nupes-elections/nupes/scripts/import_geographie.py

import json
from pathlib import Path

import requests
from sqlalchemy import Engine, select
from sqlalchemy.orm import Session
from tqdm import tqdm

from nupes.cache import get_file
from nupes.models.geographie import BureauVote, Circonscription, Commune, Departement, Region


def importer_regions(engine: Engine, verbose: bool = False) -> None:
    etag = requests.get(
        "https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
        "/georef-france-region?select=data_processed").json()['data_processed']
    file = get_file("https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
                    "/georef-france-region/exports/geojson?lang=fr&timezone=Europe%2FParis",
                    "georef-france-region.geojson", etag)

    with file.open('r') as f:
        features = json.load(f)['features']

    with Session(engine) as session:
        for feature in tqdm(features, desc="Régions", disable=not verbose):
            region_dict = feature['properties']
            code_region = region_dict['reg_code'][0]
            nom_region = region_dict['reg_name'][0]

            if region := session.execute(select(Region).filter_by(code_insee=code_region)).scalar_one_or_none():
                region.libelle = nom_region
                region.geometry = feature['geometry']
            else:
                region = Region(code_insee=code_region, libelle=nom_region, geometry=feature['geometry'])
                session.add(region)

        if reg := session.execute(select(Region).filter_by(code_insee="ZZ")).scalar_one_or_none():
            reg.libelle = "Français⋅es de l'étranger"
        else:
            session.add(Region(code_insee="ZZ", libelle="Français⋅es de l'étranger", geometry={}))

        session.commit()


def importer_departements(engine: Engine, verbose: bool = False) -> None:
    etag = requests.get(
        "https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
        "/georef-france-departement?select=data_processed").json()['data_processed']
    file = get_file("https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
                    "/georef-france-departement/exports/geojson?lang=fr&timezone=Europe%2FParis",
                    "georef-france-departement.geojson", etag)

    with file.open('r') as f:
        features = json.load(f)['features']

    with Session(engine) as session:
        for feature in tqdm(features, desc="Départements", disable=not verbose):
            dpt_dict = feature['properties']
            code_dpt = dpt_dict['dep_code'][0]
            nom_dpt = dpt_dict['dep_name'][0]

            if dpt := session.execute(select(Departement).filter_by(code_insee=code_dpt)).scalar_one_or_none():
                dpt.libelle = nom_dpt
                dpt.region_code = dpt_dict['reg_code'][0]
                dpt.geometry = feature['geometry']
            else:
                dpt = Departement(code_insee=code_dpt, libelle=nom_dpt, region_code=dpt_dict['reg_code'][0],
                                  geometry=feature['geometry'])
                session.add(dpt)

        reg_etranger = session.execute(select(Region).filter_by(code_insee="ZZ")).scalar_one_or_none()
        if dpt := session.execute(select(Departement).filter_by(code_insee="ZZ")).scalar_one_or_none():
            dpt.libelle = "Français⋅es de l'étranger"
            dpt.region_code = reg_etranger.code_insee
        else:
            session.add(Departement(code_insee="ZZ", region_code=reg_etranger.code_insee,
                                    libelle="Français⋅es de l'étranger", geometry={}))

        session.commit()


def importer_circonscriptions(engine: Engine, verbose: bool = False) -> None:
    file = get_file("https://www.data.gouv.fr/fr/datasets/r/67c0f382-dc8d-4d1f-8a76-1162c53b9dfe",
                    "circonscriptions-legislatives-p10.geojson")

    with file.open('r') as f:
        features = json.load(f)['features']

    with Session(engine) as session:
        for feature in tqdm(features, desc="Circonscriptions", disable=not verbose):
            circo_dict = feature['properties']
            code_circo = circo_dict['codeCirconscription']
            code_dpt = circo_dict['codeDepartement']

            match code_dpt:
                case "ZA":
                    code_dpt = "971"
                case "ZB":
                    code_dpt = "972"
                case "ZC":
                    code_dpt = "973"
                case "ZD":
                    code_dpt = "974"
                case "ZS":
                    code_dpt = "975"
                case "ZM":
                    code_dpt = "976"

            numero_circo = int(code_circo[len(code_dpt):])
            circo_id = f"{code_dpt}-{numero_circo:02d}"

            if not session.execute(select(Departement).filter_by(code_insee=code_dpt)).scalar_one_or_none():
                print("Département non trouvé avec le code", code_dpt)
                continue

