# -*- coding: utf-8 -*- # Copyright(C) 2012-2013 Romain Bignon # # This file is part of weboob. # # weboob is free software: you can redistribute it and/or modify # it under the terms of the GNU Affero General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # # weboob is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU Affero General Public License for more details. # # You should have received a copy of the GNU Affero General Public License # along with weboob. If not, see . from datetime import timedelta import urllib import re from weboob.tools.date import LinearDateGuesser from weboob.tools.browser import BaseBrowser, BrowserIncorrectPassword, BasePage, BrokenPageError from .pages.accounts import AccountsListPage, CPTHistoryPage, CardHistoryPage __all__ = ['HSBC'] class NotLoggedPage(BasePage): pass class HSBC(BaseBrowser): DOMAIN = 'client.hsbc.fr' PROTOCOL = 'https' CERTHASH = '3f8d5765792a7a83b8e82b121d87adf2c25f2358c3a39d4716c09854e5be791a' ENCODING = None # refer to the HTML encoding PAGES = {'https://client.hsbc.fr/session_absente.html': NotLoggedPage, 'https://client.hsbc.fr/cgi-bin/emcgi\?.*debr=COMPTES_PAN': AccountsListPage, 'https://client.hsbc.fr/cgi-bin/emcgi\?.*CPT_IdPrestation=.*': CPTHistoryPage, 'https://client.hsbc.fr/cgi-bin/emcgi\?.*CB_IdPrestation=.*': CardHistoryPage, } _session = None def home(self): self.login() def is_logged(self): return self._session is not None and not self.is_on_page(NotLoggedPage) def login(self): assert isinstance(self.username, basestring) assert isinstance(self.password, basestring) assert self.password.isdigit() data = {'Ident': self.username} r = self.readurl('https://client.hsbc.fr/cgi-bin/emcgi?Appl=WEBACC', urllib.urlencode(data), if_fail='raise') m = re.search('sessionid=([^ "]+)', r, flags=re.MULTILINE) if not m: raise BrowserIncorrectPassword() self._session = m.group(1) data = {'Secret': self.password} r = self.readurl('https://client.hsbc.fr/cgi-bin/emcgi?sessionid=%s' % self._session, urllib.urlencode(data), if_fail='raise') if r.find('Erreur Identification') >= 0: raise BrowserIncorrectPassword() m = re.search('url = "/cgi-bin/emcgi\?sessionid=([^& "]+)&debr="', r, flags=re.MULTILINE) if not m: raise BrokenPageError('Unable to find session token') self._session = m.group(1) def get_accounts_list(self): self.location(self.buildurl('/cgi-bin/emcgi', sessionid=self._session, debr='COMPTES_PAN')) return self.page.get_list() def get_account(self, id): assert isinstance(id, basestring) if not self.is_on_page(AccountsListPage): l = self.get_accounts_list() else: l = self.page.get_list() for a in l: if a.id == id: return a return None def get_history(self, account): if account._link_id is None: return for tr in self._get_history(account._link_id): yield tr for card in account._card_links: for tr in self._get_history(card): yield tr def _get_history(self, link): num_page = 0 guesser = LinearDateGuesser(date_max_bump=timedelta(45)) while link is not None: self.location(link) if self.page is None: return for tr in self.page.get_operations(num_page, guesser): yield tr link = self.page.get_next_link() num_page += 1