName] && (COMMON_SVG_AND_HTML_ELEMENTS[tagName] || !ALL_SVG_TAGS[tagName]);
+        return _checkHtmlNamespace(tagName, parent, parentTagName);
       }
       // For XHTML and XML documents that support custom namespaces
       if (PARSER_MEDIA_TYPE === 'application/xhtml+xml' && ALLOWED_NAMESPACES[element.namespaceURI]) {
@@ -942,7 +1080,74 @@
         // eslint-disable-next-line unicorn/prefer-dom-node-remove
         getParentNode(node).removeChild(node);
       } catch (_) {
+        /* The normal detach failed — this is reached for a parentless node
+           (getParentNode() is null, so .removeChild throws). Element.prototype
+           .remove() is itself a spec no-op on a parentless node, so a recorded
+           "removal" would otherwise hand the caller back an intact,
+           payload-bearing node (e.g. a detached IN_PLACE root the mXSS canary or
+           the style-with-element-child rule decided to kill). Fail closed by
+           throwing — exactly as a clobbered root does at the IN_PLACE entry —
+           rather than trying to "neutralize" the node via its own methods.
+           Neutralizing would mean calling getAttributeNames()/removeAttribute()
+           on the node, both of which a <form> root can clobber via a named child
+           (and _isClobbered does not even probe getAttributeNames), so the
+           neutralize step could itself be silently defeated, leaving the payload
+           intact. A throw touches only the cached, clobber-safe remove() and
+           getParentNode(). Generalizes GHSA-r47g-fvhr-h676 (clobbered-form root)
+           to every root-kill reason. REPORT-3.
+                  This lives inside the catch, so it never fires for a normally-removed
+           in-tree node: those have a parent, removeChild() succeeds, and the
+           catch is not entered. Only a kept (parentless) root reaches here. */
         remove(node);
+        if (!getParentNode(node)) {
+          throw typeErrorCreate('a node selected for removal could not be detached from its tree ' + 'and cannot be safely returned; refusing to sanitize in place');
+        }
+      }
+    };
+    /**
+     * _neutralizeRoot
+     *
+     * Fail-closed teardown of an in-place root after the sanitize walk aborts
+     * (campaign-3 F2). An internal throw mid-walk — e.g. a page-registered
+     * custom element's reaction detaches a node so `_forceRemove`'s deliberate
+     * parentless guard throws, or any other re-entrant engine mutation — would
+     * otherwise leave the caller's *live* tree half-sanitized, with everything
+     * after the abort point still carrying its handlers. There is no safe way
+     * to resume the walk (the tree mutated under us), so we strip the root bare:
+     * remove every child and every attribute, then let the caller's catch see
+     * the original error. Clobber-safe (cached `remove`/`childNodes`/`attributes`
+     * getters; the root was already clobber-pre-flighted at the IN_PLACE entry).
+     *
+     * @param root the in-place root to empty
+     */
+    const _neutralizeRoot = function _neutralizeRoot(root) {
+      const childNodes = getChildNodes(root);
+      if (childNodes) {
+        const snapshot = [];
+        arrayForEach(childNodes, child => {
+          arrayPush(snapshot, child);
+        });
+        arrayForEach(snapshot, child => {
+          try {
+            remove(child);
+          } catch (_) {
+            /* Best-effort teardown; a still-attached child is handled below */
+          }
+        });
+      }
+      const attributes = getAttributes(root);
+      if (attributes) {
+        for (let i = attributes.length - 1; i >= 0; --i) {
+          const attribute = attributes[i];
+          const name = attribute && attribute.name;
+          if (typeof name === 'string') {
+            try {
+              root.removeAttribute(name);
+            } catch (_) {
+              /* Clobbered removeAttribute — ignore (fail-closed best effort) */
+            }
+          }
+        }
       }
     };
     /**
@@ -977,6 +1182,72 @@
         }
       }
     };
+    /**
+     * _stripDisallowedAttributes
+     *
+     * Removes every attribute the active configuration does not allow from a
+     * single element, using the same allowlist as the main attribute pass (so
+     * `on*` handlers go, but no `/^on/` blocklist is introduced). Used only to
+     * neutralise nodes that are being discarded from an in-place tree.
