support repositories to manage backends (closes #747)

modules/cragr/__init__.py

@@ -0,0 +1,23 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+from .backend import CragrBackend
+
+__all__ = ['CragrBackend']

modules/cragr/backend.py

@@ -0,0 +1,105 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 Romain Bignon, Christophe Benz
+#
+# 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 <http://www.gnu.org/licenses/>.
+
+
+from weboob.capabilities.bank import ICapBank, AccountNotFound
+from weboob.tools.backend import BaseBackend, BackendConfig
+from weboob.tools.ordereddict import OrderedDict
+from weboob.tools.value import ValueBackendPassword, Value
+
+from .browser import Cragr
+
+
+__all__ = ['CragrBackend']
+
+
+class CragrBackend(BaseBackend, ICapBank):
+    NAME = 'cragr'
+    MAINTAINER = 'Xavier Guerrin'
+    EMAIL = 'xavier@tuxfamily.org'
+    VERSION = '0.a'
+    DESCRIPTION = 'Credit Agricole french bank\'s website'
+    LICENSE = 'AGPLv3+'
+    website_choices = OrderedDict([(k, u'%s (%s)' % (v, k)) for k, v in sorted({
+        'm.ca-alpesprovence.fr': u'Alpes Provence',
+        'm.ca-anjou-maine.fr': u'Anjou Maine',
+        'm.ca-atlantique-vendee.fr': u'Atlantique Vendée',
+        'm.ca-aquitaine.fr': u'Aquitaine',
+        'm.ca-briepicardie.fr': u'Brie Picardie',
+        'm.ca-centrest.fr': u'Centre Est',
+        'm.ca-centrefrance.fr': u'Centre France',
+        'm.ca-centreloire.fr': u'Centre Loire',
+        'm.ca-centreouest.fr': u'Centre Ouest',
+        'm.ca-cb.fr': u'Champagne Bourgogne',
+        'm.ca-charente-perigord.fr': u'Charente Périgord',
+        'm.ca-cmds.fr': u'Charente-Maritime Deux-Sèvres',
+        'm.ca-corse.fr': u'Corse',
+        'm.ca-cotesdarmor.fr': u'Côtes d\'Armor',
+        'm.ca-des-savoie.fr': u'Des Savoie',
+        'm.ca-finistere.fr': u'Finistere',
+        'm.ca-paris.fr': u'Ile-de-France',
+        'm.ca-illeetvilaine.fr': u'Ille-et-Vilaine',
+        'm.ca-languedoc.fr': u'Languedoc',
+        'm.ca-loirehauteloire.fr': u'Loire Haute Loire',
+        'm.ca-lorraine.fr': u'Lorraine',
+        'm.ca-martinique.fr': u'Martinique Guyane',
+        'm.ca-morbihan.fr': u'Morbihan',
+        'm.ca-norddefrance.fr': u'Nord de France',
+        'm.ca-nord-est.fr': u'Nord Est',
+        'm.ca-nmp.fr': u'Nord Midi-Pyrénées',
+        'm.ca-normandie.fr': u'Normandie',
+        'm.ca-normandie-seine.fr': u'Normandie Seine',
+        'm.ca-pca.fr': u'Provence Côte d\'Azur',
+        'm.lefil.com': u'Pyrénées Gascogne',
+        'm.ca-reunion.fr': u'Réunion',
+        'm.ca-sudrhonealpes.fr': u'Sud Rhône Alpes',
+        'm.ca-sudmed.fr': u'Sud Méditerranée',
+        'm.ca-toulouse31.fr': u'Toulouse 31', # m.ca-toulousain.fr redirects here
+        'm.ca-tourainepoitou.fr': u'Tourraine Poitou',
+        }.iteritems())])
+    CONFIG = BackendConfig(Value('website',  label='Website to use', choices=website_choices),
+                           ValueBackendPassword('login',    label='Account ID', masked=False),
+                           ValueBackendPassword('password', label='Password'))
+    BROWSER = Cragr
+
+    def create_default_browser(self):
+        return self.create_browser(self.config['website'].get(),
+                                   self.config['login'].get(),
+                                   self.config['password'].get())
+
+    def iter_accounts(self):
+        for account in self.browser.get_accounts_list():
+            yield account
+
+    def get_account(self, _id):
+        if not _id.isdigit():
+            raise AccountNotFound()
+        account = self.browser.get_account(_id)
+        if account:
+            return account
+        else:
+            raise AccountNotFound()
+
+    def iter_history(self, account):
+        for history in self.browser.get_history(account):
+            yield history
+
+
+    def transfer(self, account, to, amount, reason=None):
+        return self.browser.do_transfer(account, to, amount, reason)

