| Summary: | Text recognition results appear on wrong position when the parents elements containing <video> element is brought fullscreen | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Ethan Wong <bunnnywong> | ||||||
| Component: | Media | Assignee: | Nobody <webkit-unassigned> | ||||||
| Status: | NEW --- | ||||||||
| Severity: | Normal | CC: | webkit-bug-importer, wenson_hsieh | ||||||
| Priority: | P2 | Keywords: | InRadar | ||||||
| Version: | Safari 16 | ||||||||
| Hardware: | Mac (Apple Silicon) | ||||||||
| OS: | macOS 13 | ||||||||
| Attachments: |
|
||||||||
Created attachment 467526 [details]
A screen recording of this issue
Oh, I think the issue is that we don't correctly account for margin when positioning the native VisionKit overlay view, in the case of fullscreen video. Part of a fix so far involves extracting `inlineVideoFrame()` out of `VideoFullscreenManager.mm`, and using it in `WebPage::beginTextRecognitionForVideoInElementFullScreen` when determining the video rect in root view coordinates. Unfortunately, even after this, the live text button ends up appearing over the built-in video controls... needs further investigation to figure out how to avoid this subsequent issue. |
Created attachment 467525 [details] A page that produces this issue Text recognition results may appear on wrong position when the parent element containing <video> elements is brought fullscreen. See the attached example and screen recording for this issue.