+     *
+     * @param element the element to strip
+     */
+    const _stripDisallowedAttributes = function _stripDisallowedAttributes(element) {
+      const attributes = getAttributes(element);
+      if (!attributes) {
+        return;
+      }
+      for (let i = attributes.length - 1; i >= 0; --i) {
+        const attribute = attributes[i];
+        const name = attribute && attribute.name;
+        if (typeof name !== 'string' || ALLOWED_ATTR[transformCaseFunc(name)]) {
+          continue;
+        }
+        try {
+          element.removeAttribute(name);
+        } catch (_) {
+          /* Clobbered removeAttribute on a doomed node — ignore */
+        }
+      }
+    };
+    /**
+     * _neutralizeSubtree
+     *
+     * Completes the audit-5 F1 fix across every removal path. The KEEP_CONTENT
+     * move-hoist neutralises only disallowed-tag removals; clobber, mXSS-canary,
+     * namespace, comment, processing-instruction and KEEP_CONTENT:false removals
+     * all drop their subtree wholesale via `_forceRemove`. On the IN_PLACE path
+     * those dropped nodes are detached from the caller's LIVE tree but a
+     * handler-bearing original among them (an `<img onerror>`/`<video>` that was
+     * loading) keeps its queued resource event, which fires in page scope after
+     * sanitize returns. This walks a removed subtree and strips every attribute
+     * the active configuration does not allow — so `on*` handlers are cancelled
+     * through the SAME allowlist that governs kept nodes, not a separate `/^on/`
+     * blocklist. Run synchronously before sanitize returns, i.e. before any
+     * queued event can fire. Hook-free by design: these nodes leave the output,
+     * so firing attribute hooks for them would be surprising. Clobber-safe reads;
+     * a doomed clobbered node may shadow `removeAttribute` (its own attributes are
+     * irrelevant — it is discarded — while its non-clobbered descendants, e.g.
+     * the `<img>`, are reached and scrubbed).
+     *
+     * @param root the root of a removed subtree to neutralise
+     */
+    const _neutralizeSubtree = function _neutralizeSubtree(root) {
+      const stack = [root];
+      while (stack.length > 0) {
+        const node = stack.pop();
+        const nodeType = getNodeType ? getNodeType(node) : node.nodeType;
+        if (nodeType === NODE_TYPE.element) {
+          _stripDisallowedAttributes(node);
+        }
+        const childNodes = getChildNodes(node);
+        if (childNodes) {
+          for (let i = childNodes.length - 1; i >= 0; --i) {
+            stack.push(childNodes[i]);
+          }
+        }
+      }
+    };
     /**
      * _initDocument
      *
@@ -1038,6 +1309,20 @@
       // eslint-disable-next-line no-bitwise
       NodeFilter.SHOW_ELEMENT | NodeFilter.SHOW_COMMENT | NodeFilter.SHOW_TEXT | NodeFilter.SHOW_PROCESSING_INSTRUCTION | NodeFilter.SHOW_CDATA_SECTION, null);
     };
+    /**
+     * Replace template expression syntax (mustache, ERB, template
+     * literal) with a space; shared by all SAFE_FOR_TEMPLATES scrub
+     * sites. Order matters: mustache, then ERB, then template literal.
+     *
+     * @param value the string to scrub
+     * @returns the scrubbed string
+     */
+    const _stripTemplateExpressions = function _stripTemplateExpressions(value) {
+      value = stringReplace(value, MUSTACHE_EXPR$1, ' ');
+      value = stringReplace(value, ERB_EXPR$1, ' ');
+      value = stringReplace(value, TMPLIT_EXPR$1, ' ');
+      return value;
+    };
     /**
      * Strip template-engine expressions ({{...}}, ${...}, <%...%>) from the
      * character data of an element subtree. Used as the final safety net for
@@ -1058,29 +1343,27 @@
      * @param node The root element whose character data should be scrubbed.
      */
     const _scrubTemplateExpressions2 = function _scrubTemplateExpressions(node) {
-      var _node$querySelectorAl, _node$querySelectorAl2;
+      var _node$querySelectorAl;
       node.normalize();
       const walker = createNodeIterator.call(node.ownerDocument || node, node,
       // eslint-disable-next-line no-bitwise
       NodeFilter.SHOW_TEXT | NodeFilter.SHOW_COMMENT | NodeFilter.SHOW_CDATA_SECTION | NodeFilter.SHOW_PROCESSING_INSTRUCTION, null);
       let currentNode = walker.nextNode();
       while (currentNode) {
-        let data = currentNode.data;
-        arrayForEach([MUSTACHE_EXPR$1, ERB_EXPR$1, TMPLIT_EXPR$1], expr => {
-          data = stringReplace(data, expr, ' ');
-        });
-        currentNode.data = data;
+        currentNode.data = _stripTemplateExpressions(currentNode.data);
         currentNode = walker.nextNode();
       }
       // NodeIterator does not descend into <template>.content per the DOM spec,
       // so we must explicitly recurse into each template's content fragment,
       // mirroring the approach used by _sanitizeShadowDOM.
-      const templates = (_node$querySelectorAl = (_node$querySelectorAl2 = node.querySelectorAll) === null || _node$querySelectorAl2 === void 0 ? void 0 : _node$querySelectorAl2.call(node, 'template')) !== null && _node$querySelectorAl !== void 0 ? _node$querySelectorAl : [];
-      arrayForEach(Array.from(templates), tmpl => {
-        if (_isDocumentFragment(tmpl.content)) {
-          _scrubTemplateExpressions2(tmpl.content);
-        }
-      });
+      const templates = (_node$querySelectorAl = node.querySelectorAll) === null || _node$querySelectorAl === void 0 ? void 0 : _node$querySelectorAl.call(node, 'template');
+      if (templates) {
+        arrayForEach(templates, tmpl => {
+          if (_isDocumentFragment(tmpl.content)) {
+            _scrubTemplateExpressions2(tmpl.content);
+          }
+        });
+      }
     };
     /**
      * _isClobbered
@@ -1176,10 +1459,104 @@
       }
     };
     function _executeHooks(hooks, currentNode, data) {
+      if (hooks.length === 0) {
+        return;
+      }
       arrayForEach(hooks, hook => {
         hook.call(DOMPurify, currentNode, data, CONFIG);
       });
     }
+    /**
+     * Structural-threat checks that condemn a node regardless of the
+     * allowlists: mXSS via namespace confusion, risky CSS construction,
+     * processing instructions, markup-bearing comments. Pure predicate;
+     * the caller removes. Check order is load-bearing.