modules/cragr/browser.py

@@ -0,0 +1,259 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2009-2011  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 <http://www.gnu.org/licenses/>.
+
+
+from weboob.tools.browser import BaseBrowser, BrowserIncorrectPassword
+from weboob.capabilities.bank import Transfer, TransferError
+from cragr import pages
+import mechanize
+from datetime import datetime
+import re
+
+class Cragr(BaseBrowser):
+    PROTOCOL = 'https'
+    ENCODING = 'utf-8'
+    USER_AGENT = BaseBrowser.USER_AGENTS['wget']
+    # a session id that is sometimes added, and should be ignored when matching pages
+    SESSION_REGEXP = '(?:|%s[A-Z0-9]+)' % re.escape(r';jsessionid=')
+
+    is_logging = False
+
+    def __init__(self, website, *args, **kwargs):
+        self.DOMAIN = website
+        self.PAGES = {'https://[^/]+/':                               pages.LoginPage,
+                      'https://[^/]+/.*\.c.*':                        pages.AccountsList,
+                      'https://[^/]+/login/process%s' % self.SESSION_REGEXP:   pages.AccountsList,
+                      'https://[^/]+/accounting/listAccounts':        pages.AccountsList,
+                      'https://[^/]+/accounting/listOperations':      pages.AccountsList,
+                      'https://[^/]+/accounting/showAccountDetail.+': pages.AccountsList,
+                      'https://[^/]+/accounting/showMoreAccountOperations.*': pages.AccountsList,
+                     }
+        BaseBrowser.__init__(self, *args, **kwargs)
+
+    def viewing_html(self):
+        """
+        As the fucking HTTP server returns a document in unknown mimetype
+        'application/vnd.wap.xhtml+xml' it is not recognized by mechanize.
+
+        So this is a fucking hack.
+        """
+        return True
+
+    def is_logged(self):
+        logged = self.page and self.page.is_logged() or self.is_logging
+        self.logger.debug('logged: %s' % (logged and 'yes' or 'no'))
+        return logged
+
+    def login(self):
+        """
+        Attempt to log in.
+        Note: this method does nothing if we are already logged in.
+        """
+        assert isinstance(self.username, basestring)
+        assert isinstance(self.password, basestring)
+
+        # Do we really need to login?
+        if self.is_logged():
+            self.logger.debug('already logged in')
+            return
+
+        self.is_logging = True
+        # Are we on the good page?
+        if not self.is_on_page(pages.LoginPage):
+            self.logger.debug('going to login page')
+            BaseBrowser.home(self)
+        self.logger.debug('attempting to log in')
+        self.page.login(self.username, self.password)
+        self.is_logging = False
+
+        if not self.is_logged():
+            raise BrowserIncorrectPassword()
+
+    def get_accounts_list(self):
+        self.logger.debug('accounts list required')
+        self.home()
+        return self.page.get_list()
+
+    def home(self):
+        """
+        Ensure we are both logged and on the accounts list.
+        """
+        self.logger.debug('accounts list page required')
+        if self.is_on_page(pages.AccountsList) and self.page.is_accounts_list():
+            self.logger.debug('already on accounts list')
+            return
+
+        # simply go to http(s)://the.doma.in/
+        BaseBrowser.home(self)
+
+        if self.is_on_page(pages.LoginPage):
+            if not self.is_logged():
+                # So, we are not logged on the login page -- what about logging ourselves?
+                self.login()
+                # we assume we are logged in
+            # for some regions, we may stay on the login page once we're
+            # logged in, without being redirected...
+            if self.is_on_page(pages.LoginPage):
+                # ... so we have to move by ourselves
+                self.move_to_accounts_list()
+
+    def move_to_accounts_list(self):
+        """
+        For regions where you can stay on http(s)://the.doma.in/ while you are
+        logged in, move to the accounts list
+        """
+        self.location('%s://%s/accounting/listAccounts' % (self.PROTOCOL, self.DOMAIN))
+
+    def get_account(self, id):
+        assert isinstance(id, basestring)
+
+        l = self.get_accounts_list()
+        for a in l:
+            if a.id == ('%s' % id):
+                return a
+
+        return None
+
+    def get_history(self, account):
+        history_url = account.link_id
+        operations_count = 0
+
+        # 1st, go on the account page
+        self.logger.debug('going on: %s' % history_url)
+        self.location('https://%s%s' % (self.DOMAIN, history_url))
+
+        # Some regions have a "Show more" (well, actually "Voir les 25
+        # suivants") link we have to use to get all the operations.
+        # However, it does not show only the 25 next results, it *adds* them
+        # to the current view. Therefore, we have to parse each new page using
+        # an offset, in order to ignore all already-fetched operations.
+        # This especially occurs on CA Centre.
+        use_expand_url = bool(self.page.expand_history_page_url())
+        while (history_url):
