Add ignore_fields option

[jerr0328]

We have a use-case where we need to avoid camelizing/underscorizing certain field data. For example:

{"extraData": {"FOO1": "sample", "FOO2": "example"}}

is underscorized as:

{"extra_data": {"f_o_o_1": "sample", "f_o_o_2": "example"}}

Instead, by adding extra_data or extraData to ignore_fields, we can now get the expected behavior of underscorizing/camelizing:

{"extra_data": {"FOO1": "sample", "FOO2": "example"}}

This relates to another PR that did this by passing the option to the constructor, but here we use the settings file to configure the option.

[JonasKs]

@vbabiy , any status on this one? I'm upvoting it atleast, would be a nice feature to have. 👍

[vbabiy]

@JonasKs sorry, I am going to test and merge today.

[JonasKs]

@vbabiy , no rush, just asking :)

[lucas-bremond]

That'd be a highly desirable feature indeed, any update on this?

[lucas-bremond]

Shouldn't this be:

new_dict[key] = value

instead?

Otherwise, the new key (post transformation) is applied, which is not what is expected.

For example, using __typename in ignore_fields won't work here.

[lucas-bremond]

What about something like this instead?

def camelize (data, **options):

    ignore_fields = options.get('ignore_fields') or ()

    if isinstance(data, Promise):
        data = force_text(data)

    if isinstance(data, dict):

        if isinstance(data, ReturnDict):
            new_dict = ReturnDict(serializer = data.serializer)
        else:
            new_dict = OrderedDict()

        for (key, value) in data.items():

            if isinstance(key, Promise):
                key = force_text(key)

            if key in ignore_fields:
                new_dict[key] = camelize(value, **options)

            else:

                if isinstance(key, str) and ('_' in key):
                    new_key = re.sub(camelize_re, underscore_to_camel, key)
                else:
                    new_key = key

                new_dict[new_key] = camelize(value, **options)

        return new_dict

    if is_iterable(data) and not isinstance(data, str):
        return [camelize(item, **options) for item in data]

    return data

[jerr0328]

Shouldn't this be:

new_dict[key] = value

instead?

Otherwise, the new key (post transformation) is applied, which is not what is expected.

For example, using __typename in ignore_fields won't work here.

Thanks for your comment! It could be that the naming is a bit confusing. What we really are doing is ignoring fields under the given key. In other words, we don't want to recursively modify any of the data under the specified key. We might need to split the option of not changing specific key names (like __typename in your example) with not modifying any data under certain key names (like extraData or extra_data in my example, where the field name should be changed but the others shouldn't).

[vbabiy]

I am okay with this being called ignore_fields if we want to support the other model maybe call it ignore_keys