+     *
+     * @param currentNode the node to inspect
+     * @param tagName the node's transformCaseFunc'd tag name
+     * @return true if the node must be removed
+     */
+    const _isUnsafeNode = function _isUnsafeNode(currentNode, tagName) {
+      /* Detect mXSS attempts abusing namespace confusion */
+      if (SAFE_FOR_XML && currentNode.hasChildNodes() && !_isNode(currentNode.firstElementChild) && regExpTest(ELEMENT_MARKUP_PROBE, currentNode.textContent) && regExpTest(ELEMENT_MARKUP_PROBE, currentNode.innerHTML)) {
+        return true;
+      }
+      /* Remove risky CSS construction leading to mXSS */
+      if (SAFE_FOR_XML && currentNode.namespaceURI === HTML_NAMESPACE && tagName === 'style' && _isNode(currentNode.firstElementChild)) {
+        return true;
+      }
+      /* Remove any occurrence of processing instructions */
+      if (currentNode.nodeType === NODE_TYPE.processingInstruction) {
+        return true;
+      }
+      /* Remove any kind of possibly harmful comments */
+      if (SAFE_FOR_XML && currentNode.nodeType === NODE_TYPE.comment && regExpTest(COMMENT_MARKUP_PROBE, currentNode.data)) {
+        return true;
+      }
+      return false;
+    };
+    /**
+     * Handle a node whose tag is forbidden or not allowlisted: keep
+     * allowed custom elements (false return exits _sanitizeElements
+     * early - namespace/fallback checks and the afterSanitizeElements
+     * hook are intentionally skipped for kept custom elements), else
+     * hoist content per KEEP_CONTENT and remove.
+     *
+     * @param currentNode the disallowed node
+     * @param tagName the node's transformCaseFunc'd tag name
+     * @return true if the node was removed, false if kept
+     */
+    const _sanitizeDisallowedNode = function _sanitizeDisallowedNode(currentNode, tagName) {
+      /* Check if we have a custom element to handle */
+      if (!FORBID_TAGS[tagName] && _isBasicCustomElement(tagName)) {
+        if (CUSTOM_ELEMENT_HANDLING.tagNameCheck instanceof RegExp && regExpTest(CUSTOM_ELEMENT_HANDLING.tagNameCheck, tagName)) {
+          return false;
+        }
+        if (CUSTOM_ELEMENT_HANDLING.tagNameCheck instanceof Function && CUSTOM_ELEMENT_HANDLING.tagNameCheck(tagName)) {
+          return false;
+        }
+      }
+      /* Keep content except for bad-listed elements.
+           Use the cached prototype getters exclusively — the previous code
+           had `|| currentNode.parentNode` / `|| currentNode.childNodes`
+           fallbacks, but the cached getters always return the canonical
+           value (or null for a real parent-less node), so the fallback
+           path was dead in safe cases and a clobbering surface in unsafe
+           ones. Falsy cached results stay falsy; the `if (childNodes &&
+           parentNode)` check already gates correctly. */
+      if (KEEP_CONTENT && !FORBID_CONTENTS[tagName]) {
+        const parentNode = getParentNode(currentNode);
+        const childNodes = getChildNodes(currentNode);
+        if (childNodes && parentNode) {
+          const childCount = childNodes.length;
+          /* In-place: hoist the *original* children so the iterator visits
+               and sanitises them through the same allowlist pass as every other
+               node. The caller built the tree in the live document, so the
+               originals carry already-queued resource events (`<img onerror>`,
+               `<video>`/`<audio>` error, lazy/`onload`, …); cloning would leave
+               those originals detached but still armed, firing in page scope
+               while the returned tree looked clean. Moving is safe in-place: the
+               root is pre-validated as an allowed tag and so is never the node
+               being removed, which keeps `parentNode` inside the iterator root
+               and the relocated child inside the serialised tree.
+                        Otherwise (string / DOM-copy paths): clone. The iterator is rooted
+               at — and the result serialised from — `body`, so a restrictive
+               ALLOWED_TAGS that removes `body` itself must leave its content in
+               place, which only cloning does; and those paths parse into an
+               inert document, so their discarded originals never had a queued
+               event to neutralise.