            if circo := session.execute(select(Circonscription).filter_by(id=circo_id)).scalar_one_or_none():
                circo.departement_code = code_dpt
                circo.numero = numero_circo
                circo.geometry = feature['geometry']
            else:
                circo = Circonscription(id=circo_id, departement_code=code_dpt, numero=numero_circo,
                                        geometry=feature['geometry'])
                session.add(circo)

        circos_manquantes = [
            {"id": "977-01", "dpt_id": "977", "numero": 1, "geometry": {}},
            {"id": "986-01", "dpt_id": "986", "numero": 1, "geometry": {}},
            {"id": "987-01", "dpt_id": "987", "numero": 1, "geometry": {}},
            {"id": "987-02", "dpt_id": "987", "numero": 2, "geometry": {}},
            {"id": "987-03", "dpt_id": "987", "numero": 3, "geometry": {}},
            {"id": "988-01", "dpt_id": "988", "numero": 1, "geometry": {}},
            {"id": "988-02", "dpt_id": "988", "numero": 2, "geometry": {}},
            {"id": "ZZ-01", "dpt_id": "ZZ", "numero": 1, "geometry": {}},
            {"id": "ZZ-02", "dpt_id": "ZZ", "numero": 2, "geometry": {}},
            {"id": "ZZ-03", "dpt_id": "ZZ", "numero": 3, "geometry": {}},
            {"id": "ZZ-04", "dpt_id": "ZZ", "numero": 4, "geometry": {}},
            {"id": "ZZ-05", "dpt_id": "ZZ", "numero": 5, "geometry": {}},
            {"id": "ZZ-06", "dpt_id": "ZZ", "numero": 6, "geometry": {}},
            {"id": "ZZ-07", "dpt_id": "ZZ", "numero": 7, "geometry": {}},
            {"id": "ZZ-08", "dpt_id": "ZZ", "numero": 8, "geometry": {}},
            {"id": "ZZ-09", "dpt_id": "ZZ", "numero": 9, "geometry": {}},
            {"id": "ZZ-10", "dpt_id": "ZZ", "numero": 10, "geometry": {}},
            {"id": "ZZ-11", "dpt_id": "ZZ", "numero": 11, "geometry": {}},
        ]
        for circo_dict in circos_manquantes:
            circo_geometry_file = Path(__file__).parent.parent / "initial_data" \
                                  / f"geometry_circo_{circo_dict['id']}.json"
            if circo_geometry_file.is_file():
                with circo_geometry_file.open('r') as f:
                    circo_dict['geometry'] = json.load(f)

            if not session.execute(select(Departement).filter_by(code_insee=circo_dict['dpt_id'])).scalar_one_or_none():
                print("Département non trouvé avec le code", circo_dict['dpt_id'])
                continue

            if circo := session.execute(select(Circonscription).filter_by(id=circo_dict['id'])).scalar_one_or_none():
                circo.departement_code = circo_dict['dpt_id']
                circo.numero = circo_dict['numero']
                circo.geometry = circo_dict['geometry']
            else:
                circo = Circonscription(id=circo_dict['id'], departement_code=circo_dict['dpt_id'],
                                        numero=circo_dict['numero'], geometry=circo_dict['geometry'])
                session.add(circo)

        session.commit()


def importer_communes(engine: Engine, verbose: bool = False) -> None:
    etag = requests.get(
        "https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
        "/georef-france-commune-arrondissement-municipal?select=data_processed").json()['data_processed']
    file = get_file("https://public.opendatasoft.com/api/explore/v2.1/catalog/datasets"
                    "/georef-france-commune-arrondissement-municipal/exports/geojson?lang=fr&timezone=Europe%2FParis",
                    "georef-france-commune-arrondissement-municipal.geojson", etag)

    with file.open('r') as f:
        features = json.load(f)['features']

    with Session(engine) as session:
        for feature in tqdm(features, desc="Communes", disable=not verbose):
            commune_dict = feature['properties']
            code_commune = commune_dict['com_arm_code'][0]
            nom_commune = commune_dict['com_name'][0]

            if commune := session.execute(select(Commune).filter_by(code_insee=code_commune)).scalar_one_or_none():
                commune.libelle = nom_commune
                commune.departement_code = commune_dict['dep_code'][0]
                commune.geometry = feature['geometry']
            else:
                commune = Commune(code_insee=code_commune, libelle=nom_commune,
                                  departement_code=commune_dict['dep_code'][0], geometry=feature['geometry'])
                session.add(commune)

