react-spectrum

6ea044ee - fix(@react-aria/datepicker): don't steal focus into a date segment from selectionchange (Firefox) (#10260)

fix(@react-aria/datepicker): don't steal focus into a date segment from selectionchange (Firefox) (#10260) * fix(@react-aria/datepicker): don't steal focus into a date segment from selectionchange while another element is focused In Firefox, the caret inside a sibling <input> is not reflected in the document selection, so a stale anchor can remain inside a date segment after focus moves away. useDateSegment's document-level 'selectionchange' listener then calls Selection.collapse(segment), and in Firefox collapsing onto a contentEditable node moves focus to it -- stealing focus into the segment while the user types in a neighbouring input. Gate the collapse on the segment actually being the active element. This preserves the Android-Chrome composition behaviour the handler was written for (which only applies while the segment is focused) and removes the cross-element focus steal. Fixes #10259 * test: make DateField selectionchange focus-steal test fail without the fix The previous test relied on a stale selection anchor surviving inside the segment after focus moved to a sibling input, but Chromium/WebKit collapse the document selection on focus change so the guarded branch was never reached -- the test passed with and without the fix. Reproduce the guarded precondition deterministically instead: keep genuine focus on a sibling element (so getActiveElement() !== the segment) and stub window.getSelection for a single selectionchange so the anchor is inside the segment, then assert the handler does not call Selection.collapse onto it. Now fails in every browser without the active-element guard. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore: fix formatting in DateField.browser.test.tsx Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test: move DateField focus-steal regression test to jsdom The selectionchange focus steal is Firefox-specific browser behaviour, but the logic the fix changed -- whether the handler calls Selection.collapse onto the segment while another element is focused -- is environment independent. A real browser can't reproduce the steal through synthetic events anyway, so replace the browser test with a focused jsdom unit test that stubs window.getSelection to model the stale Firefox anchor and asserts collapse is not called. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * chore: trim comment on selectionchange focus-steal guard Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * ci: retrigger CI (s2-docs parcel-resolver panic) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: Robert Snow <rsnow@adobe.com>