+            # we skip "operations_count" operations on each page if we are in the case described above
+            operations_offset = operations_count if use_expand_url else 0
+            for page_operation in self.page.get_history(operations_count, operations_offset):
+                operations_count += 1
+                yield page_operation
+            history_url = self.page.expand_history_page_url() if use_expand_url else self.page.next_page_url()
+            self.logger.debug('going on: %s' % history_url)
+            self.location('https://%s%s' % (self.DOMAIN, history_url))
+
+    def dict_find_value(self, dictionary, value):
+        """
+            Returns the first key pointing on the given value, or None if none
+            is found.
+        """
+        for k, v in dictionary.iteritems():
+            if v == value:
+                return k
+        return None
+
+    def do_transfer(self, account, to, amount, reason=None):
+        """
+            Transfer the given amount of money from an account to another,
+            tagging the transfer with the given reason.
+        """
+        # access the transfer page
+        transfer_page_unreachable_message = u'Could not reach the transfer page.'
+        self.home()
+        if not self.page.is_accounts_list():
+            raise TransferError(transfer_page_unreachable_message)
+
+        operations_url = self.page.operations_page_url()
+
+        self.location('https://%s%s' % (self.DOMAIN, operations_url))
+        transfer_url = self.page.transfer_page_url()
+
+        abs_transfer_url = 'https://%s%s' % (self.DOMAIN, transfer_url)
+        self.location(abs_transfer_url)
+        if not self.page.is_transfer_page():
+            raise TransferError(transfer_page_unreachable_message)
+
+        source_accounts = self.page.get_transfer_source_accounts()
+        target_accounts = self.page.get_transfer_target_accounts()
+
+        # check that the given source account can be used
+        if not account in source_accounts.values():
+            raise TransferError('You cannot use account %s as a source account.' % account)
+
+        # check that the given source account can be used
+        if not to in target_accounts.values():
+            raise TransferError('You cannot use account %s as a target account.' % to)
+
+        # separate euros from cents
+        amount_euros = int(amount)
+        amount_cents = int((amount * 100) - (amount_euros * 100))
+
+        # let's circumvent https://github.com/jjlee/mechanize/issues/closed#issue/17
+        # using http://wwwsearch.sourceforge.net/mechanize/faq.html#usage
+        adjusted_response = self.response().get_data().replace('<br/>', '<br />')
+        response = mechanize.make_response(adjusted_response, [('Content-Type', 'text/html')], abs_transfer_url, 200, 'OK')
+        self.set_response(response)
+
+        # fill the form
+        self.select_form(nr=0)
+        self['numCompteEmetteur']     = ['%s' % self.dict_find_value(source_accounts, account)]
+        self['numCompteBeneficiaire'] = ['%s' % self.dict_find_value(target_accounts, to)]
+        self['montantPartieEntiere']  = '%s'   % amount_euros
+        self['montantPartieDecimale'] = '%02d' % amount_cents
+        if reason != None:
+            self['libelle'] = reason
+        self.submit()
+
+        # look for known errors
+        content = unicode(self.response().get_data(), 'utf-8')
+        insufficient_amount_message     = u'Montant insuffisant.'
+        maximum_allowed_balance_message = u'Solde maximum autorisé dépassé.'
+
+        if content.find(insufficient_amount_message) != -1:
+            raise TransferError('The amount you tried to transfer is too low.')
+
+        if content.find(maximum_allowed_balance_message) != -1:
+            raise TransferError('The maximum allowed balance for the target account has been / would be reached.')
+
+        # look for the known "all right" message
+        ready_for_transfer_message = u'Vous allez effectuer un virement'
+        if not content.find(ready_for_transfer_message):
+            raise TransferError('The expected message "%s" was not found.' % ready_for_transfer_message)
+
+        # submit the last form
+        self.select_form(nr=0)
+        submit_date = datetime.now()
+        self.submit()
+
+        # look for the known "everything went well" message
+        content = unicode(self.response().get_data(), 'utf-8')
+        transfer_ok_message = u'Vous venez d\'effectuer un virement du compte'
+        if not content.find(transfer_ok_message):
+            raise TransferError('The expected message "%s" was not found.' % transfer_ok_message)
+
+        # We now have to return a Transfer object
+        # the final page does not provide any transfer id, so we'll use the submit date
+        transfer = Transfer(submit_date.strftime('%Y%m%d%H%M%S'))
+        transfer.amount = amount
+        transfer.origin = account
+        transfer.recipient = to
+        transfer.date = submit_date
+        return transfer
+
+    #def get_coming_operations(self, account):
+    #    if not self.is_on_page(pages.AccountComing) or self.page.account.id != account.id:
+    #        self.location('/NS_AVEEC?ch4=%s' % account.link_id)
+    #    return self.page.get_operations()

