feat: Hog transformation comparer

[benjackwhite]

Problem

We want to safely roll out the migration to hog functions but the question is - how?

Changes

• Adds a comparer step that compares the plugin transforms to the hog version
• Adds a migration script to migrate all plugin ones over to hog ones
• Adds an env var for sampling this so we don't hit the ingestion pipeline too hard
• Fixes the migrator to work for transformations as well
• Fixes the UI so that we have a way of saying an input doesn't support templating (so that hog bytecode isn't generated)

👉 Stay up-to-date with PostHog coding conventions for a smoother review.

Does this work well for both Cloud and self-hosted?

How did you test this code?

[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}

[benjackwhite]

Wasn't sure of the naming here but basically wanted one function that does the transform and another that does it and produces the kafka messages

[greptile-apps]

PR Summary

This PR introduces a comprehensive system for safely migrating from plugin transformations to hog functions. Here's a summary of the key changes:

• Added comparison mechanism in compareToHogTransformStep.ts to validate hog transformations against existing plugin outputs with configurable sampling rate
• Introduced templating: false property across transformation plugin templates to explicitly control template processing behavior
• Modified HogTransformerService to separate message production from event transformation logic and improve error handling
• Added migration scripts with --kind parameter to support selective migration of 'destination' or 'transformation' plugins
• Added HOG_TRANSFORMATIONS_COMPARISON_PERCENTAGE config to control sampling rate and prevent excessive load on ingestion pipeline

Key points to review:

• The hardcoded project name in createInvocationGlobals needs to be addressed
• The execute method in HogTransformerService doesn't await hogExecutor.execute result which could cause timing issues
• The downsampling plugin template has a simplified return event implementation that may not fully replicate original functionality
• Error handling in transformEvent continues execution after errors instead of failing fast

_{22 file(s) reviewed, 13 comment(s)
Edit PR Review Bot Settings | Greptile}

[pl]

Looked at the TS code, LGTM. I haven't worked on the migration bit at all, so it's hard for me to do a review yet.

[sentry-io]

Suspect Issues

This pull request was deployed and Sentry observed the following issues:

• ‼️ ValidationError: {"template_id":["Transformation functions must be created from a template."]} posthog.cdp.migrations in migrate_legacy_plugins View Issue
• ‼️ TypeError: Cannot create distinct fields once a slice has been taken. posthog.cdp.migrations in migrate_batch View Issue
• ‼️ ValidationError: {"template_id":["Transformation functions must be created from a template."]} posthog.cdp.migrations in migrate_batch View Issue
• ‼️ KeyError: 117775 posthog.cdp.migrations in migrate_batch View Issue
• ‼️ KeyboardInterrupt posthog.hogql.ast in __init__ View Issue

_{Did you find this useful? React with a 👍 or 👎}