fix(experiments): Apply correct filters to "View recordings"

[danielbachhuber]

See https://posthog.slack.com/archives/C087ZBJBZ7C/p1737128365755459

Also reported in a direct email:

I’ve been running another no-code experiment on PostHog. The recordings show that not everyone is seeing the variant I set, even when I filter based on the experiment and variant. For example, if my control is ABC and the variant is CBA, the recordings for CBA sometimes show the control instead. This makes me question the accuracy of the stats... Is this a bug or user error on my part?

Changes

Introduces a getViewRecordingFilters() function with tests to generate the various permutations of filters based on the metric. Note: the screencasts are solely to show the filters being applied; I don't actually have any recordings because I generated the events synthetically.

If no filters could be identified, then the button will be disabled:

Funnel with events

CleanShot.2025-01-23.at.15.48.50.mp4

Funnel with action

CleanShot.2025-01-23.at.15.49.42.mp4

Trend with default exposure

CleanShot.2025-01-23.at.15.52.18.mp4

How did you test this code?

Tests should pass + manual review.

[github-actions]

Size Change: -5 B (0%)

Total Size: 1.16 MB

ℹ️ View Unchanged

Filename	Size	Change
frontend/dist/toolbar.js	1.16 MB	-5 B (0%)

_{compressed-size-action}

[andehen]

Nice one!

Never tried this button myself for the same reasons (no recordings locally). But it is a good reminder that we should dog food ourselves more and run actual experiments ourselves. Even dummy ones would be good, just to try out the functionality in a more realistic setting.