Sanitize while parsing by noamr · Pull Request #12756 · whatwg/html

noamr · 2026-08-04T18:19:56Z

Instead of parsing into a fragment and then statically sanitizing that fragment, sanitize as we parse.
This entails the following changes:

The sanitizer config is propagated to the parser and kept as a new parser flag ("parser sanitizer configuration").
The "safe" flavor of sanitization is kept as a boolean flag in the parser ("remove javascript navigation URLs"), as that is a special safety check not kept in the sanitizer configuration.
The fragment created to hold the result is created in the inert document when sanitizing to avoid creation-time side effects (see Nodes are made non-inert before sanitizing #12560).
Instead of a recursive "sanitize" algorithm operating on a node tree, we define a "sanitize" algorithm that operates on a single Element, along with two helper checks: "sanitizer config allows comments" and "sanitizer config allows processing instruction target" (other node types do not need to check the sanitizer).
For "replace with children", we replaced the parser's "root insertion target" flag with an "insertion target redirection map" (mapping a node to an insertion location). This map is used to redirect insertions from the root dummy element to the target DocumentFragment, and also to redirect elements to their nearest non-replaced ancestor when a child is stripped but its children are kept.

Note that this still leaves out important follow ups, where we need to sometimes sanitize before the element is even created (e.g. declarative shadow roots and is), and having a live sanitizer config while parsing gives us the infrastructure for that.

Closes #12560
Closes #12543

At least two implementers are interested (and none opposed):
- Chromium
- WebKit (as per @annevk)
- Mozilla (as per @mozfreddyb)
Tests are written and can be reviewed and commented upon at:
- This mostly keeps existing behavior. The observable difference is coalescing of text nodes while parsing. This was modified in tests in Remove special text node behavior from streaming sanitizer web-platform-tests/wpt#59541.
- https://wpt.fyi/results/sanitizer-api/sanitizer-inert-document.html checks that images are created as part of an inert document
Implementation bugs are filed:
- Chromium: works already
- Gecko: see https://bugzilla.mozilla.org/show_bug.cgi?id=2043150
- WebKit: doesn't require its own bug report
MDN issue is filed: N/A, no behavior change
The top of this comment includes a clear commit message to use.

(See WHATWG Working Mode: Changes for more details.)

/dom.html ( diff )
/dynamic-markup-insertion.html ( diff )
/iframe-embed-object.html ( diff )
/infrastructure.html ( diff )
/parsing.html ( diff )
/timers-and-user-prompts.html ( diff )

zcorpan

Missing null check; nits.

zcorpan · 2026-08-14T08:44:34Z

+   <li><p>Create a new <span>HTML parser</span> whose <span>parser sanitizer configuration</span> is
+   <var>configuration</var>, <span>allow declarative shadow roots</span> is
+   <var>allowDeclarativeShadowRoots</var>, <span>scripting mode</span> is
+   <var>scriptingMode</var>, and <span>remove javascript navigation URLs</span>
+   is <var>safe</var>, and associate it with <var>document</var>.</p></li>

   <li><p>If <var>contextDocument</var>'s <span data-x="concept-n-script">scripting is
   disabled</span>, then set <var>scriptingMode</var> to <span
   data-x="scripting-mode-disabled">Disabled</span>.</p></li>


From chatgpt's re-review:

The context document’s scripting-disabled state is still applied only after constructing the parser. The following step mutates the local scriptingMode; it does not update the parser created in the previous step. Please calculate the final mode before parser construction, or explicitly update the parser’s stored mode afterward. This commit should be correct independently of #12758. A <noscript> fragment test with scripting disabled would catch the ordering error.

Switched the order, should be OK now (this is also changed in the next PR in the stack)

keithamus · 2026-08-14T14:42:55Z

+     <li><p>If <var>lastNode</var> is non-null, insert whatever <var>lastNode</var> ended up being
+     in the previous step at the <span>adjusted insertion location</span> given
+     <var>insertionLocation</var>.</p></li>


Does this insert lastNode without a sanitizer check?

data:text/html,<!DOCTYPE html><body><div id=div></div><script>div.setHTMLUnsafe('<b><div><b>x', { sanitizer: {removeElements: ['div'] }});</script>

<b> is pushed

<div> is created, sanitize removes, so insert is aborted but its still pushed

</b> runs AAA, furthestBlock is removed div, commonAncestor is b.

