pages.py

# -*- coding: utf-8 -*-
# Copyright(C) 2010-2011  Romain Bignon, Pierre Mazière
#
# This file is part of a weboob module.
#
# This weboob module is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This weboob module 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 Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public License
# along with this weboob module. If not, see <http://www.gnu.org/licenses/>.

# flake8: compatible

from __future__ import unicode_literals, division

import re
import requests
import base64
import math
import random
from decimal import Decimal
from io import BytesIO
from datetime import datetime, timedelta

from dateutil.relativedelta import relativedelta

from weboob.capabilities.base import empty, find_object, NotAvailable
from weboob.capabilities.bank import (
    Account, Recipient, TransferError, TransferBankError, Transfer,
    AccountOwnership, AddRecipientBankError,
)
from weboob.capabilities.wealth import Investment, MarketOrder, MarketOrderDirection, MarketOrderType
from weboob.capabilities.bill import Document, Subscription, DocumentTypes
from weboob.capabilities.profile import Person, ProfileMissing
from weboob.capabilities.contact import Advisor
from weboob.browser.elements import method, ListElement, TableElement, ItemElement, DictElement
from weboob.browser.exceptions import ServerError
from weboob.browser.pages import LoggedPage, HTMLPage, JsonPage, FormNotFound, pagination, PartialHTMLPage
from weboob.browser.filters.html import Attr, Link, TableCell, AttributeNotFound, AbsoluteLink
from weboob.browser.filters.standard import (
    CleanText, Field, Regexp, Format, Date, CleanDecimal, Map, AsyncLoad, Async, Env, Slugify,
    BrowserURL, Eval, Currency, Base, Coalesce, MapIn, Lower,
)
from weboob.browser.filters.json import Dict
from weboob.exceptions import BrowserUnavailable, BrowserIncorrectPassword, ActionNeeded, ParseError
from weboob.tools.capabilities.bank.transactions import FrenchTransaction, parse_with_patterns
from weboob.tools.captcha.virtkeyboard import MappedVirtKeyboard, VirtKeyboardError
from weboob.tools.compat import unicode, urlparse, parse_qs, urljoin
from weboob.tools.html import html2text
from weboob.tools.date import parse_french_date
from weboob.tools.capabilities.bank.investments import is_isin_valid, IsinCode


def MyDecimal(*args, **kwargs):
    kwargs.update(replace_dots=True, default=Decimal(0))
    return CleanDecimal(*args, **kwargs)


def myXOR(value, seed):
    s = ''
    for i in range(len(value)):
        s += chr(seed ^ ord(value[i]))
    return s


class LCLBasePage(HTMLPage):
    def get_from_js(self, pattern, end, is_list=False):
        """
        find a pattern in any javascript text
        """
        value = None
        for script in self.doc.xpath('//script'):
            txt = script.text
            if txt is None:
                continue

            start = txt.find(pattern)
            if start < 0:
                continue

            while True:
                if value is None:
                    value = ''
                else:
                    value += ','
                value += txt[start + len(pattern):start + txt[start + len(pattern):].find(end) + len(pattern)]

                if not is_list:
                    break

                txt = txt[start + len(pattern) + txt[start + len(pattern):].find(end):]

                start = txt.find(pattern)
                if start < 0:
                    break
            return value


class LCLVirtKeyboard(MappedVirtKeyboard):
    symbols = {
        '0': '9da2724133f2221482013151735f033c',
        '1': '873ab0087447610841ae1332221be37b',
        '2': '93ce6c330393ff5980949d7b6c800f77',
        '3': 'b2d70c69693784e1bf1f0973d81223c0',
        '4': '498c8f5d885611938f94f1c746c32978',
        '5': '359bcd60a9b8565917a7bf34522052c3',
        '6': 'aba912172f21f78cd6da437cfc4cdbd0',
        '7': 'f710190d6b947869879ec02d8e851dfa',
        '8': 'b42cc25e1539a15f767aa7a641f3bfec',
        '9': 'cc60e5894a9d8e12ee0c2c104c1d5490',
    }

    url = "/outil/UAUT/Clavier/creationClavier?random="

    color = (255, 255, 255, 255)

    def __init__(self, basepage):
        img = basepage.doc.find("//img[@id='idImageClavier']")
        random.seed()
        self.url += "%s" % str(int(math.floor(int(random.random() * 1000000000000000000000))))
        super(LCLVirtKeyboard, self).__init__(
            BytesIO(basepage.browser.open(self.url).content),
            basepage.doc, img, self.color, "id")
        self.check_symbols(self.symbols, basepage.browser.responses_dirname)

    def get_symbol_code(self, md5sum):
        code = MappedVirtKeyboard.get_symbol_code(self, md5sum)
        return code[-2:]

    def get_string_code(self, string):
        code = ''
        for c in string:
            code += self.get_symbol_code(self.symbols[c])
        return code


