Skip to content
Extraits de code Groupes Projets
Valider e2a3d024 rédigé par Bastien Le Querrec's avatar Bastien Le Querrec
Parcourir les fichiers

pref69: ajout de la préfecture du Rhône

parent 222ac499
Aucune branche associée trouvée
Aucune étiquette associée trouvée
Aucune requête de fusion associée trouvée
......@@ -238,3 +238,31 @@ test_pref62:
- data/pref62/*.txt
- output.log
expire_in: 1 hour
test_pref69:
stage: test
image: registry.git.laquadrature.net/bastien/raaspotter/base:latest
tags:
- unprivileged
needs: [install]
script:
- curl --silent --location --output artifacts.zip "${CI_SERVER_PROTOCOL}://${CI_SERVER_HOST}:${CI_SERVER_PORT}/api/v4/projects/${CI_PROJECT_ID}/jobs/artifacts/${CI_COMMIT_BRANCH}/download?job=${CI_JOB_NAME}&job_token=${CI_JOB_TOKEN}" || true
- unzip -q artifacts.zip data/pref69/* || true
- rm artifacts.zip || true
- source bin/activate
- /etc/init.d/tor start
- python ./cli.py --pref pref69
retry: 2
only:
- main
cache:
key: $CI_COMMIT_REF_SLUG
paths:
- bin/
- lib/
- pyvenv.cfg
artifacts:
paths:
- data/pref69/*.txt
- output.log
expire_in: 1 hour
make: ppparis pref06 pref13 pref35 pref38 pref59 pref62
make: ppparis pref06 pref13 pref35 pref38 pref59 pref62 pref69
ppparis:
python cli.py --pref ppparis
pref06:
......@@ -13,3 +13,5 @@ pref59:
python cli.py --pref pref59
pref62:
python cli.py --pref pref62
pref69:
python cli.py --pref pref69
import os, re, ssl
import os, re, ssl, sys
import subprocess
import logging
import requests
......@@ -136,6 +136,27 @@ class RAAspotter:
sub_pages.append(url)
return sub_pages
def get_sub_pages_with_pager(self, page, sub_page_element, pager_element, host=""):
pages = []
page_content = self.get_page(page, 'get').content
# On initialise le parser
soup = BeautifulSoup(page_content, 'html.parser')
# On recherche les sous-pages
sub_pages = soup.select(sub_page_element)
for sub_page in sub_pages:
if sub_page.get('href'):
pages.append(f"{host}{sub_page['href']}")
# On recherche un pager, et si on le trouve on le suit
pager = soup.select(pager_element)
if pager[0] and pager[0].get('href'):
for sub_page in self.get_sub_pages_with_pager(f"{host}{pager[0]['href']}", sub_page_element, pager_element, host):
pages.append(sub_page)
return pages
def get_raa_with_pager(self, pages_list, pager_element, host=""):
elements = []
# On parse chaque page passée en paramètre
......
import os, sys
import datetime
from bs4 import BeautifulSoup
from urllib.parse import unquote
from RAAspotter import RAAspotter
class RAAspotter_pref69(RAAspotter):
# Config
__HOST = 'https://www.rhone.gouv.fr'
__RAA_PAGE = {'2024': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2024',
'2023': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2023',
'2022': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2022',
'2021': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2021',
'2020': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2020',
'2019': f'{__HOST}/Publications/Recueil-des-actes-administratifs-du-Rhone-RAA/Recueils-de-2019'}
__USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/115.0'
full_name = 'Préfecture du Rhône'
short_code = 'pref69'
def __init__(self, data_dir):
super().__init__(data_dir, self.__USER_AGENT)
self.enable_tor(20)
def get_raa(self, keywords):
self.print_output('RAAspotter_pref69')
self.print_output(f'Termes recherchés: {keywords}')
self.print_output('')
pages_to_parse = []
if self.not_before.year <= 2024:
pages_to_parse.append(self.__RAA_PAGE['2024'])
if self.not_before.year <= 2023:
pages_to_parse.append(self.__RAA_PAGE['2023'])
if self.not_before.year <= 2022:
pages_to_parse.append(self.__RAA_PAGE['2022'])
if self.not_before.year <= 2021:
pages_to_parse.append(self.__RAA_PAGE['2021'])
if self.not_before.year <= 2020:
pages_to_parse.append(self.__RAA_PAGE['2020'])
if self.not_before.year <= 2019:
pages_to_parse.append(self.__RAA_PAGE['2019'])
sub_pages_to_parse = []
for raa_page in pages_to_parse:
sub_pages = self.get_sub_pages_with_pager(raa_page,
"div.fr-card__body div.fr-card__content h2.fr-card__title a.fr-card__link",
"ul.fr-pagination__list li a.fr-pagination__link--next",
self.__HOST)[::-1]
for sub_page in sub_pages:
sub_pages_to_parse.append(sub_page)
elements = []
for sub_page_to_parse in sub_pages_to_parse:
page_content = self.get_page(sub_page_to_parse, 'get').content
for element in self.get_raa_elements(page_content)[::-1]:
elements.append(element)
self.parse_raa(elements, keywords.split(','))
self.mailer()
def get_raa_elements(self, page_content):
elements = []
# On charge le parser
soup = BeautifulSoup(page_content, 'html.parser')
# On récupère chaque balise a
for a in soup.select('a.fr-link.fr-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').previous_sibling.replace('Télécharger ', '').strip()
date = datetime.datetime.strptime(a.find('span').get_text().split(' - ')[-1].strip(), '%d/%m/%Y')
filename = url.split('/')[-1]
raa = RAAspotter.RAA(url, date, name, filename)
elements.append(raa)
return elements
......@@ -38,6 +38,7 @@ Il est possible de ne lancer l'analyse que pour une seule administration, avec l
- Préfecture d'Ille-et-Vilaine (identifiant : `pref35`)
- Préfecture du Nord (identifiant : `pref59`)
- Préfecture du Pas-de-Calais (identifiant : `pref62`)
- Préfecture du Rhône (identifiant : `pref69`)
## Licence
......
......@@ -41,7 +41,8 @@ available_prefs = [
'pref35',
'pref38',
'pref59',
'pref62'
'pref62',
'pref69'
]
# Début du script
......
0% Chargement en cours ou .
You are about to add 0 people to the discussion. Proceed with caution.
Terminez d'abord l'édition de ce message.
Veuillez vous inscrire ou vous pour commenter