mapping-geojson-osm/etalab_data/planing_familial/scrap.py

import json
import geopandas as gpd
from bs4 import BeautifulSoup
from shapely.geometry import Point

# Charger le fichier HTML
with open("liste.html", "r") as file:
    html = file.read()

# Analyser le code HTML avec BeautifulSoup
soup = BeautifulSoup(html, "html.parser")

# Trouver toutes les balises <article> qui contiennent les informations des antennes
articles = soup.find_all("article", class_="node node--type-hp-antenne node--view-mode-map-result")

# Liste pour stocker les données de chaque antenne
antennes = []

# Parcourir chaque article et extraire les informations
for article in articles:
    titre = article.find("button", class_="nsb action-title").text.strip()
    adresse = article.find("p", class_="address").text.strip()
    website = article.find("a", class_="icon arrow").get("href").strip()
    violences = article.find("li", class_="icon violences")
    sexualities = article.find("li", class_="icon sexualities")
    detection = article.find("li", class_="icon detection")
    contraception = article.find("li", class_="icon contraception")
    abortion = article.find("li", class_="icon abortion")

    # Gérer les articles qui n'ont pas de numéro de téléphone
    telephone = ""
    try:
        telephone = article.find("p", class_="tel").text.strip()
    except AttributeError:
        pass

    latitude = float(article["data-latlng"].split(",")[0].replace("[", "").replace('"', ""))
    longitude = float(article["data-latlng"].split(",")[1].replace("]", "").replace('"', ""))
    geometry = Point(longitude, latitude)

    antenne = {
        "type": "Feature",
        "geometry":geometry,
        "nom": titre,
        "adresse": adresse,
        "contact:phone": telephone,
        "contact:website": 'https://www.planning-familial.org'+website,
        "family_planning:handles:violences": ('yes' if violences else 'no'),
        "family_planning:handles:sexualities": ('yes' if sexualities else 'no'),
        "family_planning:handles:detection": ('yes' if detection else 'no'),
        "family_planning:handles:abortion": ('yes' if abortion else 'no'),
        "family_planning:handles:contraception": ('yes' if contraception else 'no'),
    }

#     print(antenne)

    antennes.append(antenne)

# Convertir la liste des antennes en un GeoDataFrame
gdf = gpd.GeoDataFrame(antennes)

# Exporter le GeoDataFrame au format GeoJSON
gdf.to_file("antennes_planning_familial.json", driver="GeoJSON")