diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml index f755737088e2287166536bbb11ee4552f3385e1a..43a71409bbf2521a57cfff810ce605cb8649f783 100644 --- a/.gitlab-ci.yml +++ b/.gitlab-ci.yml @@ -147,6 +147,11 @@ test_pref38: PREF: "pref38" extends: .default_pref +test_pref42: + variables: + PREF: "pref42" + extends: .default_pref + test_pref59: variables: PREF: "pref59" diff --git a/Makefile b/Makefile index 9b98fcaa8fa48ca601e3005f236da4d2f9fad509..af3a6fbe675b632c29fc4e2da72f7babd375d2b2 100644 --- a/Makefile +++ b/Makefile @@ -1,4 +1,4 @@ -make: ppparis pref04 pref05 pref06 pref09 pref13 pref31 pref34 pref35 pref38 pref59 pref62 pref65 pref66 pref69 pref80 pref81 pref83 pref87 pref976 +make: ppparis pref04 pref05 pref06 pref09 pref13 pref31 pref34 pref35 pref38 pref42 pref59 pref62 pref65 pref66 pref69 pref80 pref81 pref83 pref87 pref976 ppparis: python cli.py --pref ppparis pref04: diff --git a/RAAspotter_pref42.py b/RAAspotter_pref42.py new file mode 100644 index 0000000000000000000000000000000000000000..254f560f50302680bf32d2458a26813de1accb1e --- /dev/null +++ b/RAAspotter_pref42.py @@ -0,0 +1,83 @@ +import os +import datetime +import re + +from bs4 import BeautifulSoup +from urllib.parse import unquote + +from RAAspotter import RAAspotter + + +class RAAspotter_pref42(RAAspotter): + + # Config + __HOST = 'https://www.loire.gouv.fr' + __RAA_PAGE = f'{__HOST}/Publications/Publications-legales/Recueil-des-Actes-Administratifs' + __USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/115.0' + full_name = 'Préfecture de de la Loire' + short_code = 'pref42' + + def __init__(self, data_dir): + super().__init__(data_dir, self.__USER_AGENT) + self.enable_tor(10) + + def get_raa(self, keywords): + self.print_output('RAAspotter_pref42') + self.print_output(f'Termes recherchés: {keywords}') + self.print_output('') + + year_pages_to_parse = [] + + # On détermine quelles pages d'année parser + year_pages = self.get_sub_pages_with_pager( + self.__RAA_PAGE, + 'div.fr-card.fr-card--horizontal.fr-card--sm.fr-enlarge-link.fr-mb-3w 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.fr-pagination__link--next.fr-pagination__link--lg-label', + 'div.fr-card.fr-card--horizontal.fr-card--sm.fr-enlarge-link.fr-mb-3w div.fr-card__body div.fr-card__content div.fr-card__end p.fr-card__detail', + self.__HOST + ) + for year_page in year_pages: + year = 9999 + try: + year = int(re.search('([0-9]{4})', year_page['name'], re.IGNORECASE).group(1)) + if year is None: + year = 9999 + except Exception as exc: + logger.warning(f"Impossible de deviner l\'année de la page {year_page['name']}") + year = 9999 + + if year >= self.not_before.year: + year_pages_to_parse.append(year_page['url']) + + elements = [] + # Pour chaque année, on parse les RAA + for year_page in year_pages_to_parse: + page_content = self.get_page(year_page, 'get').content + for element in self.get_raa_elements(page_content)[::-1]: + elements.append(element) + + # On parse les RAA + 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('div.fr-downloads-group.fr-downloads-group--bordered ul li a'): + 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 diff --git a/README.md b/README.md index 9c5dee9caf373be2468b84799eebd92b6119d69b..6d8a32ead9c44f177aeea8cea92cad2218ca1f82 100644 --- a/README.md +++ b/README.md @@ -42,6 +42,7 @@ Il est possible de ne lancer l'analyse que pour une seule administration, avec l - Préfecture de l'Hérault (identifiant : `pref34`) - Préfecture d'Ille-et-Vilaine (identifiant : `pref35`) - Préfecture de l'Isère (identifiant : `pref38`) +- Préfecture de la Loire (identifiant : `pref42`) - Préfecture du Nord (identifiant : `pref59`) - Préfecture du Pas-de-Calais (identifiant : `pref62`) - Préfecture des Pyrénées-Atlantiques (identifiant : `pref64`) diff --git a/cli.py b/cli.py index 390c082855bddecbadd9b98535d83a3c8ace25eb..ec19ba4ea698596d5b47e39f78f1b7714b31d6c0 100755 --- a/cli.py +++ b/cli.py @@ -48,6 +48,7 @@ available_prefs = [ 'pref34', 'pref35', 'pref38', + 'pref42', 'pref59', 'pref62', 'pref64',