Page MenuHomePhabricator

[SPIKE] Investigate feasibility of using "abusefilter-analyzer" to bootstrap new Edit Checks from existing AbuseFilters
Open, Needs TriagePublic

Description

With abusefilter-analyzer, @Msz2001 re-implemented the AbuseFilter rules engine in JavaScript.

Why?

To help analyze why a filter matches (or doesn't) a specific edit. [i]

In doing so, @Msz2001 created the potential for the Editing Team to load up pre-existing abuse filter rules, and (using this code), apply them in the (completely new) EditCheck context.

Emphasis on potential above as this task is intended to help the Editing Team determine how technically viable doing the above would be.

Open question(s)

  1. What would need to be true about abusefilter-analyzer and/or Edit Check for the Editing Team to be able to repurpose existing AbuseFilter rules, and (using this code), implement new Edit Checks from/with them?

Done

  • Answers to all === Open question(s) are documented on this ticket

i. https://pl.wikipedia.org/wiki/Wikipedia:Kawiarenka/Kwestie_techniczne#Analizator_wywołań_filtra_nadużyć


Thank you to @matmarex for making the Editing Team aware of the work @Msz2001 is doing and its potential relevance to Edit Check 🙏🏼

Event Timeline

For reference, here is some description of the script as well: https://meta.wikimedia.org/wiki/User:Msz2001/AbuseFilter_analyzer

A better documentation is still in to do phase but in case of any questions, I'll be happy to answer (and create/expand the docs). While building the library, I wanted it to be repurposeable and extensible but I hadn't an occasion yet to check it in practice :)