+                        `childNodes` is live; a tail-to-head walk keeps `childNodes[i]`
+               valid whether we move (drops the trailing entry) or clone (leaves
+               the list intact). */
+          for (let i = childCount - 1; i >= 0; --i) {
+            const hoisted = IN_PLACE ? childNodes[i] : cloneNode(childNodes[i], true);
+            parentNode.insertBefore(hoisted, getNextSibling(currentNode));
+          }
+        }
+      }
+      _forceRemove(currentNode);
+      return true;
+    };
     /**
      * _sanitizeElements
      *
@@ -1190,7 +1567,6 @@
      * @return true if node was killed, false if left alive
      */
     const _sanitizeElements = function _sanitizeElements(currentNode) {
-      let content = null;
       /* Execute a hook if present */
       _executeHooks(hooks.beforeSanitizeElements, currentNode, null);
       /* Check if element is clobbered or can clobber */
@@ -1205,58 +1581,14 @@
         tagName,
         allowedTags: ALLOWED_TAGS
       });
-      /* Detect mXSS attempts abusing namespace confusion */
-      if (SAFE_FOR_XML && currentNode.hasChildNodes() && !_isNode(currentNode.firstElementChild) && regExpTest(/<[/\w!]/g, currentNode.innerHTML) && regExpTest(/<[/\w!]/g, currentNode.textContent)) {
-        _forceRemove(currentNode);
-        return true;
-      }
-      /* Remove risky CSS construction leading to mXSS */
-      if (SAFE_FOR_XML && currentNode.namespaceURI === HTML_NAMESPACE && tagName === 'style' && _isNode(currentNode.firstElementChild)) {
-        _forceRemove(currentNode);
-        return true;
-      }
-      /* Remove any occurrence of processing instructions */
-      if (currentNode.nodeType === NODE_TYPE.progressingInstruction) {
-        _forceRemove(currentNode);
-        return true;
-      }
-      /* Remove any kind of possibly harmful comments */
-      if (SAFE_FOR_XML && currentNode.nodeType === NODE_TYPE.comment && regExpTest(/<[/\w]/g, currentNode.data)) {
+      /* Remove mXSS vectors, processing instructions and risky comments */
+      if (_isUnsafeNode(currentNode, tagName)) {
         _forceRemove(currentNode);
         return true;
       }
       /* Remove element if anything forbids its presence */
       if (FORBID_TAGS[tagName] || !(EXTRA_ELEMENT_HANDLING.tagCheck instanceof Function && EXTRA_ELEMENT_HANDLING.tagCheck(tagName)) && !ALLOWED_TAGS[tagName]) {
-        /* Check if we have a custom element to handle */
-        if (!FORBID_TAGS[tagName] && _isBasicCustomElement(tagName)) {
-          if (CUSTOM_ELEMENT_HANDLING.tagNameCheck instanceof RegExp && regExpTest(CUSTOM_ELEMENT_HANDLING.tagNameCheck, tagName)) {
-            return false;
-          }
-          if (CUSTOM_ELEMENT_HANDLING.tagNameCheck instanceof Function && CUSTOM_ELEMENT_HANDLING.tagNameCheck(tagName)) {
-            return false;
-          }
-        }
-        /* Keep content except for bad-listed elements.
-           Use the cached prototype getters exclusively — the previous code
-           had `|| currentNode.parentNode` / `|| currentNode.childNodes`
-           fallbacks, but the cached getters always return the canonical
-           value (or null for a real parent-less node), so the fallback
-           path was dead in safe cases and a clobbering surface in unsafe
-           ones. Falsy cached results stay falsy; the `if (childNodes &&
-           parentNode)` check already gates correctly. */
-        if (KEEP_CONTENT && !FORBID_CONTENTS[tagName]) {
-          const parentNode = getParentNode(currentNode);
-          const childNodes = getChildNodes(currentNode);
-          if (childNodes && parentNode) {
-            const childCount = childNodes.length;
-            for (let i = childCount - 1; i >= 0; --i) {
-              const childClone = cloneNode(childNodes[i], true);
-              parentNode.insertBefore(childClone, getNextSibling(currentNode));
-            }
-          }
-        }
-        _forceRemove(currentNode);
-        return true;
+        return _sanitizeDisallowedNode(currentNode, tagName);
       }
       /* Check whether element has a valid namespace.