Inner loop breaks, lastNode is furthestBlock

This step insets removed div into commonAncestor.

Perhaps worth a test at least.

Good catch! This is fixed now (Adding tests as part of https://chromium-review.git.corp.google.com/c/chromium/src/+/8250959, once they're in I'll add them to the un-tentative PR)

Added multiple test cases for this: web-platform-tests/wpt@4409cc2

(They're not tentative as they should be sanitized with the post-processing sanitizer as well)

keithamus · 2026-08-14T14:46:28Z

+   <li><p>If the parser's <span>insertion target redirection map</span>[<var>targetParent</var>]
+   does not <span data-x="map exists">exist</span>, then return
+   <var>adjustedInsertionLocation</var>.</p></li>


So here, as removed elements never end up in the map, this can return early and - as the adjusted insertion location becomes a detached element - all children end up being added to that. While it's unobservable, it's a bit slow - but if we added a map of removed elements we could do something like:

If the parser's sanitizer-removed elements contains targetParent, then return null.

To no-op sooner, when targetParent is detatched. Of course implementations could do this and still be conformant but it may be nice to either specify or at least comment & hint at it.

I don't know about adding more non-observable optimizations to the spec just for the purpose of AAA... seems like an additional potential point of failure.

Sanitization should take place while performing the adoption agency algorithm, otherwise some nodes can be missed. Since adoption agency often adds an intermediate element, the ReplaceWithChildren directive essentially negates it, while Remove reparents the children into the intermediate element and removes all of them. This is in line with spec changes made as part of whatwg/html#12756 Bug: 498272014 Change-Id: Ieb755555635e197f8ed7daf06a08679a0f2d38c3 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/8250959 Commit-Queue: Noam Rosenthal <nrosenthal@google.com> Reviewed-by: Daniel Vogelheim <vogelheim@chromium.org> Cr-Commit-Position: refs/heads/main@{#1680770}

This behavior was never shipped, due to compatibility issues. Effectively reverts 5f626a1 & 7c8fb86.

…owIfMissing polarity

…x TT throwIfMissing polarity" This reverts commit 7898aca.

…ocumentFragment

…ensure the adjusted insertion location is correct

…ency edge cases, a=testonly Automatic update from web-platform-tests Fix some streaming-sanitizer/adoption-agency edge cases Sanitization should take place while performing the adoption agency algorithm, otherwise some nodes can be missed. Since adoption agency often adds an intermediate element, the ReplaceWithChildren directive essentially negates it, while Remove reparents the children into the intermediate element and removes all of them. This is in line with spec changes made as part of whatwg/html#12756 Bug: 498272014 Change-Id: Ieb755555635e197f8ed7daf06a08679a0f2d38c3 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/8250959 Commit-Queue: Noam Rosenthal <nrosenthal@google.com> Reviewed-by: Daniel Vogelheim <vogelheim@chromium.org> Cr-Commit-Position: refs/heads/main@{#1680770} -- wpt-commits: 4409cc231d390712ee52ffb5529177ebb0cd0da3 wpt-pr: 61987

noamr closed this Aug 4, 2026

noamr reopened this Aug 4, 2026

noamr mentioned this pull request Aug 4, 2026

Sanitize while parsing #12645

Closed

5 tasks

noamr force-pushed the noamr/streaming-sanitizer branch from 7b105e3 to e24c23b Compare August 5, 2026 09:31

zcorpan mentioned this pull request Aug 5, 2026

Integrate fragment parser options with trusted types #12583

Open

5 tasks

noamr force-pushed the noamr/streaming-sanitizer branch 3 times, most recently from 4ff3705 to 8551681 Compare August 7, 2026 20:17

