track topics by product for use by AI-based classifier agent

[escattone]

This PR adds a YAML file that details the topic hierarchies for each product, including descriptions. It will be useful for building the prompt for AI agents that will automatically classify questions by topic.

It was generated by the script below (run in production), and then manually adjusted to remove a few remaining legacy topics, and add a few missing topics. Finally, it was manually verified against the "golden" list of topics defined in the Taxonomy Map PDF.

TODO: We should update this file with the more detailed descriptions that Josh is working-on once they are finalized.

from itertools import chain

import yaml

from kitsune.products.models import Product, Topic


def get_taxonomy(product_slug=None, **kwargs):

    def clean(text):
        return text.replace("\u2019", "'")

    result = dict(products=[])

    if product_slug:
        products = [Product.active.get(slug=product_slug)]
    else:
        products = chain(
            Product.active.filter(visible=True),
            Product.active.filter(slug="mozilla-account"),
        )

    for product in products:
        pdict = dict(
            title=product.title, description=clean(product.description), topics=[]
        )
        result["products"].append(pdict)
        for t1 in Topic.active.filter(
            products=product, parent=None, visible=True, **kwargs
        ):
            t1_dict = dict(
                title=t1.title, description=clean(t1.description), subtopics=[]
            )
            pdict["topics"].append(t1_dict)
            for t2 in Topic.active.filter(
                products=product, parent=t1, visible=True, **kwargs
            ):
                t2_dict = dict(
                    title=t2.title, description=clean(t2.description), subtopics=[]
                )
                t1_dict["subtopics"].append(t2_dict)
                for t3 in Topic.active.filter(
                    products=product, parent=t2, visible=True, **kwargs
                ):
                    t2_dict["subtopics"].append(
                        dict(title=t3.title, description=clean(t3.description))
                    )

    return print(yaml.dump(result, sort_keys=False))

[akatsoulas]

Let's add the script too under the scripts folder and move the produced yml file either under docs or docs/taxonomy