class LoginPage(HTMLPage):
    def on_load(self):
        try:
            form = self.get_form(xpath='//form[@id="setInfosCGS" or @name="form"]')
        except FormNotFound:
            return

        form.submit()

    def login(self, login, passwd):
        try:
            vk = LCLVirtKeyboard(self)
        except VirtKeyboardError as err:
            self.logger.exception(err)
            return False

        password = vk.get_string_code(passwd)

        seed = -1
        s = "var aleatoire = "
        for script in self.doc.findall("//script"):
            if script.text is None or len(script.text) == 0:
                continue
            offset = script.text.find(s)
            if offset != -1:
                seed = int(script.text[offset + len(s) + 1:offset + len(s) + 2])
                break
        if seed == -1:
            raise ParseError("Variable 'aleatoire' not found")

        form = self.get_form('//form[@id="formAuthenticate"]')
        form['identifiant'] = login
        form['postClavierXor'] = base64.b64encode(
            myXOR(password, seed).encode("utf-8")
        )
        try:
            form['identifiantRouting'] = self.browser.IDENTIFIANT_ROUTING
        except AttributeError:
            pass

        form.submit(allow_redirects=False)

    def check_error(self):
        errors = self.doc.xpath(u'//*[@class="erreur" or @class="messError"]')
        if not errors or self.doc.xpath('//a[@href="/outil/UWHO/Accueil/"]'):
            return

        for error in errors:
            error_text = CleanText(error.xpath('./div/text()'))(self.doc)
            if 'Suite à la saisie de plusieurs identifiant / code erronés' in error_text:
                raise ActionNeeded(error_text)
            if 'Votre identifiant ou votre code personnel est incorrect' in error_text:
                raise BrowserIncorrectPassword(error_text)
        raise BrowserIncorrectPassword()


class RedirectPage(LoginPage, PartialHTMLPage):
    def is_here(self):
        # During login a form submit with an allow_redirects=False is done
        # The submit request can be done on contract urls following by a redirection
        # So if we get a 302 this new class avoids misleading on_load
        return self.response.status_code == 302


class ContractsPage(LoginPage, PartialHTMLPage):
    def on_load(self):
        # after login we are redirect in ContractsPage even if there is an error at login
        # I let the error check code here to simplify
        # a better solution will be to put error check on browser.py and error parsing in pages.py
        self.check_error()

        # To avoid skipping contract page the first time we see it,
        # and to be able to get the contracts list from it
        if self.browser.parsed_contracts:
            self.select_contract()

    def get_contracts_list(self):
        return self.doc.xpath('//input[@name="contratId"]/@value')

    def select_contract(self, id_contract=None):
        link = self.doc.xpath('//a[contains(text(), "Votre situation globale")]')
        if not id_contract and len(link):
            self.browser.location(link[0].attrib['href'])
        else:
            form = self.get_form(nr=0)
            if 'contratId' in form:
                if id_contract:
                    form['contratId'] = id_contract
                self.browser.current_contract = form['contratId']
            form.submit()


class ContractsChoicePage(ContractsPage):
    def on_load(self):
        self.check_error()
        if not self.logged and not self.browser.current_contract:
            self.select_contract()


class OwnedItemElement(ItemElement):
    def get_ownership(self, owner):
        if re.search(r'(m|mr|me|mme|mlle|mle|ml)\.? (.*)\bou (m|mr|me|mme|mlle|mle|ml)\b(.*)', owner, re.IGNORECASE):
            return AccountOwnership.CO_OWNER
        elif all(n in owner for n in self.env['name'].split()):
            return AccountOwnership.OWNER
        return AccountOwnership.ATTORNEY


class AccountsPage(LoggedPage, HTMLPage):
    def on_load(self):
        warn = self.doc.xpath('//div[@id="attTxt"]')
        if len(warn) > 0:
            raise BrowserIncorrectPassword(warn[0].text)

    def get_name(self):
        return CleanText('//li[@id="nomClient"]/p')(self.doc)

    @method
    class get_accounts_list(ListElement):

        # XXX Ugly Hack to replace account by second occurrence.
        # LCL pro website sometimes display the same account twice and only second link is valid to fetch transactions.
        def store(self, obj):
            assert obj.id
            if obj.id in self.objects:
                self.logger.warning('There are two objects with the same ID! %s' % obj.id)
            self.objects[obj.id] = obj
            return obj

        item_xpath = '//tr[contains(@onclick, "redirect")]'
        flush_at_end = True

        class account(OwnedItemElement):
            klass = Account

            def condition(self):
                return '/outil/UWLM/ListeMouvement' in self.el.attrib['onclick']

            def load_details(self):
                link_id = Field('_link_id')(self)
                if link_id:
                    account_url = urljoin(self.page.browser.BASEURL, link_id)
                    return self.page.browser.async_open(url=account_url)
                return NotAvailable

