Newer
Older
import os
import datetime
from bs4 import BeautifulSoup
from urllib.parse import unquote
from Attrap import Attrap
class Attrap_prefIdf(Attrap):
# Config
__HOST = 'https://www.prefectures-regions.gouv.fr'
__RAA_PAGE = {
'2024': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2024',
'2023': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2023',
'2022': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2022',
'2021': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2021',
'2020': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2020',
'2019': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2019',
'2018': f'{__HOST}/ile-de-france/ile-de-france/ile-de-france/Documents-publications/Recueil-des-actes-administratifs/RAA-de-la-region-Ile-de-France-2018'
}
__USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/115.0'
full_name = 'Préfecture d\'Île-de-France'
short_code = 'prefIdf'
def __init__(self, data_dir):
super().__init__(data_dir, self.__USER_AGENT)
self.enable_tor(10)
def get_raa(self, keywords):
year_pages_to_parse = []
# Les RAA de l'Île-de-France sont éparpillés sur des sous-pages par mois.
# Donc on parse la page principale à la recherche des sous-pages.
if self.not_before.year <= 2024:
year_pages_to_parse.append(self.__RAA_PAGE['2024'])
if self.not_before.year <= 2023:
year_pages_to_parse.append(self.__RAA_PAGE['2023'])
if self.not_before.year <= 2022:
year_pages_to_parse.append(self.__RAA_PAGE['2022'])
if self.not_before.year <= 2021:
year_pages_to_parse.append(self.__RAA_PAGE['2021'])
if self.not_before.year <= 2020:
year_pages_to_parse.append(self.__RAA_PAGE['2020'])
if self.not_before.year <= 2019:
year_pages_to_parse.append(self.__RAA_PAGE['2019'])
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
pages_to_parse = []
for year_page in year_pages_to_parse:
page_content = self.get_page(year_page, 'get').content
year = BeautifulSoup(page_content, 'html.parser').select('div.breadcrumb div.container p span.active')[0].get_text().split('-')[-1].strip()
month_pages = self.get_sub_pages(
page_content,
'div.sommaire-bloc div.sommaire-content ol li a',
self.__HOST,
False
)[::-1]
for month_page in month_pages:
month_date = Attrap.guess_date(f"{month_page['name']} {year}", "(.*)").replace(day=1)
if month_date >= self.not_before.replace(day=1):
pages_to_parse.append(month_page['url'])
elements = []
for page in pages_to_parse:
page_content = self.get_page(page, 'get').content
for element in self.get_raa_elements(page_content):
elements.append(element)
self.parse_raa(elements, keywords)
self.mailer()
def get_raa_elements(self, page_content):
elements = []
# On charge le parser
soup = BeautifulSoup(page_content, 'html.parser')
# Pour chaque balise a, on regarde si c'est un PDF, et si oui on le parse
for a in soup.select('main div.container.main-container div.col-main article.article div.texte div a.link-download'):
if a.get('href') and a['href'].endswith('.pdf'):
if a['href'].startswith('/'):
url = f"{self.__HOST}{a['href']}"
else:
url = a['href']
url = unquote(url)
name = a.find('span').get_text().strip()
# On devine la date du RAA à partir du nom de fichier
guessed = Attrap.guess_date(name, '((?:[0-9]{2}(?:-|\\.)[0-9]{2}(?:-|\\.)20[0-9]{2})|(?:20[0-9]{2}(?:-|\\.)[0-9]{2}(?:-|\\.)[0-9]{2})\\D*^)')
if (guessed == datetime.datetime(9999, 1, 1, 0, 0)):
date = None
else:
date = guessed
raa = Attrap.RAA(url, date, name)
elements.append(raa)
return elements