          Realm-safe check (GHSA-hpcv-96wg-7vj8): use the cached Node.prototype
@@ -1270,17 +1602,14 @@
         return true;
       }
       /* Make sure that older browsers don't get fallback-tag mXSS */
-      if ((tagName === 'noscript' || tagName === 'noembed' || tagName === 'noframes') && regExpTest(/<\/no(script|embed|frames)/i, currentNode.innerHTML)) {
+      if ((tagName === 'noscript' || tagName === 'noembed' || tagName === 'noframes') && regExpTest(FALLBACK_TAG_CLOSE, currentNode.innerHTML)) {
         _forceRemove(currentNode);
         return true;
       }
       /* Sanitize element content to be template-safe */
       if (SAFE_FOR_TEMPLATES && currentNode.nodeType === NODE_TYPE.text) {
         /* Get the element's text content */
-        content = currentNode.textContent;
-        arrayForEach([MUSTACHE_EXPR$1, ERB_EXPR$1, TMPLIT_EXPR$1], expr => {
-          content = stringReplace(content, expr, ' ');
-        });
+        const content = _stripTemplateExpressions(currentNode.textContent);
         if (currentNode.textContent !== content) {
           arrayPush(DOMPurify.removed, {
             element: currentNode.cloneNode()
@@ -1315,7 +1644,7 @@
           (https://html.spec.whatwg.org/multipage/dom.html#embedding-custom-non-visible-data-with-the-data-*-attributes)
           XML-compatible (https://html.spec.whatwg.org/multipage/infrastructure.html#xml-compatible and http://www.w3.org/TR/xml/#d0e804)
           We don't need to check the value; it's always URI safe. */
-      if (ALLOW_DATA_ATTR && !FORBID_ATTR[lcName] && regExpTest(DATA_ATTR$1, lcName)) ; else if (ALLOW_ARIA_ATTR && regExpTest(ARIA_ATTR$1, lcName)) ; else if (!nameIsPermitted || FORBID_ATTR[lcName]) {
+      if (ALLOW_DATA_ATTR && regExpTest(DATA_ATTR$1, lcName)) ; else if (ALLOW_ARIA_ATTR && regExpTest(ARIA_ATTR$1, lcName)) ; else if (!nameIsPermitted) {
         if (
         // First condition does a very basic check if a) it's basically a valid custom element tagname AND
         // b) if the tagName passes whatever the user has configured for CUSTOM_ELEMENT_HANDLING.tagNameCheck
@@ -1347,6 +1676,63 @@
     const _isBasicCustomElement = function _isBasicCustomElement(tagName) {
       return !RESERVED_CUSTOM_ELEMENT_NAMES[stringToLowerCase(tagName)] && regExpTest(CUSTOM_ELEMENT$1, tagName);
     };
+    /**
+     * Wrap an attribute value in the matching Trusted Types object when
+     * the active policy requires it. Namespaced attributes pass through
+     * unchanged (no TT support yet, see
+     * https://bugs.chromium.org/p/chromium/issues/detail?id=1305293).
+     *
+     * @param lcTag lowercase tag name of the containing element
+     * @param lcName lowercase attribute name
+     * @param namespaceURI the attribute's namespace, if any
+     * @param value the attribute value to wrap
+     * @return the value, wrapped when Trusted Types demand it
+     */
+    const _applyTrustedTypesToAttribute = function _applyTrustedTypesToAttribute(lcTag, lcName, namespaceURI, value) {
+      if (trustedTypesPolicy && typeof trustedTypes === 'object' && typeof trustedTypes.getAttributeType === 'function' && !namespaceURI) {
+        switch (trustedTypes.getAttributeType(lcTag, lcName)) {
+          case 'TrustedHTML':
+            {
+              return _createTrustedHTML(value);
+            }
+          case 'TrustedScriptURL':
+            {
+              return _createTrustedScriptURL(value);
+            }
+        }
+      }
+      return value;
+    };
+    /**
+     * Write a modified attribute value back onto the element. On
+     * success, re-probe for clobbering introduced by the new value and
+     * remove the element when found; otherwise pop the removal entry
+     * recorded by the earlier _removeAttribute (long-standing pairing
+     * with the SANITIZE_NAMED_PROPS path - do not "fix" casually). On
+     * failure, remove the attribute instead.
+     *
+     * @param currentNode the element carrying the attribute
+     * @param name the attribute name as present on the element
+     * @param namespaceURI the attribute's namespace, if any
+     * @param value the new attribute value
+     */
+    const _setAttributeValue = function _setAttributeValue(currentNode, name, namespaceURI, value) {
+      try {
+        if (namespaceURI) {
+          currentNode.setAttributeNS(namespaceURI, name, value);
+        } else {
+          /* Fallback to setAttribute() for browser-unrecognized namespaces e.g. "x-schema". */
+          currentNode.setAttribute(name, value);
+        }
+        if (_isClobbered(currentNode)) {
+          _forceRemove(currentNode);
+        } else {
+          arrayPop(DOMPurify.removed);
+        }
+      } catch (_) {
+        _removeAttribute(name, currentNode);
+      }
+    };
     /**
      * _sanitizeAttributes
      *
@@ -1373,6 +1759,7 @@
         forceKeepAttr: undefined
       };
       let l = attributes.length;
