Conversion HTML arrêté "ZIPVA" en geojson

jorf2geojson.py

#! /usr/bin/python3

# script de conversion du code HTML de l'arrêté https://www.legifrance.gouv.fr/jorf/id/JORFTEXT000048234967
# au format geojson
# !!! MANQUE MURUROA non délimité dans le tableau d'origine !!!

import re, json
from bs4 import BeautifulSoup

cols = ['','immatriculation','communes','nom','dep','limites','ministere','zone_aero']

# regex pour extraire les coordonnées
dms = re.compile(r'(\d*)° (\d*)[\'′] ([\d,\.]*)["”] ?([NSEWO])')
geojson = {'type': 'FeatureCollection', 'features' : []}

with open('zipva.html','r') as html:
    soup = BeautifulSoup(html, 'html.parser')
    for tr in soup.find('table').find_all('tr'):
        j = {}
        g = {'type':'Polygon', 'coordinates':[]}
        c = 0
        for td in tr.find_all('td'):
            c += 1
            n = 0
            for txt in td.strings:
                if 'Zone' in txt:
                    if len(g['coordinates']) > 0:
                        # on termine le polygone avec le premier point
                        g['coordinates'].append(g['coordinates'][0])
                        g['coordinates'] = [g['coordinates']]
                        geojson['features'].append({'type':'Feature','properties': j,'geometry': g})
                        g = {'type':'Polygon', 'coordinates':[]}


                n += 1
                if cols[c] != 'limites': # properties
                    j[cols[c]] = txt
                else:
                    point = []
                    for coord in re.findall(dms, txt.replace("''",'"')):
                        deg = round(float(coord[0]) + float(coord[1])/60 + float(coord[2].replace(',','.'))/3600,6)
                        if coord[3] in ['S','W','O']:
                            deg = -deg
                        point.append(deg)
                        if len(point) == 2:
                            g['coordinates'].append(point)
                            point = []

        if len(g['coordinates'])>0:
            # on termine le polygone avec le premier point
            g['coordinates'].append(g['coordinates'][0])
            g['coordinates'] = [g['coordinates']]
            geojson['features'].append({'type':'Feature','properties': j,'geometry': g})

# dump du geojson final
print(json.dumps(geojson, ensure_ascii=False))