            NATURE2TYPE = {
                '001': Account.TYPE_SAVINGS,
                '004': Account.TYPE_CHECKING,
                '005': Account.TYPE_CHECKING,
                '006': Account.TYPE_CHECKING,
                '007': Account.TYPE_SAVINGS,
                '012': Account.TYPE_SAVINGS,
                '023': Account.TYPE_CHECKING,
                '036': Account.TYPE_SAVINGS,
                '046': Account.TYPE_SAVINGS,
                '047': Account.TYPE_SAVINGS,
                '049': Account.TYPE_SAVINGS,
                '058': Account.TYPE_CHECKING,
                '068': Account.TYPE_PEA,
                '069': Account.TYPE_SAVINGS,
            }

            obj__link_id = Format('%s&mode=190', Regexp(CleanText('./@onclick'), "'(.*)'"))
            obj__agence = Regexp(Field('_link_id'), r'.*agence=(\w+)')
            obj__compte = Regexp(Field('_link_id'), r'compte=(\w+)')
            obj_id = Format('%s%s', Field('_agence'), Field('_compte'))
            obj__transfer_id = Format('%s0000%s', Field('_agence'), Field('_compte'))
            obj_label = CleanText('.//div[@class="libelleCompte"]')
            obj_currency = FrenchTransaction.Currency('.//td[has-class("right")]')
            obj_type = Map(Regexp(Field('_link_id'), r'.*nature=(\w+)'), NATURE2TYPE, default=Account.TYPE_UNKNOWN)
            obj__market_link = None
            obj_number = Field('id')

            def obj_balance(self):
                balance = None
                if 'professionnels' in self.page.browser.url and Field('type')(self) == Account.TYPE_CHECKING:
                    # for pro accounts with comings, balance without comings must be fetched on details page
                    async_page = Async('details').loaded_page(self)
                    balance = async_page.get_balance_without_comings_main()
                    # maybe the next get_balance can be removed
                    # sometimes it returns the sum of transactions for last x days (47 ?)
                    if empty(balance):
                        self.logger.info('GET_BALANCE_MAIN EMPTY')
                        balance = async_page.get_balance_without_comings()
                if not empty(balance):
                    return balance
                return CleanDecimal.French('.//td[has-class("right")]')(self)

            def obj_ownership(self):
                async_page = Async('details').loaded_page(self)
                owner = CleanText('//h5[contains(text(), "Titulaire")]')(async_page.doc)
                return self.get_ownership(owner)

    def get_deferred_cards(self):
        trs = self.doc.xpath('//tr[contains(@onclick, "EncoursCB")]')
        links = []

        for tr in trs:
            parent_id = Regexp(CleanText('./@onclick'), r'.*AGENCE=(\w+).*COMPTE=(\w+).*CLE=(\w+)', r'\1\2\3')(tr)
            link = Regexp(CleanText('./@onclick'), "'(.*)'")(tr)
            links.append((parent_id, link))

        return links

    @method
    class get_advisor(ItemElement):
        klass = Advisor

        obj_name = CleanText('//div[@id="contacterMaBqMenu"]//p[@id="itemNomContactMaBq"]/span')
        obj_email = obj_mobile = obj_fax = NotAvailable
        obj_phone = Regexp(
            CleanText('//div[@id="contacterMaBqMenu"]//p[contains(text(), "Tel")]', replace=[(' ', '')]),
            r'([\s\d]+)',
            default=NotAvailable
        )
        obj_agency = CleanText('//div[@id="sousContentAgence"]//p[@class="itemSousTitreMenuMaBq"][1]')

        def obj_address(self):
            address = CleanText(
                '//div[@id="sousContentAgence"]//p[@class="itemSousTitreMenuMaBq"][2]',
                default=None
            )(self)
            city = CleanText(
                '//div[@id="sousContentAgence"]//p[@class="itemSousTitreMenuMaBq"][3]',
                default=None
            )(self)
            if not (address and city):
                return NotAvailable
            return "%s %s" % (address, city)


class LoansPage(LoggedPage, HTMLPage):
    @method
    class get_list(TableElement):
        item_xpath = '//table[.//th[contains(text(), "Emprunteur")]]/tbody/tr[td[3]]'
        head_xpath = '//table[.//th[contains(text(), "Emprunteur")]]/thead/tr/th'
        flush_at_end = True

        col_id = re.compile('Emprunteur')
        col_balance = [u'Capital restant dû', re.compile('Sommes totales restant dues'), re.compile('Montant disponible')]

        class account(ItemElement):
            klass = Account

            obj_balance = CleanDecimal(TableCell('balance'), replace_dots=True, sign='-')
            obj_currency = FrenchTransaction.Currency(TableCell('balance'))
            obj_type = Account.TYPE_LOAN
            obj_id = Env('id')
            obj__transfer_id = None
            obj_number = Regexp(CleanText(TableCell('id'), replace=[(' ', ''), ('-', '')]), r'(\d{11}[A-Z])')

