fix(surveys): re-translate displayed surveys when user language changes

[nandinitiw]

Reopening #4195 as a fresh PR — the original was auto-closed by the stale bot after a week of no activity, not for cause. lucasheriques's last review was positive ("the re-translate path holds up") and all requested changes were already addressed; the GitHub reopen API is failing with an opaque validation error, so opening fresh from the same branch instead. Same commits, same diff.

What

When a survey popover is on screen, it was never re-translated if the user's language changed mid-session. This happened in two scenarios:

1. posthog.identify() / setPersonPropertiesForFlags() sets a new language person property
2. The browser fires a languagechange event (OS/browser language switch)

The TODO comment in survey-translations.ts already called this out as a known gap.

Also fixes a gap this surfaced: $survey_questions[].question and $survey_language were read at capture time, not answer time — if the language changed mid-survey, submitted events showed the new language's question text even for questions the user answered under the old language.

How

In SurveyManager:

• Register a window.languagechange listener in the constructor
• Subscribe to posthog.onFeatureFlags() (which fires after identify() reloads flags)
• Both call _onLanguageChange(), which:
  • No-ops if no survey is in focus, or if the survey hasn't actually rendered yet (gates against a delayed survey — surveyPopupDelaySeconds — being forced on screen early by a mid-delay reload, skipping the eligibility recheck the delay timeout does)
  • Re-detects the resolved language
  • No-ops if the language is unchanged (idempotent, safe to call on every flag reload)
  • Otherwise updates _currentLanguage and re-renders the popover via renderPopover()
• Listeners are cleaned up on destroy()
• _currentLanguage is cleared when a survey is dismissed

questionSnapshots: onNextButtonClick now snapshots the currently-displayed question text into questionSnapshots: Record<string, string> (question id → text) before advancing, and persists it to InProgressSurveyState in localStorage. buildSurveyResponseProperties in @posthog/core now accepts this optional param and prefers snapshot text over the current (possibly re-translated) survey question text, so $survey_questions reflects what the user actually saw when they answered. $survey_language on dismissed now similarly prefers the in-progress language over the current display-time language.

The iOS reference implementation in posthog-ios #686 follows the same pattern.

Tests

Unit tests covering:

• Re-renders when language differs
• Does not re-render when language is unchanged
• No-ops when no survey is in focus
• Does not re-render a survey that's in focus but not yet rendered (pending delay)
• Clears _currentLanguage on dismiss
• Removes languagechange listener on destroy()
• questionSnapshots recorded and used in place of current question text
• A real end-to-end render test: renders a popup via the unmocked handlePopoverSurvey, types an answer, fires a real languagechange event, and asserts both that the rendered question text updates and that the typed answer survives the re-render

Closes #4174

[turnipdabeets]

Thanks for reopening this and for the earlier rounds. Confirmed the diff here is identical to #4195.

A few things need a push before we can merge:

1. Need to address merge conflict
2. The end-to-end test you added now fails on this head. It passes in isolation but fails in the full file — main added tests to that describe and the ordering shifted.
3. #4532 also introduced SurveyManager.dispose(), which is what posthog-surveys.ts actually calls on teardown — destroy() has no production caller anymore. Could you move the listener/subscription cleanup into dispose() so the languagechange listener and onFeatureFlags subscription don't leak on reset?
4. All your commits need to be verified and signed to merge the PR.

I'll help run CI once the above is pushed.

[turnipdabeets]

Confirmed with a run on this head: jest src/__tests__/extensions/surveys.test.ts fails with TypeError: this._posthog.get_property is not a function inside _onLanguageChange, while the same test passes on its own via -t "updates the rendered question text". It passed at the pre-merge head (92/92), so main's newer tests shifted the ordering. This afterEach only destroys whatever is in surveyManager at that moment — the nested describes reassign it (new SurveyManager(mockPH)) and orphan the outer instance, whose languagechange listener stays attached. I think we need to track every instance and destroy each one here.

[turnipdabeets]

destroy() has no production caller — teardown goes through dispose(), which posthog-surveys.ts calls as this._surveyManager?.dispose?.(). dispose() landed in #4532 after this branch was written, so as it stands the languagechange listener and the onFeatureFlags subscription both survive a posthog.surveys.dispose(). I think we need to move this cleanup into dispose().

[turnipdabeets]

[question] _handleWidget never calls _addSurveyToFocus, so _surveyInFocus stays null for feedback_button surveys and this returns on the first line — and the render below builds a SurveyPopup rather than a FeedbackWidget. #4174 mentions widgets too; is popover-only the intended scope here?

[turnipdabeets]

[suggestion] "displayed surveys" reads as every survey type, but widgets aren't covered (see the _surveyInFocus gate). Could we name popovers explicitly so the changelog doesn't over-promise?

Suggested change

fix(surveys): re-translate displayed surveys when the user's language changes mid-session — handles both browser `languagechange` events and `posthog.identify()` calls that update the `language` person property. Also snapshots each question's displayed text and language at answer time (`questionSnapshots`), so `$survey_questions` and `$survey_language` reflect what the user actually saw rather than the language active when the event was sent.

Re-translate popover surveys when the display language changes while the survey is on screen, either from a browser `languagechange` event or from `identify()` updating the `language` person property. In-progress answers are preserved. `$survey_questions[].question` and `$survey_language` on `survey sent` / `survey dismissed` now report the text and language the user saw when they answered, not the language active when the event fired. Feedback-button (widget) surveys are unchanged.

[turnipdabeets]

[nit] ?? only catches null/undefined, so a question with id: '' gives '' ?? question.question → ''. Undefined ids fall through fine. questionSnapshots?.[question.id ?? ''] ?? question.question would cover it, if empty ids are reachable.

[nandinitiw]

Thanks for the detailed review! Addressed the first three:

1. Merge conflict: rebased on latest main and resolved. Main had reworked the delayed-render path into renderAfterDelay/renderIfStillEligible helpers (for delay-resumption-across-navigations) since this diff was last synced, so _surveyIsRendered = true now lives in the new renderAfterDelay function.

2. The end-to-end test failing in the full file: traced this down — several other SurveyManager instances constructed elsewhere in the same test file were never disposed, so their languagechange window listeners stayed attached and fired on my test's real dispatchEvent call. Found and fixed three separate leaks: (a) several nested describes reassign the shared surveyManager variable without disposing the previous instance — added a createSurveyManager() helper that tracks every constructed instance so cleanup covers all of them; (b) a jest.clearAllMocks() in the "timeout management" describe left a jest.spyOn(global, 'clearTimeout') spy installed rather than restored, which went stale once fake timers tore down — switched to jest.restoreAllMocks(); (c) generateSurveys()'s own test never disposed the manager it returns — now does, in a finally.

3. dispose() vs destroy(): moved the languagechange listener removal and onFeatureFlags unsubscribe into dispose() and removed destroy() entirely, since it has no production caller.

4. Signed commits: I don't have commit signing configured on this machine yet — need to sort that out on my end before these can be verified. Will follow up once that's set up.

Full survey test suite (268 tests across surveys.test.ts, extensions/surveys.test.ts, survey-popup.test.tsx, posthog-surveys.test.ts) passes, along with lint and prettier.