[Python] allow caller to determine "Accept" content-type

Description

The Python client requests the Accept header from a list, calling method select_header_accept. This does not work well when a single API call needs to be done with "application/xml" rather than "application/json". Work-arounds I have found do not work well with asynchronous models.

Swagger-codegen version

3.0.24 and 3.0.25

Swagger declaration file content or url

The operationId "get/almaws/v1/bibs/{mms_id}" will return different information if "application/json" is selected as opposed to "application/xml". The Machine Readable Catalog (MARC) specification supports an XML serialization, but there is no JSON serialization.

https://developers.exlibrisgroup.com/wp-content/uploads/alma/openapi/bibs.yaml

Command line used for generation

My process is pretty complicated - see https://github.com/NLM-OCCS/python-alma-sdk for a public version that does not disclose additional magic for handling API keys. As a summary, here is the command-line used:

java -jar lib/swagger-codegen-cli.jar generate -l python -c api-config.json -i https://developers.exlibrisgroup.com/wp-content/uploads/alma/openapi/bibs.yaml --resolve-fully -o target

Some notes on the arguments:

• api-config.json is below:

{
  "packageName": "alma_api",
  "modelPackage": "models",
  "apiPackage": "apis"
}

Steps to reproduce

1. Install Apache Ant and place in path
2. Install CPython 3.6+
3. Clone the repository: git clone git@github.com:NLM-OCCS/python-alma-sdk.git
4. Run ant bootstrap to install apache ivy: ant bootstrap
5. Run ant resolve to download swagger-codegen-cli: ant resolve
6. Generate the API as described above - java -jar lib/swagger-codegen-cli.jar generate -l python -c api-config -i https://developers.exlibrisgroup.com/wp-content/uploads/alma/openapi/bibs.yaml --resolve-fully -o target
7. Change into the target directory.
8. Create a virtualenv, activate it, and install the requirements and the ipython (interactive python) environment.
9. Run the following in an ipython shell

from alma_api import Configuration, ApiClient, CatalogApi

config = Configuration()
config.api_key['apikey'] = 'an appropriate key'
client = ApiClient(config)
catapi = CatalogApi(client)

r = catapi.getalmawsv1bibsmms_id('991572500000541', _preload_content=False)

Suggest a fix/enhancement

Change modules/swagger-codegen/src/main/resources/python/api.mustache to add an additional _accept_header s argument to each generated {operation_id}_with_http_info(), such as the generated code:

        all_params.append('_accept_header')

Change the code to determine the accept header as follows:

        # HTTP header `Accept`
        accept_header = all_params.pop('_accept_header', None)
        header_params['Accept'] = accept_header if accept_header else self.api_client.select_header_accept(
            ['application/json', 'application/xml'])  # noqa: E501

Start with modules/swagger-codegen/src/main/resources/python/api.mustache and reproduce generation using the linked Alma OpenAPI document and command. Check the generated getalmawsv1bibsmms_id_with_http_info() path and verify that callers can select application/xml for one operation while the existing default Accept selection still works.