class item(Transaction.TransactionElement):
condition = lambda self: len(self.el.xpath('./td')) >= 4
obj_date = DateGuesser(CleanText(TableCell("date")),
Env("date_guesser"))
obj_vdate = DateGuesser(CleanText(TableCell("date")),
Env("date_guesser"))
def parse(self, obj):
self.env['date'] = DateGuesser(CleanText('./td[1]'), Env('date_guesser'))(self)
self.env['vdate'] = NotAvailable
if CleanText('//table[@class="ca-table"][caption[span[b[text()="Historique des opérations"]]]]//tr[count(td) = 4]')(self):
# History table with 4 columns
self.env['raw'] = CleanText('./td[2]', children=False)(self)
self.env['amount'] = CleanDecimal.French('./td[last()]')(self)
elif CleanText('//table[@class="ca-table"][caption[span[b[text()="Historique des opérations"]]]]//tr[count(td) = 5]')(self):
# History table with 5 columns
self.env['raw'] = CleanText('./td[3]', children=False)(self)
self.env['amount'] = CleanDecimal.French('./td[last()]')(self)
elif CleanText('//table[@class="ca-table"][caption[span[b[text()="Historique des opérations"]]]]//tr[count(td) = 6]')(self):
# History table with 6 columns (contains vdate)
self.env['raw'] = CleanText('./td[4]', children=False)(self)
self.env['vdate'] = DateGuesser(CleanText('./td[2]'), Env('date_guesser'))(self)
self.env['amount'] = CleanDecimal.French('./td[last()]')(self)
elif CleanText('//table[@class="ca-table"][caption[span[b[text()="Historique des opérations"]]]]//tr[count(td) = 7]')(self):
# History table with 7 columns
self.env['amount'] = Coalesce(
CleanDecimal.French('./td[6]', sign=lambda x: -1, default=None),
CleanDecimal.French('./td[7]', default=None)
)(self)
if CleanText('//table[@class="ca-table"][caption[span[b[text()="Historique des opérations"]]]]//th[a[contains(text(), "Valeur")]]')(self):
# With vdate column ('Valeur')
self.env['raw'] = CleanText('./td[4]', children=False)(self)
self.env['vdate'] = DateGuesser(CleanText('./td[2]'), Env('date_guesser'))(self)
else:
# Without any vdate column
self.env['raw'] = CleanText('./td[3]', children=False)(self)
else:
assert False, 'This type of history table is not handled yet!'
def obj_date(self):
# debit date is guessed in text such as 'Opérations débitées le 05/07'
guessed_date = DateGuesser(
Regexp(
CleanText(
self.xpath(
'./preceding-sibling::tr[.//a[contains(text(), "Opérations débitées le")]][1]'
)), r'(\d{2}/\d{2})'), Env("date_guesser"))(self)
# Handle the case where the guessed debit date would be before the rdate (happens when
# the debit date is in january whereas the rdate is in december).
if guessed_date < Field('rdate')(self):
return guessed_date.replace(year=guessed_date.year + 1)
return guessed_date
def get_card_transactions(self, latest_date, ongoing_coming):
for item in self.doc.xpath('//table[@class="ca-table"][2]//tr[td]'):
if CleanText('./td[2]/b')(item):
# This node is a summary containing the 'date' for all following transactions.
raw_date = Regexp(CleanText('./td[2]/b/text()'),
r'le (.*) :')(item)
if latest_date and parse_french_date(
raw_date).date() > latest_date:
# This summary has already been fetched
continue
latest_date = parse_french_date(raw_date).date()
if latest_date < ongoing_coming:
# This summary is anterior to the ongoing_coming so we create a transaction from it
tr = FrenchTransaction()
tr.date = tr.rdate = latest_date
tr.raw = tr.label = CleanText('./td[2]/b/text()')(item)
tr.amount = -CleanDecimal.French(
'./td[position()=last()]')(item)
tr.type = FrenchTransaction.TYPE_CARD_SUMMARY
yield tr
else:
# This node is a real transaction.
# Its 'date' is the date of the most recently encountered summary node.
tr = FrenchTransaction()
tr.date = latest_date
date_guesser = LinearDateGuesser(latest_date)
tr.rdate = tr.bdate = DateGuesser(
CleanText('./td[1]//text()'),
date_guesser=date_guesser)(item)
tr.label = tr.raw = CleanText('./td[2]')(item)
tr.amount = CleanDecimal.French('./td[last()]')(item)
tr.type = FrenchTransaction.TYPE_DEFERRED_CARD
yield tr
def obj_date(self):
date = CleanText('./td[1]/font//text()')(self)
if len(date) == 10:
return Date(CleanText('./td[1]/font//text()'), dayfirst=True)(self)
elif len(date) == 5:
# Date has no indicated year.
return DateGuesser(CleanText('./td[1]//text()'), Env('date_guesser'))(self)
def obj_date(self):
# Dates in the first column may appear as '12/01/2019' or '12/01'
date = CleanText('./td[1]/font//text()')(self)
if len(date) == 10:
return Date(CleanText('./td[1]/font//text()'), dayfirst=True)(self)
elif len(date) == 5:
# Date has no indicated year.
return DateGuesser(CleanText('./td[1]//text()'), Env('date_guesser'))(self)
class item(ItemElement):
klass = Transaction
obj_date = DateGuesser(CleanText('.//span[contains(., "/")]'), LinearDateGuesser(date_max_bump=timedelta(45)))
obj_label = CleanText('.//h3/strong')
obj_amount = MyDecimal('./td[@class="al-r"]/div/span[has-class("badge")]')
def obj_type(self):
amount = Field('amount')(self)
if amount < 0:
return Transaction.TYPE_CARD
else:
return Transaction.TYPE_TRANSFER
def obj_date(self):
return DateGuesser(
Regexp(
CleanText(self.page.doc.xpath('//table/tr[2]/td[1]')),
r'(\d{2}/\d{2})'), Env("date_guesser"))(self)
def date(selector):
return DateGuesser(CleanText(selector), Env('date_guesser')) | Transaction.Date(selector)