: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 { Histogram } from 'prom-client'
import { HogTransformer } from '~/common/hog-transformations/hog-transformer.interface'
import { PipelineResult, ok } from '~/ingestion/framework/results'
import { ProcessingStep } from '~/ingestion/framework/steps'
const backgroundTaskHogTransformerDuration = new Histogram({
name: 'ingestion_background_task_hog_transformer_duration_seconds',
help: 'Time waiting for hog transformer invocation results in the background task',
buckets: [0.01, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10],
})
/**
* afterBatch step that drains the hog transformer's accumulated invocation
* results once per batch. `transformEventAndProduceMessages` buffers results
* on every call; without a per-batch drain that buffer grows unbounded over the
* consumer's lifetime. Returns the drain as a side effect so it's awaited
* alongside the batch's other scheduled work rather than blocking the pipeline,
* timing it on the way out so the duration stays observable.
*/
export function createFlushHogTransformerStep<T>(
hogTransformer: Pick<HogTransformer, 'processInvocationResults'>
): ProcessingStep<T, T> {
return function flushHogTransformerStep(input: T): Promise<PipelineResult<T>> {
const stopTimer = backgroundTaskHogTransformerDuration.startTimer()
const drain = hogTransformer.processInvocationResults().finally(stopTimer)
return Promise.resolve(ok(input, [drain]))
}
}