            def obj_label(self):
                has_type = CleanText('./ancestor::table[.//th[contains(text(), "Type")]]', default=None)(self)
                if has_type:
                    return CleanText('./td[2]')(self)
                else:
                    return CleanText('./ancestor::table/preceding-sibling::div[1]')(self).split(' - ')[0]

            def obj_ownership(self):
                pattern = re.compile(
                    r'(m|mr|me|mme|mlle|mle|ml)\.? (.*)\b(ou)? (m|mr|me|mme|mlle|mle|ml)\b(.*)',
                    re.IGNORECASE
                )
                if pattern.search(CleanText(TableCell('id'))(self)):
                    return AccountOwnership.CO_OWNER
                return AccountOwnership.OWNER

            def parse(self, el):
                label = Field('label')(self)
                trs = self.xpath(
                    '//td[contains(text(), $label)]/ancestor::tr[1] | ./ancestor::table[1]/tbody/tr',
                    label=label
                )
                i = [i for i in range(len(trs)) if el == trs[i]]
                if i:
                    i = i[0]
                else:
                    i = 0
                label = label.replace(' ', '')
                self.env['id'] = "%s%s%s" % (
                    Regexp(CleanText(TableCell('id')), r'(\w+)\s-\s(\w+)', r'\1\2')(self),
                    label.replace(' ', ''),
                    i,
                )


class LoansProPage(LoggedPage, HTMLPage):
    @method
    class get_list(TableElement):
        item_xpath = '//table[.//th[contains(text(), "Emprunteur")]]/tbody/tr[td[3]]'
        head_xpath = '//table[.//th[contains(text(), "Emprunteur")]]/thead/tr/th'
        flush_at_end = True

        col_id = re.compile('Emprunteur')
        col_balance = [u'Capital restant dû', re.compile('Sommes totales restant dues')]

        class account(ItemElement):
            klass = Account

            obj_balance = CleanDecimal(TableCell('balance'), replace_dots=True, sign='-')
            obj_currency = FrenchTransaction.Currency(TableCell('balance'))
            obj_type = Account.TYPE_LOAN
            obj_id = Env('id')
            obj__transfer_id = None
            obj_number = Regexp(CleanText(TableCell('id'), replace=[(' ', ''), ('-', '')]), r'(\d{11}[A-Z])')

            def obj_label(self):
                has_type = CleanText('./ancestor::table[.//th[contains(text(), "Nature libell")]]', default=None)(self)
                if has_type:
                    return CleanText('./td[3]')(self)
                else:
                    return CleanText('./ancestor::table/preceding-sibling::div[1]')(self).split(' - ')[0]

            def parse(self, el):
                label = Field('label')(self)
                trs = self.xpath(
                    '//td[contains(text(), $label)]/ancestor::tr[1] | ./ancestor::table[1]/tbody/tr',
                    label=label
                )
                i = [i for i in range(len(trs)) if el == trs[i]]
                if i:
                    i = i[0]
                else:
                    i = 0
                label = label.replace(' ', '')
                self.env['id'] = "%s%s%s" % (
                    Regexp(CleanText(TableCell('id')), r'(\w+)\s-\s(\w+)', r'\1\2')(self),
                    label.replace(' ', ''),
                    i,
                )


