feat(selectors): replace pierceFrames() with frameLocator() - #42395
Merged
Conversation
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
Contributor
|
Hi, I'm the Playwright bot and I took a first look at the CI failures here. 🟡 One failure, can't pin it on this PR either way
DetailsThis PR is a selector-engine refactor ( Uncertain
Not triaged
Triaged by the Playwright bot - agent run |
This comment has been minimized.
This comment has been minimized.
Dmitry Gozman (dgozman)
force-pushed
the
any-frame
branch
from
August 26, 2026 08:15
9fd1ca2 to
20ca261
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
Devin Rousso (dcrousso)
approved these changes
Aug 27, 2026
page.anyFrame() and frame.anyFrame() start the match in any frame of the subtree, and the rest of the selector resolves inside that single frame, just like a regular locator. This drops the mid-chain frame piercing that pierceFrames() did, together with the restrictions it required: composite locators, *-capture and mid-chain nth are all allowed now. Also removes the pierceFrames context and test option. Codegen prefers an anyFrame() prefix whenever it saves at least two frameLocator() calls and still pinpoints the target frame.
A nested "any-frame" token used to reach the injected control engine and
match nothing, so page.locator('button').or(page.anyFrame().locator('a'))
silently dropped the second operand.
The token applies to the whole locator, so an operand never carries one:
page.anyFrame().locator('.classname').or(page.getByRole('button'))
matches ".classname or button" inside every frame. Skip a leading
"any-frame" when matching the nested frame prefix, so that an operand may
still repeat the frame chain, and reject any leftover token.
parseSelector() strips the frame prefix a nested selector shares with the
outer one, but left "capture" pointing at the old part index, so
frameLocator('#f').locator('body')
.locator(frameLocator('#f').locator('*css=section >> span'))
resolved to the span instead of the section. Shift the capture along with
the removed parts, and reject a capture that falls inside the prefix.
Calling page.frameLocator() or frame.frameLocator() without a selector now starts the match in any frame of the subtree, replacing anyFrame(). The selector format and the behavior are unchanged - only the API shape. Locator descriptions and codegen render the token as frameLocator() in all languages, which also makes it round-trip through the locator parser, unlike the bare any_frame/AnyFrame property forms.
Dmitry Gozman (dgozman)
force-pushed
the
any-frame
branch
from
August 27, 2026 19:17
e69eda3 to
796831d
Compare
Dmitry Gozman (dgozman)
changed the title
Contributor
Test results for "MCP"1 failed 8258 passed, 1361 skipped Merge workflow run. |
Contributor
Test results for "tests 1"6 flaky51218 passed, 1237 skipped Merge workflow run. |
Contributor
|
Hi, I'm the Playwright bot and I took a look at the CI failures here. 🟢 The one failure is a known flake — this PR is clear
DetailsThis PR replaces Pre-existing flake / infra
Not triaged
Triaged by the Playwright bot - agent run |
Dmitry Gozman (dgozman)
merged commit Aug 28, 2026
6748802
into
microsoft:main
44 of 45 checks passed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
page.frameLocator()/frame.frameLocator()without a selector replacepierceFrames(): they pick the frame the match starts in, and the rest of the selector resolves inside that single frame like a regular locator.*-capture and mid-chainnthnow work — and replaces the recursive piercing resolver with a per-frame run of the existing chained resolution.pierceFramescontext option and the corresponding test option, along with theno-pierce-framesopt-out token.frameLocator()prefix whenever it saves at least twoframeLocator(selector)calls and still pinpoints the target frame, falling back to the full chain otherwise.