mailcat/mail_routes.py

"""Shared mail routing policy for mailcat."""

import email.utils
import json
import re
from datetime import datetime
from functools import cache
from pathlib import Path

PREFIX = "INBOX."
MAILINGLIST_ROUTES_FILE = Path(__file__).parent / "mailinglist_routes.json"

DOMAIN_ROUTES: list[tuple[list[str], str]] = [
    (["nl.abnamro.com", "abnamro.nl"], "Financieel.Bank.ABN AMRO"),
    (["icscards.nl", "icsmarketing.icscards.nl", "service.icscards.nl"], "Financieel.Bank.ICS"),
    (["npex.nl"], "Financieel.Bank.NPEX"),
    (["snelstart.nl"], "Financieel.Bank.SnelStart"),
    (["tellow.nl"], "Financieel.Bank.Tellow"),
    (["unive.nl", "nieuwsbrief.unive.nl"], "Financieel.Verzekering.Univé"),
    (["heinenoord.nl"], "Financieel.Verzekering.VEZA"),
    (["flextender.nl"], "Werk.Opdrachten.Flextender"),
    (["onestopsourcing.nl"], "Werk.Opdrachten.OneStopSourcing"),
    (["immens.nu"], "Werk.Opdrachten.IMMENS"),
    (["circle8.nl"], "Werk.Opdrachten.Circle8"),
    (["getthere.nl"], "Werk.Opdrachten.Get There"),
    (["huckr.ai"], "Werk.Opdrachten.Huckr"),
    (["hey.meetjamie.ai"], "Werk.Opdrachten.Jamie"),
    (["futurexl.nl"], "Werk.Opdrachten.FutureXL"),
    (["boip.int"], "Werk.Overig.BOIP"),
    (["groningen.nl", "groningenbereikbaar.nl"], "Werk.Overig.Gemeente Groningen"),
    (["eneco-emobility.com", "eneco.nl"], "Mobiliteit.EV.Eneco eMobility"),
    (["shellrecharge.com"], "Mobiliteit.EV.Shell Recharge"),
    (["fastned.nl"], "Mobiliteit.EV.FastNed"),
    (["opngo.com"], "Mobiliteit.EV.OPnGO"),
    (["tesla.com"], "Mobiliteit.EV.Tesla"),
    (["mmsn.nl"], "Mobiliteit.EV.Mitsubishi"),
    (["alfen.com"], "Mobiliteit.EV.Alfen"),
    (["ns.nl", "email.ns.nl"], "Mobiliteit.OV & Reizen.NS"),
    (["booking.com"], "Mobiliteit.OV & Reizen.Booking"),
    (["flitsmeister.nl"], "Mobiliteit.OV & Reizen.Flitsmeister"),
    (["postnl.nl", "edm.postnl.nl", "notificatie.postnl.nl"], "Bestellingen.PostNL"),
    (["dhlparcel.nl", "dhlecommerce.nl"], "Bestellingen.DHL"),
    (["gls-netherlands.com"], "Bestellingen.GLS"),
    (["bol.com", "feedback.bol.com"], "Bestellingen.bol"),
    (["coolblue.nl", "coolblue.eu"], "Bestellingen.Coolblue"),
    (["allekabels.nl"], "Bestellingen.Allekabels"),
    (["vikingdirect.nl", "delivery.vikingdirect.nl", "online.vikingdirect.nl"], "Bestellingen.Viking"),
    (["123inkt.nl", "m.123inkt.nl"], "Bestellingen.123inkt"),
    (["officecentre.nl"], "Bestellingen.Office Centre"),
    (["sligro.nl"], "Bestellingen.Sligro"),
    (["makro.nl"], "Bestellingen.Makro"),
    (["smartphonehoesjes.nl"], "Bestellingen.Smartphonehoesjes"),
    (["strato.com", "news.strato.com", "marketingradar.strato.com"], "Diensten.Hosting.STRATO"),
    (["transip.nl"], "Diensten.Hosting.TransIP"),
    (["letsencrypt.org"], "Diensten.Hosting.Lets Encrypt"),
    (["backblaze.com"], "Diensten.Hosting.Backblaze"),
    (["kpn.com", "campagnes.kpn.com"], "Diensten.Telecom.KPN"),
    (["zakelijk.vodafone.nl", "vodafone.nl"], "Diensten.Telecom.Vodafone"),
    (["cursor.com"], "Diensten.Software.Cursor"),
    (["1password.com"], "Diensten.Software.1Password"),
    (["home.sophos.com", "cleverbridge.com"], "Diensten.Software.Sophos"),
    (["cogsciapps.com"], "Diensten.Software.CogSci Apps"),
    (["claude.com"], "Diensten.Software.Claude"),
    (["microsoft365.com"], "Diensten.Software.Microsoft Teams"),
    (["getmatter.com"], "Diensten.Software.Matter"),
    (["mindnode.com"], "Diensten.Software.MindNode"),
    (["papersapp.com"], "Diensten.Software.Papers"),
    (["team.kahoot.com"], "Diensten.Software.Kahoot"),
    (["perplexity.ai"], "Diensten.Software.Perplexity"),
    (["coursera.org", "email.coursera.org", "m.mail.coursera.org", "m.learn.coursera.org"], "Nieuwsbrieven.Leren.Coursera"),
    (["technologyreview.com", "fulfillment.technologyreview.com"], "Nieuwsbrieven.Leren.MIT Technology Review"),
    (["mit.edu"], "Nieuwsbrieven.Leren.MIT Open Learning"),
    (["storytellingwithdata.com"], "Nieuwsbrieven.Leren.Storytelling with Data"),
    (["linkingyourthinking.com"], "Nieuwsbrieven.Leren.Nick Milo"),
    (["evakeiffenheim.com"], "Nieuwsbrieven.Leren.Eva Keiffenheim"),
    (["tiggelaar.nl"], "Nieuwsbrieven.Leren.Ben Tiggelaar"),
    (["briankeating.com"], "Nieuwsbrieven.Leren.Brian Keating"),
    (["beehiiv.com", "mail.beehiiv.com"], "Nieuwsbrieven.Leren.AI Report"),
    (["elizabethbutlermd.com"], "Nieuwsbrieven.Leren.Elizabeth Butler"),
    (["crisismanager.nl"], "Nieuwsbrieven.Nieuws & Vakbladen.Vakblad Crisismanager"),
    (["jaarbeurs.nl"], "Nieuwsbrieven.Nieuws & Vakbladen.Computable"),
    (["presentation-guru.com"], "Nieuwsbrieven.Nieuws & Vakbladen.The Presentation Guru"),
    (["lendahand.com"], "Nieuwsbrieven.Overig.Lendahand"),
    (["thegoodroll.nl"], "Nieuwsbrieven.Overig.The Good Roll"),
    (["dmarc.postmarkapp.com"], "Technisch.DMARC"),
]