modules/cragr/pages/__init__.py

@@ -0,0 +1,24 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+from .accounts_list import AccountsList
+from .login import LoginPage
+
+__all__ = ['AccountsList', 'LoginPage']

modules/cragr/pages/accounts_list.py

@@ -0,0 +1,283 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+import re
+from weboob.capabilities.bank import Account
+from .base import CragrBasePage
+from weboob.capabilities.bank import Operation
+
+def clean_amount(amount):
+    """
+        Removes weird characters and converts to a float
+        >>> clean_amount(u'1 000,00 $')
+        1000.0
+    """
+    data = amount.replace(',', '.').replace(' ', '').replace(u'\xa0', '')
+    matches = re.findall('^(-?[0-9]+\.[0-9]{2}).*$', data)
+    return float(matches[0]) if (matches) else 0.0
+
+class AccountsList(CragrBasePage):
+
+    def get_list(self):
+        """
+            Returns the list of available bank accounts
+        """
+        l = []
+
+        for div in self.document.getiterator('div'):
+            if div.attrib.get('class', '') == 'dv' and div.getchildren()[0].tag in ('a', 'br'):
+                account = Account()
+                if div.getchildren()[0].tag == 'a':
+                    # This is at least present on CA Nord-Est
+                    account.label = ' '.join(div.find('a').text.split()[:-1])
+                    account.link_id = div.find('a').get('href', '')
+                    account.id = div.find('a').text.split()[-1]
+                    s = div.find('div').find('b').find('span').text
+                else:
+                    # This is at least present on CA Toulouse
+                    account.label = div.find('a').text.strip()
+                    account.link_id = div.find('a').get('href', '')
+                    account.id = div.findall('br')[1].tail.strip()
+                    s = div.find('div').find('b').text
+                account.balance = clean_amount(s)
+                l.append(account)
+        return l
+
+    def is_accounts_list(self):
+        """
+            Returns True if the current page appears to be the page dedicated to
+            list the accounts.
+        """
+        # we check for the presence of a "mes comptes titres" link_id
+        link = self.document.xpath('/html/body//a[contains(text(), "comptes titres")]')
+        return bool(link)
+
+    def is_account_page(self):
+        """
+            Returns True if the current page appears to be a page dedicated to list
+            the history of a specific account.
+        """
+        # tested on CA Lorraine, Paris, Toulouse
+        title_spans = self.document.xpath('/html/body//div[@class="dv"]/span')
+        for title_span in title_spans:
+            title_text = title_span.text_content().strip().replace("\n", '')
+            if (re.match('.*Compte.*n.*[0-9]+.*au.*', title_text)):
+                return True
+        return False
+
+    def is_transfer_page(self):
+        """
+            Returns True if the current page appears to be the page dedicated to
+            order transfers between accounts.
+        """
+        source_account_select_field = self.document.xpath('/html/body//form//select[@name="numCompteEmetteur"]')
+        target_account_select_field  = self.document.xpath('/html/body//form//select[@name="numCompteBeneficiaire"]')
+        return bool(source_account_select_field) and bool(target_account_select_field)
+
+    def get_transfer_accounts(self, select_name):
+        """
+            Returns the accounts proposed for a transfer in a select field.
+            This method assumes the current page is the one dedicated to transfers.
+            select_name is the name of the select field to analyze
+        """
+        if not self.is_transfer_page():
+          return False
+        source_accounts = {}
+        source_account_options = self.document.xpath('/html/body//form//select[@name="%s"]/option' % select_name)
+        for option in source_account_options:
+            source_account_value = option.get('value', -1)
+            if (source_account_value != -1):
+                matches = re.findall('^[A-Z0-9]+.*([0-9]{11}).*$', self.extract_text(option))
+                if matches:
+                    source_accounts[source_account_value] = matches[0]
+        return source_accounts
+
+    def get_transfer_source_accounts(self):
+        return self.get_transfer_accounts('numCompteEmetteur')
+
+    def get_transfer_target_accounts(self):
+        return self.get_transfer_accounts('numCompteBeneficiaire')
+
+    def expand_history_page_url(self):
+        """
+            When on a page dedicated to list the history of a specific account (see
+            is_account_page), returns the link to expand the history with 25 more results,
+            or False if the link is not present.
+        """
+        # tested on CA centre france
+        a = self.document.xpath('/html/body//div[@class="navlink"]//a[contains(text(), "Voir les 25 suivants")]')
+        if not a:
+            return False
+        else:
+            return a[0].get('href', '')
+
+    def next_page_url(self):
+        """
+            When on a page dedicated to list the history of a specific account (see
+            is_account_page), returns the link to the next page, or False if the
+            link is not present.
+        """
+        # tested on CA Lorraine, Paris, Toulouse
+        a = self.document.xpath('/html/body//div[@class="navlink"]//a[contains(text(), "Suite")]')
+        if not a:
+            return False
+        else:
+            return a[0].get('href', '')
+
+    def operations_page_url(self):
+        """
+            Returns the link to the "Opérations" page. This function assumes the
+            current page is the accounts list (see is_accounts_list)
+        """
+        link = self.document.xpath(u'/html/body//a[contains(text(), "Opérations")]')
+        return link[0].get('href')
+
+    def transfer_page_url(self):
+        """
+            Returns the link to the "Virements" page. This function assumes the
+            current page is the operations list (see operations_page_url)
+        """
+        link = self.document.xpath('/html/body//a[@accesskey=1]/@href')
+        return link[0]
+
+    def is_right_aligned_div(self, div_elmt):
+        """
+            Returns True if the given div element is right-aligned
+        """
+        return(re.match('.*text-align: ?right.*', div_elmt.get('style', '')))
+
+    def extract_text(self, xml_elmt):
+        """
+            Given an XML element, returns its inner text in a reasonably readable way
+        """
+        data = u''
+        for text in xml_elmt.itertext():
+            data = data + u'%s ' % text
+        data = re.sub(' +', ' ', data.replace("\n", ' ').strip())
+        return data
+
+    def get_history(self, start_index = 0, start_offset = 0):
+        """
+            Returns the history of a specific account. Note that this function
+            expects the current page to be the one dedicated to this history.
+            start_index is the id used for the first created operation.
+            start_offset allows ignoring the `n' first Operations on the page.
+        """
+        # tested on CA Lorraine, Paris, Toulouse
+        # avoir parsing the page as an account-dedicated page if it is not the case
+        if not self.is_account_page():
+            return
+
+        index = start_index
+        operation = False
+        skipped = 0
+
+        body_elmt_list = self.document.xpath('/html/body/*')
+
+        # type of separator used in the page
+        separators = 'hr'
+        # How many <hr> elements do we have under the <body>?
+        sep_expected = len(self.document.xpath('/html/body/hr'))
+        if (not sep_expected):
+            # no <hr>? Then how many class-less <div> used as separators instead?
+            sep_expected = len(self.document.xpath('/html/body/div[not(@class) and not(@style)]'))
+            separators = 'div'
+
+        # the interesting divs are after the <hr> elements