zcorpan reviewed Aug 11, 2026

View reviewed changes

Comment thread source Outdated

Comment thread source Outdated

Comment thread source Outdated

Comment thread source Outdated

Comment thread source

Comment thread source Outdated

Comment thread source Outdated

Comment thread source

Comment thread source Outdated

keithamus mentioned this pull request Aug 11, 2026

Sanitize while Parsing mozilla/standards-positions#1443

Closed

noamr force-pushed the noamr/streaming-sanitizer branch from c374a0b to e8c7cba Compare August 11, 2026 20:09

zcorpan reviewed Aug 12, 2026

View reviewed changes

Comment thread source Outdated

Comment thread source

Comment thread source

noamr force-pushed the noamr/streaming-sanitizer branch 2 times, most recently from 1d9e935 to 4453c2a Compare August 12, 2026 16:07

noamr closed this Aug 13, 2026

noamr reopened this Aug 13, 2026

zcorpan approved these changes Aug 13, 2026

View reviewed changes

zcorpan requested changes Aug 14, 2026

View reviewed changes

noamr force-pushed the noamr/streaming-sanitizer branch from 4453c2a to 931141c Compare August 14, 2026 08:52

noamr requested a review from zcorpan August 14, 2026 08:57

keithamus reviewed Aug 14, 2026

View reviewed changes

noamr force-pushed the noamr/streaming-sanitizer branch from 32e5b9a to 1b77b8d Compare August 15, 2026 10:04

chromium-wpt-export-bot mentioned this pull request Aug 17, 2026

Fix some streaming-sanitizer/adoption-agency edge cases web-platform-tests/wpt#61987

Merged

Remove cross-origin check for simple dialogs

40814eb

This behavior was never shipped, due to compatibility issues. Effectively reverts 5f626a1 & 7c8fb86.

noamr added 28 commits August 19, 2026 16:18

Rebaseline to new TT changes

3054bf6

Handle default sanitizer correctly when safe is false, and fix TT thr…

7e101ea

…owIfMissing polarity

Revert "Handle default sanitizer correctly when safe is false, and fi…

ac55946

…x TT throwIfMissing polarity" This reverts commit 7898aca.

Handle null in TrustedParserOptions

31085c1

Fix ccf

73d20d8

Add some null checks

2640ad2

positional

eaf5d51

Remove spurious

2665b73

Align helper algorithms with call site

477dadb

Rebase

a87f819

Clean up helper algos

3882899

Throw when parent is non-null and neither an element or DocFrag

294d5e8

Throw when using before/after/replaceWith and the parent is a plain D…

3cbd9dd

…ocumentFragment

nits

3ef18c9

nits

1c8a2a8

Disallow inserting HTML to children of template

d007ac9

typo

cbed385

Streaming sanitizer

c9e14e4

Fix attribute merging

47524e6

Add XXX notes

e59eff3

Some nits

ef56a86

Check for script root

9f45d6f

nit

d9909b3

Enable sanitizer integration in adoption agency algorithm

81ef71b

Use null instead of fake DF

8f974ec

Switch order

942bb22

Sanitize the initial furthestBlock in Adoption Agency Algorithm, and …

7953578

…ensure the adjusted insertion location is correct

indent

2eb51b7

noamr force-pushed the noamr/streaming-sanitizer branch from 1b77b8d to 2eb51b7 Compare August 19, 2026 16:22

Uh oh!

Conversation

noamr commented Aug 4, 2026 • edited by pr-preview Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zcorpan left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

zcorpan Aug 14, 2026

Choose a reason for hiding this comment

Uh oh!

noamr Aug 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

keithamus Aug 14, 2026

Choose a reason for hiding this comment

Uh oh!

noamr Aug 15, 2026

Choose a reason for hiding this comment

Uh oh!

noamr Aug 18, 2026

Choose a reason for hiding this comment

Uh oh!

keithamus Aug 14, 2026

Choose a reason for hiding this comment

Uh oh!

noamr Aug 15, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

3 participants

noamr commented Aug 4, 2026 •

edited by pr-preview Bot

Loading

noamr Aug 14, 2026 •

edited

Loading