+      const lcTag = transformCaseFunc(currentNode.nodeName);
       /* Go backwards over all attributes; safely remove bad ones */
       while (l--) {
         const attr = attributes[l];
@@ -1410,7 +1797,7 @@
           _removeAttribute(name, currentNode);
           continue;
         }
-        /* Did the hooks approve of the attribute? */
+        /* Did the hooks force-keep the attribute? */
         if (hookEvent.forceKeepAttr) {
           continue;
         }
@@ -1420,56 +1807,24 @@
           continue;
         }
         /* Work around a security issue in jQuery 3.0 */
-        if (!ALLOW_SELF_CLOSE_IN_ATTR && regExpTest(/\/>/i, value)) {
+        if (!ALLOW_SELF_CLOSE_IN_ATTR && regExpTest(SELF_CLOSING_TAG, value)) {
           _removeAttribute(name, currentNode);
           continue;
         }
         /* Sanitize attribute content to be template-safe */
         if (SAFE_FOR_TEMPLATES) {
-          arrayForEach([MUSTACHE_EXPR$1, ERB_EXPR$1, TMPLIT_EXPR$1], expr => {
-            value = stringReplace(value, expr, ' ');
-          });
+          value = _stripTemplateExpressions(value);
         }
         /* Is `value` valid for this attribute? */
-        const lcTag = transformCaseFunc(currentNode.nodeName);
         if (!_isValidAttribute(lcTag, lcName, value)) {
           _removeAttribute(name, currentNode);
           continue;
         }
         /* Handle attributes that require Trusted Types */
-        if (trustedTypesPolicy && typeof trustedTypes === 'object' && typeof trustedTypes.getAttributeType === 'function') {
-          if (namespaceURI) ; else {
-            switch (trustedTypes.getAttributeType(lcTag, lcName)) {
-              case 'TrustedHTML':
-                {
-                  value = _createTrustedHTML(value);
-                  break;
-                }
-              case 'TrustedScriptURL':
-                {
-                  value = trustedTypesPolicy.createScriptURL(value);
-                  break;
-                }
-            }
-          }
-        }
+        value = _applyTrustedTypesToAttribute(lcTag, lcName, namespaceURI, value);
         /* Handle invalid data-* attribute set by try-catching it */
         if (value !== initValue) {
-          try {
-            if (namespaceURI) {
-              currentNode.setAttributeNS(namespaceURI, name, value);
-            } else {
-              /* Fallback to setAttribute() for browser-unrecognized namespaces e.g. "x-schema". */
-              currentNode.setAttribute(name, value);
-            }
-            if (_isClobbered(currentNode)) {
-              _forceRemove(currentNode);
-            } else {
-              arrayPop(DOMPurify.removed);
-            }
-          } catch (_) {
-            _removeAttribute(name, currentNode);
-          }
+          _setAttributeValue(currentNode, name, namespaceURI, value);
         }
       }
       /* Execute a hook if present */
@@ -1511,9 +1866,9 @@
            iterator also surfaces. */
         const shadowNodeType = getNodeType ? getNodeType(shadowNode) : shadowNode.nodeType;
         if (shadowNodeType === NODE_TYPE.element) {
-          const innerSr = getShadowRoot ? getShadowRoot(shadowNode) : shadowNode.shadowRoot;
+          const innerSr = getShadowRoot(shadowNode);
           if (_isDocumentFragment(innerSr)) {
-            _sanitizeAttachedShadowRoots2(innerSr);
+            _sanitizeAttachedShadowRoots(innerSr);
             _sanitizeShadowDOM2(innerSr);
           }
         }
@@ -1540,46 +1895,81 @@
      *
      * @param root the subtree root to walk for attached shadow roots
      */
-    const _sanitizeAttachedShadowRoots2 = function _sanitizeAttachedShadowRoots(root) {
-      const nodeType = getNodeType ? getNodeType(root) : root.nodeType;
-      if (nodeType === NODE_TYPE.element) {
-        const sr = getShadowRoot ? getShadowRoot(root) : root.shadowRoot;
-        // Realm-safe check (GHSA-hpcv-96wg-7vj8): use nodeType-based
-        // detection rather than `instanceof DocumentFragment`, which is
-        // realm-bound and silently skipped shadow roots whose host element
-        // belonged to a foreign realm (e.g. iframe.contentDocument
-        // attachShadow). A foreign-realm ShadowRoot extends the foreign
-        // realm's DocumentFragment, not ours, so the old instanceof check
-        // returned false and the shadow subtree was never walked.
-        if (_isDocumentFragment(sr)) {
-          // Recurse first so that nested shadow roots are reached even if
-          // _sanitizeShadowDOM removes hosts at this level.
-          _sanitizeAttachedShadowRoots2(sr);
-          _sanitizeShadowDOM2(sr);
-        }
-      }
-      // Snapshot children before recursing. Sanitization of one subtree
-      // (e.g. via an uponSanitizeShadowNode hook) may detach siblings,
-      // and naive nextSibling traversal would silently skip the rest of
-      // the list once a node is detached.