+        interesting_divs = []
+        right_div_count = 0
+        left_div_count = 0
+        sep_found = 0
+        for body_elmt in body_elmt_list:
+            if (separators == 'hr' and body_elmt.tag == 'hr'):
+                sep_found += 1
+            elif (separators == 'div' and body_elmt.tag == 'div' and body_elmt.get('class', 'nope') == 'nope'):
+                sep_found += 1
+            elif (sep_found >= sep_expected and body_elmt.tag == 'div'):
+                # we just want <div> with dv class and a style attribute
+                if (body_elmt.get('class', '') != 'dv'):
+                    continue
+                if (body_elmt.get('style', 'nope') == 'nope'):
+                    continue
+                interesting_divs.append(body_elmt)
+                if (self.is_right_aligned_div(body_elmt)):
+                    right_div_count += 1
+                else:
+                    left_div_count += 1
+
+        # new layout that is somewhat easier to parse (found at Toulouse)
+        table_layout = len(self.document.xpath("id('operationsHeader')")) > 0
+        # So, how are data laid out?
+        alternate_layout = (left_div_count == 2 * right_div_count)
+        # we'll have: one left-aligned div for the date, one right-aligned
+        # div for the amount, and one left-aligned div for the label. Each time.
+
+        if table_layout:
+            lines = self.document.xpath('id("operationsContent")//table[@class="tb"]/tr')
+            for line in lines:
+                if skipped < start_offset:
+                    skipped += 1
+                    continue
+                operation = Operation(index)
+                index += 1
+                operation.date = self.extract_text(line[0])
+                operation.label = self.extract_text(line[1])
+                operation.amount = clean_amount(self.extract_text(line[2]))
+                yield operation
+        elif (not alternate_layout):
+            for body_elmt in interesting_divs:
+                if skipped < start_offset:
+                    if self.is_right_aligned_div(body_elmt):
+                        skipped += 1 
+                    continue
+                if (self.is_right_aligned_div(body_elmt)):
+                    # this is the second line of an operation entry, displaying the amount
+                    operation.amount = clean_amount(self.extract_text(body_elmt))
+                    yield operation
+                else:
+                    # this is the first line of an operation entry, displaying the date and label
+                    data = self.extract_text(body_elmt)
+                    matches = re.findall('^([012][0-9]|3[01])/(0[1-9]|1[012]).(.+)$', data)
+                    operation = Operation(index)
+                    index += 1
+                    if (matches):
+                        operation.date  = u'%s/%s' % (matches[0][0], matches[0][1])
+                        operation.label = u'%s'    % matches[0][2]
+                    else:
+                        operation.date  = u'01/01'
+                        operation.label = u'Unknown'
+        else:
+            for i in range(0, len(interesting_divs)/3):
+                if skipped < start_offset:
+                    skipped += 1
+                    continue
+                operation = Operation(index)
+                index += 1
+                # amount
+                operation.amount = clean_amount(self.extract_text(interesting_divs[(i*3)+1]))
+                # date
+                data = self.extract_text(interesting_divs[i*3])
+                matches = re.findall('^([012][0-9]|3[01])/(0[1-9]|1[012])', data)
+                operation.date = u'%s/%s' % (matches[0][0], matches[0][1]) if (matches) else u'01/01'
+                #label
+                data = self.extract_text(interesting_divs[(i*3)+2])
+                data = re.sub(' +', ' ', data)
+                operation.label = u'%s' % data
+                yield operation

