Contributors for v2

[saleweaver]

Please comment here to coordinate contributing towards v2

[adeelsohailahmed]

I don't have a lot of time to spare, but I'd love to contribute towards v2 wherever and however I can!

[darshandevelopes]

Hello, I would like to contribute.

[saleweaver]

Hi @adeelsohailahmed , @darshandevelopes

thanks for getting back! I have pushed a first version of v2 to
https://github.com/saleweaver/python-amazon-sp-api/tree/v2-01

I'll publish a contribution guideline shortly. For now, if you have any questions, feel free to ask!

Note that the entire library is autogenerated. Never change anything in sp_api.api - everything gets overriden daily. The main work is in sdk_generator at the moment. We parse all models and generate the sdk from there.

Thanks for contributing!

[estebanthi]

Hi, I would like to contribute! I'm using the v2 in a new project and I'm already seeing a few issues I would be happy to fix.

[estebanthi]

ItemSummaryByMarketplace and ListingsItemSubmissionResponse in listings_items_2021_08_01/common.py share the same StatusEnum when they should not. They both have a status field, so I would say duplicate model names within a client are probably not handled.

I fixed this for enums by providing model context when parsing properties to generate unique enum names such as ItemSummaryByMarketplaceStatusEnum. I will wait for your contribution guidelines to open the PR.

Maybe we should apply the same logic when generating reference names if they are not unique among a client but we will end up with long and redundant names (ItemOfferByMarketplace will become ItemOffersItemOfferByMarketplace, ItemSummaryByMarketplace will become ItemSummariesItemSummaryByMarketplace).

[saleweaver]

Thanks for the feedback!

You can open the PR if you want, we definitely need this, and the way you've named the enums is exactly how I would have. I think I saw this occurring with a different endpoint as well.

[saleweaver]

By the way - the pydantic model generation used to generate the endpoint models should work for product type definition schemas used by listing items as well.

My plan is to create a CLI tool to autogenerate models for creating and updating listing items. Since you seem to be working on listing items, you might want to give it a shot.

[estebanthi]

My plan is to create a CLI tool to autogenerate models for creating and updating listing items. Since you seem to be working on listing items, you might want to give it a shot.

I work a lot on listings items, I would be happy to give it a shot.

I noticed a few more issues. I will list them here but maybe we should open issues for v2 directly in the repo for better tracking?

• Payload parsing doesn't handle the cases where the Pydantic model contains a payload, that is a list of Pydantic models (e.g., in GetMarketplaceParticipationsResponse). There are some other issues with payload parsing, sometimes nested Pydantic models are parsed as dict, maybe related to lists parsing too (e.g., ItemSearchResults items).
• Some endpoint requests are malformed — the actual request data is wrapped in a field when it should not be (e.g., for ListingOffersBatchRequest, data is in a field getListingOffersBatchRequestBody so when the body is parsed it contains this).
• Some Pydantic models are not generated correctly (e.g. ListingOffersRequestBody has no fields).
• Not sure about this one — rate-limit errors don't seem to be handled correctly, the returned ApiResponse is empty instead of containing error details.

[saleweaver]

Thanks for the PR!

I've already handled the payload parsing, the models amazon provides have a lot of errors and indicate different responses from what actually is returned. I used model_compile instead of ...validate because of this, but this needs to change to handle the nested parsing. (Will push this the next couple of days)

I was thinking of forking the amazon repository and fixing the models there, that's faster than waiting for amazon to do something.

If you can create issues and label them with v2 would be perfect, then they can be worked on and fixed.