:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import { type ExperimentExposureCriteria, NodeKind } from '~/queries/schema/schema-general'
import type { Experiment } from '~/types'
import { NEW_EXPERIMENT } from 'products/experiments/frontend/constants'
import { prefillScannerForExperiment } from 'products/replay_vision/frontend/replay_scanners/experimentTargeting'
import type { ReplayScanner } from 'products/replay_vision/frontend/replay_scanners/types'
import { experimentScannerBody } from './replayVisionScanner'
describe('replayVisionScanner', () => {
describe('saved scanner config', () => {
// The scanner API shows the config to anyone who can view the scanner, so the experiment's
// hypothesis must stay out of it. Each scan adds the hypothesis after an access check.
it.each([
{
name: 'the scanner wizard prefill',
build: (experiment: Experiment): unknown =>
prefillScannerForExperiment(
{ name: 'Frustration score' } as ReplayScanner,
{ experiment, variantKey: null },
true
).scanner_config,
},
{
name: 'the experiment form checkbox',
build: (experiment: Experiment): unknown => experimentScannerBody(experiment).scanner_config,
},
])('$name keeps the experiment hypothesis out of it', ({ build }) => {
const experiment: Experiment = {
...NEW_EXPERIMENT,
id: 123,
name: 'Checkout redesign',
description: 'Private hypothesis about one-page checkout',
}
expect(JSON.stringify(build(experiment))).not.toContain('Private hypothesis')
})
})
describe('experimentScannerBody', () => {
// The experiment is a draft at this point, so a scanner saved on, or without
// start_on_launch, is refused by the API or never starts.
it.each([
{
name: 'a default exposure event',
exposure_criteria: undefined,
expectedFilterTestAccounts: true,
},
{
name: 'a custom exposure event',
exposure_criteria: {
exposure_config: {
kind: NodeKind.ExperimentEventExposureConfig,
event: 'backend_assigned',
properties: [],
},
} satisfies ExperimentExposureCriteria as ExperimentExposureCriteria,
expectedFilterTestAccounts: true,
},
{
name: 'an experiment that keeps test accounts',
exposure_criteria: { filterTestAccounts: false } satisfies ExperimentExposureCriteria,
expectedFilterTestAccounts: false,
},
])(
'saves an experiment scanner that waits for launch, with no filters: $name',
({ exposure_criteria, expectedFilterTestAccounts }) => {
const experiment: Experiment = {
...NEW_EXPERIMENT,
id: 123,
name: 'Checkout redesign',
exposure_criteria,
}
const body = experimentScannerBody(experiment)
expect(body).toMatchObject({
scanner_type: 'experiment',
scanner_config: { experiment_id: 123, variants: null, start_on_launch: true },
enabled: false,
})
expect(body.experiment_targeting).toBeUndefined()
expect(body.query).toEqual({
kind: NodeKind.RecordingsQuery,
filter_test_accounts: expectedFilterTestAccounts,
})
}
)
// Both entry points feed the same server-derived exposure filter, so the population they
// build must stay identical.
it('builds the same population as the scanner wizard prefill', () => {
const experiment: Experiment = { ...NEW_EXPERIMENT, id: 123, name: 'Checkout redesign' }
const prefilled = prefillScannerForExperiment(
{ name: 'Frustration score' } as ReplayScanner,
{ experiment, variantKey: null },
true
)
const body = experimentScannerBody(experiment)
expect(prefilled.scanner_type).toEqual(body.scanner_type)
expect(prefilled.scanner_config).toMatchObject({ experiment_id: 123, variants: null })
expect(body.query).toEqual(prefilled.query)
})
})
})