class Transaction(FrenchTransaction):
    PATTERNS = [
        (
            re.compile(r'^(?P<category>CB) (?P<text>RETRAIT) DU (?P<dd>\d+)/(?P<mm>\d+)'),
            FrenchTransaction.TYPE_WITHDRAWAL,
        ),
        (re.compile(r'^(?P<category>(PRLV|PE)( SEPA)?) (?P<text>.*)'), FrenchTransaction.TYPE_ORDER),
        (re.compile(r'^(?P<category>CHQ\.) (?P<text>.*)'), FrenchTransaction.TYPE_CHECK),
        (re.compile(r'^(?P<category>RELEVE CB) AU (\d+)/(\d+)/(\d+)'), FrenchTransaction.TYPE_CARD),
        (
            re.compile(r'^(?P<category>CB) (?P<text>.*) (?P<dd>\d+)/(?P<mm>\d+)/(?P<yy>\d+)'),
            FrenchTransaction.TYPE_CARD,
        ),
        (re.compile(r'^(?P<category>(PRELEVEMENT|TELEREGLEMENT|TIP)) (?P<text>.*)'), FrenchTransaction.TYPE_ORDER),
        (re.compile(r'^(?P<category>(ECHEANCE\s*)?PRET)(?P<text>.*)'), FrenchTransaction.TYPE_LOAN_PAYMENT),
        (
            re.compile(r'^(TP-\d+-)?(?P<category>(EVI|VIR(EM(EN)?)?T?)(.PERMANENT)? ((RECU|FAVEUR) TIERS|SEPA RECU)?)( /FRM)?(?P<text>.*)'),
            FrenchTransaction.TYPE_TRANSFER,
        ),
        (re.compile(r'^(?P<category>REMBOURST)(?P<text>.*)'), FrenchTransaction.TYPE_PAYBACK),
        (re.compile(r'^(?P<category>COM(MISSIONS?)?)(?P<text>.*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'^(?P<text>(?P<category>REMUNERATION).*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'^(?P<text>(?P<category>ABON.*?)\s*.*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'^(?P<text>(?P<category>RESULTAT .*?)\s*.*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'^(?P<text>(?P<category>TRAIT\..*?)\s*.*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'(?P<text>(?P<category>COTISATION).*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'(?P<text>(?P<category>INTERETS).*)'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'^(?P<category>REM CHQ) (?P<text>.*)'), FrenchTransaction.TYPE_DEPOSIT),
        (re.compile(r'^VIREMENT.*'), FrenchTransaction.TYPE_TRANSFER),
        (re.compile(r'.*(PRELEVEMENTS|PRELVT|TIP).*'), FrenchTransaction.TYPE_ORDER),
        (re.compile(r'.*CHEQUE.*'), FrenchTransaction.TYPE_CHECK),
        (re.compile(r'.*ESPECES.*'), FrenchTransaction.TYPE_DEPOSIT),
        (re.compile(r'.*(CARTE|CB).*'), FrenchTransaction.TYPE_CARD),
        (re.compile(r'.*(AGIOS|ANNULATIONS|IMPAYES|CREDIT).*'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'.*(FRAIS DE TENUE DE COMPTE).*'), FrenchTransaction.TYPE_BANK),
        (re.compile(r'.*\b(RETRAIT)\b.*'), FrenchTransaction.TYPE_WITHDRAWAL),
    ]


class Pagination(object):
    def next_page(self):
        links = self.page.doc.xpath('//div[@class="pagination"] /a')
        if len(links) == 0:
            return
        for link in links:
            if link.xpath('./span')[0].text == 'Page suivante':
                return link.attrib.get('href')
        return


class AccountHistoryPage(LoggedPage, HTMLPage):
    class _get_operations(Pagination, Transaction.TransactionsElement):
        item_xpath = '//table[has-class("tagTab") and (not(@style) or @style="")]/tr'
        head_xpath = '//table[has-class("tagTab") and (not(@style) or @style="")]/tr/th'

        col_raw = [u'Vos opérations', u'Libellé']

        class item(Transaction.TransactionElement):
            def fill_env(self, page, parent=None):
                # This *Element's parent has only the dateguesser in its env and we want to
                # use the same object, not copy it.
                self.env = parent.env

            def obj_rdate(self):
                rdate = self.obj.rdate
                date = Field('date')(self)

                if rdate > date:
                    date_guesser = Env('date_guesser')(self)
                    return date_guesser.guess_date(rdate.day, rdate.month)

                return rdate

            def obj__el(self):
                return self.el

            def condition(self):
                return (
                    self.parent.get_colnum('date') is not None
                    and len(self.el.findall('td')) >= 3
                    and self.el.get('class')
                    and 'tableTr' not in self.el.get('class')
                )

    def open_transaction_page(self, tr):
        # Those are summary for deferred card transactions,
        # they do not have details.
        if CleanText('./td[contains(text(), "RELEVE CB")]')(tr._el):
            return None

        row = Attr('.', 'id', default=None)(tr._el)
        assert row, 'HTML format of transactions details changed'

        if not re.match(r'\d+', row):
            return self.browser.open(
                Attr('.', 'href')(tr._el),
                method='POST',
            )

        return self.browser.open(
            '/outil/UWLM/ListeMouvementsParticulier/accesDetailsMouvement?element=%s' % row,
            method='POST',
        )

    def fix_transaction_stuff(self, obj, tr_page):
        if obj.category == 'RELEVE CB':
            obj.type = Transaction.TYPE_CARD_SUMMARY

        raw = obj.raw
        if tr_page:
            # TODO move this xpath to the relevant page class
            raw = CleanText(
                '//td[contains(text(), "Libellé")]/following-sibling::*[1]|//td[contains(text(), "Nom du donneur")]/following-sibling::*[1]',
            )(tr_page.doc)

        if raw:
            if (
                obj.raw in raw
                or raw in obj.raw
                or ' ' not in obj.raw
            ):
                obj.raw = raw
                obj.label = raw
            else:
                obj.label = '%s %s' % (obj.raw, raw)
                obj.raw = '%s %s' % (obj.raw, raw)

            m = re.search(r'\d+,\d+COM (\d+,\d+)', raw)
            if m:
                obj.commission = -CleanDecimal(replace_dots=True).filter(m.group(1))

