"""
Configuration constants for the DVF data pipeline.

Single source of truth for paths, URLs, thresholds, and mappings.
"""

from pathlib import Path

# ---------------------------------------------------------------------------
# Paths
# ---------------------------------------------------------------------------
ROOT_DIR = Path(__file__).resolve().parent.parent
DATA_DIR = ROOT_DIR / "data"
RAW_DIR = DATA_DIR / "raw"
PROCESSED_DIR = DATA_DIR / "processed"
AGGREGATED_DIR = DATA_DIR / "aggregated"
SECTIONS_DIR = AGGREGATED_DIR / "sections"

# ---------------------------------------------------------------------------
# DVF data source
# ---------------------------------------------------------------------------
DVF_BASE_URL = "https://files.data.gouv.fr/geo-dvf/latest/csv"
DVF_YEARS = list(range(2020, 2026))  # 2020-2025: geo-dvf only available for last 5 years

def dvf_url(year: int) -> str:
    """Return download URL for a given year's national DVF geolocalized CSV."""
    return f"{DVF_BASE_URL}/{year}/full.csv.gz"

# ---------------------------------------------------------------------------
# Columns we actually need (saves memory on load)
# ---------------------------------------------------------------------------
DVF_COLUMNS = [
    "id_mutation",
    "date_mutation",
    "nature_mutation",
    "valeur_fonciere",
    "code_postal",
    "code_commune",
    "nom_commune",
    "code_departement",
    "id_parcelle",
    "code_type_local",
    "type_local",
    "surface_reelle_bati",
    "nombre_pieces_principales",
    "nombre_lots",
    "longitude",
    "latitude",
]

# ---------------------------------------------------------------------------
# Filtering thresholds
# ---------------------------------------------------------------------------
VALID_NATURE_MUTATION = "Vente"

VALID_TYPE_LOCAL = ["Appartement", "Maison"]  # Residential only per Carlos's feedback

TYPE_LOCAL_SHORT = {
    "Appartement": "appartement",
    "Maison": "maison",
}

# ---------------------------------------------------------------------------
# Temporal weighting parameters
# ---------------------------------------------------------------------------
REFERENCE_DATE = "2025-01-01"  # Anchor date for temporal decay
TEMPORAL_LAMBDA = 0.97         # Monthly decay factor (half-life ~23 months)
TRIM_FRACTION = 0.20           # Trim 20% from each tail for trimmed mean

# Price per m² bounds for outlier removal
PRICE_M2_MIN = 200       # €/m² — below this is almost certainly an error
PRICE_M2_MAX = 25_000    # €/m² — above this is extreme luxury / error
SURFACE_MIN = 9           # m² — below 9m² is legally not habitable in France
SURFACE_MAX = 1000        # m² — above this for a single unit is suspect

# ---------------------------------------------------------------------------
# Department → Region mapping (2016 reform)
# ---------------------------------------------------------------------------
DEPT_TO_REGION: dict[str, str] = {}
_REGION_DEPTS = {
    "84-Auvergne-Rhône-Alpes": "01,03,07,15,26,38,42,43,63,69,73,74",
    "27-Bourgogne-Franche-Comté": "21,25,39,58,70,71,89,90",
    "53-Bretagne": "22,29,35,56",
    "24-Centre-Val de Loire": "18,28,36,37,41,45",
    "94-Corse": "2A,2B",
    "44-Grand Est": "08,10,51,52,54,55,57,67,68,88",
    "32-Hauts-de-France": "02,59,60,62,80",
    "11-Île-de-France": "75,77,78,91,92,93,94,95",
    "28-Normandie": "14,27,50,61,76",
    "75-Nouvelle-Aquitaine": "16,17,19,23,24,33,40,47,64,79,86,87",
    "76-Occitanie": "09,11,12,30,31,32,34,46,48,65,66,81,82",
    "52-Pays de la Loire": "44,49,53,72,85",
    "93-Provence-Alpes-Côte d'Azur": "04,05,06,13,83,84",
    "01-Guadeloupe": "971",
    "02-Martinique": "972",
    "03-Guyane": "973",
    "04-La Réunion": "974",
    "06-Mayotte": "976",
}

REGION_NAMES: dict[str, str] = {}
for key, depts_str in _REGION_DEPTS.items():
    code, name = key.split("-", 1)
    REGION_NAMES[code] = name
    for d in depts_str.split(","):
        DEPT_TO_REGION[d.strip()] = code

# Departments with no DVF data (Alsace-Moselle + Mayotte)
NO_DVF_DEPARTMENTS = {"57", "67", "68", "976"}

# ---------------------------------------------------------------------------
# Top 10 cities by commune population (INSEE code → name)
# Source: INSEE Recensement de la population
# https://www.data.gouv.fr/datasets/population-municipale-des-communes-france-entiere
# Note: Strasbourg (67482) is in Alsace-Moselle — no DVF data available
# ---------------------------------------------------------------------------
TOP_10_CITIES: dict[str, str] = {
    "75056": "Paris",
    "13055": "Marseille",
    "69123": "Lyon",
    "31555": "Toulouse",
    "06088": "Nice",
    "44109": "Nantes",
    "34172": "Montpellier",
    "67482": "Strasbourg",
    "33063": "Bordeaux",
    "59350": "Lille",
}

# Paris, Lyon, Marseille have arrondissements — we need to map them back
ARRONDISSEMENT_MAPPING: dict[str, str] = {}
# Paris: 75101-75120 → 75056
for i in range(1, 21):
    ARRONDISSEMENT_MAPPING[f"751{i:02d}"] = "75056"
# Lyon: 69381-69389 → 69123
for i in range(1, 10):
    ARRONDISSEMENT_MAPPING[f"6938{i}"] = "69123"
# Marseille: 13201-13216 → 13055
for i in range(1, 17):
    ARRONDISSEMENT_MAPPING[f"132{i:02d}"] = "13055"

# ---------------------------------------------------------------------------
# Aggregation levels
# ---------------------------------------------------------------------------
AGGREGATION_LEVELS = [
    "country",
    "region",
    "department",
    "commune",
    "postcode",
    "section",
]