from __future__ import annotations

from django.db import transaction

from apps.jobs.services.geocoding import LocationMatchResult, resolve_cached_location

from .models import ProfileRevision, SearchProfile


@transaction.atomic
def resolve_profile_home_location(profile: SearchProfile) -> LocationMatchResult:
    """Resolve a user-entered postal code without exposing coordinate fields in the UI."""
    query = profile.home_postal_code.strip()
    result = (
        resolve_cached_location(query) if query else LocationMatchResult(query=query, location=None)
    )
    location = result.location
    profile.home_municipality = location.municipality or "" if location else ""
    profile.home_latitude = location.point.latitude if location and location.point else None
    profile.home_longitude = location.point.longitude if location and location.point else None
    profile.save(
        update_fields=[
            "home_municipality",
            "home_latitude",
            "home_longitude",
            "updated_at",
        ]
    )
    return result


@transaction.atomic
def save_profile_revision(profile: SearchProfile, *, reason: str) -> ProfileRevision:
    latest = profile.revisions.order_by("-version").first()
    next_version = (latest.version if latest else 0) + 1
    if profile.version != next_version:
        profile.version = next_version
        profile.save(update_fields=["version", "updated_at"])
    return ProfileRevision.objects.create(
        profile=profile,
        version=next_version,
        snapshot=profile.snapshot(),
        reason=reason,
    )


@transaction.atomic
def apply_feedback_delta(
    profile: SearchProfile,
    feature: str,
    delta: float,
    *,
    min_weight: float = 0.0,
    max_weight: float = 40.0,
) -> SearchProfile:
    if not profile.learning_enabled:
        return profile
    weights = dict(profile.weights)
    current = float(weights.get(feature, 0.0))
    weights[feature] = round(max(min_weight, min(max_weight, current + delta)), 2)
    profile.weights = weights
    profile.save(update_fields=["weights", "updated_at"])
    save_profile_revision(profile, reason=f"feedback_delta:{feature}:{delta:+.2f}")
    return profile