        elif not obj.raw:
            # Empty transaction label
            # TODO move this xpath to the relevant page class
            if tr_page:
                obj.raw = obj.label = CleanText(
                    """//td[contains(text(), "Nature de l'opération")]/following-sibling::*[1]"""
                )(tr_page.doc)

        if not obj.date:
            if tr_page:
                obj.date = Date(
                    CleanText(
                        """//td[contains(text(), "Date de l'opération")]/following-sibling::*[1]""",
                        default=''
                    ),
                    dayfirst=True,
                    default=NotAvailable
                )(tr_page.doc)

            obj.rdate = obj.date

            if tr_page:
                # TODO move this xpath to the relevant page class
                obj.vdate = Date(
                    CleanText(
                        '//td[contains(text(), "Date de valeur")]/following-sibling::*[1]',
                        default=''
                    ),
                    dayfirst=True,
                    default=NotAvailable
                )(tr_page.doc)

                # TODO move this xpath to the relevant page class
                obj.amount = CleanDecimal(
                    '//td[contains(text(), "Montant")]/following-sibling::*[1]',
                    replace_dots=True,
                    default=NotAvailable
                )(tr_page.doc)

        # ugly hack to fix broken html
        # sometimes transactions have really an amount of 0...
        if not obj.amount and CleanDecimal(TableCell('credit'), default=None)(self) is None:
            if tr_page:
                # TODO move this xpath to the relevant page class
                obj.amount = CleanDecimal(
                    u'//td[contains(text(), "Montant")]/following-sibling::*[1]',
                    replace_dots=True,
                    default=NotAvailable
                )(tr_page.doc)

        obj.type = Transaction.TYPE_UNKNOWN
        if tr_page:
            typestring = CleanText(
                """//td[contains(text(), "Nature de l'opération")]/following-sibling::*[1]"""
            )(tr_page.doc)
            if typestring:
                for pattern, trtype in Transaction.PATTERNS:
                    match = pattern.match(typestring)
                    if match:
                        obj.type = trtype
                        break

        # Some transactions have no details, but we can find the type of the transaction,
        # the label and the category from the raw label.
        if obj.type == Transaction.TYPE_UNKNOWN:
            parse_with_patterns(obj.raw, obj, Transaction.PATTERNS)

    @pagination
    def get_operations(self, date_guesser):
        return self._get_operations(self)(date_guesser=date_guesser)

    def get_balance_without_comings_main(self):
        return CleanDecimal.French(
            '//span[@class="mtSolde"]',
            default=NotAvailable
        )(self.doc)

    def get_balance_without_comings(self):
        return CleanDecimal.French(
            '//span[contains(text(), "Opérations effectuées")]//ancestor::div[1]/following-sibling::div',
            default=NotAvailable
        )(self.doc)


class CardsPage(LoggedPage, HTMLPage):
    def deferred_date(self):
        deferred_date = Regexp(
            CleanText('//div[@class="date"][contains(text(), "Carte")]'),
            r'le ([^:]+)',
            default=None
        )(self.doc)

        assert deferred_date, 'Cannot find deferred_date'
        return parse_french_date(deferred_date).date()

    def get_card_summary(self):
        amount = CleanDecimal.French('//div[@class="montantEncours"]')(self.doc)

        if amount:
            t = Transaction()
            t.date = t.rdate = self.deferred_date()
            t.type = Transaction.TYPE_CARD_SUMMARY
            t.label = t.raw = CleanText('//div[@class="date"][contains(text(), "Carte")]')(self.doc)
            t.amount = abs(amount)
            return t

    def format_url(self, url):
        cb_type = re.match(r'.*(UWCBEncours.*)/.*', url).group(1)
        return '/outil/UWCB/%s/listeOperations' % cb_type

    @method
    class iter_multi_cards(TableElement):
        head_xpath = '//table[@class="tagTab"]/tr/th'
        item_xpath = '//table[@class="tagTab"]//tr[position()>1]'

        col_label = re.compile('Type')
        col_number = re.compile('Numéro')
        col_owner = re.compile('Titulaire')
        col_coming = re.compile('Montant')

        class Item(ItemElement):
            klass = Account

            obj_type = Account.TYPE_CARD
            obj_balance = Decimal(0)
            obj_parent = Env('parent_account')
            obj_coming = CleanDecimal.French(TableCell('coming'))
            obj_currency = Currency(TableCell('coming'))
            obj__transfer_id = None

            obj__cards_list = CleanText(Env('cards_list'))

            def obj__transactions_link(self):
                link = Attr('.', 'onclick')(self)
                url = re.match('.*\'(.*)\'\\.*', link).group(1)
                return self.page.format_url(url)

