luci-base: form.js: fix stripTags NotFoundError and modify source node

[jjm2473]

Pull request details

Description

This PR addresses two issues:

1. NotFoundError:
   For example s='<a><br></a>', then x=dom.parse('<div><a><br></a></div>'), this br is not a direct child node of x, the subsequent call x.replaceChild(document.createTextNode('\n'), br) will result in an error:

Uncaught (in promise) NotFoundError: Failed to execute 'replaceChild' on 'Node': The node to be replaced is not a child of this node.
  at eval (form.js:312:6)
  at NodeList.forEach (<anonymous>)
  at ClassConstructor.stripTags (form.js:311:28)
  at ClassConstructor.renderFrame (form.js:4434:30)

2. modify source node
   stripTags modifies the source node, causing switch port status missing <br> on first render.
   Clone source node as AI suggests: luci-mod-network: switch: fix port status missing <br> on first render #8937 (comment)

Screenshot or video of changes (if applicable)

Maintainer (preferred)

@systemcrash

---

Tested on

OpenWrt version: iStoreOS 25.12.5 2026081215
LuCI version: LuCI istoreos-25.12 branch 26.195.42598~4a1ef8c
Web browser(s): Chrome (151.0.7922.76)

---

Checklist

• (Nice to have) Includes what Issue it closes (e.g. openwrt/luci#issue-number).
• (Nice to have) Includes what it depends on (e.g. openwrt/packages#pr-number in sister repo).

[openwrt-ai]

Reviewed 1 new commit; one inline simplification suggested. The commit message accurately describes the change, and the underlying NotFoundError diagnosis is correct.

---

Generated by Claude Code

[openwrt-ai]

The branch is unnecessary — br.parentNode is the correct target in every case and makes the fallback redundant.

br always comes from x.querySelectorAll('br'), so it is always a descendant of x and always has a parent node. Two situations exist:

• x is an element (the dom.parse(
  ${s}
  ) path, or an element passed in via dom.elem(s)): the parent is an element, so parentElement === parentNode.
• x is a DocumentFragment and the <br> is a direct child of it: parentElement is null — this is the only case the else reaches — and parentNode is the fragment, i.e. exactly what x.replaceChild() uses. Note dom.elem() only tests for nodeType, so a fragment does reach here.

Using parentNode unconditionally covers both, and matches how the rest of the tree does node replacement (e.g. cbi.js:366, tools/network.js:1383).``

Suggested change

	if (br.parentElement)
	br.parentElement.replaceChild(document.createTextNode('\n'), br);
	else
	x.replaceChild(document.createTextNode('\n'), br);
	br.parentNode.replaceChild(document.createTextNode('\n'), br);

---

Generated by Claude Code

[openwrt-ai]

Reviewed 2 new commits (the branch was force-pushed, so 61436bb was rewritten as 19a5954, which adopts the br.parentNode simplification from the previous review).

Both changes look correct:

• 19a5954 — br.parentNode.replaceChild() is right for every case, since br always comes from x.querySelectorAll('br') and therefore always has a parent, whether x is an element or a DocumentFragment.
• 619ce76 — cloneNode(true) on the dom.elem(s) branch is the right scope for the mutation fix: only that branch receives a caller-owned node, while the dom.parse() branch already builds a throwaway tree. textContent on the detached clone is unaffected by the clone.

Both commit messages match their diffs. One nit inline about the PR description.

---

Generated by Claude Code