        geometry_paris_file = Path(__file__).parent.parent / "initial_data" / "geometry_paris.json"
        geometry_lyon_file = Path(__file__).parent.parent / "initial_data" / "geometry_lyon.json"
        geometry_marseille_file = Path(__file__).parent.parent / "initial_data" / "geometry_marseille.json"

        with geometry_paris_file.open('r') as f:
            geometry_paris = json.load(f)
        with geometry_lyon_file.open('r') as f:
            geometry_lyon = json.load(f)
        with geometry_marseille_file.open('r') as f:
            geometry_marseille = json.load(f)

        paris_lyon_marseille = [
            {'code_insee': "75056", 'libelle': "Paris", 'departement_code': "75", 'geometry': geometry_paris},
            {'code_insee': "69123", 'libelle': "Lyon", 'departement_code': "69", 'geometry': geometry_lyon},
            {'code_insee': "13055", 'libelle': "Marseille", 'departement_code': "13", 'geometry': geometry_marseille},
        ]

        for commune_dict in paris_lyon_marseille:
            if not session.execute(select(Commune).filter_by(code_insee=commune_dict['code_insee'])) \
                    .scalar_one_or_none():
                commune = Commune(code_insee=commune_dict['code_insee'], libelle=commune_dict['libelle'],
                                  departement_code=commune_dict['departement_code'],
                                  geometry=commune_dict['geometry'])
                session.add(commune)

        session.commit()


def importer_bureaux_vote(engine: Engine, verbose: bool = False) -> None:
    file = get_file("https://files.data.gouv.fr/reu/contours-france-entiere-latest-v2.geojson",
                    "contours-france-entiere-latest-v2.geojson")

    with file.open('r') as f:
        features = json.load(f)['features']

    with Session(engine) as session:
        for feature in tqdm(features, desc="Bureaux de vote", disable=not verbose):
            bv_dict = feature['properties']
            code_commune = bv_dict['id_bv'].split('_')[0]
            code_bv = bv_dict['numeroBureauVote']
            dpt_code = bv_dict['codeDepartement']

            match dpt_code:
                case "ZA":
                    dpt_code = "971"
                case "ZB":
                    dpt_code = "972"
                case "ZC":
                    dpt_code = "973"
                case "ZD":
                    dpt_code = "974"
                case "ZS":
                    dpt_code = "975"
                case "ZM":
                    dpt_code = "976"

            numero_circo = int(bv_dict['codeCirconscription'][len(dpt_code):])
            code_circo = f"{dpt_code}-{numero_circo:02d}"
            bv_id = bv_dict['id_bv'].split()[0]

            if code_commune == "57447":
                # Petites erreurs sur la commune de Marly
                code_bv = int(bv_id.split('_')[1])
            elif bv_id == "59508_3":
                # Petite erreur à Roncq
                code_bv = 103

            bv_libelle = f"Bureau {code_bv} de {bv_dict['nomCommune']}"

            if not session.execute(select(Commune).filter_by(code_insee=code_commune)).scalar_one_or_none():
                print("Commune non trouvée avec le code", code_commune, "et le nom", bv_dict['nomCommune'])
                continue

            if not session.execute(select(Circonscription).filter_by(id=code_circo)).scalar_one_or_none():
                session.add(Circonscription(id=code_circo, departement_code=dpt_code, numero=numero_circo))

            if bv := session.execute(select(BureauVote).filter_by(id=bv_id)).scalar_one_or_none():
                bv.commune_code = code_commune
                bv.code_bureau = code_bv
                bv.circo_code = code_circo
                bv.libelle = bv_libelle
                bv.geometry = feature['geometry']
            else:
                bv = BureauVote(id=bv_id, commune_code=code_commune, code_bureau=code_bv, circo_code=code_circo,
                                libelle=bv_libelle, adresse="", geometry=feature['geometry'])
                session.add(bv)

        session.commit()


def run(engine: Engine, verbose: bool = False) -> None:
    importer_regions(engine, verbose)
    importer_departements(engine, verbose)
    importer_circonscriptions(engine, verbose)
    importer_communes(engine, verbose)
    importer_bureaux_vote(engine, verbose)