-      const childNodes = getChildNodes ? getChildNodes(root) : root.childNodes;
-      if (!childNodes) {
-        return;
-      }
-      const snapshot = [];
-      arrayForEach(childNodes, child => {
-        arrayPush(snapshot, child);
-      });
-      for (const child of snapshot) {
-        _sanitizeAttachedShadowRoots2(child);
-      }
-      /* When the root is a <template>, also descend into root.content */
-      if (nodeType === NODE_TYPE.element) {
-        const rootName = getNodeName ? getNodeName(root) : null;
-        if (typeof rootName === 'string' && transformCaseFunc(rootName) === 'template') {
-          const content = root.content;
-          if (_isDocumentFragment(content)) {
-            _sanitizeAttachedShadowRoots2(content);
+    const _sanitizeAttachedShadowRoots = function _sanitizeAttachedShadowRoots(root) {
+      /* Iterative (explicit stack) rather than per-child recursion. DOM APIs
+         impose no depth cap, so an attacker-shaped tree (JSON/CRDT/editor data
+         built straight into the DOM — the IN_PLACE surface) deeper than the JS
+         call-stack budget would otherwise overflow native recursion here and
+         throw at the IN_PLACE entry pre-pass, before a single node is
+         sanitized, leaving the caller's live tree untouched (fail-open). See
+         campaign-3 F4. A heap stack keeps depth off the call stack.
+              Each work item is either a node to descend into, or a deferred
+         `_sanitizeShadowDOM` for an already-walked shadow root. The deferred
+         form preserves the original post-order discipline: a shadow root's
+         nested shadow roots are discovered before the outer shadow is
+         sanitized (which may remove hosts). Pushes are in reverse of the
+         desired processing order (LIFO): template content, then children, then
+         the shadow-sanitize, then the shadow walk — so the order matches the
+         previous recursion exactly. */
+      const stack = [{
+        node: root,
+        shadow: null
+      }];
+      while (stack.length > 0) {
+        const item = stack.pop();
+        /* Deferred shadow-DOM sanitisation: runs after its subtree was walked. */
+        if (item.shadow) {
+          _sanitizeShadowDOM2(item.shadow);
+          continue;
+        }
+        const node = item.node;
+        const nodeType = getNodeType ? getNodeType(node) : node.nodeType;
+        const isElement = nodeType === NODE_TYPE.element;
+        /* (pushed last → processed first) Children, snapshotted in reverse so
+           the first child is processed first. Snapshotting matters because a
+           hook may detach siblings mid-walk. */
+        const childNodes = getChildNodes(node);
+        if (childNodes) {
+          for (let i = childNodes.length - 1; i >= 0; --i) {
+            stack.push({
+              node: childNodes[i],
+              shadow: null
+            });
+          }
+        }
+        /* (pushed before children → processed after them, matching the old
+           "template content last" order) When the node is a <template>,
+           descend into its content. */
+        if (isElement) {
+          const rootName = getNodeName ? getNodeName(node) : null;
+          if (typeof rootName === 'string' && transformCaseFunc(rootName) === 'template') {
+            const content = node.content;
+            if (_isDocumentFragment(content)) {
+              stack.push({
+                node: content,
+                shadow: null
+              });
+            }
+          }
+        }
+        /* Shadow root (processed first): walk its subtree, then sanitise it.
+           Realm-safe check (GHSA-hpcv-96wg-7vj8): nodeType-based detection
+           rather than `instanceof DocumentFragment`, which is realm-bound and
+           silently skipped foreign-realm shadow roots (e.g.
+           iframe.contentDocument attachShadow). */
+        if (isElement) {
+          const sr = getShadowRoot(node);
+          if (_isDocumentFragment(sr)) {
+            /* Push the deferred sanitise first so it pops after the shadow
+               walk we push next, i.e. nested shadow roots are discovered
+               before this one is sanitised. */
+            stack.push({
+              node: null,
+              shadow: sr
+            }, {
+              node: sr,
+              shadow: null
+            });
           }
         }
       }
@@ -1615,11 +2005,14 @@
       }
       /* Clean up removed elements */
       DOMPurify.removed = [];
-      /* Check if dirty is correctly typed for IN_PLACE */
-      if (typeof dirty === 'string') {
-        IN_PLACE = false;
-      }
-      if (IN_PLACE) {
+      /* Resolve IN_PLACE for this call without mutating persistent config.
+         Writing the IN_PLACE closure variable here leaks under setConfig(),
+         where _parseConfig is skipped on later calls: a single string call would
+         disable in-place mode for every subsequent node call, returning a
+         sanitized copy while leaving the caller's node — which in-place callers
+         keep using and whose return value they ignore — unsanitized. REPORT-2. */
+      const inPlace = IN_PLACE && typeof dirty !== 'string' && _isNode(dirty);
+      if (inPlace) {
         /* Do some early pre-sanitization to avoid unsafe root nodes.
            Read nodeName through the cached prototype getter — a clobbering
            child named "nodeName" on the form root would otherwise shadow
@@ -1646,8 +2039,16 @@
           throw typeErrorCreate('root node is clobbered and cannot be sanitized in-place');
         }
         /* Sanitize attached shadow roots before the main iterator runs.
