backend/app/categories.py

"""Document categories for the retail back-office taxonomy.

Five business categories the UI and DB organize around. Each maps to one or more
internal `doc_type`s the pipeline classifies/extracts.
"""
from __future__ import annotations

CATEGORIES = [
    {"id": "orders", "label": "Orders", "icon": "🧾",
     "doc_types": ["purchase_order", "complex_form"],
     "description": "Purchase orders, order forms, and multi-vendor requisitions."},
    {"id": "receipts", "label": "Receipts", "icon": "🧾",
     "doc_types": ["receipt"],
     "description": "Store and supplier receipts."},
    {"id": "invoices", "label": "Invoices", "icon": "📄",
     "doc_types": ["invoice"],
     "description": "Accounts-payable invoices."},
    {"id": "contracts", "label": "Contracts", "icon": "📜",
     "doc_types": ["contract"],
     "description": "Master service agreements, NDAs, supplier contracts."},
    {"id": "subscription_memo", "label": "Subscription Memos", "icon": "🔁",
     "doc_types": ["subscription_memo"],
     "description": "Recurring subscription / renewal memos and notices."},
    {"id": "other", "label": "Other", "icon": "🗂️",
     "doc_types": ["other"],
     "description": "Unclassified documents."},
]

_DOCTYPE_TO_CATEGORY = {
    dt: c["id"] for c in CATEGORIES for dt in c["doc_types"]
}

CATEGORY_IDS = [c["id"] for c in CATEGORIES]


def category_for_doc_type(doc_type: str | None) -> str:
    return _DOCTYPE_TO_CATEGORY.get(doc_type or "", "other")


def category_meta(category_id: str) -> dict | None:
    return next((c for c in CATEGORIES if c["id"] == category_id), None)


def list_categories() -> list[dict]:
    return CATEGORIES