- Feb 12, 2020
-
-
There are loans which we did not handle.
-
The xpaths changed on the website.
-
-
The website is sometimes really slow, this timeout allows to wait for a slow response from the website.
-
Also try to get rid of some "JSON()" calls.
-
We only fetched 10 orders a year, there is a pagination for each year. Some orders have no price because they are purchased using "audio credits". In this case the xpath for the date is also different.
-
We slide 3 months window to paginate documents but if there are no documents on the first page, we couldn't take 3 months in the past as the date wasn't initialized. Now, take current month instead.
-
The Attr()() returns a string, coerce to int.
-
While navigating, we fetch a token on differents places to access the website, but in some case we are unable to find it at given location, so I add one more url that works for almost all the cases.
-
-
On the bnppere module, some customers put letters + digits in their passwords, making the VK crash. However, if we remove these letters and there are exactly 6 digits remaining, we sometimes manage to log in. We cannot use a regex in module.py because bnppere has two browsers (Personeo & Visiogo) and the Visiogo website does not use a VK.
-
-
Summary transactions come from the parent checking account and their dates have no time, whereas card transactions had a time, making the comparison with sorted_transactions() crash. Since the times are just '12:00:00 AM', we exclude them in order to fix iter_card_history.
-
raise AuthMethodNotImplemented for the moment
-
Some transactions have no details but have a raw label which we can parse the type, category and label from.
-
It's not a mistake, there are really `AUTHENTICATION_FAILED` and `FAILED_AUTHENTICATION` ...
-
Even if the 30 is on the website, return NotAvailable ...
-
DateTime for date is removed because the time is always 12:00:00 AM. If the rdate is not found with the patterns, we try to set it to dateOperation if dateOperation is before dateValeur.
-
-
-
The string that we were fetching was not the right one.
-
SMS are sent with background worker. We avoid it.
-
Previous xpath didn't get every case, sometimes the tag was h3 instead of h2.
-
A logger named only "iter_accounts" was not explicit enough anyway.
-
-
All the performances (1 3 5 years) are available on this CPR details page.
-
-
The new login mechanism is like transfer authentication and use same URLs and pages. There is 'context' in json response to know if it's a login or transfer, but this key is not always here when authentication fail and we don't know if it's for login or transfer. Refactorize all authentication mechanism to have the context through the code.
-
Some invests have no diff value, but the website fills the associated json field with the investment valuation.
-
Error parsing was not done properly on LoginPage on_load() method. Different messages are used to inform user of AppValidation. Certain connections don't have 2FA enabmed yet, they might bypass it, or not.
-
Some transactions have a rdate in the future in their label, which cause the transaction rdate to be after the date.
-
-
Sometimes the website displays us a page asking us to validate something before being able to go to the market page.
-
-
The multi-accounts was not handled yet so I adapted the browsing and parsing to retrieve transactions & investments.
-
-
-
The previous xpath returned the asset category twice.
-
-
hydrargyrum authored
-