            def obj_number(self):
                card_number = re.match('((XXXX ){3}X ([0-9]{3}))', CleanText(TableCell('number'))(self))
                return card_number.group(1)[0:16] + card_number.group(1)[-3:]

            def obj_label(self):
                return '%s %s %s' % (
                    CleanText(TableCell('label'))(self),
                    CleanText(TableCell('owner'))(self),
                    Field('number')(self),
                )

            def obj_id(self):
                card_number = re.match('((XXXX ){3}X([0-9]{3}))', CleanText(Field('number'))(self))
                return '%s-%s' % (Env('parent_account')(self).id, card_number.group(3))

    def get_single_card(self, parent_account):
        account = Account()

        card_info = CleanText('//select[@id="selectCard"]/option/text()')(self.doc)
        # ex: VISA INFINITE DD M FIRSTNAME LASTNAME N°XXXX XXXX XXXX X103
        regex = '(.*)N°((XXXX ){3}X([0-9]{3})).*'
        card_infos = re.match(regex, card_info)

        coming = CleanDecimal.French('//div[@class="montantEncours"]/text()')(self.doc)

        account.id = '%s-%s' % (parent_account.id, card_infos.group(4))
        account.type = Account.TYPE_CARD
        account.parent = parent_account
        account.balance = Decimal('0')
        account.coming = coming
        account.number = card_infos.group(2)
        account.label = card_info
        account.currency = parent_account.currency
        account._transactions_link = self.format_url(self.url)
        account._transfer_id = None
        # We need to store this url. It will be useful later to get the transactions.
        account._cards_list = self.url
        return account

    def get_child_cards(self, parent_account):
        # There is a selector with only one entry when there is only one card
        # But not when there are multiple card.
        if self.doc.xpath('//select[@id="selectCard"]'):
            return [self.get_single_card(parent_account)]
        return list(self.iter_multi_cards(parent_account=parent_account, cards_list=self.url))

    @method
    class iter_transactions(TableElement):

        item_xpath = '//tr[contains(@class, "ligne")]'
        head_xpath = '//th'

        col_date = re.compile('Date')
        col_label = re.compile('Libellé')
        col_amount = re.compile('Montant')

        class item(ItemElement):

            klass = Transaction

            obj_rdate = obj_bdate = Date(CleanText(TableCell('date')), dayfirst=True)
            obj_type = Transaction.TYPE_DEFERRED_CARD
            obj_raw = obj_label = CleanText(TableCell('label'))
            obj_amount = CleanDecimal.French(TableCell('amount'))

            def obj_date(self):
                return self.page.deferred_date()

            def condition(self):
                if Field('date')(self) < Field('rdate')(self):
                    self.logger.error(
                        'skipping transaction with rdate(%s) > date(%s) for label(%s)',
                        Field('rdate')(self), Field('date')(self), Field('label')(self)
                    )
                    return False
                return True


class BoursePage(LoggedPage, HTMLPage):
    ENCODING = 'latin-1'
    REFRESH_MAX = 0

    TYPES = {
        'plan épargne en actions': Account.TYPE_PEA,
        "plan d'épargne en actions": Account.TYPE_PEA,
        'plan épargne en actions bourse': Account.TYPE_PEA,
        "plan d'épargne en actions bourse": Account.TYPE_PEA,
        'pea pme bourse': Account.TYPE_PEA,
        'pea pme': Account.TYPE_PEA,
    }

    def on_load(self):
        """
        Sometimes we are directed towards a prior html page before accessing Bourse Page.
        Submit the form to access the page that contains the Bourse Page's session cookie.
        """
        try:
            form = self.get_form(id='form')
        except FormNotFound:  # already on the targetted page
            pass
        else:
            form.submit()

        super(BoursePage, self).on_load()

    def open_iframe(self):
        # should be done always (in on_load)?
        for iframe in self.doc.xpath('//iframe[@id="mainIframe"]'):
            self.browser.location(iframe.attrib['src'])
            break

    def password_required(self):
        return CleanText(
            '//b[contains(text(), "Afin de sécuriser vos transactions, nous vous invitons à créer un mot de passe trading")]'
        )(self.doc)

    def get_next(self):
        if 'onload' in self.doc.xpath('.//body')[0].attrib:
            return re.search('"(.*?)"', self.doc.xpath('.//body')[0].attrib['onload']).group(1)

    def get_fullhistory(self):
        form = self.get_form(id="historyFilter")
        form['cashFilter'] = "ALL"
        # We can't go above 2 years
        form['beginDayfilter'] = (datetime.strptime(form['endDayfilter'], '%d/%m/%Y') - timedelta(days=730)).strftime('%d/%m/%Y')
        form.submit()