modules/cragr/pages/base.py

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+from weboob.tools.browser import BasePage
+from weboob.tools.browser import BrowserUnavailable
+
+class CragrBasePage(BasePage):
+    def on_loaded(self):
+        # Check for an error
+        for div in self.document.getiterator('div'):
+            if div.attrib.get('class', '') == 'dv' and div.getchildren()[0].tag in ('img') and div.getchildren()[0].attrib.get('alt', '') == 'Attention':
+                # Try to find a detailed error message
+                if div.getchildren()[1].tag == 'span':
+                    raise BrowserUnavailable(div.find('span').find('b').text)
+                elif div.getchildren()[1].tag == 'b':
+                    # I haven't encountered this variation in the wild,
+                    # but I wouldn't be surprised if it existed
+                    # given the similar differences between regions.
+                    raise BrowserUnavailable(div.find('b').find('span').text)
+                raise BrowserUnavailable()
+
+    def is_logged(self):
+        return not self.document.xpath('/html/body//form//input[@name = "code"]') and \
+           not self.document.xpath('/html/body//form//input[@name = "userPassword"]')

modules/cragr/pages/login.py

@@ -0,0 +1,51 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+from weboob.tools.mech import ClientForm
+ControlNotFoundError = ClientForm.ControlNotFoundError
+
+from .base import CragrBasePage
+
+
+__all__ = ['LoginPage']
+
+
+class LoginPage(CragrBasePage):
+    def login(self, login, password):
+        self.browser.select_form(nr=0)
+        try:
+            self.browser['numero'] = login
+            self.browser['code'] = password
+        except ControlNotFoundError:
+            try:
+                self.browser['userLogin'] = login
+                self.browser['userPassword'] = password
+            except ControlNotFoundError:
+                self.browser.controls.append(ClientForm.TextControl('text', 'numero', {'value': ''}))
+                self.browser.controls.append(ClientForm.TextControl('text', 'code', {'value': ''}))
+                self.browser.controls.append(ClientForm.TextControl('text', 'userLogin', {'value': ''}))
+                self.browser.controls.append(ClientForm.TextControl('text', 'userPassword', {'value': ''}))
+                self.browser.set_all_readonly(False)
+                self.browser['numero'] = login
+                self.browser['code'] = password
+                self.browser['userLogin'] = login
+                self.browser['userPassword'] = password
+
+        self.browser.submit()

modules/cragr/test.py

@@ -0,0 +1,30 @@
+# -*- coding: utf-8 -*-
+
+# Copyright(C) 2010-2011 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 <http://www.gnu.org/licenses/>.
+
+
+from weboob.tools.test import BackendTest
+
+class CrAgrTest(BackendTest):
+    BACKEND = 'cragr'
+
+    def test_cragr(self):
+        l = list(self.backend.iter_accounts())
+        if len(l) > 0:
+            a = l[0]
+            list(self.backend.iter_history(a))