-           The iterator does not descend into shadow trees. */
-        _sanitizeAttachedShadowRoots2(dirty);
+           The iterator does not descend into shadow trees. Same fail-closed
+           barrier as the main walk (campaign-3 F2): a custom-element reaction
+           inside a shadow root could abort this pre-pass before the walk runs,
+           which would otherwise leave the entire live tree unsanitized. */
+        try {
+          _sanitizeAttachedShadowRoots(dirty);
+        } catch (error) {
+          _neutralizeRoot(dirty);
+          throw error;
+        }
       } else if (_isNode(dirty)) {
         /* If dirty is a DOM element, append to an empty document to avoid
            elements being stripped by the parser */
@@ -1667,7 +2068,7 @@
            descend into shadow trees. The walk routes every read through a
            cached prototype getter so clobbering descendants on a form root
            cannot hide a shadow host from this pass. */
-        _sanitizeAttachedShadowRoots2(importedNode);
+        _sanitizeAttachedShadowRoots(importedNode);
       } else {
         /* Exit directly if we have nothing to do */
         if (!RETURN_DOM && !SAFE_FOR_TEMPLATES && !WHOLE_DOCUMENT &&
@@ -1687,23 +2088,50 @@
         _forceRemove(body.firstChild);
       }
       /* Get node iterator */
-      const nodeIterator = _createNodeIterator(IN_PLACE ? dirty : body);
-      /* Now start iterating over the created document */
-      while (currentNode = nodeIterator.nextNode()) {
-        /* Sanitize tags and elements */
-        _sanitizeElements(currentNode);
-        /* Check attributes next */
-        _sanitizeAttributes(currentNode);
-        /* Shadow DOM detected, sanitize it.
-           Realm-safe check (GHSA-hpcv-96wg-7vj8): nodeType-based detection
-           instead of instanceof, so foreign-realm <template>.content is
-           walked correctly. */
-        if (_isDocumentFragment(currentNode.content)) {
-          _sanitizeShadowDOM2(currentNode.content);
+      const nodeIterator = _createNodeIterator(inPlace ? dirty : body);
+      /* Now start iterating over the created document.
+         The walk runs inside an exception barrier (campaign-3 F2): a re-entrant
+         engine/custom-element mutation can detach a node mid-walk so
+         `_forceRemove`'s parentless guard throws, aborting the loop. Without the
+         barrier the caller's in-place tree would be left half-sanitized with the
+         unvisited tail still armed. On any throw we fail closed — strip the
+         in-place root bare — then rethrow so the existing throw contract is
+         preserved. (String/DOM-copy paths never return the partial body, so the
+         propagating throw is already fail-closed there.) */
+      try {
+        while (currentNode = nodeIterator.nextNode()) {
+          /* Sanitize tags and elements */
+          _sanitizeElements(currentNode);
+          /* Check attributes next */
+          _sanitizeAttributes(currentNode);
+          /* Shadow DOM detected, sanitize it.
+             Realm-safe check (GHSA-hpcv-96wg-7vj8): nodeType-based detection
+             instead of instanceof, so foreign-realm <template>.content is
+             walked correctly. */
+          if (_isDocumentFragment(currentNode.content)) {
+            _sanitizeShadowDOM2(currentNode.content);
+          }
+        }
+      } catch (error) {
+        if (inPlace) {
+          _neutralizeRoot(dirty);
         }
+        throw error;
       }
       /* If we sanitized `dirty` in-place, return it. */
-      if (IN_PLACE) {
+      if (inPlace) {
+        /* Fail-closed completion of the audit-5 F1 fix: every node removed from
+           the caller's live tree is detached but may still hold a queued
+           resource-event handler that fires in page scope after we return. The
+           move-hoist covers only disallowed-tag KEEP_CONTENT removals; strip the
+           non-allow-listed attributes off every other removed subtree (clobber,
+           mXSS, namespace, comments, KEEP_CONTENT:false, …) so those handlers are
+           cancelled before any event can fire. Runs synchronously, pre-return. */
+        arrayForEach(DOMPurify.removed, entry => {
+          if (entry.element) {
+            _neutralizeSubtree(entry.element);
+          }
+        });
         if (SAFE_FOR_TEMPLATES) {
           _scrubTemplateExpressions2(dirty);
         }
@@ -1742,9 +2170,7 @@
       }
       /* Sanitize final string template-safe */
       if (SAFE_FOR_TEMPLATES) {
-        arrayForEach([MUSTACHE_EXPR$1, ERB_EXPR$1, TMPLIT_EXPR$1], expr => {
-          serializedHTML = stringReplace(serializedHTML, expr, ' ');
-        });
+        serializedHTML = _stripTemplateExpressions(serializedHTML);
       }
       return trustedTypesPolicy && RETURN_TRUSTED_TYPE ? _createTrustedHTML(serializedHTML) : serializedHTML;
     };
@@ -1756,6 +2182,12 @@
     DOMPurify.clearConfig = function () {
       CONFIG = null;
       SET_CONFIG = false;
+      // Drop any caller-supplied Trusted Types policy so it cannot poison later
+      // `RETURN_TRUSTED_TYPE` output. The internal default policy (cached, and
+      // never recreated — Trusted Types throws on duplicate names) is restored by
+      // the next `_parseConfig`. See GHSA-vxr8-fq34-vvx9.
+      trustedTypesPolicy = defaultTrustedTypesPolicy;
+      emptyHTML = '';
     };
     DOMPurify.isValidAttribute = function (tag, attr, value) {
       /* Initialize shared config vars if necessary. */