    @method
    class get_list(TableElement):
        item_xpath = '//table[has-class("tableau_comptes_details")]//tr[td and not(parent::tfoot)]'
        head_xpath = '//table[has-class("tableau_comptes_details")]/thead/tr/th'

        col_label = 'Comptes'
        col_owner = re.compile('Titulaire')
        col_titres = re.compile('Valorisation')
        col_especes = re.compile('Solde espèces')

        class item(OwnedItemElement):
            klass = Account

            load_details = Field('_market_link') & AsyncLoad

            obj__especes = CleanDecimal(TableCell('especes'), replace_dots=True, default=0)
            obj__titres = CleanDecimal(TableCell('titres'), replace_dots=True, default=0)
            obj_valuation_diff = Async('details') & CleanDecimal(
                '//td[contains(text(), "value latente")]/following-sibling::td[1]',
                replace_dots=True,
            )
            obj__market_id = Regexp(Attr(TableCell('label'), 'onclick'), r'nump=(\d+:\d+)')
            obj__market_link = Regexp(Attr(TableCell('label'), 'onclick'), r"goTo\('(.*?)'")
            obj__link_id = Async('details') & Link(u'//a[text()="Historique"]')
            obj__transfer_id = None
            obj_balance = Field('_titres')
            obj_currency = Currency(CleanText(TableCell('titres')))

            def obj_number(self):
                number = CleanText((TableCell('label')(self)[0]).xpath('./div[not(b)]'))(self).replace(' - ', '')
                m = re.search(r'(\d{11,})[A-Z]', number)
                if m:
                    number = m.group(0)
                return number

            def obj_id(self):
                return "%sbourse" % Field('number')(self)

            def obj_label(self):
                return "%s Bourse" % CleanText((TableCell('label')(self)[0]).xpath('./div[b]'))(self)

            def obj_type(self):
                _label = ' '.join(Field('label')(self).split()[:-1]).lower()
                for key in self.page.TYPES:
                    if key in _label:
                        return self.page.TYPES.get(key)
                return Account.TYPE_MARKET

            def obj_ownership(self):
                owner = CleanText(TableCell('owner'))(self)
                return self.get_ownership(owner)

    def get_logout_link(self):
        return Link('//a[contains(text(), "Retour aux comptes")]')(self.doc)

    @method
    class iter_investment(ListElement):
        item_xpath = '//table[@id="tableValeurs"]/tbody/tr[@id and count(descendant::td) > 1]'

        class item(ItemElement):
            klass = Investment

            obj_label = CleanText('.//td[2]/div/a')
            obj_code = (
                CleanText('.//td[2]/div/br/following-sibling::text()')
                & Regexp(pattern='^([^ ]+).*', default=NotAvailable)
            )
            obj_quantity = MyDecimal('.//td[3]/span')
            obj_diff = MyDecimal('.//td[7]/span')
            obj_valuation = MyDecimal('.//td[5]')

            def obj_code_type(self):
                code = Field('code')(self)
                if code and is_isin_valid(code):
                    return Investment.CODE_TYPE_ISIN
                return NotAvailable

            def obj_unitvalue(self):
                if "%" in CleanText('.//td[4]')(self) and "%" in CleanText('.//td[6]')(self):
                    return NotAvailable
                return MyDecimal('.//td[4]/text()')(self)

            def obj_unitprice(self):
                if "%" in CleanText('.//td[4]')(self) and "%" in CleanText('.//td[6]')(self):
                    return NotAvailable
                return MyDecimal('.//td[6]')(self)

    @pagination
    @method
    class iter_history(TableElement):
        item_xpath = '//table[@id="historyTable" and thead]/tbody/tr'
        head_xpath = '//table[@id="historyTable" and thead]/thead/tr/th'

        col_date = 'Date'
        col_label = u'Opération'
        col_quantity = u'Qté'
        col_code = u'Libellé'
        col_amount = 'Montant'

        def next_page(self):
            form = self.page.get_form(id="historyFilter")
            form['PAGE'] = int(form['PAGE']) + 1
            if self.page.doc.xpath('//*[@data-page = $page]', page=form['PAGE']):
                return requests.Request("POST", form.url, data=dict(form))

        class item(ItemElement):
            klass = Transaction

            obj_date = Date(CleanText(TableCell('date')), dayfirst=True)
            obj_type = Transaction.TYPE_BANK
            obj_amount = CleanDecimal(TableCell('amount'), replace_dots=True)
            obj_investments = Env('investments')

            def obj_label(self):
                return TableCell('label')(self)[0].xpath('./text()')[0].strip()

            def parse(self, el):
                i = None
                self.env['investments'] = []

                if CleanText(TableCell('code'))(self):
                    i = Investment()
                    i.label = Field('label')(self)
                    i.code = unicode(TableCell('code')(self)[0].xpath('./text()[last()]')[0]).strip()
                    i.quantity = MyDecimal(TableCell('quantity'))(self)
                    i.valuation = Field('amount')(self)
                    i.vdate = Field('date')(self)

                    self.env['investments'] = [i]