DOMAIN_LOOKUP: dict[str, str] = {
    domain.lower(): target
    for domains, target in DOMAIN_ROUTES
    for domain in domains
}

FACTUUR_KEYWORDS = [
    "factuur", "invoice", "rekening", "nota", "betaalverzoek",
    "receipt", "orderbevestiging", "order confirmation",
    "betaling ontvangen", "payment received",
    "uw factuur", "your invoice", "je factuur",
]

EXCLUDED_SOURCE_FOLDERS = {
    "INBOX.Sent", "Sent",
    "INBOX.Drafts", "Drafts",
    "INBOX.Trash", "Trash",
    "INBOX.Spam", "Spam",
}

ALREADY_SORTED_FOLDERS = {"INBOX.Facturen - verwerkt"}


@cache
def mailinglist_routes() -> list[dict]:
    """Return normalized mailing-list routing rules."""
    if not MAILINGLIST_ROUTES_FILE.exists():
        return []
    return json.loads(MAILINGLIST_ROUTES_FILE.read_text(encoding="utf-8"))


def mailinglist_destination(from_addr: str) -> str | None:
    """Return a mailing-list destination based on normalized policy rules."""
    domain = sender_domain(from_addr)
    from_lo = from_addr.lower()

    for rule in mailinglist_routes():
        domains = {item.lower() for item in rule.get("domains", [])}
        if domain and domain in domains:
            return rule["mailbox"]

        for needle in rule.get("from_contains", []):
            if needle.lower() in from_lo:
                return rule["mailbox"]

    return None


def sender_domain(from_addr: str) -> str:
    match = re.search(r"@([\w.\-]+)", from_addr)
    return match.group(1).lower() if match else ""


def quarter_folder(date_str: str) -> str:
    """Return the invoice quarter target based on an email Date header."""
    try:
        dt = email.utils.parsedate_to_datetime(date_str)
        year = dt.year
        quarter = (dt.month - 1) // 3 + 1
        return f"Financieel.Facturen.{year}.Q{quarter}"
    except Exception:
        year = datetime.now().year
        return f"Financieel.Facturen.{year}.Q1"


def route_from_subject(from_addr: str, subject: str) -> str | None:
    """Return a destination folder, __FACTUUR_DATE__, or None."""
    subj_lo = subject.lower()
    if any(keyword in subj_lo for keyword in FACTUUR_KEYWORDS):
        return "__FACTUUR_DATE__"
    if destination := mailinglist_destination(from_addr):
        return destination
    return DOMAIN_LOOKUP.get(sender_domain(from_addr))


def is_source_folder(folder: str) -> bool:
    """True if a folder should be scanned as a source mailbox."""
    return folder not in EXCLUDED_SOURCE_FOLDERS and folder not in ALREADY_SORTED_FOLDERS


def destination_folders(start_year: int = 2020, end_year: int = 2026) -> list[str]:
    """Return all destination folders without the INBOX. prefix."""
    folders = {target for _, target in DOMAIN_ROUTES}
    folders.update(
        rule["mailbox"]
        for rule in mailinglist_routes()
        if rule.get("mailbox")
    )
    folders.update(
        f"Financieel.Facturen.{year}.Q{quarter}"
        for year in range(start_year, end_year + 1)
        for quarter in range(1, 5)
    )

    with_parents = set(folders)
    for folder in list(folders):
        parts = folder.split(".")
        for i in range(1, len(parts)):
            with_parents.add(".".join(parts[:i]))

    return sorted(with_parents)