Gentle heads up to @GriceTurrble, who reviewed the mentioned PRs.

@adamantike do you have an example of a call that failed before this fix - so I can do a live test to check ?

@Bobspadger, sure! A simple example is trying to get the second page of a ListOrders call (using ListOrdersByNextToken). The NextToken tends to have characters that are escaped by quote(), so you can test it like this:

import datetime
import mws

next_token = None
last_updated_after = datetime.datetime(2020, 1, 1)

orders_api = mws.Orders(
    access_key="...",
    secret_key="...",
    account_id="...",
    auth_token="...",
    region="US",
)

for _ in range(2):
    response = orders_api.list_orders(
        marketplaceids=("ATVPDKIKX0DER",),
        lastupdatedafter=last_updated_after,
        max_results=1,
        next_token=next_token,
    )
    next_token = response.parsed["NextToken"]
    assert next_token

For the second call, with version 0.8.13, I'm getting the following error response:

<ErrorResponse xmlns="https://mws.amazonservices.com/Orders/2013-09-01">
  <Error>
    <Type>Sender</Type>
    <Code>NextTokenCorrupted</Code>
    <Message>We could not decode your NextToken. Possible reasons include: a transmission error, improper quoting or a truncation problem.</Message>
  </Error>
  <RequestId>...</RequestId>
</ErrorResponse>

My apologies for the delay on my end. Home life and such. Not ignoring this, I promise!

@GriceTurrble, no worries! Just to confirm, we have forked 0.8.13 and applied this patch without any issues in our systems (besides changing date strings to actual datetime objects to avoid #259).

@Bobspadger @GriceTurrble, gentle ping on this one!

I also just hit this corner of the internet while working for my employer. @adamantike is the mentioned fork actually publicly available?

@ihucos, I haven't made it publicly available, as I would prefer for this fix to quickly be merged, and avoid more people from start depending on a fork. It's literally just this PR applied on top of 0.8.13, in case you want to temporarily fork it.

@adamantike thanks for the message. We decided to wait for a fix upstream. I could to something to work towards it if it makes sense.

Hi Sorry, I've not forgotten about this but work has got quite hectic so not had the time to really look into this.

I am hoping to in the next week or so and try to look at what is truly happening, as this is only a new issue since there were some items back-ported from the develop branch.

I'd like to not be in a situation where we keep releasing point updates for essentially the same bug,

@GriceTurrble, if there are no planned changes in the short term, do you mind tagging and releasing a new patch version, to have this fix available for current 0.8.x users? :)

Thanks for the new release, @GriceTurrble!

cc @ihucos, this change is now available in version 0.8.14.

0.8.14 is still not published to PyPI, though.

First time for me to actually perform the PyPI release, so that was a learning process.

However, 0.8.14 is now available on PyPI: https://pypi.org/project/mws/0.8.14/